End of training
Browse files- all_results.json +10 -10
- eval_results.json +6 -6
- runs/Dec22_23-06-37_129-146-176-120/events.out.tfevents.1671779100.129-146-176-120.850347.2 +3 -0
- train_results.json +5 -5
- trainer_state.json +1850 -107
all_results.json
CHANGED
|
@@ -1,12 +1,12 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second": 1.
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
-
"eval_wer":
|
| 8 |
-
"train_loss": 0.
|
| 9 |
-
"train_runtime":
|
| 10 |
-
"train_samples_per_second":
|
| 11 |
-
"train_steps_per_second": 0.
|
| 12 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 528.76,
|
| 3 |
+
"eval_loss": 0.56689453125,
|
| 4 |
+
"eval_runtime": 156.403,
|
| 5 |
+
"eval_samples_per_second": 1.739,
|
| 6 |
+
"eval_steps_per_second": 0.109,
|
| 7 |
+
"eval_wer": 9.899702823179792,
|
| 8 |
+
"train_loss": 0.00025936167769961886,
|
| 9 |
+
"train_runtime": 28092.3517,
|
| 10 |
+
"train_samples_per_second": 10.252,
|
| 11 |
+
"train_steps_per_second": 0.32
|
| 12 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_loss": 0.
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second": 1.
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
-
"eval_wer":
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 528.76,
|
| 3 |
+
"eval_loss": 0.56689453125,
|
| 4 |
+
"eval_runtime": 156.403,
|
| 5 |
+
"eval_samples_per_second": 1.739,
|
| 6 |
+
"eval_steps_per_second": 0.109,
|
| 7 |
+
"eval_wer": 9.899702823179792
|
| 8 |
}
|
runs/Dec22_23-06-37_129-146-176-120/events.out.tfevents.1671779100.129-146-176-120.850347.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:eab3c8a5fb58537fd8ed93d72dad79534604a89d26750c9845c15102817d15b2
|
| 3 |
+
size 358
|
train_results.json
CHANGED
|
@@ -1,7 +1,7 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
-
"train_samples_per_second":
|
| 6 |
-
"train_steps_per_second": 0.
|
| 7 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 528.76,
|
| 3 |
+
"train_loss": 0.00025936167769961886,
|
| 4 |
+
"train_runtime": 28092.3517,
|
| 5 |
+
"train_samples_per_second": 10.252,
|
| 6 |
+
"train_steps_per_second": 0.32
|
| 7 |
}
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric":
|
| 3 |
-
"best_model_checkpoint": "./checkpoint-
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -129,395 +129,2138 @@
|
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 30.88,
|
| 132 |
-
"learning_rate": 2.
|
| 133 |
"loss": 0.0018,
|
| 134 |
"step": 525
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 32.35,
|
| 138 |
-
"learning_rate": 2.
|
| 139 |
"loss": 0.0021,
|
| 140 |
"step": 550
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 33.82,
|
| 144 |
-
"learning_rate": 2.
|
| 145 |
"loss": 0.0017,
|
| 146 |
"step": 575
|
| 147 |
},
|
| 148 |
{
|
| 149 |
"epoch": 35.29,
|
| 150 |
-
"learning_rate": 2.
|
| 151 |
"loss": 0.0018,
|
| 152 |
"step": 600
|
| 153 |
},
|
| 154 |
{
|
| 155 |
"epoch": 36.76,
|
| 156 |
-
"learning_rate": 2.
|
| 157 |
"loss": 0.0016,
|
| 158 |
"step": 625
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 38.24,
|
| 162 |
-
"learning_rate": 2.
|
| 163 |
"loss": 0.0017,
|
| 164 |
"step": 650
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 39.71,
|
| 168 |
-
"learning_rate": 2.
|
| 169 |
"loss": 0.0016,
|
| 170 |
"step": 675
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 41.18,
|
| 174 |
-
"learning_rate": 2.
|
| 175 |
"loss": 0.0013,
|
| 176 |
"step": 700
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 42.65,
|
| 180 |
-
"learning_rate": 2.
|
| 181 |
-
"loss": 0.
|
| 182 |
"step": 725
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 44.12,
|
| 186 |
-
"learning_rate": 2.
|
| 187 |
"loss": 0.0012,
|
| 188 |
"step": 750
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 45.59,
|
| 192 |
-
"learning_rate": 2.
|
| 193 |
-
"loss": 0.
|
| 194 |
"step": 775
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 47.06,
|
| 198 |
-
"learning_rate": 2.
|
| 199 |
-
"loss": 0.
|
| 200 |
"step": 800
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 48.53,
|
| 204 |
-
"learning_rate": 2.
|
| 205 |
"loss": 0.0013,
|
| 206 |
"step": 825
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 50.0,
|
| 210 |
-
"learning_rate": 2.
|
| 211 |
"loss": 0.0011,
|
| 212 |
"step": 850
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 51.47,
|
| 216 |
-
"learning_rate": 2.
|
| 217 |
"loss": 0.0013,
|
| 218 |
"step": 875
|
| 219 |
},
|
| 220 |
{
|
| 221 |
"epoch": 52.94,
|
| 222 |
-
"learning_rate": 2.
|
| 223 |
"loss": 0.0014,
|
| 224 |
"step": 900
|
| 225 |
},
|
| 226 |
{
|
| 227 |
"epoch": 54.41,
|
| 228 |
-
"learning_rate": 2.
|
| 229 |
"loss": 0.0011,
|
| 230 |
"step": 925
|
| 231 |
},
|
| 232 |
{
|
| 233 |
"epoch": 55.88,
|
| 234 |
-
"learning_rate": 2.
|
| 235 |
-
"loss": 0.
|
| 236 |
"step": 950
|
| 237 |
},
|
| 238 |
{
|
| 239 |
"epoch": 57.35,
|
| 240 |
-
"learning_rate": 2.
|
| 241 |
"loss": 0.0014,
|
| 242 |
"step": 975
|
| 243 |
},
|
| 244 |
{
|
| 245 |
"epoch": 58.82,
|
| 246 |
-
"learning_rate": 2.
