End of training
Browse files- all_results.json +11 -11
- eval_results.json +6 -6
- train_results.json +6 -6
- trainer_state.json +1493 -494
all_results.json
CHANGED
|
@@ -1,13 +1,13 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 32.
|
| 6 |
-
"eval_samples_per_second": 138.
|
| 7 |
-
"eval_steps_per_second": 4.
|
| 8 |
-
"total_flos":
|
| 9 |
-
"train_loss": 0.
|
| 10 |
-
"train_runtime":
|
| 11 |
-
"train_samples_per_second":
|
| 12 |
-
"train_steps_per_second": 0.
|
| 13 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.7280604310153299,
|
| 4 |
+
"eval_loss": 0.4742942750453949,
|
| 5 |
+
"eval_runtime": 32.4474,
|
| 6 |
+
"eval_samples_per_second": 138.717,
|
| 7 |
+
"eval_steps_per_second": 4.345,
|
| 8 |
+
"total_flos": 1.0051627680166625e+19,
|
| 9 |
+
"train_loss": 0.5403492726857149,
|
| 10 |
+
"train_runtime": 6645.2491,
|
| 11 |
+
"train_samples_per_second": 60.946,
|
| 12 |
+
"train_steps_per_second": 0.476
|
| 13 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_accuracy": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 32.
|
| 6 |
-
"eval_samples_per_second": 138.
|
| 7 |
-
"eval_steps_per_second": 4.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"eval_accuracy": 0.7280604310153299,
|
| 4 |
+
"eval_loss": 0.4742942750453949,
|
| 5 |
+
"eval_runtime": 32.4474,
|
| 6 |
+
"eval_samples_per_second": 138.717,
|
| 7 |
+
"eval_steps_per_second": 4.345
|
| 8 |
}
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"total_flos":
|
| 4 |
-
"train_loss": 0.
|
| 5 |
-
"train_runtime":
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 0.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 9.98,
|
| 3 |
+
"total_flos": 1.0051627680166625e+19,
|
| 4 |
+
"train_loss": 0.5403492726857149,
|
| 5 |
+
"train_runtime": 6645.2491,
|
| 6 |
+
"train_samples_per_second": 60.946,
|
| 7 |
+
"train_steps_per_second": 0.476
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,1015 +1,2014 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "swin-tiny-patch4-window7-224-img_orientation/checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.03,
|
| 13 |
-
"learning_rate":
|
| 14 |
-
"loss": 1.
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.06,
|
| 19 |
-
"learning_rate":
|
| 20 |
-
"loss": 1.
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
-
"epoch": 0.
|
| 25 |
-
"learning_rate":
|
| 26 |
-
"loss": 1.
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.13,
|
| 31 |
-
"learning_rate":
|
| 32 |
-
"loss": 1.
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.16,
|
| 37 |
-
"learning_rate":
|
| 38 |
-
"loss": 1.
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.19,
|
| 43 |
-
"learning_rate":
|
| 44 |
-
"loss": 1.
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.22,
|
| 49 |
-
"learning_rate":
|
| 50 |
-
"loss": 1.
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.25,
|
| 55 |
-
"learning_rate":
|
| 56 |
-
"loss": 1.
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
-
"epoch": 0.
|
| 61 |
-
"learning_rate":
|
| 62 |
-
"loss":
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.32,
|
| 67 |
-
"learning_rate":
|
| 68 |
-
"loss":
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.35,
|
| 73 |
-
"learning_rate":
|
| 74 |
-
"loss":
|
| 75 |
"step": 110
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.38,
|
| 79 |
-
"learning_rate":
|
| 80 |
-
"loss": 0.
|
| 81 |
"step": 120
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.41,
|
| 85 |
-
"learning_rate":
|
| 86 |
-
"loss": 0.
|
| 87 |
"step": 130
|
| 88 |
},
|
| 89 |
{
|
| 90 |
-
"epoch": 0.
|
| 91 |
-
"learning_rate":
|
| 92 |
-
"loss": 0.
|
| 93 |
"step": 140
|
| 94 |
},
|
| 95 |
{
|
| 96 |
-
"epoch": 0.
|
| 97 |
-
"learning_rate":
|
| 98 |
-
"loss": 0.
|
| 99 |
"step": 150
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.51,
|
| 103 |
-
"learning_rate":
|
| 104 |
-
"loss": 0.
|
| 105 |
"step": 160
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.54,
|
| 109 |
-
"learning_rate":
|
| 110 |
-
"loss": 0.
|
| 111 |
"step": 170
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.57,
|
| 115 |
-
"learning_rate":
|
| 116 |
-
"loss": 0.
|
| 117 |
"step": 180
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.6,
|
| 121 |
-
"learning_rate":
|
| 122 |
-
"loss": 0.
|
| 123 |
"step": 190
|
| 124 |
},
|
| 125 |
{
|
| 126 |
-
"epoch": 0.
|
| 127 |
-
"learning_rate":
|
| 128 |
-
"loss": 0.
|
| 129 |
"step": 200
|
| 130 |
},
|
| 131 |
{
|
| 132 |
-
"epoch": 0.
|
| 133 |
-
"learning_rate":
|
| 134 |
-
"loss": 0.
|
| 135 |
"step": 210
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.7,
|
| 139 |
-
"learning_rate":
|
| 140 |
-
"loss": 0.
|
| 141 |
"step": 220
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.73,
|
| 145 |
-
"learning_rate":
|
| 146 |
-
"loss": 0.
|
| 147 |
"step": 230
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.76,
|
| 151 |
-
"learning_rate":
|
| 152 |
-
"loss": 0.
|
| 153 |
"step": 240
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.79,
|
| 157 |
-
"learning_rate":
|
| 158 |
-
"loss": 0.
|
| 159 |
"step": 250
|
| 160 |
},
|
| 161 |
{
|
| 162 |
-
"epoch": 0.
|
| 163 |
-
"learning_rate": 4.
|
| 164 |
-
"loss": 0.
|
| 165 |
"step": 260
|
| 166 |
},
|
| 167 |
{
|
| 168 |
-
"epoch": 0.
|
| 169 |
-
"learning_rate": 4.
|
| 170 |
-
"loss": 0.
|
| 171 |
"step": 270
|
| 172 |
},
|
| 173 |
{
|
| 174 |
-
"epoch": 0.
|
| 175 |
-
"learning_rate": 4.
|
| 176 |
-
"loss": 0.
|
| 177 |
"step": 280
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.92,
|
| 181 |
-
"learning_rate": 4.
|
| 182 |
-
"loss": 0.
|
| 183 |
"step": 290
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.95,
|
| 187 |
-
"learning_rate": 4.
|
| 188 |
-
"loss": 0.
|
| 189 |
"step": 300
|
| 190 |
},
|
| 191 |
{
|
| 192 |
-
"epoch": 0.
|
| 193 |
-
"learning_rate": 4.
|
| 194 |
-
"loss": 0.
|
| 195 |
"step": 310
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 1.0,
|
| 199 |
-
"eval_accuracy": 0.
|
| 200 |
-
"eval_loss": 0.
|
| 201 |
-
"eval_runtime": 33.
|
| 202 |
-
"eval_samples_per_second":
|
| 203 |
-
"eval_steps_per_second": 4.
|
| 204 |
-
"step":
|
| 205 |
},
|
| 206 |
{
|
| 207 |
-
"epoch": 1.
|
| 208 |
-
"learning_rate": 4.
|
| 209 |
-
"loss": 0.
|
| 210 |
"step": 320
|
| 211 |
},
|
| 212 |
{
|
| 213 |
-
"epoch": 1.
|
| 214 |
-
"learning_rate": 4.
|
| 215 |
-
"loss": 0.
|
| 216 |
"step": 330
|
| 217 |
},
|
| 218 |
{
|
| 219 |
-
"epoch": 1.
|
| 220 |
-
"learning_rate": 4.
|
| 221 |
-
"loss": 0.
|
| 222 |
"step": 340
|
| 223 |
},
|
| 224 |
{
|
| 225 |
"epoch": 1.11,
|
| 226 |
-
"learning_rate": 4.
|
| 227 |
-
"loss": 0.
|
| 228 |
"step": 350
|
| 229 |
},
|
| 230 |
{
|
| 231 |
"epoch": 1.14,
|
| 232 |
-
"learning_rate": 4.
|
| 233 |
-
"loss": 0.
|
| 234 |
"step": 360
|
| 235 |
},
|
| 236 |
{
|
| 237 |
-
"epoch": 1.
|
| 238 |
-
"learning_rate": 4.
|
| 239 |
-
"loss": 0.
|
| 240 |
"step": 370
|
| 241 |
},
|
| 242 |
{
|
| 243 |
-
"epoch": 1.
|
| 244 |
-
"learning_rate": 4.
|
| 245 |
-
"loss": 0.
|
| 246 |
"step": 380
|
| 247 |
},
|
| 248 |
{
|
| 249 |
-
"epoch": 1.
|
| 250 |
-
"learning_rate": 4.
|
| 251 |
-
"loss": 0.
|
| 252 |
"step": 390
|
| 253 |
},
|
| 254 |
{
|
| 255 |
-
"epoch": 1.
|
| 256 |
-
"learning_rate": 4.
|
| 257 |
-
"loss": 0.
|
| 258 |
"step": 400
|
| 259 |
},
|
| 260 |
{
|
| 261 |
"epoch": 1.3,
|
| 262 |
-
"learning_rate": 4.
|
| 263 |
-
"loss": 0.
|
| 264 |
"step": 410
|
| 265 |
},
|
| 266 |
{
|
| 267 |
-
"epoch": 1.
|
| 268 |
-
"learning_rate": 4.
|
| 269 |
-
"loss": 0.
|
| 270 |
"step": 420
|
| 271 |
},
|
| 272 |
{
|
| 273 |
-
"epoch": 1.
|
| 274 |
-
"learning_rate": 4.
|
| 275 |
-
"loss": 0.
|
| 276 |
"step": 430
|
| 277 |
},
|
| 278 |
{
|
| 279 |
-
"epoch": 1.
|
| 280 |
-
"learning_rate":
|
| 281 |
-
"loss": 0.
|
| 282 |
"step": 440
|
| 283 |
},
|
| 284 |
{
|
| 285 |
-
"epoch": 1.
|
| 286 |
-
"learning_rate":
|
| 287 |
-
"loss": 0.
|
| 288 |
"step": 450
|
| 289 |
},
|
| 290 |
{
|
| 291 |
-
"epoch": 1.
|
| 292 |
-
"learning_rate":
|
| 293 |
-
"loss": 0.
|
| 294 |
"step": 460
|
| 295 |
},
|
| 296 |
{
|
| 297 |
-
"epoch": 1.
|
| 298 |
-
"learning_rate":
|
| 299 |
-
"loss": 0.
|
| 300 |
"step": 470
|
| 301 |
},
|
| 302 |
{
|
| 303 |
-
"epoch": 1.
|
| 304 |
-
"learning_rate":
|
| 305 |
-
"loss": 0.
|
| 306 |
"step": 480
|
| 307 |
},
|
| 308 |
{
|
| 309 |
-
"epoch": 1.
|
| 310 |
-
"learning_rate":
|
| 311 |
-
"loss": 0.
|
| 312 |
"step": 490
|
| 313 |
},
|
| 314 |
{
|
| 315 |
-
"epoch": 1.
|
| 316 |
-
"learning_rate":
|
| 317 |
-
"loss": 0.
|
| 318 |
"step": 500
|
| 319 |
},
|
| 320 |
{
|
| 321 |
-
"epoch": 1.
|
| 322 |
-
"learning_rate":
|
| 323 |
-
"loss": 0.
|
| 324 |
"step": 510
|
| 325 |
},
|
| 326 |
{
|
| 327 |
-
"epoch": 1.
|
| 328 |
-
"learning_rate":
|
| 329 |
-
"loss": 0.
|
| 330 |
"step": 520
|
| 331 |
},
|
| 332 |
{
|
| 333 |
-
"epoch": 1.
|
| 334 |
-
"learning_rate":
|
| 335 |
-
"loss": 0.
|
| 336 |
"step": 530
|
| 337 |
},
|
| 338 |
{
|
| 339 |
-
"epoch": 1.
|
| 340 |
-
"learning_rate":
|
| 341 |
-
"loss": 0.
|
| 342 |
"step": 540
|
| 343 |
},
|
| 344 |
{
|
| 345 |
-
"epoch": 1.
|
| 346 |
-
"learning_rate":
|
| 347 |
-
"loss": 0.
|
| 348 |
"step": 550
|
| 349 |
},
|
| 350 |
{
|
| 351 |
-
"epoch": 1.
|
| 352 |
-
"learning_rate":
|
| 353 |
-
"loss": 0.
|
| 354 |
"step": 560
|
| 355 |
},
|
| 356 |
{
|
| 357 |
-
"epoch": 1.
|
| 358 |
-
"learning_rate":
|
| 359 |
-
"loss": 0.
|
| 360 |
"step": 570
|
| 361 |
},
|
| 362 |
{
|
| 363 |
-
"epoch": 1.
|
| 364 |
-
"learning_rate":
|
| 365 |
-
"loss": 0.
|
| 366 |
"step": 580
|
| 367 |
},
|
| 368 |
{
|
| 369 |
-
"epoch": 1.
|
| 370 |
-
"learning_rate":
|
| 371 |
-
"loss": 0.
|
| 372 |
"step": 590
|
| 373 |
},
|
| 374 |
{
|
| 375 |
-
"epoch": 1.
|
| 376 |
-
"learning_rate":
|
| 377 |
-
"loss": 0.
|
| 378 |
"step": 600
|
| 379 |
},
|
| 380 |
{
|
| 381 |
-
"epoch": 1.
|
| 382 |
-
"learning_rate":
|
| 383 |
-
"loss": 0.
|
| 384 |
"step": 610
|
| 385 |
},
|
| 386 |
{
|
| 387 |
-
"epoch": 1.
|
| 388 |
-
"learning_rate":
|
| 389 |
-
"loss": 0.
|
| 390 |
"step": 620
|
| 391 |
},
|
| 392 |
{
|
| 393 |
-
"epoch":
|
| 394 |
-
"
|
| 395 |
-
"
|
| 396 |
-
"
|
| 397 |
-
"eval_samples_per_second": 136.643,
|
| 398 |
-
"eval_steps_per_second": 4.278,
|
| 399 |
-
"step": 629
|
| 400 |
},
|
| 401 |
{
|
| 402 |
"epoch": 2.0,
|
| 403 |
-
"
|
| 404 |
-
"
|
| 405 |
-
"
|
|
|
|
|
|
|
|
|
|
| 406 |
},
|
| 407 |
{
|
| 408 |
-
"epoch": 2.
