paul
commited on
Commit
•
884b0dc
1
Parent(s):
3cfb633
End of training
Browse files- all_results.json +6 -6
- pytorch_model.bin +1 -1
- runs/Jan16_16-52-03_teesta/events.out.tfevents.1673868138.teesta.13849.0 +2 -2
- train_results.json +6 -6
- trainer_state.json +1288 -322
all_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": 8.0,
|
| 3 |
+
"total_flos": 4.6098795710207476e+19,
|
| 4 |
+
"train_loss": 0.3678270006343594,
|
| 5 |
+
"train_runtime": 6722.1212,
|
| 6 |
+
"train_samples_per_second": 88.513,
|
| 7 |
+
"train_steps_per_second": 0.345
|
| 8 |
}
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 343282353
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9fffa524f1376fd63a09c60d8acc1a21ee47558b1dd183bb48fe5dc47cfff97f
|
| 3 |
size 343282353
|
runs/Jan16_16-52-03_teesta/events.out.tfevents.1673868138.teesta.13849.0
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3d691ffa8a775bc0dee4e53b8c31f9689293ca9ed5c8bbfc07e704235673cd26
|
| 3 |
+
size 44458
|
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": 8.0,
|
| 3 |
+
"total_flos": 4.6098795710207476e+19,
|
| 4 |
+
"train_loss": 0.3678270006343594,
|
| 5 |
+
"train_runtime": 6722.1212,
|
| 6 |
+
"train_samples_per_second": 88.513,
|
| 7 |
+
"train_steps_per_second": 0.345
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,547 +1,1513 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 0.
|
| 3 |
-
"best_model_checkpoint": "google-vit-base-patch16-224-face/checkpoint-
|
| 4 |
-
"epoch": 7.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
-
"epoch": 0.
|
| 12 |
-
"learning_rate":
|
| 13 |
-
"loss":
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
-
"epoch": 0.
|
| 18 |
-
"learning_rate":
|
| 19 |
-
"loss":
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
-
"epoch": 0.
|
| 24 |
-
"learning_rate":
|
| 25 |
-
"loss":
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
-
"epoch": 0.
|
| 30 |
-
"learning_rate":
|
| 31 |
-
"loss":
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
-
"epoch": 0.
|
| 36 |
-
"learning_rate":
|
| 37 |
-
"loss":
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
-
"epoch": 0.
|
| 42 |
-
"learning_rate":
|
| 43 |
-
"loss":
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
-
"epoch": 0.
|
| 48 |
-
"learning_rate":
|
| 49 |
-
"loss":
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
-
"epoch": 0.
|
| 54 |
-
"learning_rate":
|
| 55 |
-
"loss":
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
-
"epoch": 0.
|
| 60 |
-
"
|
| 61 |
-
"
|
| 62 |
-
"eval_loss": 0.9452694654464722,
|
| 63 |
-
"eval_precision": 0.6461904026185493,
|
| 64 |
-
"eval_recall": 0.648380355276907,
|
| 65 |
-
"eval_runtime": 32.7545,
|
| 66 |
-
"eval_samples_per_second": 175.304,
|
| 67 |
-
"eval_steps_per_second": 2.748,
|
| 68 |
-
"step": 89
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 1.01,
|
| 72 |
-
"learning_rate": 0.000116625,
|
| 73 |
-
"loss": 0.9165,
|
| 74 |
"step": 90
|
| 75 |
},
|
| 76 |
{
|
| 77 |
-
"epoch":
|
| 78 |
-
"learning_rate":
|
| 79 |
-
"loss": 0.
|
| 80 |
"step": 100
|
| 81 |
},
|
| 82 |
{
|
| 83 |
-
"epoch":
|
| 84 |
-
"learning_rate":
|
| 85 |
-
"loss":
|
| 86 |
"step": 110
|
| 87 |
},
|
| 88 |
{
|
| 89 |
-
"epoch":
|
| 90 |
-
"learning_rate":
|
| 91 |
-
"loss": 0.
|
| 92 |
"step": 120
|
| 93 |
},
|
| 94 |
{
|
| 95 |
-
"epoch":
|
| 96 |
-
"learning_rate":
|
| 97 |
-
"loss": 0.
|
| 98 |
"step": 130
|
| 99 |
},
|
| 100 |
{
|
| 101 |
-
"epoch":
|
| 102 |
-
"learning_rate":
|
| 103 |
-
"loss": 0.
|
| 104 |
"step": 140
|
| 105 |
},
|
| 106 |
{
|
| 107 |
-
"epoch":
|
| 108 |
-
"learning_rate":
|
| 109 |
-
"loss": 0.
|
| 110 |
"step": 150
|
| 111 |
},
|
| 112 |
{
|
| 113 |
-
"epoch":
|
| 114 |
-
"learning_rate":
|
| 115 |
-
"loss": 0.
|
| 116 |
"step": 160
|
| 117 |
},
|
| 118 |
{
|
| 119 |
-
"epoch":
|
| 120 |
-
"learning_rate":
|
| 121 |
-
"loss": 0.
|
| 122 |
"step": 170
|
| 123 |
},
|
| 124 |
{
|
| 125 |
-
"epoch":
|
| 126 |
-
"
|
| 127 |
-
"
|
| 128 |
-
"eval_loss": 0.8876001834869385,
|
| 129 |
-
"eval_precision": 0.679362397138447,
|
| 130 |
-
"eval_recall": 0.6778126088470916,
|
| 131 |
-
"eval_runtime": 37.9557,
|
| 132 |
-
"eval_samples_per_second": 151.282,
|
| 133 |
-
"eval_steps_per_second": 2.371,
|
| 134 |
-
"step": 178
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 2.02,
|
| 138 |
-
"learning_rate": 9.975000000000001e-05,
|
| 139 |
-
"loss": 0.7192,
|
| 140 |
"step": 180
|
| 141 |
},
|
| 142 |
{
|
| 143 |
-
"epoch":
|
| 144 |
-
"learning_rate": 9.
|
| 145 |
-
"loss": 0.
|
| 146 |
"step": 190
|
| 147 |
},
|
| 148 |
{
|
| 149 |
-
"epoch":
|
| 150 |
-
"learning_rate":
|
| 151 |
-
"loss": 0.
|
| 152 |
"step": 200
|
| 153 |
},
|
| 154 |
{
|
| 155 |
-
"epoch":
|
| 156 |
-
"learning_rate":
|
| 157 |
-
"loss": 0.
|
| 158 |
"step": 210
|
| 159 |
},
|
| 160 |
{
|
| 161 |
-
"epoch":
|
| 162 |
-
"learning_rate":
|
| 163 |
-
"loss": 0.
|
| 164 |
"step": 220
|
| 165 |
},
|
| 166 |
{
|
| 167 |
-
"epoch":
|
| 168 |
-
"learning_rate":
|
| 169 |
-
"loss": 0.
|
| 170 |
"step": 230
|
| 171 |
},
|
| 172 |
{
|
| 173 |
-
"epoch":
|
| 174 |
-
"learning_rate":
|
| 175 |
-
"loss": 0.
|
| 176 |
"step": 240
|
| 177 |
},
|
| 178 |
{
|
| 179 |
-
"epoch":
|
| 180 |
-
"learning_rate":
|
| 181 |
-
"loss": 0.
|
| 182 |
"step": 250
|
| 183 |
},
|
| 184 |
{
|
| 185 |
-
"epoch":
|
| 186 |
-
"learning_rate":
|
| 187 |
-
"loss": 0.
|
| 188 |
"step": 260
|
| 189 |
},
|
| 190 |
{
|
| 191 |
-
"epoch":
|
| 192 |
-
"
|
| 193 |
-
"
|
| 194 |
-
"eval_loss": 0.9042980670928955,
|
| 195 |
-
"eval_precision": 0.6906579200512205,
|
| 196 |
-
"eval_recall": 0.6872169975618252,
|
| 197 |
-
"eval_runtime": 32.8137,
|
| 198 |
-
"eval_samples_per_second": 174.988,
|
| 199 |
-
"eval_steps_per_second": 2.743,
|
| 200 |
-
"step": 267
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 3.03,
|
| 204 |
-
"learning_rate": 8.287500000000001e-05,
|
| 205 |
-
"loss": 0.4772,
|
| 206 |
"step": 270
|
| 207 |
},
|
| 208 |
{
|
| 209 |
-
"epoch":
|
| 210 |
-
"learning_rate":
|
| 211 |
-
"loss": 0.
|
| 212 |
"step": 280
|
| 213 |
},
|
| 214 |
{
|
| 215 |
-
"epoch":
|
| 216 |
-
"learning_rate":
|
| 217 |
-
"loss": 0.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 218 |
"step": 290
|
| 219 |
},
|
| 220 |
{
|
| 221 |
-
"epoch":
|
| 222 |
-
"learning_rate":
|
| 223 |
-
"loss": 0.
|
| 224 |
"step": 300
|
| 225 |
},
|
| 226 |
{
|
| 227 |
-
"epoch":
|
| 228 |
-
"learning_rate":
|
| 229 |
-
"loss": 0.
|
| 230 |
"step": 310
|
| 231 |
},
|
| 232 |
{
|
| 233 |
-
"epoch":
|
| 234 |
-
"learning_rate":
|
| 235 |
-
"loss": 0.