|
| 247 |
"loss": 0.0014,
|
| 248 |
"step": 1000
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 58.82,
|
| 252 |
-
"eval_loss": 0.
|
| 253 |
-
"eval_runtime":
|
| 254 |
-
"eval_samples_per_second": 1.
|
| 255 |
-
"eval_steps_per_second": 0.
|
| 256 |
"eval_wer": 10.36404160475483,
|
| 257 |
"step": 1000
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 60.29,
|
| 261 |
-
"learning_rate":
|
| 262 |
-
"loss": 0.
|
| 263 |
"step": 1025
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 61.76,
|
| 267 |
-
"learning_rate":
|
| 268 |
"loss": 0.0009,
|
| 269 |
"step": 1050
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 63.24,
|
| 273 |
-
"learning_rate":
|
| 274 |
"loss": 0.0012,
|
| 275 |
"step": 1075
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 64.71,
|
| 279 |
-
"learning_rate":
|
| 280 |
"loss": 0.0009,
|
| 281 |
"step": 1100
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 66.18,
|
| 285 |
-
"learning_rate":
|
| 286 |
-
"loss": 0.
|
| 287 |
"step": 1125
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 67.65,
|
| 291 |
-
"learning_rate":
|
| 292 |
"loss": 0.0012,
|
| 293 |
"step": 1150
|
| 294 |
},
|
| 295 |
{
|
| 296 |
"epoch": 69.12,
|
| 297 |
-
"learning_rate":
|
| 298 |
"loss": 0.0009,
|
| 299 |
"step": 1175
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 70.59,
|
| 303 |
-
"learning_rate":
|
| 304 |
-
"loss": 0.
|
| 305 |
"step": 1200
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 72.06,
|
| 309 |
-
"learning_rate":
|
| 310 |
"loss": 0.0011,
|
| 311 |
"step": 1225
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 73.53,
|
| 315 |
-
"learning_rate":
|
| 316 |
-
"loss": 0.
|
| 317 |
"step": 1250
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"epoch": 75.0,
|
| 321 |
-
"learning_rate":
|
| 322 |
"loss": 0.0008,
|
| 323 |
"step": 1275
|
| 324 |
},
|
| 325 |
{
|
| 326 |
"epoch": 76.47,
|
| 327 |
-
"learning_rate":
|
| 328 |
-
"loss": 0.
|
| 329 |
"step": 1300
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 77.94,
|
| 333 |
-
"learning_rate":
|
| 334 |
-
"loss": 0.
|
| 335 |
"step": 1325
|
| 336 |
},
|
| 337 |
{
|
| 338 |
"epoch": 79.41,
|
| 339 |
-
"learning_rate":
|
| 340 |
"loss": 0.0009,
|
| 341 |
"step": 1350
|
| 342 |
},
|
| 343 |
{
|
| 344 |
"epoch": 80.88,
|
| 345 |
-
"learning_rate":
|
| 346 |
"loss": 0.0011,
|
| 347 |
"step": 1375
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 82.35,
|
| 351 |
-
"learning_rate":
|
| 352 |
"loss": 0.0007,
|
| 353 |
"step": 1400
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 83.82,
|
| 357 |
-
"learning_rate":
|
| 358 |
-
"loss": 0.
|
| 359 |
"step": 1425
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"epoch": 85.29,
|
| 363 |
-
"learning_rate":
|
| 364 |
"loss": 0.001,
|
| 365 |
"step": 1450
|
| 366 |
},
|
| 367 |
{
|
| 368 |
"epoch": 86.76,
|
| 369 |
-
"learning_rate":
|
| 370 |
"loss": 0.0008,
|
| 371 |
"step": 1475
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 88.24,
|
| 375 |
-
"learning_rate":
|
| 376 |
-
"loss": 0.
|
| 377 |
"step": 1500
|
| 378 |
},
|
| 379 |
{
|
| 380 |
"epoch": 89.71,
|
| 381 |
-
"learning_rate":
|
| 382 |
-
"loss": 0.
|
| 383 |
"step": 1525
|
| 384 |
},
|
| 385 |
{
|
| 386 |
"epoch": 91.18,
|
| 387 |
-
"learning_rate":
|
| 388 |
-
"loss": 0.
|
| 389 |
"step": 1550
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 92.65,
|
| 393 |
-
"learning_rate":
|
| 394 |
"loss": 0.0006,
|
| 395 |
"step": 1575
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 94.12,
|
| 399 |
-
"learning_rate":
|
| 400 |
"loss": 0.0008,
|
| 401 |
"step": 1600
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 95.59,
|
| 405 |
-
"learning_rate":
|
| 406 |
"loss": 0.0009,
|
| 407 |
"step": 1625
|
| 408 |
},
|
| 409 |
{
|
| 410 |
"epoch": 97.06,
|
| 411 |
-
"learning_rate":
|
| 412 |
-
"loss": 0.
|
| 413 |
"step": 1650
|
| 414 |
},
|
| 415 |
{
|
| 416 |
"epoch": 98.53,
|
| 417 |
-
"learning_rate":
|
| 418 |
-
"loss": 0.
|
| 419 |
"step": 1675
|
| 420 |
},
|
| 421 |
{
|
| 422 |
"epoch": 100.0,
|
| 423 |
-
"learning_rate":
|
| 424 |
"loss": 0.0008,
|
| 425 |
"step": 1700
|
| 426 |
},
|
| 427 |
{
|
| 428 |
"epoch": 101.47,
|
| 429 |
-
"learning_rate":
|
| 430 |
-
"loss": 0.
|
| 431 |
"step": 1725
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 102.94,
|
| 435 |
-
"learning_rate":
|
| 436 |
-
"loss": 0.
|
| 437 |
"step": 1750
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 104.41,
|
| 441 |
-
"learning_rate":
|
| 442 |
-
"loss": 0.
|
| 443 |
"step": 1775
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"epoch": 105.88,
|
| 447 |
-
"learning_rate":
|
| 448 |
"loss": 0.0006,
|
| 449 |
"step": 1800
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 107.35,
|
| 453 |
-
"learning_rate":
|
| 454 |
"loss": 0.0007,
|
| 455 |
"step": 1825
|
| 456 |
},
|
| 457 |
{
|
| 458 |
"epoch": 108.82,
|
| 459 |
-
"learning_rate":
|
| 460 |
-
"loss": 0.