|
| 409 |
-
"learning_rate":
|
| 410 |
-
"loss": 0.
|
| 411 |
"step": 640
|
| 412 |
},
|
| 413 |
{
|
| 414 |
-
"epoch": 2.
|
| 415 |
-
"learning_rate":
|
| 416 |
-
"loss": 0.
|
| 417 |
"step": 650
|
| 418 |
},
|
| 419 |
{
|
| 420 |
-
"epoch": 2.
|
| 421 |
-
"learning_rate":
|
| 422 |
-
"loss": 0.
|
| 423 |
"step": 660
|
| 424 |
},
|
| 425 |
{
|
| 426 |
-
"epoch": 2.
|
| 427 |
-
"learning_rate":
|
| 428 |
-
"loss": 0.
|
| 429 |
"step": 670
|
| 430 |
},
|
| 431 |
{
|
| 432 |
-
"epoch": 2.
|
| 433 |
-
"learning_rate":
|
| 434 |
-
"loss": 0.
|
| 435 |
"step": 680
|
| 436 |
},
|
| 437 |
{
|
| 438 |
-
"epoch": 2.
|
| 439 |
-
"learning_rate":
|
| 440 |
-
"loss": 0.
|
| 441 |
"step": 690
|
| 442 |
},
|
| 443 |
{
|
| 444 |
-
"epoch": 2.
|
| 445 |
-
"learning_rate":
|
| 446 |
-
"loss": 0.
|
| 447 |
"step": 700
|
| 448 |
},
|
| 449 |
{
|
| 450 |
-
"epoch": 2.
|
| 451 |
-
"learning_rate":
|
| 452 |
-
"loss": 0.
|
| 453 |
"step": 710
|
| 454 |
},
|
| 455 |
{
|
| 456 |
-
"epoch": 2.
|
| 457 |
-
"learning_rate":
|
| 458 |
-
"loss": 0.
|
| 459 |
"step": 720
|
| 460 |
},
|
| 461 |
{
|
| 462 |
-
"epoch": 2.
|
| 463 |
-
"learning_rate":
|
| 464 |
-
"loss": 0.
|
| 465 |
"step": 730
|
| 466 |
},
|
| 467 |
{
|
| 468 |
-
"epoch": 2.
|
| 469 |
-
"learning_rate":
|
| 470 |
-
"loss": 0.
|
| 471 |
"step": 740
|
| 472 |
},
|
| 473 |
{
|
| 474 |
-
"epoch": 2.
|
| 475 |
-
"learning_rate":
|
| 476 |
-
"loss": 0.
|
| 477 |
"step": 750
|
| 478 |
},
|
| 479 |
{
|
| 480 |
-
"epoch": 2.
|
| 481 |
-
"learning_rate":
|
| 482 |
-
"loss": 0.
|
| 483 |
"step": 760
|
| 484 |
},
|
| 485 |
{
|
| 486 |
-
"epoch": 2.
|
| 487 |
-
"learning_rate":
|
| 488 |
-
"loss": 0.
|
| 489 |
"step": 770
|
| 490 |
},
|
| 491 |
{
|
| 492 |
-
"epoch": 2.
|
| 493 |
-
"learning_rate":
|
| 494 |
-
"loss": 0.
|
| 495 |
"step": 780
|
| 496 |
},
|
| 497 |
{
|
| 498 |
-
"epoch": 2.
|
| 499 |
-
"learning_rate":
|
| 500 |
-
"loss": 0.
|
| 501 |
"step": 790
|
| 502 |
},
|
| 503 |
{
|
| 504 |
-
"epoch": 2.
|
| 505 |
-
"learning_rate":
|
| 506 |
-
"loss": 0.
|
| 507 |
"step": 800
|
| 508 |
},
|
| 509 |
{
|
| 510 |
-
"epoch": 2.
|
| 511 |
-
"learning_rate":
|
| 512 |
-
"loss": 0.
|
| 513 |
"step": 810
|
| 514 |
},
|
| 515 |
{
|
| 516 |
-
"epoch": 2.
|
| 517 |
-
"learning_rate":
|
| 518 |
-
"loss": 0.
|
| 519 |
"step": 820
|
| 520 |
},
|
| 521 |
{
|
| 522 |
-
"epoch": 2.
|
| 523 |
-
"learning_rate":
|
| 524 |
-
"loss": 0.
|
| 525 |
"step": 830
|
| 526 |
},
|
| 527 |
{
|
| 528 |
-
"epoch": 2.
|
| 529 |
-
"learning_rate":
|
| 530 |
-
"loss": 0.
|
| 531 |
"step": 840
|
| 532 |
},
|
| 533 |
{
|
| 534 |
-
"epoch": 2.
|
| 535 |
-
"learning_rate":
|
| 536 |
-
"loss": 0.
|
| 537 |
"step": 850
|
| 538 |
},
|
| 539 |
{
|
| 540 |
-
"epoch": 2.
|
| 541 |
-
"learning_rate":
|
| 542 |
-
"loss": 0.
|
| 543 |
"step": 860
|
| 544 |
},
|
| 545 |
{
|
| 546 |
-
"epoch": 2.
|
| 547 |
-
"learning_rate":
|
| 548 |
-
"loss": 0.
|
| 549 |
"step": 870
|
| 550 |
},
|
| 551 |
{
|
| 552 |
-
"epoch": 2.
|
| 553 |
-
"learning_rate":
|
| 554 |
-
"loss": 0.
|
| 555 |
"step": 880
|
| 556 |
},
|
| 557 |
{
|
| 558 |
-
"epoch": 2.
|
| 559 |
-
"learning_rate":
|
| 560 |
-
"loss": 0.
|
| 561 |
"step": 890
|
| 562 |
},
|
| 563 |
{
|
| 564 |
-
"epoch": 2.
|
| 565 |
-
"learning_rate":
|
| 566 |
-
"loss": 0.
|
| 567 |
"step": 900
|
| 568 |
},
|
| 569 |
{
|
| 570 |
-
"epoch": 2.
|
| 571 |
-
"learning_rate":
|
| 572 |
-
"loss": 0.
|
| 573 |
"step": 910
|
| 574 |
},
|
| 575 |
{
|
| 576 |
-
"epoch": 2.
|
| 577 |
-
"learning_rate":
|
| 578 |
-
"loss": 0.
|
| 579 |
"step": 920
|
| 580 |
},
|
| 581 |
{
|
| 582 |
-
"epoch": 2.
|
| 583 |
-
"learning_rate":
|
| 584 |
-
"loss": 0.
|
| 585 |
"step": 930
|
| 586 |
},
|
| 587 |
{
|
| 588 |
-
"epoch": 2.
|
| 589 |
-
"learning_rate":
|
| 590 |
-
"loss": 0.
|
| 591 |
"step": 940
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 3.0,
|
| 595 |
-
"eval_accuracy": 0.
|
| 596 |
-
"eval_loss": 0.
|
| 597 |
-
"eval_runtime":
|
| 598 |
-
"eval_samples_per_second":
|
| 599 |
-
"eval_steps_per_second": 4.
|
| 600 |
-
"step":
|
| 601 |
},
|
| 602 |
{
|
| 603 |
-
"epoch": 3.
|
| 604 |
-
"learning_rate":
|
| 605 |
-
"loss": 0.
|
| 606 |
"step": 950
|
| 607 |
},
|
| 608 |
{
|
| 609 |
-
"epoch": 3.
|
| 610 |
-
"learning_rate":
|
| 611 |
-
"loss": 0.
|
| 612 |
"step": 960
|
| 613 |
},
|
| 614 |
{
|
| 615 |
-
"epoch": 3.
|
| 616 |
-
"learning_rate":
|
| 617 |
-
"loss": 0.
|
| 618 |
"step": 970
|
| 619 |
},
|
| 620 |
{
|
| 621 |
-
"epoch": 3.
|
| 622 |
-
"learning_rate":
|
| 623 |
-
"loss": 0.
|
| 624 |
"step": 980
|
| 625 |
},
|
| 626 |
{
|
| 627 |
-
"epoch": 3.
|
| 628 |
-
"learning_rate":
|
| 629 |
-
"loss": 0.
|
| 630 |
"step": 990
|
| 631 |
},
|
| 632 |
{
|
| 633 |
-
"epoch": 3.
|
| 634 |
-
"learning_rate":
|
| 635 |
-
"loss": 0.
|
| 636 |
"step": 1000
|
| 637 |
},
|
| 638 |
{
|
| 639 |
-
"epoch": 3.
|
| 640 |
-
"learning_rate":
|
| 641 |
-
"loss": 0.
|
| 642 |
"step": 1010
|
| 643 |
},
|
| 644 |
{
|
| 645 |
-
"epoch": 3.
|
| 646 |
-
"learning_rate":
|
| 647 |
-
"loss": 0.
|
| 648 |
"step": 1020
|
| 649 |
},
|
| 650 |
{
|
| 651 |
-
"epoch": 3.
|
| 652 |
-
"learning_rate":
|
| 653 |
-
"loss": 0.
|
| 654 |
"step": 1030
|
| 655 |
},
|
| 656 |
{
|
| 657 |
-
"epoch": 3.
|
| 658 |
-
"learning_rate":
|
| 659 |
-
"loss": 0.
|
| 660 |
"step": 1040
|
| 661 |
},
|
| 662 |
{
|
| 663 |
-
"epoch": 3.
|
| 664 |
-
"learning_rate":
|
| 665 |
-
"loss": 0.
|
| 666 |
"step": 1050
|
| 667 |
},
|
| 668 |
{
|
| 669 |
-
"epoch": 3.
|
| 670 |
-
"learning_rate":
|
| 671 |
-
"loss": 0.
|
| 672 |
"step": 1060
|
| 673 |
},
|
| 674 |
{
|
| 675 |
-
"epoch": 3.
|
| 676 |
-
"learning_rate":
|
| 677 |
-
"loss": 0.
|
| 678 |
"step": 1070
|
| 679 |
},
|
| 680 |
{
|
| 681 |
-
"epoch": 3.
|
| 682 |
-
"learning_rate":
|
| 683 |
-
"loss": 0.
|
| 684 |
"step": 1080
|
| 685 |
},
|
| 686 |
{
|
| 687 |
-
"epoch": 3.
|
| 688 |
-
"learning_rate":
|
| 689 |
-
"loss": 0.
|
| 690 |
"step": 1090
|
| 691 |
},
|
| 692 |
{
|
| 693 |
-
"epoch": 3.
|
| 694 |
-
"learning_rate":
|
| 695 |
-
"loss": 0.
|
| 696 |
"step": 1100
|
| 697 |
},
|
| 698 |
{
|
| 699 |
-
"epoch": 3.
|
| 700 |
-
"learning_rate":
|
| 701 |
-
"loss": 0.
|
| 702 |
"step": 1110
|
| 703 |
},
|
| 704 |
{
|
| 705 |
-
"epoch": 3.
|
| 706 |
-
"learning_rate":
|
| 707 |
-
"loss": 0.
|
| 708 |
"step": 1120
|
| 709 |
},
|
| 710 |
{
|
| 711 |
-
"epoch": 3.
|
| 712 |
-
"learning_rate":
|
| 713 |
-
"loss": 0.
|
| 714 |
"step": 1130
|
| 715 |
},
|
| 716 |
{
|
| 717 |
-
"epoch": 3.
|
| 718 |
-
"learning_rate":
|
| 719 |
-
"loss": 0.
|
| 720 |
"step": 1140
|
| 721 |
},
|
| 722 |
{
|
| 723 |
-
"epoch": 3.
|
| 724 |
-
"learning_rate":
|
| 725 |
-
"loss": 0.
|
| 726 |
"step": 1150
|
| 727 |
},
|
| 728 |
{
|
| 729 |
-
"epoch": 3.
|
| 730 |
-
"learning_rate":
|
| 731 |
-
"loss": 0.
|
| 732 |
"step": 1160
|
| 733 |
},
|
| 734 |
{
|
| 735 |
-
"epoch": 3.
|
| 736 |
-
"learning_rate":
|
| 737 |
-
"loss": 0.
|
| 738 |
"step": 1170
|
| 739 |
},
|
| 740 |
{
|
| 741 |
-
"epoch": 3.
|
| 742 |
-
"learning_rate":
|
| 743 |
-
"loss": 0.
|
| 744 |
"step": 1180
|
| 745 |
},
|
| 746 |
{
|
| 747 |
-
"epoch": 3.
|
| 748 |
-
"learning_rate":
|
| 749 |
-
"loss": 0.
|
| 750 |
"step": 1190
|
| 751 |
},
|
| 752 |
{
|
| 753 |
-
"epoch": 3.
|
| 754 |
-
"learning_rate":
|
| 755 |
-
"loss": 0.
|
| 756 |
"step": 1200
|
| 757 |
},
|
| 758 |
{
|
| 759 |
-
"epoch": 3.
|
| 760 |
-
"learning_rate":
|
| 761 |
-
"loss": 0.
|
| 762 |
"step": 1210
|
| 763 |
},
|
| 764 |
{
|
| 765 |
-
"epoch": 3.
|
| 766 |
-
"learning_rate":
|
| 767 |
-
"loss": 0.
|
| 768 |
"step": 1220
|
| 769 |
},
|
| 770 |
{
|
| 771 |
-
"epoch": 3.
|
| 772 |
-
"learning_rate":
|
| 773 |
-
"loss": 0.
|
| 774 |
"step": 1230
|
| 775 |
},
|
| 776 |
{
|
| 777 |
-
"epoch": 3.
|
| 778 |
-
"learning_rate":
|
| 779 |
-
"loss": 0.
|
| 780 |
"step": 1240
|
| 781 |
},
|
| 782 |
{
|
| 783 |
-
"epoch": 3.
|
| 784 |
-
"learning_rate":
|
| 785 |
-
"loss": 0.
|
| 786 |
"step": 1250
|
| 787 |
},
|
| 788 |
{
|
| 789 |
-
"epoch":
|
| 790 |
-
"
|
| 791 |
-
"
|
| 792 |
-
"
|
| 793 |
-
"eval_samples_per_second": 139.805,
|
| 794 |
-
"eval_steps_per_second": 4.377,
|
| 795 |
-
"step": 1258
|
| 796 |
},
|
| 797 |
{
|
| 798 |
-
"epoch": 4.
|
| 799 |
-
"
|
| 800 |
-
"
|
| 801 |
-
"
|
|
|
|
|
|
|
|
|
|
| 802 |
},
|
| 803 |
{
|
| 804 |
-
"epoch": 4.
|
| 805 |
-
"learning_rate":
|
| 806 |
-
"loss": 0.
|
| 807 |
"step": 1270
|
| 808 |
},
|
| 809 |
{
|
| 810 |
-
"epoch": 4.