|
| 236 |
"step": 320
|
| 237 |
},
|
| 238 |
{
|
| 239 |
-
"epoch":
|
| 240 |
-
"learning_rate":
|
| 241 |
-
"loss": 0.
|
| 242 |
"step": 330
|
| 243 |
},
|
| 244 |
{
|
| 245 |
-
"epoch":
|
| 246 |
-
"learning_rate":
|
| 247 |
-
"loss": 0.
|
| 248 |
"step": 340
|
| 249 |
},
|
| 250 |
{
|
| 251 |
-
"epoch":
|
| 252 |
-
"learning_rate":
|
| 253 |
-
"loss": 0.
|
| 254 |
"step": 350
|
| 255 |
},
|
| 256 |
{
|
| 257 |
-
"epoch":
|
| 258 |
-
"
|
| 259 |
-
"
|
| 260 |
-
"eval_loss": 0.9865238666534424,
|
| 261 |
-
"eval_precision": 0.6808262458026351,
|
| 262 |
-
"eval_recall": 0.6847788227098572,
|
| 263 |
-
"eval_runtime": 34.0061,
|
| 264 |
-
"eval_samples_per_second": 168.852,
|
| 265 |
-
"eval_steps_per_second": 2.647,
|
| 266 |
-
"step": 356
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 4.04,
|
| 270 |
-
"learning_rate": 6.6e-05,
|
| 271 |
-
"loss": 0.2485,
|
| 272 |
"step": 360
|
| 273 |
},
|
| 274 |
{
|
| 275 |
-
"epoch":
|
| 276 |
-
"learning_rate":
|
| 277 |
-
"loss": 0.
|
| 278 |
"step": 370
|
| 279 |
},
|
| 280 |
{
|
| 281 |
-
"epoch":
|
| 282 |
-
"learning_rate":
|
| 283 |
-
"loss": 0.
|
| 284 |
"step": 380
|
| 285 |
},
|
| 286 |
{
|
| 287 |
-
"epoch":
|
| 288 |
-
"learning_rate":
|
| 289 |
-
"loss": 0.
|
| 290 |
"step": 390
|
| 291 |
},
|
| 292 |
{
|
| 293 |
-
"epoch":
|
| 294 |
-
"learning_rate":
|
| 295 |
-
"loss": 0.
|
| 296 |
"step": 400
|
| 297 |
},
|
| 298 |
{
|
| 299 |
-
"epoch":
|
| 300 |
-
"learning_rate":
|
| 301 |
-
"loss": 0.
|
| 302 |
"step": 410
|
| 303 |
},
|
| 304 |
{
|
| 305 |
-
"epoch":
|
| 306 |
-
"learning_rate":
|
| 307 |
-
"loss": 0.
|
| 308 |
"step": 420
|
| 309 |
},
|
| 310 |
{
|
| 311 |
-
"epoch":
|
| 312 |
-
"learning_rate":
|
| 313 |
-
"loss": 0.
|
| 314 |
"step": 430
|
| 315 |
},
|
| 316 |
{
|
| 317 |
-
"epoch":
|
| 318 |
-
"learning_rate":
|
| 319 |
-
"loss": 0.
|
| 320 |
"step": 440
|
| 321 |
},
|
| 322 |
{
|
| 323 |
-
"epoch":
|
| 324 |
-
"
|
| 325 |
-
"
|
| 326 |
-
"eval_loss": 1.1047998666763306,
|
| 327 |
-
"eval_precision": 0.6907009316583976,
|
| 328 |
-
"eval_recall": 0.6853012887495646,
|
| 329 |
-
"eval_runtime": 32.6937,
|
| 330 |
-
"eval_samples_per_second": 175.63,
|
| 331 |
-
"eval_steps_per_second": 2.753,
|
| 332 |
-
"step": 445
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 5.06,
|
| 336 |
-
"learning_rate": 4.9125e-05,
|
| 337 |
-
"loss": 0.1018,
|
| 338 |
"step": 450
|
| 339 |
},
|
| 340 |
{
|
| 341 |
-
"epoch":
|
| 342 |
-
"learning_rate":
|
| 343 |
-
"loss": 0.
|
| 344 |
"step": 460
|
| 345 |
},
|
| 346 |
{
|
| 347 |
-
"epoch":
|
| 348 |
-
"learning_rate":
|
| 349 |
-
"loss": 0.
|
| 350 |
"step": 470
|
| 351 |
},
|
| 352 |
{
|
| 353 |
-
"epoch":
|
| 354 |
-
"learning_rate":
|
| 355 |
-
"loss": 0.
|
| 356 |
"step": 480
|
| 357 |
},
|
| 358 |
{
|
| 359 |
-
"epoch":
|
| 360 |
-
"learning_rate":
|
| 361 |
-
"loss": 0.
|
| 362 |
"step": 490
|
| 363 |
},
|
| 364 |
{
|
| 365 |
-
"epoch":
|
| 366 |
-
"learning_rate":
|
| 367 |
-
"loss": 0.
|
| 368 |
"step": 500
|
| 369 |
},
|
| 370 |
{
|
| 371 |
-
"epoch":
|
| 372 |
-
"learning_rate":
|
| 373 |
-
"loss": 0.
|
| 374 |
"step": 510
|
| 375 |
},
|
| 376 |
{
|
| 377 |
-
"epoch":
|
| 378 |
-
"learning_rate":
|
| 379 |
-
"loss": 0.
|
| 380 |
"step": 520
|
| 381 |
},
|
| 382 |
{
|
| 383 |
-
"epoch":
|
| 384 |
-
"learning_rate":
|
| 385 |
-
"loss": 0.
|
| 386 |
"step": 530
|
| 387 |
},
|
| 388 |
{
|
| 389 |
-
"epoch":
|
| 390 |
-
"
|
| 391 |
-
"
|
| 392 |
-
"eval_loss": 1.2361745834350586,
|
| 393 |
-
"eval_precision": 0.6897347831759311,
|
| 394 |
-
"eval_recall": 0.6889585510275166,
|
| 395 |
-
"eval_runtime": 34.0898,
|
| 396 |
-
"eval_samples_per_second": 168.437,
|
| 397 |
-
"eval_steps_per_second": 2.64,
|
| 398 |
-
"step": 534
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 6.07,
|
| 402 |
-
"learning_rate": 3.225e-05,
|
| 403 |
-
"loss": 0.0431,
|
| 404 |
"step": 540
|
| 405 |
},
|
| 406 |
{
|
| 407 |
-
"epoch":
|
| 408 |
-
"learning_rate":
|
| 409 |
-
"loss": 0.
|
| 410 |
"step": 550
|
| 411 |
},
|
| 412 |
{
|
| 413 |
-
"epoch":
|
| 414 |
-
"learning_rate":
|
| 415 |
-
"loss": 0.
|
| 416 |
"step": 560
|
| 417 |
},
|
| 418 |
{
|
| 419 |
-
"epoch":
|
| 420 |
-
"learning_rate":
|
| 421 |
-
"loss": 0.
|
| 422 |
"step": 570
|
| 423 |
},
|
| 424 |
{
|
| 425 |
-
"epoch":
|
| 426 |
-
"learning_rate":
|
| 427 |
-
"loss": 0.
|
| 428 |
"step": 580
|
| 429 |
},
|
| 430 |
{
|
| 431 |
-
"epoch":
|
| 432 |
-
"
|
| 433 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 434 |
"step": 590
|
| 435 |
},
|
| 436 |
{
|
| 437 |
-
"epoch":
|
| 438 |
-
"learning_rate":
|
| 439 |
-
"loss": 0.
|
| 440 |
"step": 600
|
| 441 |
},
|
| 442 |
{
|
| 443 |
-
"epoch":
|
| 444 |
-
"learning_rate":
|
| 445 |
-
"loss": 0.
|
| 446 |
"step": 610
|
| 447 |
},
|
| 448 |
{
|
| 449 |
-
"epoch":
|
| 450 |
-
"learning_rate":
|
| 451 |
-
"loss": 0.
|
| 452 |
"step": 620
|
| 453 |
},
|
| 454 |
{
|
| 455 |
-
"epoch":
|
| 456 |
-
"
|
| 457 |
-
"
|
| 458 |
-
"eval_loss": 1.3140767812728882,
|
| 459 |
-
"eval_precision": 0.6925717262836596,
|
| 460 |
-
"eval_recall": 0.6931382793451759,
|
| 461 |
-
"eval_runtime": 33.0395,
|
| 462 |
-
"eval_samples_per_second": 173.792,
|
| 463 |
-
"eval_steps_per_second": 2.724,
|
| 464 |
-
"step": 623
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 7.08,
|
| 468 |
-
"learning_rate": 1.5374999999999998e-05,
|
| 469 |
-
"loss": 0.0185,
|
| 470 |
"step": 630
|
| 471 |
},
|
| 472 |
{
|
| 473 |
-
"epoch":
|
| 474 |
-
"learning_rate":
|
| 475 |
-
"loss": 0.
|
| 476 |
"step": 640
|
| 477 |
},
|
| 478 |
{
|
| 479 |
-
"epoch":
|
| 480 |
-
"learning_rate":
|
| 481 |
-
"loss": 0.
|
| 482 |
"step": 650
|
| 483 |
},
|
| 484 |
{
|
| 485 |
-
"epoch":
|
| 486 |
-
"learning_rate": 9.
|
| 487 |
-
"loss": 0.