|
| 461 |
"step": 1850
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 110.29,
|
| 465 |
-
"learning_rate":
|
| 466 |
-
"loss": 0.
|
| 467 |
"step": 1875
|
| 468 |
},
|
| 469 |
{
|
| 470 |
"epoch": 111.76,
|
| 471 |
-
"learning_rate":
|
| 472 |
"loss": 0.0005,
|
| 473 |
"step": 1900
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 113.24,
|
| 477 |
-
"learning_rate": 1.
|
| 478 |
-
"loss": 0.
|
| 479 |
"step": 1925
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 114.71,
|
| 483 |
-
"learning_rate": 1.
|
| 484 |
"loss": 0.0006,
|
| 485 |
"step": 1950
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"epoch": 116.18,
|
| 489 |
-
"learning_rate":
|
| 490 |
-
"loss": 0.
|
| 491 |
"step": 1975
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 117.65,
|
| 495 |
-
"learning_rate":
|
| 496 |
-
"loss": 0.
|
| 497 |
"step": 2000
|
| 498 |
},
|
| 499 |
{
|
| 500 |
"epoch": 117.65,
|
| 501 |
-
"eval_loss": 0.
|
| 502 |
-
"eval_runtime":
|
| 503 |
-
"eval_samples_per_second": 1.
|
| 504 |
-
"eval_steps_per_second": 0.
|
| 505 |
-
"eval_wer": 10.
|
| 506 |
"step": 2000
|
| 507 |
},
|
| 508 |
{
|
| 509 |
-
"epoch":
|
| 510 |
-
"
|
| 511 |
-
"
|
| 512 |
-
"
|
| 513 |
-
|
| 514 |
-
|
| 515 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 516 |
}
|
| 517 |
],
|
| 518 |
-
"max_steps":
|
| 519 |
-
"num_train_epochs":
|
| 520 |
-
"total_flos":
|
| 521 |
"trial_name": null,
|
| 522 |
"trial_params": null
|
| 523 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 9.899702823179792,
|
| 3 |
+
"best_model_checkpoint": "./checkpoint-8000",
|
| 4 |
+
"epoch": 528.7647058823529,
|
| 5 |
+
"global_step": 9000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 129 |
},
|
| 130 |
{
|
| 131 |
"epoch": 30.88,
|
| 132 |
+
"learning_rate": 2.9811428571428573e-06,
|
| 133 |
"loss": 0.0018,
|
| 134 |
"step": 525
|
| 135 |
},
|
| 136 |
{
|
| 137 |
"epoch": 32.35,
|
| 138 |
+
"learning_rate": 2.959714285714286e-06,
|
| 139 |
"loss": 0.0021,
|
| 140 |
"step": 550
|
| 141 |
},
|
| 142 |
{
|
| 143 |
"epoch": 33.82,
|
| 144 |
+
"learning_rate": 2.9382857142857143e-06,
|
| 145 |
"loss": 0.0017,
|
| 146 |
"step": 575
|
| 147 |
},
|
| 148 |
{
|
| 149 |
"epoch": 35.29,
|
| 150 |
+
"learning_rate": 2.916857142857143e-06,
|
| 151 |
"loss": 0.0018,
|
| 152 |
"step": 600
|
| 153 |
},
|
| 154 |
{
|
| 155 |
"epoch": 36.76,
|
| 156 |
+
"learning_rate": 2.895428571428572e-06,
|
| 157 |
"loss": 0.0016,
|
| 158 |
"step": 625
|
| 159 |
},
|
| 160 |
{
|
| 161 |
"epoch": 38.24,
|
| 162 |
+
"learning_rate": 2.874e-06,
|
| 163 |
"loss": 0.0017,
|
| 164 |
"step": 650
|
| 165 |
},
|
| 166 |
{
|
| 167 |
"epoch": 39.71,
|
| 168 |
+
"learning_rate": 2.8525714285714285e-06,
|
| 169 |
"loss": 0.0016,
|
| 170 |
"step": 675
|
| 171 |
},
|
| 172 |
{
|
| 173 |
"epoch": 41.18,
|
| 174 |
+
"learning_rate": 2.8311428571428574e-06,
|
| 175 |
"loss": 0.0013,
|
| 176 |
"step": 700
|
| 177 |
},
|
| 178 |
{
|
| 179 |
"epoch": 42.65,
|
| 180 |
+
"learning_rate": 2.809714285714286e-06,
|
| 181 |
+
"loss": 0.0014,
|
| 182 |
"step": 725
|
| 183 |
},
|
| 184 |
{
|
| 185 |
"epoch": 44.12,
|
| 186 |
+
"learning_rate": 2.788285714285714e-06,
|
| 187 |
"loss": 0.0012,
|
| 188 |
"step": 750
|
| 189 |
},
|
| 190 |
{
|
| 191 |
"epoch": 45.59,
|
| 192 |
+
"learning_rate": 2.766857142857143e-06,
|
| 193 |
+
"loss": 0.0014,
|
| 194 |
"step": 775
|
| 195 |
},
|
| 196 |
{
|
| 197 |
"epoch": 47.06,
|
| 198 |
+
"learning_rate": 2.7454285714285716e-06,
|
| 199 |
+
"loss": 0.0014,
|
| 200 |
"step": 800
|
| 201 |
},
|
| 202 |
{
|
| 203 |
"epoch": 48.53,
|
| 204 |
+
"learning_rate": 2.724e-06,
|
| 205 |
"loss": 0.0013,
|
| 206 |
"step": 825
|
| 207 |
},
|
| 208 |
{
|
| 209 |
"epoch": 50.0,
|
| 210 |
+
"learning_rate": 2.7025714285714286e-06,
|
| 211 |
"loss": 0.0011,
|
| 212 |
"step": 850
|
| 213 |
},
|
| 214 |
{
|
| 215 |
"epoch": 51.47,
|
| 216 |
+
"learning_rate": 2.681142857142857e-06,
|
| 217 |
"loss": 0.0013,
|
| 218 |
"step": 875
|
| 219 |
},
|
| 220 |
{
|
| 221 |
"epoch": 52.94,
|
| 222 |
+
"learning_rate": 2.6597142857142857e-06,
|
| 223 |
"loss": 0.0014,
|
| 224 |
"step": 900
|
| 225 |
},
|
| 226 |
{
|
| 227 |
"epoch": 54.41,
|
| 228 |
+
"learning_rate": 2.6382857142857142e-06,
|
| 229 |
"loss": 0.0011,
|
| 230 |
"step": 925
|
| 231 |
},
|
| 232 |
{
|
| 233 |
"epoch": 55.88,
|
| 234 |
+
"learning_rate": 2.616857142857143e-06,
|
| 235 |
+
"loss": 0.0021,
|
| 236 |
"step": 950
|
| 237 |
},
|
| 238 |
{
|
| 239 |
"epoch": 57.35,
|
| 240 |
+