|
| 811 |
-
"learning_rate":
|
| 812 |
-
"loss": 0.
|
| 813 |
"step": 1280
|
| 814 |
},
|
| 815 |
{
|
| 816 |
-
"epoch": 4.
|
| 817 |
-
"learning_rate":
|
| 818 |
-
"loss": 0.
|
| 819 |
"step": 1290
|
| 820 |
},
|
| 821 |
{
|
| 822 |
-
"epoch": 4.
|
| 823 |
-
"learning_rate":
|
| 824 |
-
"loss": 0.
|
| 825 |
"step": 1300
|
| 826 |
},
|
| 827 |
{
|
| 828 |
-
"epoch": 4.
|
| 829 |
-
"learning_rate":
|
| 830 |
-
"loss": 0.
|
| 831 |
"step": 1310
|
| 832 |
},
|
| 833 |
{
|
| 834 |
-
"epoch": 4.
|
| 835 |
-
"learning_rate":
|
| 836 |
-
"loss": 0.
|
| 837 |
"step": 1320
|
| 838 |
},
|
| 839 |
{
|
| 840 |
-
"epoch": 4.
|
| 841 |
-
"learning_rate":
|
| 842 |
-
"loss": 0.
|
| 843 |
"step": 1330
|
| 844 |
},
|
| 845 |
{
|
| 846 |
-
"epoch": 4.
|
| 847 |
-
"learning_rate":
|
| 848 |
-
"loss": 0.
|
| 849 |
"step": 1340
|
| 850 |
},
|
| 851 |
{
|
| 852 |
-
"epoch": 4.
|
| 853 |
-
"learning_rate":
|
| 854 |
-
"loss": 0.
|
| 855 |
"step": 1350
|
| 856 |
},
|
| 857 |
{
|
| 858 |
-
"epoch": 4.
|
| 859 |
-
"learning_rate":
|
| 860 |
-
"loss": 0.
|
| 861 |
"step": 1360
|
| 862 |
},
|
| 863 |
{
|
| 864 |
-
"epoch": 4.
|
| 865 |
-
"learning_rate":
|
| 866 |
-
"loss": 0.
|
| 867 |
"step": 1370
|
| 868 |
},
|
| 869 |
{
|
| 870 |
-
"epoch": 4.
|
| 871 |
-
"learning_rate":
|
| 872 |
-
"loss": 0.
|
| 873 |
"step": 1380
|
| 874 |
},
|
| 875 |
{
|
| 876 |
-
"epoch": 4.
|
| 877 |
-
"learning_rate":
|
| 878 |
-
"loss": 0.
|
| 879 |
"step": 1390
|
| 880 |
},
|
| 881 |
{
|
| 882 |
-
"epoch": 4.
|
| 883 |
-
"learning_rate":
|
| 884 |
-
"loss": 0.
|
| 885 |
"step": 1400
|
| 886 |
},
|
| 887 |
{
|
| 888 |
-
"epoch": 4.
|
| 889 |
-
"learning_rate":
|
| 890 |
-
"loss": 0.
|
| 891 |
"step": 1410
|
| 892 |
},
|
| 893 |
{
|
| 894 |
-
"epoch": 4.
|
| 895 |
-
"learning_rate":
|
| 896 |
-
"loss": 0.
|
| 897 |
"step": 1420
|
| 898 |
},
|
| 899 |
{
|
| 900 |
-
"epoch": 4.
|
| 901 |
-
"learning_rate":
|
| 902 |
-
"loss": 0.
|
| 903 |
"step": 1430
|
| 904 |
},
|
| 905 |
{
|
| 906 |
-
"epoch": 4.
|
| 907 |
-
"learning_rate":
|
| 908 |
-
"loss": 0.
|
| 909 |
"step": 1440
|
| 910 |
},
|
| 911 |
{
|
| 912 |
-
"epoch": 4.
|
| 913 |
-
"learning_rate":
|
| 914 |
-
"loss": 0.
|
| 915 |
"step": 1450
|
| 916 |
},
|
| 917 |
{
|
| 918 |
-
"epoch": 4.
|
| 919 |
-
"learning_rate":
|
| 920 |
-
"loss": 0.
|
| 921 |
"step": 1460
|
| 922 |
},
|
| 923 |
{
|
| 924 |
-
"epoch": 4.
|
| 925 |
-
"learning_rate":
|
| 926 |
-
"loss": 0.
|
| 927 |
"step": 1470
|
| 928 |
},
|
| 929 |
{
|
| 930 |
-
"epoch": 4.
|
| 931 |
-
"learning_rate":
|
| 932 |
-
"loss": 0.
|
| 933 |
"step": 1480
|
| 934 |
},
|
| 935 |
{
|
| 936 |
-
"epoch": 4.
|
| 937 |
-
"learning_rate": 2.
|
| 938 |
-
"loss": 0.
|
| 939 |
"step": 1490
|
| 940 |
},
|
| 941 |
{
|
| 942 |
-
"epoch": 4.
|
| 943 |
-
"learning_rate": 2.
|
| 944 |
-
"loss": 0.
|
| 945 |
"step": 1500
|
| 946 |
},
|
| 947 |
{
|
| 948 |
-
"epoch": 4.
|
| 949 |
-
"learning_rate": 2.
|
| 950 |
-
"loss": 0.
|
| 951 |
"step": 1510
|
| 952 |
},
|
| 953 |
{
|
| 954 |
-
"epoch": 4.
|
| 955 |
-
"learning_rate":
|
| 956 |
-
"loss": 0.
|
| 957 |
"step": 1520
|
| 958 |
},
|
| 959 |
{
|
| 960 |
-
"epoch": 4.
|
| 961 |
-
"learning_rate":
|
| 962 |
-
"loss": 0.
|
| 963 |
"step": 1530
|
| 964 |
},
|
| 965 |
{
|
| 966 |
-
"epoch": 4.
|
| 967 |
-
"learning_rate":
|
| 968 |
-
"loss": 0.
|
| 969 |
"step": 1540
|
| 970 |
},
|
| 971 |
{
|
| 972 |
-
"epoch": 4.
|
| 973 |
-
"learning_rate":
|
| 974 |
-
"loss": 0.
|
| 975 |
"step": 1550
|
| 976 |
},
|
| 977 |
{
|
| 978 |
-
"epoch": 4.
|
| 979 |
-
"learning_rate":
|
| 980 |
-
"loss": 0.
|
| 981 |
"step": 1560
|
| 982 |
},
|
| 983 |
{
|
| 984 |
-
"epoch": 4.
|
| 985 |
-
"learning_rate":
|
| 986 |
-
"loss": 0.
|
| 987 |
"step": 1570
|
| 988 |
},
|
| 989 |
{
|
| 990 |
"epoch": 4.99,
|
| 991 |
-
"
|
| 992 |
-
"
|
| 993 |
-
"
|
| 994 |
-
"eval_samples_per_second": 135.451,
|
| 995 |
-
"eval_steps_per_second": 4.24,
|
| 996 |
-
"step": 1570
|
| 997 |
},
|
| 998 |
{
|
| 999 |
-
"epoch":
|
| 1000 |
-
"
|
| 1001 |
-
"
|
| 1002 |
-
"
|
| 1003 |
-
"
|
| 1004 |
-
"
|
| 1005 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1006 |
}
|
| 1007 |
],
|
| 1008 |
"logging_steps": 10,
|
| 1009 |
-
"max_steps":
|
| 1010 |
-
"num_train_epochs":
|
| 1011 |
"save_steps": 500,
|
| 1012 |
-
"total_flos":
|
| 1013 |
"trial_name": null,
|
| 1014 |
"trial_params": null
|
| 1015 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.7280604310153299,
|
| 3 |
+
"best_model_checkpoint": "swin-tiny-patch4-window7-224-img_orientation/checkpoint-3160",
|
| 4 |
+
"epoch": 9.984202211690363,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 3160,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
| 10 |
"log_history": [
|
| 11 |
{
|
| 12 |
"epoch": 0.03,
|
| 13 |
+
"learning_rate": 1.5822784810126583e-06,
|
| 14 |
+
"loss": 1.4093,
|
| 15 |
"step": 10
|
| 16 |
},
|
| 17 |
{
|
| 18 |
"epoch": 0.06,
|
| 19 |
+
"learning_rate": 3.1645569620253167e-06,
|
| 20 |
+
"loss": 1.3983,
|
| 21 |
"step": 20
|
| 22 |
},
|
| 23 |
{
|
| 24 |
+
"epoch": 0.09,
|
| 25 |
+
"learning_rate": 4.746835443037975e-06,
|
| 26 |
+
"loss": 1.402,
|
| 27 |
"step": 30
|
| 28 |
},
|
| 29 |
{
|
| 30 |
"epoch": 0.13,
|
| 31 |
+
"learning_rate": 6.329113924050633e-06,
|
| 32 |
+
"loss": 1.3799,
|
| 33 |
"step": 40
|
| 34 |
},
|
| 35 |
{
|
| 36 |
"epoch": 0.16,
|
| 37 |
+
"learning_rate": 7.911392405063292e-06,
|
| 38 |
+
"loss": 1.3363,
|
| 39 |
"step": 50
|
| 40 |
},
|
| 41 |
{
|
| 42 |
"epoch": 0.19,
|
| 43 |
+
"learning_rate": 9.49367088607595e-06,
|
| 44 |
+
"loss": 1.3071,
|
| 45 |
"step": 60
|
| 46 |
},
|
| 47 |
{
|
| 48 |
"epoch": 0.22,
|
| 49 |
+
"learning_rate": 1.1075949367088608e-05,
|
| 50 |
+
"loss": 1.2847,
|
| 51 |
"step": 70
|
| 52 |
},
|
| 53 |
{
|
| 54 |
"epoch": 0.25,
|
| 55 |
+
"learning_rate": 1.2658227848101267e-05,
|
| 56 |
+
"loss": 1.2458,
|
| 57 |
"step": 80
|
| 58 |
},
|
| 59 |
{
|
| 60 |
+
"epoch": 0.28,
|
| 61 |
+
"learning_rate": 1.4240506329113925e-05,
|
| 62 |
+
"loss": 1.1755,
|
| 63 |
"step": 90
|
| 64 |
},
|
| 65 |
{
|
| 66 |
"epoch": 0.32,
|
| 67 |
+
"learning_rate": 1.5822784810126583e-05,
|
| 68 |
+
"loss": 1.1232,
|
| 69 |
"step": 100
|
| 70 |
},
|
| 71 |
{
|
| 72 |
"epoch": 0.35,
|
| 73 |
+
"learning_rate": 1.7405063291139243e-05,
|
| 74 |
+
"loss": 1.0368,
|
| 75 |
"step": 110
|
| 76 |
},
|
| 77 |
{
|
| 78 |
"epoch": 0.38,
|
| 79 |
+
"learning_rate": 1.89873417721519e-05,
|
| 80 |
+
"loss": 0.9468,
|
| 81 |
"step": 120
|
| 82 |
},
|
| 83 |
{
|
| 84 |
"epoch": 0.41,
|
| 85 |
+
"learning_rate": 2.056962025316456e-05,
|
| 86 |
+
"loss": 0.867,
|
| 87 |
"step": 130
|
| 88 |
},
|
| 89 |
{
|
| 90 |
+
"epoch": 0.44,
|
| 91 |
+
"learning_rate": 2.2151898734177217e-05,
|
| 92 |
+
"loss": 0.8772,
|
| 93 |
"step": 140
|
| 94 |
},
|
| 95 |
{
|
| 96 |
+
"epoch": 0.47,
|
| 97 |
+
"learning_rate": 2.3734177215189873e-05,
|
| 98 |
+
"loss": 0.7954,
|
| 99 |
"step": 150
|
| 100 |
},
|
| 101 |
{
|
| 102 |
"epoch": 0.51,
|
| 103 |
+
"learning_rate": 2.5316455696202533e-05,
|
| 104 |
+
"loss": 0.7559,
|
| 105 |
"step": 160
|
| 106 |
},
|
| 107 |
{
|
| 108 |
"epoch": 0.54,
|
| 109 |
+
"learning_rate": 2.689873417721519e-05,
|
| 110 |
+
"loss": 0.8245,
|
| 111 |
"step": 170
|
| 112 |
},
|
| 113 |
{
|
| 114 |
"epoch": 0.57,
|
| 115 |
+
"learning_rate": 2.848101265822785e-05,
|
| 116 |
+
"loss": 0.7358,
|
| 117 |
"step": 180
|
| 118 |
},
|
| 119 |
{
|
| 120 |
"epoch": 0.6,
|
| 121 |
+
"learning_rate": 3.0063291139240506e-05,
|
| 122 |
+
"loss": 0.7766,
|
| 123 |
"step": 190
|
| 124 |
},
|
| 125 |
{
|
| 126 |
+
"epoch": 0.63,
|
| 127 |
+
"learning_rate": 3.1645569620253167e-05,
|
| 128 |
+
"loss": 0.7109,
|
| 129 |
"step": 200
|
| 130 |
},
|
| 131 |
{
|
| 132 |
+
"epoch": 0.66,
|
| 133 |
+
"learning_rate": 3.322784810126582e-05,
|
| 134 |
+
"loss": 0.738,
|
| 135 |
"step": 210
|
| 136 |
},
|
| 137 |
{
|
| 138 |
"epoch": 0.7,
|
| 139 |
+
"learning_rate": 3.4810126582278487e-05,
|
| 140 |
+
"loss": 0.7185,
|
| 141 |
"step": 220
|
| 142 |
},
|
| 143 |
{
|
| 144 |
"epoch": 0.73,
|
| 145 |
+
"learning_rate": 3.639240506329114e-05,
|
| 146 |
+
"loss": 0.7008,
|
| 147 |
"step": 230
|
| 148 |
},
|
| 149 |
{
|
| 150 |
"epoch": 0.76,
|
| 151 |
+
"learning_rate": 3.79746835443038e-05,
|
| 152 |
+
"loss": 0.7009,
|
| 153 |
"step": 240
|
| 154 |
},
|
| 155 |
{
|
| 156 |
"epoch": 0.79,
|
| 157 |
+
"learning_rate": 3.9556962025316456e-05,
|
| 158 |
+
"loss": 0.677,
|
| 159 |
"step": 250
|
| 160 |
},
|
| 161 |
{
|
| 162 |
+
"epoch": 0.82,
|
| 163 |
+
"learning_rate": 4.113924050632912e-05,
|
| 164 |
+
"loss": 0.6927,
|
| 165 |
"step": 260
|
| 166 |
},
|
| 167 |
{
|
| 168 |
+
"epoch": 0.85,
|
| 169 |
+
"learning_rate": 4.2721518987341776e-05,
|
| 170 |
+
"loss": 0.7135,
|
| 171 |
"step": 270
|
| 172 |
},
|
| 173 |
{
|
| 174 |
+
"epoch": 0.88,
|
| 175 |
+
"learning_rate": 4.430379746835443e-05,
|
| 176 |
+
"loss": 0.6918,
|
| 177 |
"step": 280
|
| 178 |
},
|
| 179 |
{
|
| 180 |
"epoch": 0.92,
|
| 181 |
+
"learning_rate": 4.588607594936709e-05,
|
| 182 |
+
"loss": 0.7152,
|
| 183 |
"step": 290
|
| 184 |
},
|
| 185 |
{
|
| 186 |
"epoch": 0.95,
|
| 187 |
+
"learning_rate": 4.7468354430379746e-05,
|
| 188 |
+
"loss": 0.695,
|
| 189 |
"step": 300
|
| 190 |
},
|
| 191 |
{
|
| 192 |
+
"epoch": 0.98,
|
| 193 |
+
"learning_rate": 4.905063291139241e-05,
|
| 194 |
+
"loss": 0.6659,
|
| 195 |
"step": 310
|
| 196 |
},
|
| 197 |
{
|
| 198 |
"epoch": 1.0,
|
| 199 |
+
"eval_accuracy": 0.6849588980226616,
|
| 200 |
+
"eval_loss": 0.5535573363304138,
|
| 201 |
+
"eval_runtime": 33.1556,
|
| 202 |
+
"eval_samples_per_second": 135.754,
|
| 203 |
+
"eval_steps_per_second": 4.253,
|
| 204 |
+
"step": 316
|
| 205 |
},
|
| 206 |
{
|
| 207 |
+
"epoch": 1.01,
|
| 208 |
+
"learning_rate": 4.9929676511955e-05,
|
| 209 |
+
"loss": 0.7086,
|
| 210 |
"step": 320
|
| 211 |
},
|
| 212 |
{
|
| 213 |
+
"epoch": 1.04,
|
| 214 |
+
"learning_rate": 4.975386779184248e-05,
|
| 215 |
+
"loss": 0.6302,