|
| 488 |
"step": 660
|
| 489 |
},
|
| 490 |
{
|
| 491 |
-
"epoch":
|
| 492 |
-
"learning_rate":
|
| 493 |
-
"loss": 0.
|
| 494 |
"step": 670
|
| 495 |
},
|
| 496 |
{
|
| 497 |
-
"epoch":
|
| 498 |
-
"learning_rate":
|
| 499 |
-
"loss": 0.
|
| 500 |
"step": 680
|
| 501 |
},
|
| 502 |
{
|
| 503 |
-
"epoch":
|
| 504 |
-
"learning_rate":
|
| 505 |
-
"loss": 0.
|
| 506 |
"step": 690
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch":
|
| 510 |
-
"learning_rate":
|
| 511 |
-
"loss": 0.
|
| 512 |
"step": 700
|
| 513 |
},
|
| 514 |
{
|
| 515 |
-
"epoch":
|
| 516 |
-
"learning_rate":
|
| 517 |
-
"loss": 0.
|
| 518 |
"step": 710
|
| 519 |
},
|
| 520 |
{
|
| 521 |
-
"epoch":
|
| 522 |
-
"
|
| 523 |
-
"
|
| 524 |
-
"
|
| 525 |
-
|
| 526 |
-
|
| 527 |
-
"
|
| 528 |
-
"
|
| 529 |
-
"
|
| 530 |
-
"step":
|
| 531 |
-
},
|
| 532 |
-
{
|
| 533 |
-
"epoch":
|
| 534 |
-
"
|
| 535 |
-
"
|
| 536 |
-
"
|
| 537 |
-
|
| 538 |
-
|
| 539 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 540 |
}
|
| 541 |
],
|
| 542 |
-
"max_steps":
|
| 543 |
"num_train_epochs": 8,
|
| 544 |
-
"total_flos":
|
| 545 |
"trial_name": null,
|
| 546 |
"trial_params": null
|
| 547 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 0.728460793804453,
|
| 3 |
+
"best_model_checkpoint": "google-vit-base-patch16-224-face/checkpoint-870",
|
| 4 |
+
"epoch": 7.9974204643164235,
|
| 5 |
+
"global_step": 2320,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"learning_rate": 5.172413793103449e-06,
|
| 13 |
+
"loss": 2.0401,
|
| 14 |
"step": 10
|
| 15 |
},
|
| 16 |
{
|
| 17 |
+
"epoch": 0.07,
|
| 18 |
+
"learning_rate": 1.0344827586206898e-05,
|
| 19 |
+
"loss": 1.8797,
|
| 20 |
"step": 20
|
| 21 |
},
|
| 22 |
{
|
| 23 |
+
"epoch": 0.1,
|
| 24 |
+
"learning_rate": 1.5517241379310346e-05,
|
| 25 |
+
"loss": 1.6439,
|
| 26 |
"step": 30
|
| 27 |
},
|
| 28 |
{
|
| 29 |
+
"epoch": 0.14,
|
| 30 |
+
"learning_rate": 2.0689655172413797e-05,
|
| 31 |
+
"loss": 1.4464,
|
| 32 |
"step": 40
|
| 33 |
},
|
| 34 |
{
|
| 35 |
+
"epoch": 0.17,
|
| 36 |
+
"learning_rate": 2.586206896551724e-05,
|
| 37 |
+
"loss": 1.26,
|
| 38 |
"step": 50
|
| 39 |
},
|
| 40 |
{
|
| 41 |
+
"epoch": 0.21,
|
| 42 |
+
"learning_rate": 3.103448275862069e-05,
|
| 43 |
+
"loss": 1.1469,
|
| 44 |
"step": 60
|
| 45 |
},
|
| 46 |
{
|
| 47 |
+
"epoch": 0.24,
|
| 48 |
+
"learning_rate": 3.6206896551724136e-05,
|
| 49 |
+
"loss": 1.0549,
|
| 50 |
"step": 70
|
| 51 |
},
|
| 52 |
{
|
| 53 |
+
"epoch": 0.28,
|
| 54 |
+
"learning_rate": 4.137931034482759e-05,
|
| 55 |
+
"loss": 1.047,
|
| 56 |
"step": 80
|
| 57 |
},
|
| 58 |
{
|
| 59 |
+
"epoch": 0.31,
|
| 60 |
+
"learning_rate": 4.655172413793104e-05,
|
| 61 |
+
"loss": 0.9929,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 62 |
"step": 90
|
| 63 |
},
|
| 64 |
{
|
| 65 |
+
"epoch": 0.34,
|
| 66 |
+
"learning_rate": 5.172413793103448e-05,
|
| 67 |
+
"loss": 0.9421,
|
| 68 |
"step": 100
|
| 69 |
},
|
| 70 |
{
|
| 71 |
+
"epoch": 0.38,
|
| 72 |
+
"learning_rate": 5.689655172413793e-05,
|
| 73 |
+
"loss": 1.0011,
|
| 74 |
"step": 110
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 0.41,
|
| 78 |
+
"learning_rate": 6.206896551724138e-05,
|
| 79 |
+
"loss": 0.9637,
|
| 80 |
"step": 120
|
| 81 |
},
|
| 82 |
{
|
| 83 |
+
"epoch": 0.45,
|
| 84 |
+
"learning_rate": 6.724137931034483e-05,
|
| 85 |
+
"loss": 0.9283,
|
| 86 |
"step": 130
|
| 87 |
},
|
| 88 |
{
|
| 89 |
+
"epoch": 0.48,
|
| 90 |
+
"learning_rate": 7.241379310344827e-05,
|
| 91 |
+
"loss": 0.9275,
|
| 92 |
"step": 140
|
| 93 |
},
|
| 94 |
{
|
| 95 |
+
"epoch": 0.52,
|
| 96 |
+
"learning_rate": 7.758620689655173e-05,
|
| 97 |
+
"loss": 0.9041,
|
| 98 |
"step": 150
|
| 99 |
},
|
| 100 |
{
|
| 101 |
+
"epoch": 0.55,
|
| 102 |
+
"learning_rate": 8.275862068965519e-05,
|
| 103 |
+
"loss": 0.8695,
|
| 104 |
"step": 160
|
| 105 |
},
|
| 106 |
{
|
| 107 |
+
"epoch": 0.58,
|
| 108 |
+
"learning_rate": 8.793103448275862e-05,
|
| 109 |
+
"loss": 0.9382,
|
| 110 |
"step": 170
|
| 111 |
},
|
| 112 |
{
|
| 113 |
+
"epoch": 0.62,
|
| 114 |
+
"learning_rate": 9.310344827586207e-05,
|
| 115 |
+
"loss": 0.8939,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 116 |
"step": 180
|
| 117 |
},
|
| 118 |
{
|
| 119 |
+
"epoch": 0.65,
|
| 120 |
+
"learning_rate": 9.827586206896552e-05,
|
| 121 |
+
"loss": 0.8945,
|
| 122 |
"step": 190
|
| 123 |
},
|
| 124 |
{
|
| 125 |
+
"epoch": 0.69,
|
| 126 |
+
"learning_rate": 0.00010344827586206896,
|
| 127 |
+
"loss": 0.8488,
|
| 128 |
"step": 200
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 0.72,
|
| 132 |
+
"learning_rate": 0.00010862068965517241,
|
| 133 |
+
"loss": 0.9054,
|
| 134 |
"step": 210
|
| 135 |
},
|
| 136 |
{
|
| 137 |
+
"epoch": 0.76,
|
| 138 |
+
"learning_rate": 0.00011379310344827586,
|
| 139 |
+
"loss": 0.8823,
|
| 140 |
"step": 220
|
| 141 |
},
|
| 142 |
{
|
| 143 |
+
"epoch": 0.79,
|
| 144 |
+
"learning_rate": 0.00011896551724137932,
|
| 145 |
+
"loss": 0.8898,
|
| 146 |
"step": 230
|
| 147 |
},
|
| 148 |
{
|
| 149 |
+
"epoch": 0.83,
|
| 150 |
+
"learning_rate": 0.00011954022988505747,
|
| 151 |
+
"loss": 0.8953,
|
| 152 |
"step": 240
|
| 153 |
},
|
| 154 |
{
|
| 155 |
+
"epoch": 0.86,
|
| 156 |
+
"learning_rate": 0.00011896551724137932,
|
| 157 |
+
"loss": 0.8591,
|
| 158 |
"step": 250
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 0.89,
|
| 162 |
+
"learning_rate": 0.00011839080459770115,
|
| 163 |
+
"loss": 0.8697,
|
| 164 |
"step": 260
|
| 165 |
},
|
| 166 |
{
|
| 167 |
+
"epoch": 0.93,
|
| 168 |
+
"learning_rate": 0.00011781609195402298,
|
| 169 |
+
"loss": 0.8245,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 170 |
"step": 270
|
| 171 |
},
|
| 172 |
{
|
| 173 |
+
"epoch": 0.96,
|
| 174 |
+
"learning_rate": 0.00011724137931034484,
|
| 175 |
+
"loss": 0.8431,
|
| 176 |
"step": 280
|
| 177 |
},
|
| 178 |
{
|
| 179 |
+
"epoch": 1.0,
|
| 180 |
+
"learning_rate": 0.00011666666666666667,
|
| 181 |
+
"loss": 0.8514,
|
| 182 |
+
"step": 290
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.0,
|
| 186 |
+
"eval_accuracy": 0.7047972464235776,
|
| 187 |
+
"eval_f1": 0.6909131065764619,
|
| 188 |
+
"eval_loss": 0.8464350700378418,
|
| 189 |
+
"eval_precision": 0.7034765556664452,
|
| 190 |
+
"eval_recall": 0.7047972464235776,
|
| 191 |
+
"eval_runtime": 103.9481,
|
| 192 |
+
"eval_samples_per_second": 178.878,
|
| 193 |
+
"eval_steps_per_second": 2.799,
|
| 194 |
"step": 290
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 1.03,
|
| 198 |
+
"learning_rate": 0.00011609195402298851,
|
| 199 |
+
"loss": 0.8106,
|
| 200 |
"step": 300
|
| 201 |
},
|
| 202 |
{
|
| 203 |
+
"epoch": 1.07,
|
| 204 |
+
"learning_rate": 0.00011551724137931035,
|
| 205 |
+
"loss": 0.766,
|
| 206 |
"step": 310
|
| 207 |
},
|
| 208 |
{
|
| 209 |
+
"epoch": 1.1,
|
| 210 |
+
"learning_rate": 0.00011494252873563218,
|
| 211 |
+
"loss": 0.7461,
|
| 212 |
"step": 320
|
| 213 |
},
|
| 214 |
{
|
| 215 |
+
"epoch": 1.14,
|
| 216 |
+
"learning_rate": 0.00011436781609195404,
|
| 217 |
+
"loss": 0.7814,
|
| 218 |
"step": 330
|
| 219 |
},
|
| 220 |
{
|
| 221 |
+
"epoch": 1.17,
|
| 222 |
+
"learning_rate": 0.00011379310344827586,
|
| 223 |
+
"loss": 0.741,
|
| 224 |
"step": 340
|
| 225 |
},
|
| 226 |
{
|
| 227 |
+
"epoch": 1.21,
|
| 228 |
+
"learning_rate": 0.0001132183908045977,