"learning_rate": 2.5954285714285713e-06,
|
| 241 |
"loss": 0.0014,
|
| 242 |
"step": 975
|
| 243 |
},
|
| 244 |
{
|
| 245 |
"epoch": 58.82,
|
| 246 |
+
"learning_rate": 2.574e-06,
|
| 247 |
"loss": 0.0014,
|
| 248 |
"step": 1000
|
| 249 |
},
|
| 250 |
{
|
| 251 |
"epoch": 58.82,
|
| 252 |
+
"eval_loss": 0.4951171875,
|
| 253 |
+
"eval_runtime": 168.7172,
|
| 254 |
+
"eval_samples_per_second": 1.612,
|
| 255 |
+
"eval_steps_per_second": 0.101,
|
| 256 |
"eval_wer": 10.36404160475483,
|
| 257 |
"step": 1000
|
| 258 |
},
|
| 259 |
{
|
| 260 |
"epoch": 60.29,
|
| 261 |
+
"learning_rate": 2.5542857142857143e-06,
|
| 262 |
+
"loss": 0.0013,
|
| 263 |
"step": 1025
|
| 264 |
},
|
| 265 |
{
|
| 266 |
"epoch": 61.76,
|
| 267 |
+
"learning_rate": 2.532857142857143e-06,
|
| 268 |
"loss": 0.0009,
|
| 269 |
"step": 1050
|
| 270 |
},
|
| 271 |
{
|
| 272 |
"epoch": 63.24,
|
| 273 |
+
"learning_rate": 2.5114285714285718e-06,
|
| 274 |
"loss": 0.0012,
|
| 275 |
"step": 1075
|
| 276 |
},
|
| 277 |
{
|
| 278 |
"epoch": 64.71,
|
| 279 |
+
"learning_rate": 2.49e-06,
|
| 280 |
"loss": 0.0009,
|
| 281 |
"step": 1100
|
| 282 |
},
|
| 283 |
{
|
| 284 |
"epoch": 66.18,
|
| 285 |
+
"learning_rate": 2.4685714285714284e-06,
|
| 286 |
+
"loss": 0.0014,
|
| 287 |
"step": 1125
|
| 288 |
},
|
| 289 |
{
|
| 290 |
"epoch": 67.65,
|
| 291 |
+
"learning_rate": 2.4471428571428574e-06,
|
| 292 |
"loss": 0.0012,
|
| 293 |
"step": 1150
|
| 294 |
},
|
| 295 |
{
|
| 296 |
"epoch": 69.12,
|
| 297 |
+
"learning_rate": 2.425714285714286e-06,
|
| 298 |
"loss": 0.0009,
|
| 299 |
"step": 1175
|
| 300 |
},
|
| 301 |
{
|
| 302 |
"epoch": 70.59,
|
| 303 |
+
"learning_rate": 2.404285714285714e-06,
|
| 304 |
+
"loss": 0.0008,
|
| 305 |
"step": 1200
|
| 306 |
},
|
| 307 |
{
|
| 308 |
"epoch": 72.06,
|
| 309 |
+
"learning_rate": 2.382857142857143e-06,
|
| 310 |
"loss": 0.0011,
|
| 311 |
"step": 1225
|
| 312 |
},
|
| 313 |
{
|
| 314 |
"epoch": 73.53,
|
| 315 |
+
"learning_rate": 2.3614285714285715e-06,
|
| 316 |
+
"loss": 0.0015,
|
| 317 |
"step": 1250
|
| 318 |
},
|
| 319 |
{
|
| 320 |
"epoch": 75.0,
|
| 321 |
+
"learning_rate": 2.34e-06,
|
| 322 |
"loss": 0.0008,
|
| 323 |
"step": 1275
|
| 324 |
},
|
| 325 |
{
|
| 326 |
"epoch": 76.47,
|
| 327 |
+
"learning_rate": 2.3185714285714286e-06,
|
| 328 |
+
"loss": 0.001,
|
| 329 |
"step": 1300
|
| 330 |
},
|
| 331 |
{
|
| 332 |
"epoch": 77.94,
|
| 333 |
+
"learning_rate": 2.297142857142857e-06,
|
| 334 |
+
"loss": 0.0009,
|
| 335 |
"step": 1325
|
| 336 |
},
|
| 337 |
{
|
| 338 |
"epoch": 79.41,
|
| 339 |
+
"learning_rate": 2.2757142857142856e-06,
|
| 340 |
"loss": 0.0009,
|
| 341 |
"step": 1350
|
| 342 |
},
|
| 343 |
{
|
| 344 |
"epoch": 80.88,
|
| 345 |
+
"learning_rate": 2.2542857142857146e-06,
|
| 346 |
"loss": 0.0011,
|
| 347 |
"step": 1375
|
| 348 |
},
|
| 349 |
{
|
| 350 |
"epoch": 82.35,
|
| 351 |
+
"learning_rate": 2.232857142857143e-06,
|
| 352 |
"loss": 0.0007,
|
| 353 |
"step": 1400
|
| 354 |
},
|
| 355 |
{
|
| 356 |
"epoch": 83.82,
|
| 357 |
+
"learning_rate": 2.2114285714285712e-06,
|
| 358 |
+
"loss": 0.001,
|
| 359 |
"step": 1425
|
| 360 |
},
|
| 361 |
{
|
| 362 |
"epoch": 85.29,
|
| 363 |
+
"learning_rate": 2.19e-06,
|
| 364 |
"loss": 0.001,
|
| 365 |
"step": 1450
|
| 366 |
},
|
| 367 |
{
|
| 368 |
"epoch": 86.76,
|
| 369 |
+
"learning_rate": 2.1685714285714287e-06,
|
| 370 |
"loss": 0.0008,
|
| 371 |
"step": 1475
|
| 372 |
},
|
| 373 |
{
|
| 374 |
"epoch": 88.24,
|
| 375 |
+
"learning_rate": 2.1471428571428573e-06,
|
| 376 |
+
"loss": 0.001,
|
| 377 |
"step": 1500
|
| 378 |
},
|
| 379 |
{
|
| 380 |
"epoch": 89.71,
|
| 381 |
+
"learning_rate": 2.125714285714286e-06,
|
| 382 |
+
"loss": 0.0008,
|
| 383 |
"step": 1525
|
| 384 |
},
|
| 385 |
{
|
| 386 |
"epoch": 91.18,
|
| 387 |
+
"learning_rate": 2.1042857142857143e-06,
|
| 388 |
+
"loss": 0.0008,
|
| 389 |
"step": 1550
|
| 390 |
},
|
| 391 |
{
|
| 392 |
"epoch": 92.65,
|
| 393 |
+
"learning_rate": 2.082857142857143e-06,
|
| 394 |
"loss": 0.0006,
|
| 395 |
"step": 1575
|
| 396 |
},
|
| 397 |
{
|
| 398 |
"epoch": 94.12,
|
| 399 |
+
"learning_rate": 2.0614285714285714e-06,
|
| 400 |
"loss": 0.0008,
|
| 401 |
"step": 1600
|
| 402 |
},
|
| 403 |
{
|
| 404 |
"epoch": 95.59,
|
| 405 |
+
"learning_rate": 2.0400000000000004e-06,
|
| 406 |
"loss": 0.0009,
|
| 407 |
"step": 1625
|
| 408 |
},
|
| 409 |
{
|
| 410 |
"epoch": 97.06,
|
| 411 |
+
"learning_rate": 2.0185714285714285e-06,
|
| 412 |
+
"loss": 0.0009,
|
| 413 |
"step": 1650
|
| 414 |
},
|