|
| 216 |
"step": 330
|
| 217 |
},
|
| 218 |
{
|
| 219 |
+
"epoch": 1.07,
|
| 220 |
+
"learning_rate": 4.957805907172996e-05,
|
| 221 |
+
"loss": 0.6449,
|
| 222 |
"step": 340
|
| 223 |
},
|
| 224 |
{
|
| 225 |
"epoch": 1.11,
|
| 226 |
+
"learning_rate": 4.940225035161744e-05,
|
| 227 |
+
"loss": 0.6431,
|
| 228 |
"step": 350
|
| 229 |
},
|
| 230 |
{
|
| 231 |
"epoch": 1.14,
|
| 232 |
+
"learning_rate": 4.9226441631504925e-05,
|
| 233 |
+
"loss": 0.6261,
|
| 234 |
"step": 360
|
| 235 |
},
|
| 236 |
{
|
| 237 |
+
"epoch": 1.17,
|
| 238 |
+
"learning_rate": 4.905063291139241e-05,
|
| 239 |
+
"loss": 0.7035,
|
| 240 |
"step": 370
|
| 241 |
},
|
| 242 |
{
|
| 243 |
+
"epoch": 1.2,
|
| 244 |
+
"learning_rate": 4.887482419127989e-05,
|
| 245 |
+
"loss": 0.6453,
|
| 246 |
"step": 380
|
| 247 |
},
|
| 248 |
{
|
| 249 |
+
"epoch": 1.23,
|
| 250 |
+
"learning_rate": 4.869901547116737e-05,
|
| 251 |
+
"loss": 0.624,
|
| 252 |
"step": 390
|
| 253 |
},
|
| 254 |
{
|
| 255 |
+
"epoch": 1.26,
|
| 256 |
+
"learning_rate": 4.852320675105486e-05,
|
| 257 |
+
"loss": 0.6493,
|
| 258 |
"step": 400
|
| 259 |
},
|
| 260 |
{
|
| 261 |
"epoch": 1.3,
|
| 262 |
+
"learning_rate": 4.8347398030942336e-05,
|
| 263 |
+
"loss": 0.6148,
|
| 264 |
"step": 410
|
| 265 |
},
|
| 266 |
{
|
| 267 |
+
"epoch": 1.33,
|
| 268 |
+
"learning_rate": 4.817158931082982e-05,
|
| 269 |
+
"loss": 0.6194,
|
| 270 |
"step": 420
|
| 271 |
},
|
| 272 |
{
|
| 273 |
+
"epoch": 1.36,
|
| 274 |
+
"learning_rate": 4.7995780590717305e-05,
|
| 275 |
+
"loss": 0.6322,
|
| 276 |
"step": 430
|
| 277 |
},
|
| 278 |
{
|
| 279 |
+
"epoch": 1.39,
|
| 280 |
+
"learning_rate": 4.7819971870604783e-05,
|
| 281 |
+
"loss": 0.6201,
|
| 282 |
"step": 440
|
| 283 |
},
|
| 284 |
{
|
| 285 |
+
"epoch": 1.42,
|
| 286 |
+
"learning_rate": 4.764416315049227e-05,
|
| 287 |
+
"loss": 0.6492,
|
| 288 |
"step": 450
|
| 289 |
},
|
| 290 |
{
|
| 291 |
+
"epoch": 1.45,
|
| 292 |
+
"learning_rate": 4.7468354430379746e-05,
|
| 293 |
+
"loss": 0.5862,
|
| 294 |
"step": 460
|
| 295 |
},
|
| 296 |
{
|
| 297 |
+
"epoch": 1.48,
|
| 298 |
+
"learning_rate": 4.729254571026723e-05,
|
| 299 |
+
"loss": 0.6019,
|
| 300 |
"step": 470
|
| 301 |
},
|
| 302 |
{
|
| 303 |
+
"epoch": 1.52,
|
| 304 |
+
"learning_rate": 4.7116736990154716e-05,
|
| 305 |
+
"loss": 0.6338,
|
| 306 |
"step": 480
|
| 307 |
},
|
| 308 |
{
|
| 309 |
+
"epoch": 1.55,
|
| 310 |
+
"learning_rate": 4.6940928270042194e-05,
|
| 311 |
+
"loss": 0.6159,
|
| 312 |
"step": 490
|
| 313 |
},
|
| 314 |
{
|
| 315 |
+
"epoch": 1.58,
|
| 316 |
+
"learning_rate": 4.676511954992968e-05,
|
| 317 |
+
"loss": 0.6494,
|
| 318 |
"step": 500
|
| 319 |
},
|
| 320 |
{
|
| 321 |
+
"epoch": 1.61,
|
| 322 |
+
"learning_rate": 4.6589310829817164e-05,
|
| 323 |
+
"loss": 0.606,
|
| 324 |
"step": 510
|
| 325 |
},
|
| 326 |
{
|
| 327 |
+
"epoch": 1.64,
|
| 328 |
+
"learning_rate": 4.641350210970464e-05,
|
| 329 |
+
"loss": 0.62,
|
| 330 |
"step": 520
|
| 331 |
},
|
| 332 |
{
|
| 333 |
+
"epoch": 1.67,
|
| 334 |
+
"learning_rate": 4.623769338959213e-05,
|
| 335 |
+
"loss": 0.6263,
|
| 336 |
"step": 530
|
| 337 |
},
|
| 338 |
{
|
| 339 |
+
"epoch": 1.71,
|
| 340 |
+
"learning_rate": 4.606188466947961e-05,
|
| 341 |
+
"loss": 0.5747,
|
| 342 |
"step": 540
|
| 343 |
},
|
| 344 |
{
|
| 345 |
+
"epoch": 1.74,
|
| 346 |
+
"learning_rate": 4.588607594936709e-05,
|
| 347 |
+
"loss": 0.6158,
|
| 348 |
"step": 550
|
| 349 |
},
|
| 350 |
{
|
| 351 |
+
"epoch": 1.77,
|
| 352 |
+
"learning_rate": 4.5710267229254575e-05,
|
| 353 |
+
"loss": 0.5928,
|
| 354 |
"step": 560
|
| 355 |
},
|
| 356 |
{
|
| 357 |
+
"epoch": 1.8,
|
| 358 |
+
"learning_rate": 4.553445850914206e-05,
|
| 359 |
+
"loss": 0.5981,
|
| 360 |
"step": 570
|
| 361 |
},
|
| 362 |
{
|
| 363 |
+
"epoch": 1.83,
|
| 364 |
+
"learning_rate": 4.535864978902954e-05,
|
| 365 |
+
"loss": 0.6166,
|
| 366 |
"step": 580
|
| 367 |
},
|
| 368 |
{
|
| 369 |
+
"epoch": 1.86,
|
| 370 |
+
"learning_rate": 4.518284106891702e-05,
|
| 371 |
+
"loss": 0.6215,
|
| 372 |
"step": 590
|
| 373 |
},
|
| 374 |
{
|
| 375 |
+
"epoch": 1.9,
|
| 376 |
+
"learning_rate": 4.50070323488045e-05,
|
| 377 |
+
"loss": 0.6185,
|
| 378 |
"step": 600
|
| 379 |
},
|
| 380 |
{
|
| 381 |
+
"epoch": 1.93,
|
| 382 |
+
"learning_rate": 4.4831223628691985e-05,
|
| 383 |
+
"loss": 0.5896,
|
| 384 |
"step": 610
|
| 385 |
},
|
| 386 |
{
|
| 387 |
+
"epoch": 1.96,
|
| 388 |
+
"learning_rate": 4.465541490857947e-05,
|
| 389 |
+
"loss": 0.5894,
|
| 390 |
"step": 620
|
| 391 |
},
|
| 392 |
{
|
| 393 |
+
"epoch": 1.99,
|
| 394 |
+
"learning_rate": 4.447960618846695e-05,
|
| 395 |
+
"loss": 0.5971,
|
| 396 |
+
"step": 630
|
|
|
|
|
|
|
|
|
|
| 397 |
},
|
| 398 |
{
|
| 399 |
"epoch": 2.0,
|
| 400 |
+
"eval_accuracy": 0.7169517884914464,
|
| 401 |
+
"eval_loss": 0.4985543489456177,
|
| 402 |
+
"eval_runtime": 33.2772,
|
| 403 |
+
"eval_samples_per_second": 135.258,
|
| 404 |
+
"eval_steps_per_second": 4.237,
|
| 405 |
+
"step": 633
|
| 406 |
},
|
| 407 |
{
|
| 408 |
+
"epoch": 2.02,
|
| 409 |
+
"learning_rate": 4.430379746835443e-05,
|
| 410 |
+
"loss": 0.5902,
|
| 411 |
"step": 640
|
| 412 |
},
|
| 413 |
{
|
| 414 |
+
"epoch": 2.05,
|
| 415 |
+
"learning_rate": 4.412798874824192e-05,
|
| 416 |
+
"loss": 0.5861,
|
| 417 |
"step": 650
|
| 418 |
},
|
| 419 |
{
|
| 420 |
+
"epoch": 2.09,
|
| 421 |
+
"learning_rate": 4.3952180028129396e-05,
|
| 422 |
+
"loss": 0.5626,
|
| 423 |
"step": 660
|
| 424 |
},
|
| 425 |
{
|
| 426 |
+
"epoch": 2.12,
|
| 427 |
+
"learning_rate": 4.377637130801688e-05,
|
| 428 |
+
"loss": 0.537,
|
| 429 |
"step": 670
|
| 430 |
},
|
| 431 |
{
|
| 432 |
+
"epoch": 2.15,
|
| 433 |
+
"learning_rate": 4.3600562587904366e-05,
|
| 434 |
+
"loss": 0.554,
|
| 435 |
"step": 680
|
| 436 |
},
|
| 437 |
{
|
| 438 |
+
"epoch": 2.18,
|
| 439 |
+
"learning_rate": 4.3424753867791844e-05,
|
| 440 |
+
"loss": 0.5916,
|
| 441 |
"step": 690
|
| 442 |
},
|
| 443 |
{
|
| 444 |
+
"epoch": 2.21,
|
| 445 |
+
"learning_rate": 4.324894514767933e-05,
|
| 446 |
+
"loss": 0.5525,
|
| 447 |
"step": 700
|
| 448 |
},
|
| 449 |
{
|
| 450 |
+
"epoch": 2.24,
|
| 451 |
+
"learning_rate": 4.307313642756681e-05,
|
| 452 |
+
"loss": 0.5568,
|
| 453 |
"step": 710
|
| 454 |
},
|
| 455 |
{
|
| 456 |
+
"epoch": 2.27,
|
| 457 |
+
"learning_rate": 4.289732770745429e-05,
|
| 458 |
+
"loss": 0.5474,
|
| 459 |
"step": 720
|
| 460 |
},
|
| 461 |
{
|
| 462 |
+
"epoch": 2.31,
|
| 463 |
+
"learning_rate": 4.2721518987341776e-05,
|
| 464 |
+
"loss": 0.5914,
|
| 465 |
"step": 730
|
| 466 |
},
|
| 467 |
{
|
| 468 |
+
"epoch": 2.34,
|
| 469 |
+
"learning_rate": 4.2545710267229255e-05,
|
| 470 |
+
"loss": 0.5652,
|
| 471 |
"step": 740
|
| 472 |
},
|
| 473 |
{
|
| 474 |
+
"epoch": 2.37,
|
| 475 |
+
"learning_rate": 4.236990154711674e-05,
|
| 476 |
+
"loss": 0.5814,
|
| 477 |
"step": 750
|
| 478 |
},
|
| 479 |
{
|
| 480 |
+
"epoch": 2.4,
|
| 481 |
+
"learning_rate": 4.2194092827004224e-05,
|
| 482 |
+
"loss": 0.5703,
|
| 483 |
"step": 760
|
| 484 |
},
|
| 485 |
{
|
| 486 |
+
"epoch": 2.43,
|
| 487 |
+
"learning_rate": 4.20182841068917e-05,
|
| 488 |
+
"loss": 0.564,
|
| 489 |
"step": 770
|
| 490 |
},
|
| 491 |
{
|
| 492 |
+
"epoch": 2.46,
|
| 493 |
+
"learning_rate": 4.184247538677919e-05,
|
| 494 |
+
"loss": 0.5755,
|
| 495 |
"step": 780
|
| 496 |
},
|
| 497 |
{
|
| 498 |
+
"epoch": 2.5,
|
| 499 |
+
"learning_rate": 4.166666666666667e-05,
|
| 500 |
+
"loss": 0.5765,
|
| 501 |
"step": 790
|
| 502 |
},
|
| 503 |
{
|
| 504 |
+
"epoch": 2.53,
|
| 505 |
+
"learning_rate": 4.149085794655415e-05,
|
| 506 |
+
"loss": 0.567,
|
| 507 |
"step": 800
|
| 508 |
},
|
| 509 |
{
|
| 510 |
+
"epoch": 2.56,
|
| 511 |
+
"learning_rate": 4.1315049226441635e-05,
|
| 512 |
+
"loss": 0.5398,
|
| 513 |
"step": 810
|
| 514 |
},
|
| 515 |
{
|
| 516 |
+
"epoch": 2.59,
|
| 517 |
+
"learning_rate": 4.113924050632912e-05,
|
| 518 |
+
"loss": 0.5746,
|
| 519 |
"step": 820
|
| 520 |
},
|
| 521 |
{
|
| 522 |
+
"epoch": 2.62,
|
| 523 |
+
"learning_rate": 4.09634317862166e-05,
|
| 524 |
+
"loss": 0.5443,
|
| 525 |
"step": 830
|
| 526 |
},
|
| 527 |
{
|
| 528 |
+
"epoch": 2.65,
|
| 529 |
+
"learning_rate": 4.078762306610408e-05,
|
| 530 |
+
"loss": 0.5384,
|
| 531 |
"step": 840
|
| 532 |
},
|
| 533 |
{
|
| 534 |
+
"epoch": 2.69,
|
| 535 |
+
"learning_rate": 4.061181434599156e-05,
|
| 536 |
+
"loss": 0.5861,
|
| 537 |
"step": 850
|
| 538 |
},
|
| 539 |
{
|
| 540 |
+
"epoch": 2.72,
|
| 541 |
+
"learning_rate": 4.0436005625879046e-05,
|
| 542 |
+
"loss": 0.5332,
|
| 543 |
"step": 860
|
| 544 |
},
|
| 545 |
{
|
| 546 |
+
"epoch": 2.75,
|
| 547 |
+
"learning_rate": 4.026019690576653e-05,
|
| 548 |
+
"loss": 0.5776,
|
| 549 |
"step": 870
|
| 550 |
},
|
| 551 |
{
|
| 552 |
+
"epoch": 2.78,
|
| 553 |
+
"learning_rate": 4.008438818565401e-05,
|
| 554 |
+
"loss": 0.548,
|
| 555 |
"step": 880
|
| 556 |
},
|
| 557 |
{
|
| 558 |
+
"epoch": 2.81,
|
| 559 |
+
"learning_rate": 3.9908579465541493e-05,
|
| 560 |
+
"loss": 0.5758,
|
| 561 |
"step": 890
|
| 562 |
},
|
| 563 |
{
|
| 564 |
+
"epoch": 2.84,
|
| 565 |
+
"learning_rate": 3.973277074542898e-05,
|
| 566 |
+
"loss": 0.5691,
|
| 567 |
"step": 900
|
| 568 |
},
|
| 569 |
{
|
| 570 |
+
"epoch": 2.88,
|
| 571 |
+
"learning_rate": 3.9556962025316456e-05,
|
| 572 |
+
"loss": 0.5284,
|
| 573 |
"step": 910
|
| 574 |
},
|
| 575 |
{
|
| 576 |
+
"epoch": 2.91,
|
| 577 |
+
"learning_rate": 3.938115330520394e-05,
|
| 578 |
+
"loss": 0.5518,
|
| 579 |
"step": 920
|
| 580 |
},
|
| 581 |
{
|
| 582 |
+
"epoch": 2.94,
|
| 583 |
+
"learning_rate": 3.9205344585091426e-05,
|
| 584 |
+
"loss": 0.5507,
|
| 585 |
"step": 930
|
| 586 |
},
|
| 587 |
{
|
| 588 |
+
"epoch": 2.97,
|
| 589 |
+
"learning_rate": 3.9029535864978904e-05,
|
| 590 |
+
"loss": 0.5782,
|
| 591 |
"step": 940
|
| 592 |
},
|
| 593 |
{
|
| 594 |
"epoch": 3.0,
|
| 595 |
+
"eval_accuracy": 0.717173961341924,
|
| 596 |
+
"eval_loss": 0.48247379064559937,