|
| 229 |
+
"loss": 0.7118,
|
| 230 |
"step": 350
|
| 231 |
},
|
| 232 |
{
|
| 233 |
+
"epoch": 1.24,
|
| 234 |
+
"learning_rate": 0.00011264367816091955,
|
| 235 |
+
"loss": 0.7539,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 236 |
"step": 360
|
| 237 |
},
|
| 238 |
{
|
| 239 |
+
"epoch": 1.28,
|
| 240 |
+
"learning_rate": 0.00011206896551724138,
|
| 241 |
+
"loss": 0.7633,
|
| 242 |
"step": 370
|
| 243 |
},
|
| 244 |
{
|
| 245 |
+
"epoch": 1.31,
|
| 246 |
+
"learning_rate": 0.00011149425287356322,
|
| 247 |
+
"loss": 0.7305,
|
| 248 |
"step": 380
|
| 249 |
},
|
| 250 |
{
|
| 251 |
+
"epoch": 1.34,
|
| 252 |
+
"learning_rate": 0.00011091954022988506,
|
| 253 |
+
"loss": 0.7455,
|
| 254 |
"step": 390
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 1.38,
|
| 258 |
+
"learning_rate": 0.00011034482758620689,
|
| 259 |
+
"loss": 0.724,
|
| 260 |
"step": 400
|
| 261 |
},
|
| 262 |
{
|
| 263 |
+
"epoch": 1.41,
|
| 264 |
+
"learning_rate": 0.00010977011494252874,
|
| 265 |
+
"loss": 0.7825,
|
| 266 |
"step": 410
|
| 267 |
},
|
| 268 |
{
|
| 269 |
+
"epoch": 1.45,
|
| 270 |
+
"learning_rate": 0.00010919540229885058,
|
| 271 |
+
"loss": 0.7558,
|
| 272 |
"step": 420
|
| 273 |
},
|
| 274 |
{
|
| 275 |
+
"epoch": 1.48,
|
| 276 |
+
"learning_rate": 0.00010862068965517241,
|
| 277 |
+
"loss": 0.744,
|
| 278 |
"step": 430
|
| 279 |
},
|
| 280 |
{
|
| 281 |
+
"epoch": 1.52,
|
| 282 |
+
"learning_rate": 0.00010804597701149426,
|
| 283 |
+
"loss": 0.7328,
|
| 284 |
"step": 440
|
| 285 |
},
|
| 286 |
{
|
| 287 |
+
"epoch": 1.55,
|
| 288 |
+
"learning_rate": 0.00010747126436781609,
|
| 289 |
+
"loss": 0.7385,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 290 |
"step": 450
|
| 291 |
},
|
| 292 |
{
|
| 293 |
+
"epoch": 1.58,
|
| 294 |
+
"learning_rate": 0.00010689655172413793,
|
| 295 |
+
"loss": 0.7443,
|
| 296 |
"step": 460
|
| 297 |
},
|
| 298 |
{
|
| 299 |
+
"epoch": 1.62,
|
| 300 |
+
"learning_rate": 0.00010632183908045978,
|
| 301 |
+
"loss": 0.7126,
|
| 302 |
"step": 470
|
| 303 |
},
|
| 304 |
{
|
| 305 |
+
"epoch": 1.65,
|
| 306 |
+
"learning_rate": 0.0001057471264367816,
|
| 307 |
+
"loss": 0.8166,
|
| 308 |
"step": 480
|
| 309 |
},
|
| 310 |
{
|
| 311 |
+
"epoch": 1.69,
|
| 312 |
+
"learning_rate": 0.00010517241379310345,
|
| 313 |
+
"loss": 0.7516,
|
| 314 |
"step": 490
|
| 315 |
},
|
| 316 |
{
|
| 317 |
+
"epoch": 1.72,
|
| 318 |
+
"learning_rate": 0.00010459770114942529,
|
| 319 |
+
"loss": 0.7389,
|
| 320 |
"step": 500
|
| 321 |
},
|
| 322 |
{
|
| 323 |
+
"epoch": 1.76,
|
| 324 |
+
"learning_rate": 0.00010402298850574713,
|
| 325 |
+
"loss": 0.766,
|
| 326 |
"step": 510
|
| 327 |
},
|
| 328 |
{
|
| 329 |
+
"epoch": 1.79,
|
| 330 |
+
"learning_rate": 0.00010344827586206896,
|
| 331 |
+
"loss": 0.7343,
|
| 332 |
"step": 520
|
| 333 |
},
|
| 334 |
{
|
| 335 |
+
"epoch": 1.83,
|
| 336 |
+
"learning_rate": 0.0001028735632183908,
|
| 337 |
+
"loss": 0.7132,
|
| 338 |
"step": 530
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 1.86,
|
| 342 |
+
"learning_rate": 0.00010229885057471265,
|
| 343 |
+
"loss": 0.7494,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 344 |
"step": 540
|
| 345 |
},
|
| 346 |
{
|
| 347 |
+
"epoch": 1.89,
|
| 348 |
+
"learning_rate": 0.00010172413793103449,
|
| 349 |
+
"loss": 0.7426,
|
| 350 |
"step": 550
|
| 351 |
},
|
| 352 |
{
|
| 353 |
+
"epoch": 1.93,
|
| 354 |
+
"learning_rate": 0.00010114942528735632,
|
| 355 |
+
"loss": 0.7305,
|
| 356 |
"step": 560
|
| 357 |
},
|
| 358 |
{
|
| 359 |
+
"epoch": 1.96,
|
| 360 |
+
"learning_rate": 0.00010057471264367816,
|
| 361 |
+
"loss": 0.7489,
|
| 362 |
"step": 570
|
| 363 |
},
|
| 364 |
{
|
| 365 |
+
"epoch": 2.0,
|
| 366 |
+
"learning_rate": 0.0001,
|
| 367 |
+
"loss": 0.7202,
|
| 368 |
"step": 580
|
| 369 |
},
|
| 370 |
{
|
| 371 |
+
"epoch": 2.0,
|
| 372 |
+
"eval_accuracy": 0.7282994514359471,
|
| 373 |
+
"eval_f1": 0.7111205455958668,
|
| 374 |
+
"eval_loss": 0.7790911793708801,
|
| 375 |
+
"eval_precision": 0.7297231799712814,
|
| 376 |
+
"eval_recall": 0.7282994514359471,
|
| 377 |
+
"eval_runtime": 104.6701,
|
| 378 |
+
"eval_samples_per_second": 177.644,
|
| 379 |
+
"eval_steps_per_second": 2.78,
|
| 380 |
+
"step": 580
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.03,
|
| 384 |
+
"learning_rate": 9.942528735632185e-05,
|
| 385 |
+
"loss": 0.6398,
|
| 386 |
"step": 590
|
| 387 |
},
|
| 388 |
{
|
| 389 |
+
"epoch": 2.07,
|
| 390 |
+
"learning_rate": 9.885057471264368e-05,
|
| 391 |
+
"loss": 0.5381,
|
| 392 |
"step": 600
|
| 393 |
},
|
| 394 |
{
|
| 395 |
+
"epoch": 2.1,
|
| 396 |
+
"learning_rate": 9.827586206896552e-05,
|
| 397 |
+
"loss": 0.5646,
|
| 398 |
"step": 610
|
| 399 |
},
|
| 400 |
{
|
| 401 |
+
"epoch": 2.14,
|
| 402 |
+
"learning_rate": 9.770114942528736e-05,
|
| 403 |
+
"loss": 0.5356,
|
| 404 |
"step": 620
|
| 405 |
},
|
| 406 |
{
|
| 407 |
+
"epoch": 2.17,
|
| 408 |
+
"learning_rate": 9.712643678160919e-05,
|
| 409 |
+
"loss": 0.5734,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 410 |
"step": 630
|
| 411 |
},
|
| 412 |
{
|
| 413 |
+
"epoch": 2.21,
|
| 414 |
+
"learning_rate": 9.655172413793103e-05,
|
| 415 |
+
"loss": 0.5097,
|
| 416 |
"step": 640
|
| 417 |
},
|
| 418 |
{
|
| 419 |
+
"epoch": 2.24,
|
| 420 |
+
"learning_rate": 9.597701149425288e-05,
|
| 421 |
+
"loss": 0.5701,
|
| 422 |
"step": 650
|
| 423 |
},
|
| 424 |
{
|
| 425 |
+
"epoch": 2.28,
|
| 426 |
+
"learning_rate": 9.540229885057472e-05,
|
| 427 |
+
"loss": 0.5802,
|
| 428 |
"step": 660
|
| 429 |
},
|
| 430 |
{
|
| 431 |
+
"epoch": 2.31,
|
| 432 |
+
"learning_rate": 9.482758620689656e-05,
|
| 433 |
+
"loss": 0.5794,
|
| 434 |
"step": 670
|
| 435 |
},
|
| 436 |
{
|
| 437 |
+
"epoch": 2.34,
|
| 438 |
+
"learning_rate": 9.425287356321839e-05,
|
| 439 |
+
"loss": 0.5845,
|
| 440 |
"step": 680
|
| 441 |
},
|
| 442 |
{
|
| 443 |
+
"epoch": 2.38,
|
| 444 |
+
"learning_rate": 9.367816091954023e-05,
|
| 445 |
+
"loss": 0.5591,
|
| 446 |
"step": 690
|
| 447 |
},
|
| 448 |
{
|
| 449 |
+
"epoch": 2.41,
|
| 450 |
+
"learning_rate": 9.310344827586207e-05,
|
| 451 |
+
"loss": 0.5567,
|
| 452 |
"step": 700
|
| 453 |
},
|
| 454 |
{
|
| 455 |
+
"epoch": 2.45,
|
| 456 |
+
"learning_rate": 9.25287356321839e-05,
|
| 457 |
+
"loss": 0.5524,
|
| 458 |
"step": 710
|
| 459 |
},
|
| 460 |
{
|
| 461 |
+
"epoch": 2.48,
|
| 462 |
+
"learning_rate": 9.195402298850575e-05,
|
| 463 |
+
"loss": 0.5569,
|
| 464 |
+
"step": 720
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.52,
|
| 468 |
+
"learning_rate": 9.137931034482759e-05,
|
| 469 |
+
"loss": 0.5582,
|
| 470 |
+
"step": 730
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.55,
|
| 474 |
+
"learning_rate": 9.080459770114942e-05,
|
| 475 |
+
"loss": 0.5819,
|
| 476 |
+
"step": 740
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.58,
|
| 480 |
+
"learning_rate": 9.022988505747127e-05,
|
| 481 |
+
"loss": 0.5735,
|
| 482 |
+
"step": 750
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.62,
|
| 486 |
+
"learning_rate": 8.96551724137931e-05,
|
| 487 |
+
"loss": 0.5534,
|
| 488 |
+
"step": 760
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.65,
|
| 492 |
+
"learning_rate": 8.908045977011495e-05,
|
| 493 |
+
"loss": 0.5936,
|
| 494 |
+
"step": 770