| 415 |
{
|
| 416 |
"epoch": 98.53,
|
| 417 |
+
"learning_rate": 1.997142857142857e-06,
|
| 418 |
+
"loss": 0.0008,
|
| 419 |
"step": 1675
|
| 420 |
},
|
| 421 |
{
|
| 422 |
"epoch": 100.0,
|
| 423 |
+
"learning_rate": 1.975714285714286e-06,
|
| 424 |
"loss": 0.0008,
|
| 425 |
"step": 1700
|
| 426 |
},
|
| 427 |
{
|
| 428 |
"epoch": 101.47,
|
| 429 |
+
"learning_rate": 1.9542857142857145e-06,
|
| 430 |
+
"loss": 0.0007,
|
| 431 |
"step": 1725
|
| 432 |
},
|
| 433 |
{
|
| 434 |
"epoch": 102.94,
|
| 435 |
+
"learning_rate": 1.9328571428571426e-06,
|
| 436 |
+
"loss": 0.0005,
|
| 437 |
"step": 1750
|
| 438 |
},
|
| 439 |
{
|
| 440 |
"epoch": 104.41,
|
| 441 |
+
"learning_rate": 1.9114285714285716e-06,
|
| 442 |
+
"loss": 0.0008,
|
| 443 |
"step": 1775
|
| 444 |
},
|
| 445 |
{
|
| 446 |
"epoch": 105.88,
|
| 447 |
+
"learning_rate": 1.8900000000000001e-06,
|
| 448 |
"loss": 0.0006,
|
| 449 |
"step": 1800
|
| 450 |
},
|
| 451 |
{
|
| 452 |
"epoch": 107.35,
|
| 453 |
+
"learning_rate": 1.8685714285714287e-06,
|
| 454 |
"loss": 0.0007,
|
| 455 |
"step": 1825
|
| 456 |
},
|
| 457 |
{
|
| 458 |
"epoch": 108.82,
|
| 459 |
+
"learning_rate": 1.8471428571428574e-06,
|
| 460 |
+
"loss": 0.0008,
|
| 461 |
"step": 1850
|
| 462 |
},
|
| 463 |
{
|
| 464 |
"epoch": 110.29,
|
| 465 |
+
"learning_rate": 1.8257142857142857e-06,
|
| 466 |
+
"loss": 0.0008,
|
| 467 |
"step": 1875
|
| 468 |
},
|
| 469 |
{
|
| 470 |
"epoch": 111.76,
|
| 471 |
+
"learning_rate": 1.8042857142857143e-06,
|
| 472 |
"loss": 0.0005,
|
| 473 |
"step": 1900
|
| 474 |
},
|
| 475 |
{
|
| 476 |
"epoch": 113.24,
|
| 477 |
+
"learning_rate": 1.782857142857143e-06,
|
| 478 |
+
"loss": 0.0007,
|
| 479 |
"step": 1925
|
| 480 |
},
|
| 481 |
{
|
| 482 |
"epoch": 114.71,
|
| 483 |
+
"learning_rate": 1.7614285714285715e-06,
|
| 484 |
"loss": 0.0006,
|
| 485 |
"step": 1950
|
| 486 |
},
|
| 487 |
{
|
| 488 |
"epoch": 116.18,
|
| 489 |
+
"learning_rate": 1.7399999999999999e-06,
|
| 490 |
+
"loss": 0.0005,
|
| 491 |
"step": 1975
|
| 492 |
},
|
| 493 |
{
|
| 494 |
"epoch": 117.65,
|
| 495 |
+
"learning_rate": 1.7185714285714286e-06,
|
| 496 |
+
"loss": 0.0006,
|
| 497 |
"step": 2000
|
| 498 |
},
|
| 499 |
{
|
| 500 |
"epoch": 117.65,
|
| 501 |
+
"eval_loss": 0.51806640625,
|
| 502 |
+
"eval_runtime": 168.3992,
|
| 503 |
+
"eval_samples_per_second": 1.615,
|
| 504 |
+
"eval_steps_per_second": 0.101,
|
| 505 |
+
"eval_wer": 10.280460624071322,
|
| 506 |
"step": 2000
|
| 507 |
},
|
| 508 |
{
|
| 509 |
+
"epoch": 118.47,
|
| 510 |
+
"learning_rate": 2.4642352941176473e-06,
|
| 511 |
+
"loss": 0.0007,
|
| 512 |
+
"step": 2025
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 119.94,
|
| 516 |
+
"learning_rate": 2.4554117647058827e-06,
|
| 517 |
+
"loss": 0.0006,
|
| 518 |
+
"step": 2050
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 121.41,
|
| 522 |
+
"learning_rate": 2.4465882352941176e-06,
|
| 523 |
+
"loss": 0.0003,
|
| 524 |
+
"step": 2075
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 122.88,
|
| 528 |
+
"learning_rate": 2.437764705882353e-06,
|
| 529 |
+
"loss": 0.0007,
|
| 530 |
+
"step": 2100
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 124.35,
|
| 534 |
+
"learning_rate": 2.4289411764705882e-06,
|
| 535 |
+
"loss": 0.0005,
|
| 536 |
+
"step": 2125
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 125.82,
|
| 540 |
+
"learning_rate": 2.4201176470588236e-06,
|
| 541 |
+
"loss": 0.0006,
|
| 542 |
+
"step": 2150
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 127.29,
|
| 546 |
+
"learning_rate": 2.411294117647059e-06,
|
| 547 |
+
"loss": 0.0007,
|
| 548 |
+
"step": 2175
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 128.76,
|
| 552 |
+
"learning_rate": 2.4024705882352942e-06,
|
| 553 |
+
"loss": 0.0006,
|
| 554 |
+
"step": 2200
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 130.24,
|
| 558 |
+
"learning_rate": 2.3936470588235295e-06,
|
| 559 |
+
"loss": 0.0011,
|
| 560 |
+
"step": 2225
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 131.71,
|
| 564 |
+
"learning_rate": 2.384823529411765e-06,
|
| 565 |
+
"loss": 0.0007,
|
| 566 |
+
"step": 2250
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 133.18,
|
| 570 |
+
"learning_rate": 2.376e-06,
|
| 571 |
+
"loss": 0.0006,
|
| 572 |
+
"step": 2275
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 134.65,
|
| 576 |
+
"learning_rate": 2.3671764705882355e-06,
|
| 577 |
+
"loss": 0.0005,
|
| 578 |
+
"step": 2300
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 136.12,