|
| 597 |
+
"eval_runtime": 33.4685,
|
| 598 |
+
"eval_samples_per_second": 134.485,
|
| 599 |
+
"eval_steps_per_second": 4.213,
|
| 600 |
+
"step": 949
|
| 601 |
},
|
| 602 |
{
|
| 603 |
+
"epoch": 3.0,
|
| 604 |
+
"learning_rate": 3.885372714486639e-05,
|
| 605 |
+
"loss": 0.5663,
|
| 606 |
"step": 950
|
| 607 |
},
|
| 608 |
{
|
| 609 |
+
"epoch": 3.03,
|
| 610 |
+
"learning_rate": 3.867791842475387e-05,
|
| 611 |
+
"loss": 0.5212,
|
| 612 |
"step": 960
|
| 613 |
},
|
| 614 |
{
|
| 615 |
+
"epoch": 3.06,
|
| 616 |
+
"learning_rate": 3.850210970464135e-05,
|
| 617 |
+
"loss": 0.5294,
|
| 618 |
"step": 970
|
| 619 |
},
|
| 620 |
{
|
| 621 |
+
"epoch": 3.1,
|
| 622 |
+
"learning_rate": 3.832630098452884e-05,
|
| 623 |
+
"loss": 0.525,
|
| 624 |
"step": 980
|
| 625 |
},
|
| 626 |
{
|
| 627 |
+
"epoch": 3.13,
|
| 628 |
+
"learning_rate": 3.8150492264416315e-05,
|
| 629 |
+
"loss": 0.5303,
|
| 630 |
"step": 990
|
| 631 |
},
|
| 632 |
{
|
| 633 |
+
"epoch": 3.16,
|
| 634 |
+
"learning_rate": 3.79746835443038e-05,
|
| 635 |
+
"loss": 0.5189,
|
| 636 |
"step": 1000
|
| 637 |
},
|
| 638 |
{
|
| 639 |
+
"epoch": 3.19,
|
| 640 |
+
"learning_rate": 3.7798874824191285e-05,
|
| 641 |
+
"loss": 0.5325,
|
| 642 |
"step": 1010
|
| 643 |
},
|
| 644 |
{
|
| 645 |
+
"epoch": 3.22,
|
| 646 |
+
"learning_rate": 3.762306610407876e-05,
|
| 647 |
+
"loss": 0.5308,
|
| 648 |
"step": 1020
|
| 649 |
},
|
| 650 |
{
|
| 651 |
+
"epoch": 3.25,
|
| 652 |
+
"learning_rate": 3.744725738396625e-05,
|
| 653 |
+
"loss": 0.5326,
|
| 654 |
"step": 1030
|
| 655 |
},
|
| 656 |
{
|
| 657 |
+
"epoch": 3.29,
|
| 658 |
+
"learning_rate": 3.727144866385373e-05,
|
| 659 |
+
"loss": 0.5045,
|
| 660 |
"step": 1040
|
| 661 |
},
|
| 662 |
{
|
| 663 |
+
"epoch": 3.32,
|
| 664 |
+
"learning_rate": 3.709563994374121e-05,
|
| 665 |
+
"loss": 0.5444,
|
| 666 |
"step": 1050
|
| 667 |
},
|
| 668 |
{
|
| 669 |
+
"epoch": 3.35,
|
| 670 |
+
"learning_rate": 3.6919831223628695e-05,
|
| 671 |
+
"loss": 0.5263,
|
| 672 |
"step": 1060
|
| 673 |
},
|
| 674 |
{
|
| 675 |
+
"epoch": 3.38,
|
| 676 |
+
"learning_rate": 3.674402250351618e-05,
|
| 677 |
+
"loss": 0.5251,
|
| 678 |
"step": 1070
|
| 679 |
},
|
| 680 |
{
|
| 681 |
+
"epoch": 3.41,
|
| 682 |
+
"learning_rate": 3.656821378340366e-05,
|
| 683 |
+
"loss": 0.5168,
|
| 684 |
"step": 1080
|
| 685 |
},
|
| 686 |
{
|
| 687 |
+
"epoch": 3.44,
|
| 688 |
+
"learning_rate": 3.639240506329114e-05,
|
| 689 |
+
"loss": 0.5121,
|
| 690 |
"step": 1090
|
| 691 |
},
|
| 692 |
{
|
| 693 |
+
"epoch": 3.48,
|
| 694 |
+
"learning_rate": 3.621659634317862e-05,
|
| 695 |
+
"loss": 0.5179,
|
| 696 |
"step": 1100
|
| 697 |
},
|
| 698 |
{
|
| 699 |
+
"epoch": 3.51,
|
| 700 |
+
"learning_rate": 3.6040787623066106e-05,
|
| 701 |
+
"loss": 0.5435,
|
| 702 |
"step": 1110
|
| 703 |
},
|
| 704 |
{
|
| 705 |
+
"epoch": 3.54,
|
| 706 |
+
"learning_rate": 3.586497890295359e-05,
|
| 707 |
+
"loss": 0.5158,
|
| 708 |
"step": 1120
|
| 709 |
},
|
| 710 |
{
|
| 711 |
+
"epoch": 3.57,
|
| 712 |
+
"learning_rate": 3.568917018284107e-05,
|
| 713 |
+
"loss": 0.4941,
|
| 714 |
"step": 1130
|
| 715 |
},
|
| 716 |
{
|
| 717 |
+
"epoch": 3.6,
|
| 718 |
+
"learning_rate": 3.551336146272855e-05,
|
| 719 |
+
"loss": 0.5679,
|
| 720 |
"step": 1140
|
| 721 |
},
|
| 722 |
{
|
| 723 |
+
"epoch": 3.63,
|
| 724 |
+
"learning_rate": 3.533755274261604e-05,
|
| 725 |
+
"loss": 0.5137,
|
| 726 |
"step": 1150
|
| 727 |
},
|
| 728 |
{
|
| 729 |
+
"epoch": 3.67,
|
| 730 |
+
"learning_rate": 3.516174402250352e-05,
|
| 731 |
+
"loss": 0.5221,
|
| 732 |
"step": 1160
|
| 733 |
},
|
| 734 |
{
|
| 735 |
+
"epoch": 3.7,
|
| 736 |
+
"learning_rate": 3.4985935302391e-05,
|
| 737 |
+
"loss": 0.5304,
|
| 738 |
"step": 1170
|
| 739 |
},
|
| 740 |
{
|
| 741 |
+
"epoch": 3.73,
|
| 742 |
+
"learning_rate": 3.4810126582278487e-05,
|
| 743 |
+
"loss": 0.5066,
|
| 744 |
"step": 1180
|
| 745 |
},
|
| 746 |
{
|
| 747 |
+
"epoch": 3.76,
|
| 748 |
+
"learning_rate": 3.4634317862165965e-05,
|
| 749 |
+
"loss": 0.5467,
|
| 750 |
"step": 1190
|
| 751 |
},
|
| 752 |
{
|
| 753 |
+
"epoch": 3.79,
|
| 754 |
+
"learning_rate": 3.445850914205345e-05,
|
| 755 |
+
"loss": 0.5249,
|
| 756 |
"step": 1200
|
| 757 |
},
|
| 758 |
{
|
| 759 |
+
"epoch": 3.82,
|
| 760 |
+
"learning_rate": 3.428270042194093e-05,
|
| 761 |
+
"loss": 0.5639,
|
| 762 |
"step": 1210
|
| 763 |
},
|
| 764 |
{
|
| 765 |
+
"epoch": 3.85,
|
| 766 |
+
"learning_rate": 3.410689170182841e-05,
|
| 767 |
+
"loss": 0.522,
|
| 768 |
"step": 1220
|
| 769 |
},
|
| 770 |
{
|
| 771 |
+
"epoch": 3.89,
|
| 772 |
+
"learning_rate": 3.39310829817159e-05,
|
| 773 |
+
"loss": 0.5195,
|
| 774 |
"step": 1230
|
| 775 |
},
|
| 776 |
{
|
| 777 |
+
"epoch": 3.92,
|
| 778 |
+
"learning_rate": 3.3755274261603375e-05,
|
| 779 |
+
"loss": 0.4986,
|
| 780 |
"step": 1240
|
| 781 |
},
|
| 782 |
{
|
| 783 |
+
"epoch": 3.95,
|
| 784 |
+
"learning_rate": 3.357946554149086e-05,
|
| 785 |
+
"loss": 0.5187,
|
| 786 |
"step": 1250
|
| 787 |
},
|
| 788 |
{
|
| 789 |
+
"epoch": 3.98,
|
| 790 |
+
"learning_rate": 3.3403656821378345e-05,
|
| 791 |
+
"loss": 0.5428,
|
| 792 |
+
"step": 1260
|
|
|
|
|
|
|
|
|
|
| 793 |
},
|
| 794 |
{
|
| 795 |
+
"epoch": 4.0,
|
| 796 |
+
"eval_accuracy": 0.7140635414352366,
|
| 797 |
+
"eval_loss": 0.46642085909843445,
|
| 798 |
+
"eval_runtime": 33.1603,
|
| 799 |
+
"eval_samples_per_second": 135.735,
|
| 800 |
+
"eval_steps_per_second": 4.252,
|
| 801 |
+
"step": 1266
|
| 802 |
},
|
| 803 |
{
|
| 804 |
+
"epoch": 4.01,
|
| 805 |
+
"learning_rate": 3.322784810126582e-05,
|
| 806 |
+
"loss": 0.5234,
|
| 807 |
"step": 1270
|
| 808 |
},
|
| 809 |
{
|
| 810 |
+
"epoch": 4.04,
|
| 811 |
+
"learning_rate": 3.305203938115331e-05,
|
| 812 |
+
"loss": 0.5088,
|
| 813 |
"step": 1280
|
| 814 |
},
|
| 815 |
{
|
| 816 |
+
"epoch": 4.08,
|
| 817 |
+
"learning_rate": 3.287623066104079e-05,
|
| 818 |
+
"loss": 0.4914,
|
| 819 |
"step": 1290
|
| 820 |
},
|
| 821 |
{
|
| 822 |
+
"epoch": 4.11,
|
| 823 |
+
"learning_rate": 3.270042194092827e-05,
|
| 824 |
+
"loss": 0.5056,
|
| 825 |
"step": 1300
|
| 826 |
},
|
| 827 |
{
|
| 828 |
+
"epoch": 4.14,
|
| 829 |
+
"learning_rate": 3.2524613220815756e-05,
|
| 830 |
+
"loss": 0.5277,
|
| 831 |
"step": 1310
|
| 832 |
},
|
| 833 |
{
|
| 834 |
+
"epoch": 4.17,
|
| 835 |
+
"learning_rate": 3.234880450070324e-05,
|
| 836 |
+
"loss": 0.5003,
|
| 837 |
"step": 1320
|
| 838 |
},
|
| 839 |
{
|
| 840 |
+
"epoch": 4.2,
|
| 841 |
+
"learning_rate": 3.217299578059072e-05,
|
| 842 |
+
"loss": 0.4859,
|
| 843 |
"step": 1330
|
| 844 |
},
|
| 845 |
{
|
| 846 |
+
"epoch": 4.23,
|
| 847 |
+
"learning_rate": 3.1997187060478204e-05,
|
| 848 |
+
"loss": 0.5113,
|
| 849 |
"step": 1340
|
| 850 |
},
|
| 851 |
{
|
| 852 |
+
"epoch": 4.27,
|
| 853 |
+
"learning_rate": 3.182137834036568e-05,
|
| 854 |
+
"loss": 0.5146,
|
| 855 |
"step": 1350
|
| 856 |
},
|
| 857 |
{
|
| 858 |
+
"epoch": 4.3,
|
| 859 |
+
"learning_rate": 3.1645569620253167e-05,
|
| 860 |
+
"loss": 0.522,
|
| 861 |
"step": 1360
|
| 862 |
},
|
| 863 |
{
|
| 864 |
+
"epoch": 4.33,
|
| 865 |
+
"learning_rate": 3.146976090014065e-05,
|
| 866 |
+
"loss": 0.486,
|
| 867 |
"step": 1370
|
| 868 |
},
|
| 869 |
{
|
| 870 |
+
"epoch": 4.36,
|
| 871 |
+
"learning_rate": 3.129395218002813e-05,
|
| 872 |
+
"loss": 0.5025,
|
| 873 |
"step": 1380
|
| 874 |
},
|
| 875 |
{
|
| 876 |
+
"epoch": 4.39,
|
| 877 |
+
"learning_rate": 3.111814345991561e-05,
|
| 878 |
+
"loss": 0.5099,
|
| 879 |
"step": 1390
|
| 880 |
},
|
| 881 |
{
|
| 882 |
+
"epoch": 4.42,
|
| 883 |
+
"learning_rate": 3.09423347398031e-05,
|
| 884 |
+
"loss": 0.5161,
|
| 885 |
"step": 1400
|
| 886 |
},
|
| 887 |
{
|
| 888 |
+
"epoch": 4.45,
|
| 889 |
+
"learning_rate": 3.076652601969058e-05,
|
| 890 |
+
"loss": 0.4902,
|
| 891 |
"step": 1410
|
| 892 |
},
|
| 893 |
{
|
| 894 |
+
"epoch": 4.49,
|
| 895 |
+
"learning_rate": 3.059071729957806e-05,
|
| 896 |
+
"loss": 0.5406,
|
| 897 |
"step": 1420
|
| 898 |
},
|
| 899 |
{
|
| 900 |
+
"epoch": 4.52,
|
| 901 |
+
"learning_rate": 3.0414908579465547e-05,
|
| 902 |
+
"loss": 0.5025,
|
| 903 |
"step": 1430
|
| 904 |
},
|
| 905 |
{
|
| 906 |
+
"epoch": 4.55,
|
| 907 |
+
"learning_rate": 3.0239099859353025e-05,
|
| 908 |
+
"loss": 0.4895,
|
| 909 |
"step": 1440
|
| 910 |
},
|
| 911 |
{
|
| 912 |
+
"epoch": 4.58,
|
| 913 |
+
"learning_rate": 3.0063291139240506e-05,
|
| 914 |
+
"loss": 0.5104,
|
| 915 |
"step": 1450
|
| 916 |
},
|
| 917 |
{
|
| 918 |
+
"epoch": 4.61,
|
| 919 |
+
"learning_rate": 2.9887482419127988e-05,
|
| 920 |
+
"loss": 0.5078,
|
| 921 |
"step": 1460
|
| 922 |
},
|
| 923 |
{
|
| 924 |
+
"epoch": 4.64,
|
| 925 |
+
"learning_rate": 2.9711673699015473e-05,
|
| 926 |
+
"loss": 0.5011,
|
| 927 |
"step": 1470
|
| 928 |
},
|
| 929 |
{
|
| 930 |
+
"epoch": 4.68,
|
| 931 |
+
"learning_rate": 2.9535864978902954e-05,
|
| 932 |
+
"loss": 0.5105,
|
| 933 |
"step": 1480
|
| 934 |
},
|
| 935 |
{
|
| 936 |
+
"epoch": 4.71,
|
| 937 |
+
"learning_rate": 2.9360056258790436e-05,
|
| 938 |
+
"loss": 0.5066,
|
| 939 |
"step": 1490
|
| 940 |
},
|
| 941 |
{
|
| 942 |
+
"epoch": 4.74,
|
| 943 |
+
"learning_rate": 2.9184247538677924e-05,
|
| 944 |
+
"loss": 0.4699,
|
| 945 |
"step": 1500
|
| 946 |
},
|
| 947 |
{
|
| 948 |
+
"epoch": 4.77,
|
| 949 |
+
"learning_rate": 2.9008438818565402e-05,
|
| 950 |
+
"loss": 0.5125,
|
| 951 |
"step": 1510
|
| 952 |
},
|
| 953 |
{
|
| 954 |
+
"epoch": 4.8,
|
| 955 |
+
"learning_rate": 2.8832630098452884e-05,
|
| 956 |
+
"loss": 0.4876,
|
| 957 |
"step": 1520
|
| 958 |
},
|
| 959 |
{
|
| 960 |
+
"epoch": 4.83,
|
| 961 |
+
"learning_rate": 2.8656821378340365e-05,
|
| 962 |
+
"loss": 0.5184,
|
| 963 |
"step": 1530
|
| 964 |
},
|
| 965 |
{
|
| 966 |
+
"epoch": 4.87,
|
| 967 |
+
"learning_rate": 2.848101265822785e-05,
|
| 968 |
+
"loss": 0.4957,
|
| 969 |
"step": 1540
|
| 970 |
},
|
| 971 |
{
|
| 972 |
+
"epoch": 4.9,
|
| 973 |
+