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.69,
|
| 498 |
+
"learning_rate": 8.850574712643679e-05,
|
| 499 |
+
"loss": 0.5248,
|
| 500 |
+
"step": 780
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 2.72,
|
| 504 |
+
"learning_rate": 8.793103448275862e-05,
|
| 505 |
+
"loss": 0.5574,
|
| 506 |
+
"step": 790
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 2.76,
|
| 510 |
+
"learning_rate": 8.735632183908046e-05,
|
| 511 |
+
"loss": 0.5148,
|
| 512 |
+
"step": 800
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 2.79,
|
| 516 |
+
"learning_rate": 8.67816091954023e-05,
|
| 517 |
+
"loss": 0.5914,
|
| 518 |
+
"step": 810
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 2.83,
|
| 522 |
+
"learning_rate": 8.620689655172413e-05,
|
| 523 |
+
"loss": 0.6162,
|
| 524 |
+
"step": 820
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 2.86,
|
| 528 |
+
"learning_rate": 8.563218390804599e-05,
|
| 529 |
+
"loss": 0.5406,
|
| 530 |
+
"step": 830
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 2.89,
|
| 534 |
+
"learning_rate": 8.505747126436782e-05,
|
| 535 |
+
"loss": 0.5718,
|
| 536 |
+
"step": 840
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 2.93,
|
| 540 |
+
"learning_rate": 8.448275862068965e-05,
|
| 541 |
+
"loss": 0.5683,
|
| 542 |
+
"step": 850
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 2.96,
|
| 546 |
+
"learning_rate": 8.39080459770115e-05,
|
| 547 |
+
"loss": 0.5374,
|
| 548 |
+
"step": 860
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 3.0,
|
| 552 |
+
"learning_rate": 8.333333333333333e-05,
|
| 553 |
+
"loss": 0.5455,
|
| 554 |
+
"step": 870
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 3.0,
|
| 558 |
+
"eval_accuracy": 0.728460793804453,
|
| 559 |
+
"eval_f1": 0.717108375092518,
|
| 560 |
+
"eval_loss": 0.7950025200843811,
|
| 561 |
+
"eval_precision": 0.7174203394442972,
|
| 562 |
+
"eval_recall": 0.728460793804453,
|
| 563 |
+
"eval_runtime": 109.0183,
|
| 564 |
+
"eval_samples_per_second": 170.559,
|
| 565 |
+
"eval_steps_per_second": 2.669,
|
| 566 |
+
"step": 870
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 3.03,
|
| 570 |
+
"learning_rate": 8.275862068965519e-05,
|
| 571 |
+
"loss": 0.4109,
|
| 572 |
+
"step": 880
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 3.07,
|
| 576 |
+
"learning_rate": 8.218390804597702e-05,
|
| 577 |
+
"loss": 0.3615,
|
| 578 |
+
"step": 890
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 3.1,
|
| 582 |
+
"learning_rate": 8.160919540229884e-05,
|
| 583 |
+
"loss": 0.3445,
|
| 584 |
+
"step": 900
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 3.14,
|
| 588 |
+
"learning_rate": 8.10344827586207e-05,
|
| 589 |
+
"loss": 0.3407,
|
| 590 |
+
"step": 910
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.17,
|
| 594 |
+
"learning_rate": 8.045977011494253e-05,
|
| 595 |
+
"loss": 0.3228,
|
| 596 |
+
"step": 920
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 3.21,
|
| 600 |
+
"learning_rate": 7.988505747126436e-05,
|
| 601 |
+
"loss": 0.3376,
|
| 602 |
+
"step": 930
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 3.24,
|
| 606 |
+
"learning_rate": 7.931034482758621e-05,
|
| 607 |
+
"loss": 0.3616,
|
| 608 |
+
"step": 940
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 3.28,
|
| 612 |
+
"learning_rate": 7.873563218390804e-05,
|
| 613 |
+
"loss": 0.3393,
|
| 614 |
+
"step": 950
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 3.31,
|
| 618 |
+
"learning_rate": 7.816091954022989e-05,
|
| 619 |
+
"loss": 0.324,
|
| 620 |
+
"step": 960
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 3.34,
|
| 624 |
+
"learning_rate": 7.758620689655173e-05,
|
| 625 |
+
"loss": 0.3391,
|
| 626 |
+
"step": 970
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 3.38,
|
| 630 |
+
"learning_rate": 7.701149425287356e-05,
|
| 631 |
+
"loss": 0.3589,
|
| 632 |
+
"step": 980
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.41,
|
| 636 |
+
"learning_rate": 7.643678160919541e-05,
|
| 637 |
+
"loss": 0.328,
|
| 638 |
+
"step": 990
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.45,
|
| 642 |
+
"learning_rate": 7.586206896551724e-05,
|
| 643 |
+
"loss": 0.3256,
|
| 644 |
+
"step": 1000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 3.48,
|
| 648 |
+
"learning_rate": 7.528735632183907e-05,
|
| 649 |
+
"loss": 0.338,
|
| 650 |
+
"step": 1010
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 3.52,
|
| 654 |
+
"learning_rate": 7.471264367816093e-05,
|
| 655 |
+
"loss": 0.3353,
|
| 656 |
+
"step": 1020
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 3.55,
|
| 660 |
+
"learning_rate": 7.413793103448276e-05,
|
| 661 |
+
"loss": 0.3147,
|
| 662 |
+
"step": 1030
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 3.58,
|
| 666 |
+
"learning_rate": 7.35632183908046e-05,
|
| 667 |
+
"loss": 0.3405,
|
| 668 |
+
"step": 1040
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 3.62,
|
| 672 |
+
"learning_rate": 7.298850574712644e-05,
|
| 673 |
+
"loss": 0.3409,
|
| 674 |
+
"step": 1050
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 3.65,
|
| 678 |
+
"learning_rate": 7.241379310344827e-05,
|
| 679 |
+
"loss": 0.3611,
|
| 680 |
+
"step": 1060
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 3.69,
|
| 684 |
+
"learning_rate": 7.183908045977013e-05,
|
| 685 |
+
"loss": 0.3501,
|
| 686 |
+
"step": 1070
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 3.72,
|
| 690 |
+
"learning_rate": 7.126436781609196e-05,
|
| 691 |
+
"loss": 0.3594,
|
| 692 |
+
"step": 1080
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 3.76,
|
| 696 |
+
"learning_rate": 7.068965517241379e-05,
|
| 697 |
+
"loss": 0.3507,
|
| 698 |
+
"step": 1090
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 3.79,
|
| 702 |
+
"learning_rate": 7.011494252873564e-05,
|
| 703 |
+
"loss": 0.3376,
|
| 704 |
+
"step": 1100
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 3.83,
|
| 708 |
+
"learning_rate": 6.954022988505747e-05,
|
| 709 |
+
"loss": 0.3355,
|
| 710 |
+
"step": 1110
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 3.86,
|
| 714 |
+
"learning_rate": 6.896551724137931e-05,
|
| 715 |
+
"loss": 0.3356,
|
| 716 |
+
"step": 1120
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 3.89,
|
| 720 |
+
"learning_rate": 6.839080459770116e-05,
|
| 721 |
+
"loss": 0.3489,
|
| 722 |
+
"step": 1130
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 3.93,
|
| 726 |
+
"learning_rate": 6.781609195402298e-05,
|
| 727 |
+
"loss": 0.3348,
|
| 728 |
+
"step": 1140
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 3.96,
|
| 732 |
+
"learning_rate": 6.724137931034483e-05,
|
| 733 |
+
"loss": 0.3436,
|
| 734 |
+
"step": 1150
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 4.0,
|
| 738 |
+
"learning_rate": 6.666666666666667e-05,
|
| 739 |
+
"loss": 0.334,
|
| 740 |
+
"step": 1160
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 4.0,
|
| 744 |
+
"eval_accuracy": 0.7154996235344735,
|
| 745 |
+
"eval_f1": 0.7144950276664553,
|
| 746 |
+
"eval_loss": 0.894753098487854,
|
| 747 |
+
"eval_precision": 0.715167448955365,
|
| 748 |
+
"eval_recall": 0.7154996235344735,
|
| 749 |
+
"eval_runtime": 102.1557,
|
| 750 |
+
"eval_samples_per_second": 182.016,
|
| 751 |
+
"eval_steps_per_second": 2.849,