|
| 582 |
+
"learning_rate": 2.3583529411764704e-06,
|
| 583 |
+
"loss": 0.0007,
|
| 584 |
+
"step": 2325
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 137.59,
|
| 588 |
+
"learning_rate": 2.3495294117647058e-06,
|
| 589 |
+
"loss": 0.0008,
|
| 590 |
+
"step": 2350
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 139.06,
|
| 594 |
+
"learning_rate": 2.340705882352941e-06,
|
| 595 |
+
"loss": 0.0036,
|
| 596 |
+
"step": 2375
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 140.53,
|
| 600 |
+
"learning_rate": 2.3318823529411764e-06,
|
| 601 |
+
"loss": 0.0006,
|
| 602 |
+
"step": 2400
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 142.0,
|
| 606 |
+
"learning_rate": 2.3230588235294118e-06,
|
| 607 |
+
"loss": 0.0006,
|
| 608 |
+
"step": 2425
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 143.47,
|
| 612 |
+
"learning_rate": 2.314235294117647e-06,
|
| 613 |
+
"loss": 0.0004,
|
| 614 |
+
"step": 2450
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 144.94,
|
| 618 |
+
"learning_rate": 2.3054117647058824e-06,
|
| 619 |
+
"loss": 0.0018,
|
| 620 |
+
"step": 2475
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 146.41,
|
| 624 |
+
"learning_rate": 2.2965882352941177e-06,
|
| 625 |
+
"loss": 0.0007,
|
| 626 |
+
"step": 2500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 147.88,
|
| 630 |
+
"learning_rate": 2.287764705882353e-06,
|
| 631 |
+
"loss": 0.0005,
|
| 632 |
+
"step": 2525
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 149.35,
|
| 636 |
+
"learning_rate": 2.2789411764705884e-06,
|
| 637 |
+
"loss": 0.0006,
|
| 638 |
+
"step": 2550
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 150.82,
|
| 642 |
+
"learning_rate": 2.2701176470588237e-06,
|
| 643 |
+
"loss": 0.0007,
|
| 644 |
+
"step": 2575
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 152.29,
|
| 648 |
+
"learning_rate": 2.2612941176470586e-06,
|
| 649 |
+
"loss": 0.0007,
|
| 650 |
+
"step": 2600
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 153.76,
|
| 654 |
+
"learning_rate": 2.252470588235294e-06,
|
| 655 |
+
"loss": 0.0008,
|
| 656 |
+
"step": 2625
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 155.24,
|
| 660 |
+
"learning_rate": 2.2436470588235293e-06,
|
| 661 |
+
"loss": 0.0005,
|
| 662 |
+
"step": 2650
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 156.71,
|
| 666 |
+
"learning_rate": 2.2348235294117646e-06,
|
| 667 |
+
"loss": 0.0005,
|
| 668 |
+
"step": 2675
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 158.18,
|
| 672 |
+
"learning_rate": 2.226e-06,
|
| 673 |
+
"loss": 0.0004,
|
| 674 |
+
"step": 2700
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 159.65,
|
| 678 |
+
"learning_rate": 2.2171764705882353e-06,
|
| 679 |
+
"loss": 0.0008,
|
| 680 |
+
"step": 2725
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 161.12,
|
| 684 |
+
"learning_rate": 2.2083529411764706e-06,
|
| 685 |
+
"loss": 0.0005,
|
| 686 |
+
"step": 2750
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 162.59,
|
| 690 |
+
"learning_rate": 2.199529411764706e-06,
|
| 691 |
+
"loss": 0.0005,
|
| 692 |
+
"step": 2775
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 164.06,
|
| 696 |
+
"learning_rate": 2.1907058823529413e-06,
|
| 697 |
+
"loss": 0.0005,
|
| 698 |
+
"step": 2800
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 165.53,
|
| 702 |
+
"learning_rate": 2.1818823529411766e-06,
|
| 703 |
+
"loss": 0.0006,
|
| 704 |
+
"step": 2825
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 167.0,
|
| 708 |
+
"learning_rate": 2.1730588235294115e-06,
|
| 709 |
+
"loss": 0.0006,
|
| 710 |
+
"step": 2850
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 168.47,
|
| 714 |
+
"learning_rate": 2.164235294117647e-06,
|
| 715 |
+
"loss": 0.0006,
|
| 716 |
+
"step": 2875
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 169.94,
|
| 720 |
+
"learning_rate": 2.155411764705882e-06,
|
| 721 |
+
"loss": 0.0008,
|
| 722 |
+
"step": 2900
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 171.41,
|
| 726 |
+
"learning_rate": 2.1465882352941175e-06,
|
| 727 |
+
"loss": 0.0005,
|
| 728 |
+
"step": 2925
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 172.88,
|
| 732 |
+
"learning_rate": 2.137764705882353e-06,
|
| 733 |
+
"loss": 0.0005,
|
| 734 |
+
"step": 2950
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 174.35,
|
| 738 |
+
"learning_rate": 2.128941176470588e-06,
|
| 739 |
+
"loss": 0.0004,
|
| 740 |
+