"learning_rate": 2.830520393811533e-05,
|
| 974 |
+
"loss": 0.5218,
|
| 975 |
"step": 1550
|
| 976 |
},
|
| 977 |
{
|
| 978 |
+
"epoch": 4.93,
|
| 979 |
+
"learning_rate": 2.8129395218002813e-05,
|
| 980 |
+
"loss": 0.478,
|
| 981 |
"step": 1560
|
| 982 |
},
|
| 983 |
{
|
| 984 |
+
"epoch": 4.96,
|
| 985 |
+
"learning_rate": 2.7953586497890294e-05,
|
| 986 |
+
"loss": 0.5133,
|
| 987 |
"step": 1570
|
| 988 |
},
|
| 989 |
{
|
| 990 |
"epoch": 4.99,
|
| 991 |
+
"learning_rate": 2.777777777777778e-05,
|
| 992 |
+
"loss": 0.5131,
|
| 993 |
+
"step": 1580
|
|
|
|
|
|
|
|
|
|
| 994 |
},
|
| 995 |
{
|
| 996 |
+
"epoch": 5.0,
|
| 997 |
+
"eval_accuracy": 0.7149522328371473,
|
| 998 |
+
"eval_loss": 0.47850412130355835,
|
| 999 |
+
"eval_runtime": 34.3593,
|
| 1000 |
+
"eval_samples_per_second": 130.998,
|
| 1001 |
+
"eval_steps_per_second": 4.104,
|
| 1002 |
+
"step": 1582
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 5.02,
|
| 1006 |
+
"learning_rate": 2.760196905766526e-05,
|
| 1007 |
+
"loss": 0.5038,
|
| 1008 |
+
"step": 1590
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 5.06,
|
| 1012 |
+
"learning_rate": 2.7426160337552742e-05,
|
| 1013 |
+
"loss": 0.4844,
|
| 1014 |
+
"step": 1600
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 5.09,
|
| 1018 |
+
"learning_rate": 2.7250351617440227e-05,
|
| 1019 |
+
"loss": 0.4941,
|
| 1020 |
+
"step": 1610
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 5.12,
|
| 1024 |
+
"learning_rate": 2.707454289732771e-05,
|
| 1025 |
+
"loss": 0.4742,
|
| 1026 |
+
"step": 1620
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 5.15,
|
| 1030 |
+
"learning_rate": 2.689873417721519e-05,
|
| 1031 |
+
"loss": 0.4599,
|
| 1032 |
+
"step": 1630
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 5.18,
|
| 1036 |
+
"learning_rate": 2.672292545710267e-05,
|
| 1037 |
+
"loss": 0.5227,
|
| 1038 |
+
"step": 1640
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 5.21,
|
| 1042 |
+
"learning_rate": 2.6547116736990156e-05,
|
| 1043 |
+
"loss": 0.4716,
|
| 1044 |
+
"step": 1650
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 5.24,
|
| 1048 |
+
"learning_rate": 2.6371308016877638e-05,
|
| 1049 |
+
"loss": 0.4848,
|
| 1050 |
+
"step": 1660
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 5.28,
|
| 1054 |
+
"learning_rate": 2.619549929676512e-05,
|
| 1055 |
+
"loss": 0.4654,
|
| 1056 |
+
"step": 1670
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 5.31,
|
| 1060 |
+
"learning_rate": 2.6019690576652604e-05,
|
| 1061 |
+
"loss": 0.4686,
|
| 1062 |
+
"step": 1680
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 5.34,
|
| 1066 |
+
"learning_rate": 2.5843881856540085e-05,
|
| 1067 |
+
"loss": 0.4513,
|
| 1068 |
+
"step": 1690
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 5.37,
|
| 1072 |
+
"learning_rate": 2.5668073136427567e-05,
|
| 1073 |
+
"loss": 0.4758,
|
| 1074 |
+
"step": 1700
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 5.4,
|
| 1078 |
+
"learning_rate": 2.549226441631505e-05,
|
| 1079 |
+
"loss": 0.4773,
|
| 1080 |
+
"step": 1710
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 5.43,
|
| 1084 |
+
"learning_rate": 2.5316455696202533e-05,
|
| 1085 |
+
"loss": 0.4661,
|
| 1086 |
+
"step": 1720
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 5.47,
|
| 1090 |
+
"learning_rate": 2.5140646976090015e-05,
|
| 1091 |
+
"loss": 0.5042,
|
| 1092 |
+
"step": 1730
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 5.5,
|
| 1096 |
+
"learning_rate": 2.49648382559775e-05,
|
| 1097 |
+
"loss": 0.4584,
|
| 1098 |
+
"step": 1740
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 5.53,
|
| 1102 |
+
"learning_rate": 2.478902953586498e-05,
|
| 1103 |
+
"loss": 0.4758,
|
| 1104 |
+
"step": 1750
|
| 1105 |
+
},
|
| 1106 |
+
{
|
| 1107 |
+
"epoch": 5.56,
|
| 1108 |
+
"learning_rate": 2.4613220815752462e-05,
|
| 1109 |
+
"loss": 0.4803,
|
| 1110 |
+
"step": 1760
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 5.59,
|
| 1114 |
+
"learning_rate": 2.4437412095639944e-05,
|
| 1115 |
+
"loss": 0.4805,
|
| 1116 |
+
"step": 1770
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 5.62,
|
| 1120 |
+
"learning_rate": 2.426160337552743e-05,
|
| 1121 |
+
"loss": 0.4807,
|
| 1122 |
+
"step": 1780
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 5.66,
|
| 1126 |
+
"learning_rate": 2.408579465541491e-05,
|
| 1127 |
+
"loss": 0.4921,
|
| 1128 |
+
"step": 1790
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 5.69,
|
| 1132 |
+
"learning_rate": 2.3909985935302392e-05,
|
| 1133 |
+
"loss": 0.4705,
|
| 1134 |
+
"step": 1800
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 5.72,
|
| 1138 |
+
"learning_rate": 2.3734177215189873e-05,
|
| 1139 |
+
"loss": 0.4972,
|
| 1140 |
+
"step": 1810
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 5.75,
|
| 1144 |
+
"learning_rate": 2.3558368495077358e-05,
|
| 1145 |
+
"loss": 0.4872,
|
| 1146 |
+
"step": 1820
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 5.78,
|
| 1150 |
+
"learning_rate": 2.338255977496484e-05,
|
| 1151 |
+
"loss": 0.4633,
|
| 1152 |
+
"step": 1830
|
| 1153 |
+
},
|
| 1154 |
+
{
|
| 1155 |
+
"epoch": 5.81,
|
| 1156 |
+
"learning_rate": 2.320675105485232e-05,
|
| 1157 |
+
"loss": 0.5115,
|
| 1158 |
+
"step": 1840
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 5.85,
|
| 1162 |
+
"learning_rate": 2.3030942334739806e-05,
|
| 1163 |
+
"loss": 0.4849,
|
| 1164 |
+
"step": 1850
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 5.88,
|
| 1168 |
+
"learning_rate": 2.2855133614627287e-05,
|
| 1169 |
+
"loss": 0.5129,
|
| 1170 |
+
"step": 1860
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 5.91,
|
| 1174 |
+
"learning_rate": 2.267932489451477e-05,
|
| 1175 |
+
"loss": 0.4727,
|
| 1176 |
+
"step": 1870
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 5.94,
|
| 1180 |
+
"learning_rate": 2.250351617440225e-05,
|
| 1181 |
+
"loss": 0.4904,
|
| 1182 |
+
"step": 1880
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 5.97,
|
| 1186 |
+
"learning_rate": 2.2327707454289735e-05,
|
| 1187 |
+
"loss": 0.4851,
|
| 1188 |
+
"step": 1890
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 6.0,
|
| 1192 |
+
"eval_accuracy": 0.7225061097533881,
|
| 1193 |
+
"eval_loss": 0.47057706117630005,
|
| 1194 |
+
"eval_runtime": 34.9373,
|
| 1195 |
+
"eval_samples_per_second": 128.831,
|
| 1196 |
+
"eval_steps_per_second": 4.036,
|
| 1197 |
+
"step": 1899
|
| 1198 |
+
},
|
| 1199 |
+
{
|
| 1200 |
+
"epoch": 6.0,
|
| 1201 |
+
"learning_rate": 2.2151898734177217e-05,
|
| 1202 |
+
"loss": 0.4668,
|
| 1203 |
+
"step": 1900
|
| 1204 |
+
},
|
| 1205 |
+
{
|
| 1206 |
+
"epoch": 6.03,
|
| 1207 |
+
"learning_rate": 2.1976090014064698e-05,
|
| 1208 |
+
"loss": 0.4821,
|
| 1209 |
+
"step": 1910
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 6.07,
|
| 1213 |
+
"learning_rate": 2.1800281293952183e-05,
|
| 1214 |
+
"loss": 0.4689,
|
| 1215 |
+
"step": 1920
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 6.1,
|
| 1219 |
+
"learning_rate": 2.1624472573839664e-05,
|
| 1220 |
+
"loss": 0.4384,
|
| 1221 |
+
"step": 1930
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 6.13,
|
| 1225 |
+
"learning_rate": 2.1448663853727146e-05,
|
| 1226 |
+
"loss": 0.4348,
|
| 1227 |
+
"step": 1940
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 6.16,
|
| 1231 |
+
"learning_rate": 2.1272855133614627e-05,
|
| 1232 |
+
"loss": 0.4707,
|
| 1233 |
+
"step": 1950
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 6.19,
|
| 1237 |
+
"learning_rate": 2.1097046413502112e-05,
|
| 1238 |
+
"loss": 0.4661,
|
| 1239 |
+
"step": 1960
|
| 1240 |
+
},
|
| 1241 |
+
{
|
| 1242 |
+
"epoch": 6.22,
|
| 1243 |
+
"learning_rate": 2.0921237693389594e-05,
|
| 1244 |
+
"loss": 0.4809,
|
| 1245 |
+
"step": 1970
|
| 1246 |
+
},
|
| 1247 |
+
{
|
| 1248 |
+
"epoch": 6.26,
|
| 1249 |
+
"learning_rate": 2.0745428973277075e-05,
|
| 1250 |
+
"loss": 0.4811,
|
| 1251 |
+
"step": 1980
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 6.29,
|
| 1255 |
+
"learning_rate": 2.056962025316456e-05,
|
| 1256 |
+
"loss": 0.4604,
|
| 1257 |
+
"step": 1990
|
| 1258 |
+
},
|
| 1259 |
+
{
|
| 1260 |
+
"epoch": 6.32,
|
| 1261 |
+
"learning_rate": 2.039381153305204e-05,
|
| 1262 |
+
"loss": 0.4435,
|
| 1263 |
+
"step": 2000
|
| 1264 |
+
},
|
| 1265 |
+
{
|
| 1266 |
+
"epoch": 6.35,
|
| 1267 |
+
"learning_rate": 2.0218002812939523e-05,
|
| 1268 |
+
"loss": 0.4801,
|
| 1269 |
+
"step": 2010
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 6.38,
|
| 1273 |
+
"learning_rate": 2.0042194092827004e-05,
|
| 1274 |
+
"loss": 0.4808,
|
| 1275 |
+
"step": 2020
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 6.41,
|
| 1279 |
+
"learning_rate": 1.986638537271449e-05,
|
| 1280 |
+
"loss": 0.4588,
|
| 1281 |
+
"step": 2030
|
| 1282 |
+
},
|
| 1283 |
+
{
|
| 1284 |
+
"epoch": 6.45,
|
| 1285 |
+
"learning_rate": 1.969057665260197e-05,
|
| 1286 |
+
"loss": 0.4518,
|
| 1287 |
+
"step": 2040
|
| 1288 |
+
},
|
| 1289 |
+
{
|
| 1290 |
+
"epoch": 6.48,
|
| 1291 |
+
"learning_rate": 1.9514767932489452e-05,
|
| 1292 |
+
"loss": 0.435,
|
| 1293 |
+
"step": 2050
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 6.51,
|
| 1297 |
+
"learning_rate": 1.9338959212376934e-05,
|
| 1298 |
+
"loss": 0.5161,
|
| 1299 |
+
"step": 2060
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 6.54,
|
| 1303 |
+
"learning_rate": 1.916315049226442e-05,
|
| 1304 |
+
"loss": 0.4665,
|
| 1305 |
+
"step": 2070
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 6.57,
|
| 1309 |
+
"learning_rate": 1.89873417721519e-05,
|
| 1310 |
+
"loss": 0.4403,
|
| 1311 |
+
"step": 2080
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 6.6,
|
| 1315 |
+
"learning_rate": 1.881153305203938e-05,
|
| 1316 |
+
"loss": 0.4782,
|
| 1317 |
+
"step": 2090
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 6.64,
|
| 1321 |
+
"learning_rate": 1.8635724331926866e-05,