|
| 752 |
+
"step": 1160
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.03,
|
| 756 |
+
"learning_rate": 6.60919540229885e-05,
|
| 757 |
+
"loss": 0.2132,
|
| 758 |
+
"step": 1170
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.07,
|
| 762 |
+
"learning_rate": 6.551724137931035e-05,
|
| 763 |
+
"loss": 0.1556,
|
| 764 |
+
"step": 1180
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 4.1,
|
| 768 |
+
"learning_rate": 6.494252873563218e-05,
|
| 769 |
+
"loss": 0.1607,
|
| 770 |
+
"step": 1190
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 4.14,
|
| 774 |
+
"learning_rate": 6.436781609195403e-05,
|
| 775 |
+
"loss": 0.1572,
|
| 776 |
+
"step": 1200
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 4.17,
|
| 780 |
+
"learning_rate": 6.379310344827587e-05,
|
| 781 |
+
"loss": 0.1554,
|
| 782 |
+
"step": 1210
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 4.21,
|
| 786 |
+
"learning_rate": 6.32183908045977e-05,
|
| 787 |
+
"loss": 0.156,
|
| 788 |
+
"step": 1220
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 4.24,
|
| 792 |
+
"learning_rate": 6.264367816091954e-05,
|
| 793 |
+
"loss": 0.1339,
|
| 794 |
+
"step": 1230
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.28,
|
| 798 |
+
"learning_rate": 6.206896551724138e-05,
|
| 799 |
+
"loss": 0.1565,
|
| 800 |
+
"step": 1240
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.31,
|
| 804 |
+
"learning_rate": 6.149425287356323e-05,
|
| 805 |
+
"loss": 0.1582,
|
| 806 |
+
"step": 1250
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.34,
|
| 810 |
+
"learning_rate": 6.0919540229885055e-05,
|
| 811 |
+
"loss": 0.1466,
|
| 812 |
+
"step": 1260
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 4.38,
|
| 816 |
+
"learning_rate": 6.0344827586206904e-05,
|
| 817 |
+
"loss": 0.1557,
|
| 818 |
+
"step": 1270
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 4.41,
|
| 822 |
+
"learning_rate": 5.977011494252873e-05,
|
| 823 |
+
"loss": 0.1522,
|
| 824 |
+
"step": 1280
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 4.45,
|
| 828 |
+
"learning_rate": 5.9195402298850576e-05,
|
| 829 |
+
"loss": 0.1373,
|
| 830 |
+
"step": 1290
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 4.48,
|
| 834 |
+
"learning_rate": 5.862068965517242e-05,
|
| 835 |
+
"loss": 0.1515,
|
| 836 |
+
"step": 1300
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.52,
|
| 840 |
+
"learning_rate": 5.8045977011494254e-05,
|
| 841 |
+
"loss": 0.1473,
|
| 842 |
+
"step": 1310
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 4.55,
|
| 846 |
+
"learning_rate": 5.747126436781609e-05,
|
| 847 |
+
"loss": 0.1513,
|
| 848 |
+
"step": 1320
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 4.58,
|
| 852 |
+
"learning_rate": 5.689655172413793e-05,
|
| 853 |
+
"loss": 0.1553,
|
| 854 |
+
"step": 1330
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 4.62,
|
| 858 |
+
"learning_rate": 5.6321839080459775e-05,
|
| 859 |
+
"loss": 0.158,
|
| 860 |
+
"step": 1340
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 4.65,
|
| 864 |
+
"learning_rate": 5.574712643678161e-05,
|
| 865 |
+
"loss": 0.1603,
|
| 866 |
+
"step": 1350
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 4.69,
|
| 870 |
+
"learning_rate": 5.5172413793103446e-05,
|
| 871 |
+
"loss": 0.1488,
|
| 872 |
+
"step": 1360
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 4.72,
|
| 876 |
+
"learning_rate": 5.459770114942529e-05,
|
| 877 |
+
"loss": 0.1495,
|
| 878 |
+
"step": 1370
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 4.76,
|
| 882 |
+
"learning_rate": 5.402298850574713e-05,
|
| 883 |
+
"loss": 0.1598,
|
| 884 |
+
"step": 1380
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 4.79,
|
| 888 |
+
"learning_rate": 5.344827586206897e-05,
|
| 889 |
+
"loss": 0.1468,
|
| 890 |
+
"step": 1390
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 4.83,
|
| 894 |
+
"learning_rate": 5.28735632183908e-05,
|
| 895 |
+
"loss": 0.1535,
|
| 896 |
+
"step": 1400
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 4.86,
|
| 900 |
+
"learning_rate": 5.2298850574712646e-05,
|
| 901 |
+
"loss": 0.1491,
|
| 902 |
+
"step": 1410
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 4.89,
|
| 906 |
+
"learning_rate": 5.172413793103448e-05,
|
| 907 |
+
"loss": 0.1436,
|
| 908 |
+
"step": 1420
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 4.93,
|
| 912 |
+
"learning_rate": 5.1149425287356324e-05,
|
| 913 |
+
"loss": 0.138,
|
| 914 |
+
"step": 1430
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 4.96,
|
| 918 |
+
"learning_rate": 5.057471264367816e-05,
|
| 919 |
+
"loss": 0.1439,
|
| 920 |
+
"step": 1440
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 5.0,
|
| 924 |
+
"learning_rate": 5e-05,
|
| 925 |
+
"loss": 0.1644,
|
| 926 |
+
"step": 1450
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 5.0,
|
| 930 |
+
"eval_accuracy": 0.7238894266967839,
|
| 931 |
+
"eval_f1": 0.7194230600781256,
|
| 932 |
+
"eval_loss": 1.0820467472076416,
|
| 933 |
+
"eval_precision": 0.7188913807380165,
|
| 934 |
+
"eval_recall": 0.7238894266967839,
|
| 935 |
+
"eval_runtime": 105.8352,
|
| 936 |
+
"eval_samples_per_second": 175.688,
|
| 937 |
+
"eval_steps_per_second": 2.75,
|
| 938 |
+
"step": 1450
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 5.03,
|
| 942 |
+
"learning_rate": 4.942528735632184e-05,
|
| 943 |
+
"loss": 0.0688,
|
| 944 |
+
"step": 1460
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 5.07,
|
| 948 |
+
"learning_rate": 4.885057471264368e-05,
|
| 949 |
+
"loss": 0.0619,
|
| 950 |
+
"step": 1470
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 5.1,
|
| 954 |
+
"learning_rate": 4.8275862068965517e-05,
|
| 955 |
+
"loss": 0.0502,
|
| 956 |
+
"step": 1480
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 5.14,
|
| 960 |
+
"learning_rate": 4.770114942528736e-05,
|
| 961 |
+
"loss": 0.0503,
|
| 962 |
+
"step": 1490
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 5.17,
|
| 966 |
+
"learning_rate": 4.7126436781609195e-05,
|
| 967 |
+
"loss": 0.0546,
|
| 968 |
+
"step": 1500
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 5.21,
|
| 972 |
+
"learning_rate": 4.655172413793104e-05,
|
| 973 |
+
"loss": 0.0529,
|
| 974 |
+
"step": 1510
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 5.24,
|
| 978 |
+
"learning_rate": 4.597701149425287e-05,
|
| 979 |
+
"loss": 0.0529,
|
| 980 |
+
"step": 1520
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 5.28,
|
| 984 |
+
"learning_rate": 4.540229885057471e-05,
|
| 985 |
+
"loss": 0.0499,
|
| 986 |
+
"step": 1530
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 5.31,
|
| 990 |
+
"learning_rate": 4.482758620689655e-05,
|
| 991 |
+
"loss": 0.0525,
|
| 992 |
+
"step": 1540
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 5.34,
|
| 996 |
+
"learning_rate": 4.4252873563218394e-05,
|
| 997 |
+
"loss": 0.0582,
|
| 998 |
+
"step": 1550
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 5.38,
|
| 1002 |
+
"learning_rate": 4.367816091954023e-05,
|
| 1003 |
+
"loss": 0.0598,
|
| 1004 |
+
"step": 1560
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 5.41,
|
| 1008 |
+
"learning_rate": 4.3103448275862066e-05,
|
| 1009 |
+
"loss": 0.0587,
|
| 1010 |
+
"step": 1570