"step": 2975
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 175.82,
|
| 744 |
+
"learning_rate": 2.1201176470588235e-06,
|
| 745 |
+
"loss": 0.0007,
|
| 746 |
+
"step": 3000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 175.82,
|
| 750 |
+
"eval_loss": 0.53173828125,
|
| 751 |
+
"eval_runtime": 153.6387,
|
| 752 |
+
"eval_samples_per_second": 1.77,
|
| 753 |
+
"eval_steps_per_second": 0.111,
|
| 754 |
+
"eval_wer": 10.11329866270431,
|
| 755 |
+
"step": 3000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 177.29,
|
| 759 |
+
"learning_rate": 2.112e-06,
|
| 760 |
+
"loss": 0.0005,
|
| 761 |
+
"step": 3025
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 178.76,
|
| 765 |
+
"learning_rate": 2.103176470588235e-06,
|
| 766 |
+
"loss": 0.0005,
|
| 767 |
+
"step": 3050
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 180.24,
|
| 771 |
+
"learning_rate": 2.0943529411764705e-06,
|
| 772 |
+
"loss": 0.0007,
|
| 773 |
+
"step": 3075
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 181.71,
|
| 777 |
+
"learning_rate": 2.085529411764706e-06,
|
| 778 |
+
"loss": 0.0004,
|
| 779 |
+
"step": 3100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 183.18,
|
| 783 |
+
"learning_rate": 2.076705882352941e-06,
|
| 784 |
+
"loss": 0.0004,
|
| 785 |
+
"step": 3125
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 184.65,
|
| 789 |
+
"learning_rate": 2.0678823529411765e-06,
|
| 790 |
+
"loss": 0.0005,
|
| 791 |
+
"step": 3150
|
| 792 |
+
},
|
| 793 |
+
{
|
| 794 |
+
"epoch": 186.12,
|
| 795 |
+
"learning_rate": 2.059058823529412e-06,
|
| 796 |
+
"loss": 0.0005,
|
| 797 |
+
"step": 3175
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 187.59,
|
| 801 |
+
"learning_rate": 2.050235294117647e-06,
|
| 802 |
+
"loss": 0.0004,
|
| 803 |
+
"step": 3200
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 189.06,
|
| 807 |
+
"learning_rate": 2.0414117647058825e-06,
|
| 808 |
+
"loss": 0.0005,
|
| 809 |
+
"step": 3225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 190.53,
|
| 813 |
+
"learning_rate": 2.032588235294118e-06,
|
| 814 |
+
"loss": 0.0003,
|
| 815 |
+
"step": 3250
|
| 816 |
+
},
|
| 817 |
+
{
|
| 818 |
+
"epoch": 192.0,
|
| 819 |
+
"learning_rate": 2.0237647058823527e-06,
|
| 820 |
+
"loss": 0.0006,
|
| 821 |
+
"step": 3275
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 193.47,
|
| 825 |
+
"learning_rate": 2.014941176470588e-06,
|
| 826 |
+
"loss": 0.0005,
|
| 827 |
+
"step": 3300
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 194.94,
|
| 831 |
+
"learning_rate": 2.0061176470588234e-06,
|
| 832 |
+
"loss": 0.0004,
|
| 833 |
+
"step": 3325
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 196.41,
|
| 837 |
+
"learning_rate": 1.9972941176470587e-06,
|
| 838 |
+
"loss": 0.0004,
|
| 839 |
+
"step": 3350
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 197.88,
|
| 843 |
+
"learning_rate": 1.988470588235294e-06,
|
| 844 |
+
"loss": 0.0004,
|
| 845 |
+
"step": 3375
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 199.35,
|
| 849 |
+
"learning_rate": 1.9796470588235294e-06,
|
| 850 |
+
"loss": 0.0006,
|
| 851 |
+
"step": 3400
|
| 852 |
+
},
|
| 853 |
+
{
|
| 854 |
+
"epoch": 200.82,
|
| 855 |
+
"learning_rate": 1.9708235294117647e-06,
|
| 856 |
+
"loss": 0.0005,
|
| 857 |
+
"step": 3425
|
| 858 |
+
},
|
| 859 |
+
{
|
| 860 |
+
"epoch": 202.29,
|
| 861 |
+
"learning_rate": 1.962e-06,
|
| 862 |
+
"loss": 0.0004,
|
| 863 |
+
"step": 3450
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 203.76,
|
| 867 |
+
"learning_rate": 1.9531764705882353e-06,
|
| 868 |
+
"loss": 0.0005,
|
| 869 |
+
"step": 3475
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 205.24,
|
| 873 |
+
"learning_rate": 1.9443529411764707e-06,
|
| 874 |
+
"loss": 0.0006,
|
| 875 |
+
"step": 3500
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 206.71,
|
| 879 |
+
"learning_rate": 1.9355294117647056e-06,
|
| 880 |
+
"loss": 0.0003,
|
| 881 |
+
"step": 3525
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 208.18,
|
| 885 |
+
"learning_rate": 1.926705882352941e-06,
|
| 886 |
+
"loss": 0.0004,
|
| 887 |
+
"step": 3550
|
| 888 |
+
},
|
| 889 |
+
{
|
| 890 |
+
"epoch": 209.65,
|
| 891 |
+
"learning_rate": 1.9178823529411762e-06,
|
| 892 |
+
"loss": 0.0005,
|
| 893 |
+
"step": 3575
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 211.12,
|
| 897 |
+
"learning_rate": 1.9090588235294116e-06,
|
| 898 |
+
"loss": 0.0004,
|
| 899 |
+
"step": 3600