|
| 1322 |
+
"loss": 0.4756,
|
| 1323 |
+
"step": 2100
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 6.67,
|
| 1327 |
+
"learning_rate": 1.8459915611814348e-05,
|
| 1328 |
+
"loss": 0.472,
|
| 1329 |
+
"step": 2110
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 6.7,
|
| 1333 |
+
"learning_rate": 1.828410689170183e-05,
|
| 1334 |
+
"loss": 0.4347,
|
| 1335 |
+
"step": 2120
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 6.73,
|
| 1339 |
+
"learning_rate": 1.810829817158931e-05,
|
| 1340 |
+
"loss": 0.4708,
|
| 1341 |
+
"step": 2130
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 6.76,
|
| 1345 |
+
"learning_rate": 1.7932489451476795e-05,
|
| 1346 |
+
"loss": 0.4573,
|
| 1347 |
+
"step": 2140
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 6.79,
|
| 1351 |
+
"learning_rate": 1.7756680731364274e-05,
|
| 1352 |
+
"loss": 0.4791,
|
| 1353 |
+
"step": 2150
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 6.82,
|
| 1357 |
+
"learning_rate": 1.758087201125176e-05,
|
| 1358 |
+
"loss": 0.4355,
|
| 1359 |
+
"step": 2160
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 6.86,
|
| 1363 |
+
"learning_rate": 1.7405063291139243e-05,
|
| 1364 |
+
"loss": 0.4981,
|
| 1365 |
+
"step": 2170
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 6.89,
|
| 1369 |
+
"learning_rate": 1.7229254571026725e-05,
|
| 1370 |
+
"loss": 0.4776,
|
| 1371 |
+
"step": 2180
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 6.92,
|
| 1375 |
+
"learning_rate": 1.7053445850914206e-05,
|
| 1376 |
+
"loss": 0.4564,
|
| 1377 |
+
"step": 2190
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 6.95,
|
| 1381 |
+
"learning_rate": 1.6877637130801688e-05,
|
| 1382 |
+
"loss": 0.4516,
|
| 1383 |
+
"step": 2200
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 6.98,
|
| 1387 |
+
"learning_rate": 1.6701828410689173e-05,
|
| 1388 |
+
"loss": 0.4457,
|
| 1389 |
+
"step": 2210
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 7.0,
|
| 1393 |
+
"eval_accuracy": 0.7187291712952677,
|
| 1394 |
+
"eval_loss": 0.4728682339191437,
|
| 1395 |
+
"eval_runtime": 33.7148,
|
| 1396 |
+
"eval_samples_per_second": 133.502,
|
| 1397 |
+
"eval_steps_per_second": 4.182,
|
| 1398 |
+
"step": 2215
|
| 1399 |
+
},
|
| 1400 |
+
{
|
| 1401 |
+
"epoch": 7.01,
|
| 1402 |
+
"learning_rate": 1.6526019690576654e-05,
|
| 1403 |
+
"loss": 0.4374,
|
| 1404 |
+
"step": 2220
|
| 1405 |
+
},
|
| 1406 |
+
{
|
| 1407 |
+
"epoch": 7.05,
|
| 1408 |
+
"learning_rate": 1.6350210970464135e-05,
|
| 1409 |
+
"loss": 0.4525,
|
| 1410 |
+
"step": 2230
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 7.08,
|
| 1414 |
+
"learning_rate": 1.617440225035162e-05,
|
| 1415 |
+
"loss": 0.4369,
|
| 1416 |
+
"step": 2240
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 7.11,
|
| 1420 |
+
"learning_rate": 1.5998593530239102e-05,
|
| 1421 |
+
"loss": 0.4517,
|
| 1422 |
+
"step": 2250
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 7.14,
|
| 1426 |
+
"learning_rate": 1.5822784810126583e-05,
|
| 1427 |
+
"loss": 0.4763,
|
| 1428 |
+
"step": 2260
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 7.17,
|
| 1432 |
+
"learning_rate": 1.5646976090014065e-05,
|
| 1433 |
+
"loss": 0.4582,
|
| 1434 |
+
"step": 2270
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 7.2,
|
| 1438 |
+
"learning_rate": 1.547116736990155e-05,
|
| 1439 |
+
"loss": 0.4737,
|
| 1440 |
+
"step": 2280
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 7.24,
|
| 1444 |
+
"learning_rate": 1.529535864978903e-05,
|
| 1445 |
+
"loss": 0.4456,
|
| 1446 |
+
"step": 2290
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 7.27,
|
| 1450 |
+
"learning_rate": 1.5119549929676513e-05,
|
| 1451 |
+
"loss": 0.4535,
|
| 1452 |
+
"step": 2300
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 7.3,
|
| 1456 |
+
"learning_rate": 1.4943741209563994e-05,
|
| 1457 |
+
"loss": 0.469,
|
| 1458 |
+
"step": 2310
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 7.33,
|
| 1462 |
+
"learning_rate": 1.4767932489451477e-05,
|
| 1463 |
+
"loss": 0.4567,
|
| 1464 |
+
"step": 2320
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 7.36,
|
| 1468 |
+
"learning_rate": 1.4592123769338962e-05,
|
| 1469 |
+
"loss": 0.4264,
|
| 1470 |
+
"step": 2330
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 7.39,
|
| 1474 |
+
"learning_rate": 1.4416315049226442e-05,
|
| 1475 |
+
"loss": 0.4565,
|
| 1476 |
+
"step": 2340
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 7.42,
|
| 1480 |
+
"learning_rate": 1.4240506329113925e-05,
|
| 1481 |
+
"loss": 0.4376,
|
| 1482 |
+
"step": 2350
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 7.46,
|
| 1486 |
+
"learning_rate": 1.4064697609001406e-05,
|
| 1487 |
+
"loss": 0.4415,
|
| 1488 |
+
"step": 2360
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 7.49,
|
| 1492 |
+
"learning_rate": 1.388888888888889e-05,
|
| 1493 |
+
"loss": 0.459,
|
| 1494 |
+
"step": 2370
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 7.52,
|
| 1498 |
+
"learning_rate": 1.3713080168776371e-05,
|
| 1499 |
+
"loss": 0.4562,
|
| 1500 |
+
"step": 2380
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 7.55,
|
| 1504 |
+
"learning_rate": 1.3537271448663854e-05,
|
| 1505 |
+
"loss": 0.4733,
|
| 1506 |
+
"step": 2390
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 7.58,
|
| 1510 |
+
"learning_rate": 1.3361462728551336e-05,
|
| 1511 |
+
"loss": 0.4324,
|
| 1512 |
+
"step": 2400
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 7.61,
|
| 1516 |
+
"learning_rate": 1.3185654008438819e-05,
|
| 1517 |
+
"loss": 0.4385,
|
| 1518 |
+
"step": 2410
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 7.65,
|
| 1522 |
+
"learning_rate": 1.3009845288326302e-05,
|
| 1523 |
+
"loss": 0.466,
|
| 1524 |
+
"step": 2420
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 7.68,
|
| 1528 |
+
"learning_rate": 1.2834036568213783e-05,
|
| 1529 |
+
"loss": 0.4525,
|
| 1530 |
+
"step": 2430
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 7.71,
|
| 1534 |
+
"learning_rate": 1.2658227848101267e-05,
|
| 1535 |
+
"loss": 0.4319,
|
| 1536 |
+
"step": 2440
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 7.74,
|
| 1540 |
+
"learning_rate": 1.248241912798875e-05,
|
| 1541 |
+
"loss": 0.4374,
|
| 1542 |
+
"step": 2450
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 7.77,
|
| 1546 |
+
"learning_rate": 1.2306610407876231e-05,
|
| 1547 |
+
"loss": 0.4327,
|
| 1548 |
+
"step": 2460
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 7.8,
|
| 1552 |
+
"learning_rate": 1.2130801687763714e-05,
|
| 1553 |
+
"loss": 0.4557,
|
| 1554 |
+
"step": 2470
|
| 1555 |
+
},
|
| 1556 |
+
{
|
| 1557 |
+
"epoch": 7.84,
|
| 1558 |
+
"learning_rate": 1.1954992967651196e-05,
|
| 1559 |
+
"loss": 0.4401,
|
| 1560 |
+
"step": 2480
|
| 1561 |
+
},
|
| 1562 |
+
{
|
| 1563 |
+
"epoch": 7.87,
|
| 1564 |
+
"learning_rate": 1.1779184247538679e-05,
|
| 1565 |
+
"loss": 0.4614,
|
| 1566 |
+
"step": 2490
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 7.9,
|
| 1570 |
+
"learning_rate": 1.160337552742616e-05,
|
| 1571 |
+
"loss": 0.4298,
|
| 1572 |
+
"step": 2500
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 7.93,
|
| 1576 |
+
"learning_rate": 1.1427566807313644e-05,
|
| 1577 |
+
"loss": 0.4452,
|
| 1578 |
+
"step": 2510
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 7.96,
|
| 1582 |
+
"learning_rate": 1.1251758087201125e-05,
|
| 1583 |
+
"loss": 0.4393,
|
| 1584 |
+
"step": 2520
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 7.99,
|
| 1588 |
+
"learning_rate": 1.1075949367088608e-05,
|
| 1589 |
+
"loss": 0.4407,
|
| 1590 |
+
"step": 2530
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 8.0,
|
| 1594 |
+
"eval_accuracy": 0.7207287269495668,
|
| 1595 |
+
"eval_loss": 0.47592467069625854,
|
| 1596 |
+
"eval_runtime": 33.9139,
|
| 1597 |
+
"eval_samples_per_second": 132.718,
|
| 1598 |
+
"eval_steps_per_second": 4.158,
|
| 1599 |
+
"step": 2532
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 8.03,
|
| 1603 |
+
"learning_rate": 1.0900140646976091e-05,
|
| 1604 |
+
"loss": 0.4452,
|
| 1605 |
+
"step": 2540
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 8.06,
|
| 1609 |
+
"learning_rate": 1.0724331926863573e-05,
|
| 1610 |
+
"loss": 0.473,
|
| 1611 |
+
"step": 2550
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 8.09,
|
| 1615 |
+
"learning_rate": 1.0548523206751056e-05,
|
| 1616 |
+
"loss": 0.4202,
|
| 1617 |
+
"step": 2560
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 8.12,
|
| 1621 |
+
"learning_rate": 1.0372714486638538e-05,
|
| 1622 |
+
"loss": 0.423,
|
| 1623 |
+
"step": 2570
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 8.15,
|
| 1627 |
+
"learning_rate": 1.019690576652602e-05,
|
| 1628 |
+
"loss": 0.3977,
|
| 1629 |
+
"step": 2580
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 8.18,
|
| 1633 |
+
"learning_rate": 1.0021097046413502e-05,
|
| 1634 |
+
"loss": 0.4587,
|
| 1635 |
+
"step": 2590
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 8.21,
|
| 1639 |
+
"learning_rate": 9.845288326300985e-06,
|
| 1640 |
+
"loss": 0.4175,
|
| 1641 |
+
"step": 2600
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 8.25,
|
| 1645 |
+
"learning_rate": 9.669479606188467e-06,
|
| 1646 |
+
"loss": 0.4288,
|
| 1647 |
+
"step": 2610
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 8.28,
|
| 1651 |
+
"learning_rate": 9.49367088607595e-06,
|
| 1652 |
+
"loss": 0.4552,
|
| 1653 |
+
"step": 2620
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 8.31,
|
| 1657 |
+
"learning_rate": 9.317862165963433e-06,
|
| 1658 |
+
"loss": 0.4408,
|
| 1659 |
+
"step": 2630
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 8.34,
|
| 1663 |
+
"learning_rate": 9.142053445850915e-06,
|
| 1664 |
+
"loss": 0.4519,
|
| 1665 |
+
"step": 2640
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 8.37,
|
| 1669 |