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 5.45,
|
| 1014 |
+
"learning_rate": 4.252873563218391e-05,
|
| 1015 |
+
"loss": 0.0507,
|
| 1016 |
+
"step": 1580
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 5.48,
|
| 1020 |
+
"learning_rate": 4.195402298850575e-05,
|
| 1021 |
+
"loss": 0.0565,
|
| 1022 |
+
"step": 1590
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 5.52,
|
| 1026 |
+
"learning_rate": 4.137931034482759e-05,
|
| 1027 |
+
"loss": 0.0609,
|
| 1028 |
+
"step": 1600
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 5.55,
|
| 1032 |
+
"learning_rate": 4.080459770114942e-05,
|
| 1033 |
+
"loss": 0.0459,
|
| 1034 |
+
"step": 1610
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 5.58,
|
| 1038 |
+
"learning_rate": 4.0229885057471265e-05,
|
| 1039 |
+
"loss": 0.0647,
|
| 1040 |
+
"step": 1620
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 5.62,
|
| 1044 |
+
"learning_rate": 3.965517241379311e-05,
|
| 1045 |
+
"loss": 0.0604,
|
| 1046 |
+
"step": 1630
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 5.65,
|
| 1050 |
+
"learning_rate": 3.908045977011494e-05,
|
| 1051 |
+
"loss": 0.0648,
|
| 1052 |
+
"step": 1640
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 5.69,
|
| 1056 |
+
"learning_rate": 3.850574712643678e-05,
|
| 1057 |
+
"loss": 0.0581,
|
| 1058 |
+
"step": 1650
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 5.72,
|
| 1062 |
+
"learning_rate": 3.793103448275862e-05,
|
| 1063 |
+
"loss": 0.0517,
|
| 1064 |
+
"step": 1660
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 5.76,
|
| 1068 |
+
"learning_rate": 3.7356321839080464e-05,
|
| 1069 |
+
"loss": 0.0558,
|
| 1070 |
+
"step": 1670
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 5.79,
|
| 1074 |
+
"learning_rate": 3.67816091954023e-05,
|
| 1075 |
+
"loss": 0.0464,
|
| 1076 |
+
"step": 1680
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 5.83,
|
| 1080 |
+
"learning_rate": 3.6206896551724136e-05,
|
| 1081 |
+
"loss": 0.054,
|
| 1082 |
+
"step": 1690
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 5.86,
|
| 1086 |
+
"learning_rate": 3.563218390804598e-05,
|
| 1087 |
+
"loss": 0.0609,
|
| 1088 |
+
"step": 1700
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 5.89,
|
| 1092 |
+
"learning_rate": 3.505747126436782e-05,
|
| 1093 |
+
"loss": 0.0507,
|
| 1094 |
+
"step": 1710
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 5.93,
|
| 1098 |
+
"learning_rate": 3.4482758620689657e-05,
|
| 1099 |
+
"loss": 0.0521,
|
| 1100 |
+
"step": 1720
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 5.96,
|
| 1104 |
+
"learning_rate": 3.390804597701149e-05,
|
| 1105 |
+
"loss": 0.0572,
|
| 1106 |
+
"step": 1730
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 6.0,
|
| 1110 |
+
"learning_rate": 3.3333333333333335e-05,
|
| 1111 |
+
"loss": 0.0482,
|
| 1112 |
+
"step": 1740
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 6.0,
|
| 1116 |
+
"eval_accuracy": 0.7204474561686566,
|
| 1117 |
+
"eval_f1": 0.71604121882508,
|
| 1118 |
+
"eval_loss": 1.2792034149169922,
|
| 1119 |
+
"eval_precision": 0.7144269552366298,
|
| 1120 |
+
"eval_recall": 0.7204474561686566,
|
| 1121 |
+
"eval_runtime": 99.5538,
|
| 1122 |
+
"eval_samples_per_second": 186.773,
|
| 1123 |
+
"eval_steps_per_second": 2.923,
|
| 1124 |
+
"step": 1740
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 6.03,
|
| 1128 |
+
"learning_rate": 3.275862068965518e-05,
|
| 1129 |
+
"loss": 0.0261,
|
| 1130 |
+
"step": 1750
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 6.07,
|
| 1134 |
+
"learning_rate": 3.218390804597701e-05,
|
| 1135 |
+
"loss": 0.0242,
|
| 1136 |
+
"step": 1760
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 6.1,
|
| 1140 |
+
"learning_rate": 3.160919540229885e-05,
|
| 1141 |
+
"loss": 0.0261,
|
| 1142 |
+
"step": 1770
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 6.14,
|
| 1146 |
+
"learning_rate": 3.103448275862069e-05,
|
| 1147 |
+
"loss": 0.0174,
|
| 1148 |
+
"step": 1780
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 6.17,
|
| 1152 |
+
"learning_rate": 3.0459770114942527e-05,
|
| 1153 |
+
"loss": 0.0174,
|
| 1154 |
+
"step": 1790
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 6.21,
|
| 1158 |
+
"learning_rate": 2.9885057471264367e-05,
|
| 1159 |
+
"loss": 0.0261,
|
| 1160 |
+
"step": 1800
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 6.24,
|
| 1164 |
+
"learning_rate": 2.931034482758621e-05,
|
| 1165 |
+
"loss": 0.0233,
|
| 1166 |
+
"step": 1810
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 6.28,
|
| 1170 |
+
"learning_rate": 2.8735632183908045e-05,
|
| 1171 |
+
"loss": 0.022,
|
| 1172 |
+
"step": 1820
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 6.31,
|
| 1176 |
+
"learning_rate": 2.8160919540229887e-05,
|
| 1177 |
+
"loss": 0.0264,
|
| 1178 |
+
"step": 1830
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.34,
|
| 1182 |
+
"learning_rate": 2.7586206896551723e-05,
|
| 1183 |
+
"loss": 0.0222,
|
| 1184 |
+
"step": 1840
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 6.38,
|
| 1188 |
+
"learning_rate": 2.7011494252873566e-05,
|
| 1189 |
+
"loss": 0.0216,
|
| 1190 |
+
"step": 1850
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 6.41,
|
| 1194 |
+
"learning_rate": 2.64367816091954e-05,
|
| 1195 |
+
"loss": 0.0193,
|
| 1196 |
+
"step": 1860
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 6.45,
|
| 1200 |
+
"learning_rate": 2.586206896551724e-05,
|
| 1201 |
+
"loss": 0.0222,
|
| 1202 |
+
"step": 1870
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 6.48,
|
| 1206 |
+
"learning_rate": 2.528735632183908e-05,
|
| 1207 |
+
"loss": 0.0153,
|
| 1208 |
+
"step": 1880
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 6.52,
|
| 1212 |
+
"learning_rate": 2.471264367816092e-05,
|
| 1213 |
+
"loss": 0.0142,
|
| 1214 |
+
"step": 1890
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 6.55,
|
| 1218 |
+
"learning_rate": 2.4137931034482758e-05,
|
| 1219 |
+
"loss": 0.025,
|
| 1220 |
+
"step": 1900
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 6.58,
|
| 1224 |
+
"learning_rate": 2.3563218390804597e-05,
|
| 1225 |
+
"loss": 0.0197,
|
| 1226 |
+
"step": 1910
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 6.62,
|
| 1230 |
+
"learning_rate": 2.2988505747126437e-05,
|
| 1231 |
+
"loss": 0.0212,
|
| 1232 |
+
"step": 1920
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 6.65,
|
| 1236 |
+
"learning_rate": 2.2413793103448276e-05,
|
| 1237 |
+
"loss": 0.0181,
|
| 1238 |
+
"step": 1930
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 6.69,
|
| 1242 |
+
"learning_rate": 2.1839080459770115e-05,
|
| 1243 |
+
"loss": 0.0228,
|
| 1244 |
+
"step": 1940
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 6.72,
|
| 1248 |
+
"learning_rate": 2.1264367816091954e-05,
|
| 1249 |
+
"loss": 0.015,
|
| 1250 |
+
"step": 1950
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 6.76,
|
| 1254 |
+
"learning_rate": 2.0689655172413797e-05,
|
| 1255 |
+
"loss": 0.0188,
|
| 1256 |
+
"step": 1960
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 6.79,
|
| 1260 |
+
"learning_rate": 2.0114942528735632e-05,
|
| 1261 |
+
"loss": 0.0231,
|
| 1262 |
+
"step": 1970
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 6.83,
|
| 1266 |
+