+
"learning_rate": 8.966244725738398e-06,
|
| 1670 |
+
"loss": 0.414,
|
| 1671 |
+
"step": 2650
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 8.4,
|
| 1675 |
+
"learning_rate": 8.79043600562588e-06,
|
| 1676 |
+
"loss": 0.4362,
|
| 1677 |
+
"step": 2660
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 8.44,
|
| 1681 |
+
"learning_rate": 8.614627285513362e-06,
|
| 1682 |
+
"loss": 0.4411,
|
| 1683 |
+
"step": 2670
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 8.47,
|
| 1687 |
+
"learning_rate": 8.438818565400844e-06,
|
| 1688 |
+
"loss": 0.4271,
|
| 1689 |
+
"step": 2680
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 8.5,
|
| 1693 |
+
"learning_rate": 8.263009845288327e-06,
|
| 1694 |
+
"loss": 0.4731,
|
| 1695 |
+
"step": 2690
|
| 1696 |
+
},
|
| 1697 |
+
{
|
| 1698 |
+
"epoch": 8.53,
|
| 1699 |
+
"learning_rate": 8.08720112517581e-06,
|
| 1700 |
+
"loss": 0.4188,
|
| 1701 |
+
"step": 2700
|
| 1702 |
+
},
|
| 1703 |
+
{
|
| 1704 |
+
"epoch": 8.56,
|
| 1705 |
+
"learning_rate": 7.911392405063292e-06,
|
| 1706 |
+
"loss": 0.4252,
|
| 1707 |
+
"step": 2710
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 8.59,
|
| 1711 |
+
"learning_rate": 7.735583684950775e-06,
|
| 1712 |
+
"loss": 0.4256,
|
| 1713 |
+
"step": 2720
|
| 1714 |
+
},
|
| 1715 |
+
{
|
| 1716 |
+
"epoch": 8.63,
|
| 1717 |
+
"learning_rate": 7.559774964838256e-06,
|
| 1718 |
+
"loss": 0.4429,
|
| 1719 |
+
"step": 2730
|
| 1720 |
+
},
|
| 1721 |
+
{
|
| 1722 |
+
"epoch": 8.66,
|
| 1723 |
+
"learning_rate": 7.3839662447257386e-06,
|
| 1724 |
+
"loss": 0.416,
|
| 1725 |
+
"step": 2740
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 8.69,
|
| 1729 |
+
"learning_rate": 7.208157524613221e-06,
|
| 1730 |
+
"loss": 0.416,
|
| 1731 |
+
"step": 2750
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 8.72,
|
| 1735 |
+
"learning_rate": 7.032348804500703e-06,
|
| 1736 |
+
"loss": 0.4236,
|
| 1737 |
+
"step": 2760
|
| 1738 |
+
},
|
| 1739 |
+
{
|
| 1740 |
+
"epoch": 8.75,
|
| 1741 |
+
"learning_rate": 6.8565400843881855e-06,
|
| 1742 |
+
"loss": 0.4331,
|
| 1743 |
+
"step": 2770
|
| 1744 |
+
},
|
| 1745 |
+
{
|
| 1746 |
+
"epoch": 8.78,
|
| 1747 |
+
"learning_rate": 6.680731364275668e-06,
|
| 1748 |
+
"loss": 0.4439,
|
| 1749 |
+
"step": 2780
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 8.82,
|
| 1753 |
+
"learning_rate": 6.504922644163151e-06,
|
| 1754 |
+
"loss": 0.4286,
|
| 1755 |
+
"step": 2790
|
| 1756 |
+
},
|
| 1757 |
+
{
|
| 1758 |
+
"epoch": 8.85,
|
| 1759 |
+
"learning_rate": 6.329113924050633e-06,
|
| 1760 |
+
"loss": 0.4333,
|
| 1761 |
+
"step": 2800
|
| 1762 |
+
},
|
| 1763 |
+
{
|
| 1764 |
+
"epoch": 8.88,
|
| 1765 |
+
"learning_rate": 6.153305203938116e-06,
|
| 1766 |
+
"loss": 0.45,
|
| 1767 |
+
"step": 2810
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 8.91,
|
| 1771 |
+
"learning_rate": 5.977496483825598e-06,
|
| 1772 |
+
"loss": 0.4307,
|
| 1773 |
+
"step": 2820
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 8.94,
|
| 1777 |
+
"learning_rate": 5.80168776371308e-06,
|
| 1778 |
+
"loss": 0.4156,
|
| 1779 |
+
"step": 2830
|
| 1780 |
+
},
|
| 1781 |
+
{
|
| 1782 |
+
"epoch": 8.97,
|
| 1783 |
+
"learning_rate": 5.6258790436005626e-06,
|
| 1784 |
+
"loss": 0.4636,
|
| 1785 |
+
"step": 2840
|
| 1786 |
+
},
|
| 1787 |
+
{
|
| 1788 |
+
"epoch": 9.0,
|
| 1789 |
+
"eval_accuracy": 0.7249500111086425,
|
| 1790 |
+
"eval_loss": 0.47323304414749146,
|
| 1791 |
+
"eval_runtime": 33.1442,
|
| 1792 |
+
"eval_samples_per_second": 135.8,
|
| 1793 |
+
"eval_steps_per_second": 4.254,
|
| 1794 |
+
"step": 2848
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 9.0,
|
| 1798 |
+
"learning_rate": 5.450070323488046e-06,
|
| 1799 |
+
"loss": 0.439,
|
| 1800 |
+
"step": 2850
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 9.04,
|
| 1804 |
+
"learning_rate": 5.274261603375528e-06,
|
| 1805 |
+
"loss": 0.4224,
|
| 1806 |
+
"step": 2860
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 9.07,
|
| 1810 |
+
"learning_rate": 5.09845288326301e-06,
|
| 1811 |
+
"loss": 0.4397,
|
| 1812 |
+
"step": 2870
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 9.1,
|
| 1816 |
+
"learning_rate": 4.922644163150493e-06,
|
| 1817 |
+
"loss": 0.4092,
|
| 1818 |
+
"step": 2880
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 9.13,
|
| 1822 |
+
"learning_rate": 4.746835443037975e-06,
|
| 1823 |
+
"loss": 0.4246,
|
| 1824 |
+
"step": 2890
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 9.16,
|
| 1828 |
+
"learning_rate": 4.571026722925457e-06,
|
| 1829 |
+
"loss": 0.4304,
|
| 1830 |
+
"step": 2900
|
| 1831 |
+
},
|
| 1832 |
+
{
|
| 1833 |
+
"epoch": 9.19,
|
| 1834 |
+
"learning_rate": 4.39521800281294e-06,
|
| 1835 |
+
"loss": 0.4292,
|
| 1836 |
+
"step": 2910
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 9.23,
|
| 1840 |
+
"learning_rate": 4.219409282700422e-06,
|
| 1841 |
+
"loss": 0.3956,
|
| 1842 |
+
"step": 2920
|
| 1843 |
+
},
|
| 1844 |
+
{
|
| 1845 |
+
"epoch": 9.26,
|
| 1846 |
+
"learning_rate": 4.043600562587905e-06,
|
| 1847 |
+
"loss": 0.4152,
|
| 1848 |
+
"step": 2930
|
| 1849 |
+
},
|
| 1850 |
+
{
|
| 1851 |
+
"epoch": 9.29,
|
| 1852 |
+
"learning_rate": 3.867791842475387e-06,
|
| 1853 |
+
"loss": 0.4449,
|
| 1854 |
+
"step": 2940
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 9.32,
|
| 1858 |
+
"learning_rate": 3.6919831223628693e-06,
|
| 1859 |
+
"loss": 0.4262,
|
| 1860 |
+
"step": 2950
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 9.35,
|
| 1864 |
+
"learning_rate": 3.5161744022503516e-06,
|
| 1865 |
+
"loss": 0.4303,
|
| 1866 |
+
"step": 2960
|
| 1867 |
+
},
|
| 1868 |
+
{
|
| 1869 |
+
"epoch": 9.38,
|
| 1870 |
+
"learning_rate": 3.340365682137834e-06,
|
| 1871 |
+
"loss": 0.3949,
|
| 1872 |
+
"step": 2970
|
| 1873 |
+
},
|
| 1874 |
+
{
|
| 1875 |
+
"epoch": 9.42,
|
| 1876 |
+
"learning_rate": 3.1645569620253167e-06,
|
| 1877 |
+
"loss": 0.4235,
|
| 1878 |
+
"step": 2980
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 9.45,
|
| 1882 |
+
"learning_rate": 2.988748241912799e-06,
|
| 1883 |
+
"loss": 0.4213,
|
| 1884 |
+
"step": 2990
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 9.48,
|
| 1888 |
+
"learning_rate": 2.8129395218002813e-06,
|
| 1889 |
+
"loss": 0.419,
|
| 1890 |
+
"step": 3000
|
| 1891 |
+
},
|
| 1892 |
+
{
|
| 1893 |
+
"epoch": 9.51,
|
| 1894 |
+
"learning_rate": 2.637130801687764e-06,
|
| 1895 |
+
"loss": 0.4503,
|
| 1896 |
+
"step": 3010
|
| 1897 |
+
},
|
| 1898 |
+
{
|
| 1899 |
+
"epoch": 9.54,
|
| 1900 |
+
"learning_rate": 2.4613220815752463e-06,
|
| 1901 |
+
"loss": 0.4302,
|
| 1902 |
+
"step": 3020
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 9.57,
|
| 1906 |
+
"learning_rate": 2.2855133614627286e-06,
|
| 1907 |
+
"loss": 0.429,
|
| 1908 |
+
"step": 3030
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 9.61,
|
| 1912 |
+
"learning_rate": 2.109704641350211e-06,
|
| 1913 |
+
"loss": 0.4133,
|
| 1914 |
+
"step": 3040
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 9.64,
|
| 1918 |
+
"learning_rate": 1.9338959212376937e-06,
|
| 1919 |
+
"loss": 0.4417,
|
| 1920 |
+
"step": 3050
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 9.67,
|
| 1924 |
+
"learning_rate": 1.7580872011251758e-06,
|
| 1925 |
+
"loss": 0.4096,
|
| 1926 |
+
"step": 3060
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 9.7,
|
| 1930 |
+
"learning_rate": 1.5822784810126583e-06,
|
| 1931 |
+
"loss": 0.4601,
|
| 1932 |
+
"step": 3070
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 9.73,
|
| 1936 |
+
"learning_rate": 1.4064697609001406e-06,
|
| 1937 |
+
"loss": 0.4201,
|
| 1938 |
+
"step": 3080
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 9.76,
|
| 1942 |
+
"learning_rate": 1.2306610407876232e-06,
|
| 1943 |
+
"loss": 0.4253,
|
| 1944 |
+
"step": 3090
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 9.79,
|
| 1948 |
+
"learning_rate": 1.0548523206751055e-06,
|
| 1949 |
+
"loss": 0.4238,
|
| 1950 |
+
"step": 3100
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 9.83,
|
| 1954 |
+
"learning_rate": 8.790436005625879e-07,
|
| 1955 |
+
"loss": 0.4152,
|
| 1956 |
+
"step": 3110
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 9.86,
|
| 1960 |
+
"learning_rate": 7.032348804500703e-07,
|
| 1961 |
+
"loss": 0.4091,
|
| 1962 |
+
"step": 3120
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 9.89,
|
| 1966 |
+
"learning_rate": 5.274261603375527e-07,
|
| 1967 |
+
"loss": 0.4348,
|
| 1968 |
+
"step": 3130
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 9.92,
|
| 1972 |
+
"learning_rate": 3.5161744022503516e-07,
|
| 1973 |
+
"loss": 0.4046,
|
| 1974 |
+
"step": 3140
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 9.95,
|
| 1978 |
+
"learning_rate": 1.7580872011251758e-07,
|
| 1979 |
+
"loss": 0.3884,
|
| 1980 |
+
"step": 3150
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 9.98,
|
| 1984 |
+
"learning_rate": 0.0,
|
| 1985 |
+
"loss": 0.4212,
|
| 1986 |
+
"step": 3160
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 9.98,
|
| 1990 |
+
"eval_accuracy": 0.7280604310153299,
|
| 1991 |
+
"eval_loss": 0.4742942750453949,
|
| 1992 |
+
"eval_runtime": 32.6052,
|
| 1993 |
+
"eval_samples_per_second": 138.046,
|
| 1994 |
+
"eval_steps_per_second": 4.324,
|
| 1995 |
+
"step": 3160
|
| 1996 |
+
},
|
| 1997 |
+
{
|
| 1998 |
+
"epoch": 9.98,
|
| 1999 |
+
"step": 3160,
|
| 2000 |
+
"total_flos": 1.0051627680166625e+19,
|
| 2001 |
+
"train_loss": 0.5403492726857149,
|
| 2002 |
+
"train_runtime": 6645.2491,
|
| 2003 |
+
"train_samples_per_second": 60.946,
|
| 2004 |
+
"train_steps_per_second": 0.476
|
| 2005 |
}
|
| 2006 |
],
|
| 2007 |
"logging_steps": 10,
|
| 2008 |
+
"max_steps": 3160,
|
| 2009 |
+
"num_train_epochs": 10,
|
| 2010 |
"save_steps": 500,
|
| 2011 |
+
"total_flos": 1.0051627680166625e+19,
|
| 2012 |
"trial_name": null,
|
| 2013 |
"trial_params": null
|
| 2014 |
}
|