"learning_rate": 1.954022988505747e-05,
|
| 1267 |
+
"loss": 0.0157,
|
| 1268 |
+
"step": 1980
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 6.86,
|
| 1272 |
+
"learning_rate": 1.896551724137931e-05,
|
| 1273 |
+
"loss": 0.0155,
|
| 1274 |
+
"step": 1990
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 6.89,
|
| 1278 |
+
"learning_rate": 1.839080459770115e-05,
|
| 1279 |
+
"loss": 0.0214,
|
| 1280 |
+
"step": 2000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 6.93,
|
| 1284 |
+
"learning_rate": 1.781609195402299e-05,
|
| 1285 |
+
"loss": 0.0154,
|
| 1286 |
+
"step": 2010
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 6.96,
|
| 1290 |
+
"learning_rate": 1.7241379310344828e-05,
|
| 1291 |
+
"loss": 0.0186,
|
| 1292 |
+
"step": 2020
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 7.0,
|
| 1296 |
+
"learning_rate": 1.6666666666666667e-05,
|
| 1297 |
+
"loss": 0.0236,
|
| 1298 |
+
"step": 2030
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 7.0,
|
| 1302 |
+
"eval_accuracy": 0.7278692051199311,
|
| 1303 |
+
"eval_f1": 0.7209486341359743,
|
| 1304 |
+
"eval_loss": 1.4161986112594604,
|
| 1305 |
+
"eval_precision": 0.7194977934958102,
|
| 1306 |
+
"eval_recall": 0.7278692051199311,
|
| 1307 |
+
"eval_runtime": 103.3877,
|
| 1308 |
+
"eval_samples_per_second": 179.847,
|
| 1309 |
+
"eval_steps_per_second": 2.815,
|
| 1310 |
+
"step": 2030
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 7.03,
|
| 1314 |
+
"learning_rate": 1.6091954022988507e-05,
|
| 1315 |
+
"loss": 0.0094,
|
| 1316 |
+
"step": 2040
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 7.07,
|
| 1320 |
+
"learning_rate": 1.5517241379310346e-05,
|
| 1321 |
+
"loss": 0.0117,
|
| 1322 |
+
"step": 2050
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 7.1,
|
| 1326 |
+
"learning_rate": 1.4942528735632183e-05,
|
| 1327 |
+
"loss": 0.0049,
|
| 1328 |
+
"step": 2060
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 7.14,
|
| 1332 |
+
"learning_rate": 1.4367816091954022e-05,
|
| 1333 |
+
"loss": 0.0097,
|
| 1334 |
+
"step": 2070
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 7.17,
|
| 1338 |
+
"learning_rate": 1.3793103448275862e-05,
|
| 1339 |
+
"loss": 0.0092,
|
| 1340 |
+
"step": 2080
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 7.21,
|
| 1344 |
+
"learning_rate": 1.32183908045977e-05,
|
| 1345 |
+
"loss": 0.0086,
|
| 1346 |
+
"step": 2090
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 7.24,
|
| 1350 |
+
"learning_rate": 1.264367816091954e-05,
|
| 1351 |
+
"loss": 0.008,
|
| 1352 |
+
"step": 2100
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 7.28,
|
| 1356 |
+
"learning_rate": 1.2068965517241379e-05,
|
| 1357 |
+
"loss": 0.01,
|
| 1358 |
+
"step": 2110
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 7.31,
|
| 1362 |
+
"learning_rate": 1.1494252873563218e-05,
|
| 1363 |
+
"loss": 0.0078,
|
| 1364 |
+
"step": 2120
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 7.34,
|
| 1368 |
+
"learning_rate": 1.0919540229885057e-05,
|
| 1369 |
+
"loss": 0.0093,
|
| 1370 |
+
"step": 2130
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 7.38,
|
| 1374 |
+
"learning_rate": 1.0344827586206898e-05,
|
| 1375 |
+
"loss": 0.0081,
|
| 1376 |
+
"step": 2140
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 7.41,
|
| 1380 |
+
"learning_rate": 9.770114942528736e-06,
|
| 1381 |
+
"loss": 0.0057,
|
| 1382 |
+
"step": 2150
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 7.45,
|
| 1386 |
+
"learning_rate": 9.195402298850575e-06,
|
| 1387 |
+
"loss": 0.0095,
|
| 1388 |
+
"step": 2160
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 7.48,
|
| 1392 |
+
"learning_rate": 8.620689655172414e-06,
|
| 1393 |
+
"loss": 0.0062,
|
| 1394 |
+
"step": 2170
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 7.52,
|
| 1398 |
+
"learning_rate": 8.045977011494253e-06,
|
| 1399 |
+
"loss": 0.0124,
|
| 1400 |
+
"step": 2180
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 7.55,
|
| 1404 |
+
"learning_rate": 7.471264367816092e-06,
|
| 1405 |
+
"loss": 0.008,
|
| 1406 |
+
"step": 2190
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 7.58,
|
| 1410 |
+
"learning_rate": 6.896551724137931e-06,
|
| 1411 |
+
"loss": 0.0067,
|
| 1412 |
+
"step": 2200
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 7.62,
|
| 1416 |
+
"learning_rate": 6.32183908045977e-06,
|
| 1417 |
+
"loss": 0.0085,
|
| 1418 |
+
"step": 2210
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 7.65,
|
| 1422 |
+
"learning_rate": 5.747126436781609e-06,
|
| 1423 |
+
"loss": 0.0053,
|
| 1424 |
+
"step": 2220
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 7.69,
|
| 1428 |
+
"learning_rate": 5.172413793103449e-06,
|
| 1429 |
+
"loss": 0.007,
|
| 1430 |
+
"step": 2230
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 7.72,
|
| 1434 |
+
"learning_rate": 4.5977011494252875e-06,
|
| 1435 |
+
"loss": 0.0061,
|
| 1436 |
+
"step": 2240
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 7.76,
|
| 1440 |
+
"learning_rate": 4.022988505747127e-06,
|
| 1441 |
+
"loss": 0.0054,
|
| 1442 |
+
"step": 2250
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 7.79,
|
| 1446 |
+
"learning_rate": 3.4482758620689654e-06,
|
| 1447 |
+
"loss": 0.0103,
|
| 1448 |
+
"step": 2260
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 7.83,
|
| 1452 |
+
"learning_rate": 2.8735632183908046e-06,
|
| 1453 |
+
"loss": 0.0131,
|
| 1454 |
+
"step": 2270
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 7.86,
|
| 1458 |
+
"learning_rate": 2.2988505747126437e-06,
|
| 1459 |
+
"loss": 0.0143,
|
| 1460 |
+
"step": 2280
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 7.89,
|
| 1464 |
+
"learning_rate": 1.7241379310344827e-06,
|
| 1465 |
+
"loss": 0.0075,
|
| 1466 |
+
"step": 2290
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 7.93,
|
| 1470 |
+
"learning_rate": 1.1494252873563219e-06,
|
| 1471 |
+
"loss": 0.0079,
|
| 1472 |
+
"step": 2300
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 7.96,
|
| 1476 |
+
"learning_rate": 5.747126436781609e-07,
|
| 1477 |
+
"loss": 0.0077,
|
| 1478 |
+
"step": 2310
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 8.0,
|
| 1482 |
+
"learning_rate": 0.0,
|
| 1483 |
+
"loss": 0.0049,
|
| 1484 |
+
"step": 2320
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 8.0,
|
| 1488 |
+
"eval_accuracy": 0.7248574809078198,
|
| 1489 |
+
"eval_f1": 0.7195690317790054,
|
| 1490 |
+
"eval_loss": 1.4531140327453613,
|
| 1491 |
+
"eval_precision": 0.717172031675939,
|
| 1492 |
+
"eval_recall": 0.7248574809078198,
|
| 1493 |
+
"eval_runtime": 105.0016,
|
| 1494 |
+
"eval_samples_per_second": 177.083,
|
| 1495 |
+
"eval_steps_per_second": 2.771,
|
| 1496 |
+
"step": 2320
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 8.0,
|
| 1500 |
+
"step": 2320,
|
| 1501 |
+
"total_flos": 4.6098795710207476e+19,
|
| 1502 |
+
"train_loss": 0.3678270006343594,
|
| 1503 |
+
"train_runtime": 6722.1212,
|
| 1504 |
+
"train_samples_per_second": 88.513,
|
| 1505 |
+
"train_steps_per_second": 0.345
|
| 1506 |
}
|
| 1507 |
],
|
| 1508 |
+
"max_steps": 2320,
|
| 1509 |
"num_train_epochs": 8,
|
| 1510 |
+
"total_flos": 4.6098795710207476e+19,
|
| 1511 |
"trial_name": null,
|
| 1512 |
"trial_params": null
|
| 1513 |
}
|