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 212.59,
|
| 903 |
+
"learning_rate": 1.9002352941176471e-06,
|
| 904 |
+
"loss": 0.0004,
|
| 905 |
+
"step": 3625
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 214.06,
|
| 909 |
+
"learning_rate": 1.8914117647058824e-06,
|
| 910 |
+
"loss": 0.0003,
|
| 911 |
+
"step": 3650
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 215.53,
|
| 915 |
+
"learning_rate": 1.8825882352941178e-06,
|
| 916 |
+
"loss": 0.0004,
|
| 917 |
+
"step": 3675
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 217.0,
|
| 921 |
+
"learning_rate": 1.873764705882353e-06,
|
| 922 |
+
"loss": 0.0005,
|
| 923 |
+
"step": 3700
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 218.47,
|
| 927 |
+
"learning_rate": 1.8649411764705884e-06,
|
| 928 |
+
"loss": 0.0004,
|
| 929 |
+
"step": 3725
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 219.94,
|
| 933 |
+
"learning_rate": 1.8561176470588238e-06,
|
| 934 |
+
"loss": 0.0005,
|
| 935 |
+
"step": 3750
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 221.41,
|
| 939 |
+
"learning_rate": 1.847294117647059e-06,
|
| 940 |
+
"loss": 0.0003,
|
| 941 |
+
"step": 3775
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 222.88,
|
| 945 |
+
"learning_rate": 1.838470588235294e-06,
|
| 946 |
+
"loss": 0.0004,
|
| 947 |
+
"step": 3800
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 224.35,
|
| 951 |
+
"learning_rate": 1.8296470588235293e-06,
|
| 952 |
+
"loss": 0.0003,
|
| 953 |
+
"step": 3825
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 225.82,
|
| 957 |
+
"learning_rate": 1.8208235294117646e-06,
|
| 958 |
+
"loss": 0.0003,
|
| 959 |
+
"step": 3850
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 227.29,
|
| 963 |
+
"learning_rate": 1.812e-06,
|
| 964 |
+
"loss": 0.0003,
|
| 965 |
+
"step": 3875
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 228.76,
|
| 969 |
+
"learning_rate": 1.8031764705882353e-06,
|
| 970 |
+
"loss": 0.0005,
|
| 971 |
+
"step": 3900
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 230.24,
|
| 975 |
+
"learning_rate": 1.7943529411764706e-06,
|
| 976 |
+
"loss": 0.0004,
|
| 977 |
+
"step": 3925
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 231.71,
|
| 981 |
+
"learning_rate": 1.785529411764706e-06,
|
| 982 |
+
"loss": 0.0004,
|
| 983 |
+
"step": 3950
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 233.18,
|
| 987 |
+
"learning_rate": 1.7767058823529413e-06,
|
| 988 |
+
"loss": 0.0003,
|
| 989 |
+
"step": 3975
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 234.65,
|
| 993 |
+
"learning_rate": 1.7678823529411766e-06,
|
| 994 |
+
"loss": 0.0004,
|
| 995 |
+
"step": 4000
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 234.65,
|
| 999 |
+
"eval_loss": 0.53955078125,
|
| 1000 |
+
"eval_runtime": 161.7321,
|
| 1001 |
+
"eval_samples_per_second": 1.682,
|
| 1002 |
+
"eval_steps_per_second": 0.105,
|
| 1003 |
+
"eval_wer": 10.12258543833581,
|
| 1004 |
+
"step": 4000
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 236.12,
|
| 1008 |
+
"learning_rate": 1.759764705882353e-06,
|
| 1009 |
+
"loss": 0.0003,
|
| 1010 |
+
"step": 4025
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 237.59,
|
| 1014 |
+
"learning_rate": 1.7509411764705883e-06,
|
| 1015 |
+
"loss": 0.0003,
|
| 1016 |
+
"step": 4050
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 239.06,
|
| 1020 |
+
"learning_rate": 1.7421176470588237e-06,
|
| 1021 |
+
"loss": 0.0002,
|
| 1022 |
+
"step": 4075
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 240.53,
|
| 1026 |
+
"learning_rate": 1.733294117647059e-06,
|
| 1027 |
+
"loss": 0.0004,
|
| 1028 |
+
"step": 4100
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 242.0,
|
| 1032 |
+
"learning_rate": 1.7244705882352943e-06,
|
| 1033 |
+
"loss": 0.0005,
|
| 1034 |
+
"step": 4125
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 243.47,
|
| 1038 |
+
"learning_rate": 1.7156470588235296e-06,
|
| 1039 |
+
"loss": 0.0003,
|
| 1040 |
+
"step": 4150
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 244.94,
|
| 1044 |
+
"learning_rate": 1.706823529411765e-06,
|
| 1045 |
+
"loss": 0.0003,
|
| 1046 |
+
"step": 4175
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 246.41,
|
| 1050 |
+
"learning_rate": 1.6979999999999999e-06,
|
| 1051 |
+
"loss": 0.0004,
|
| 1052 |
+
"step": 4200
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 247.88,
|
| 1056 |
+
"learning_rate": 1.6891764705882352e-06,
|
| 1057 |
+
"loss": 0.0003,
|
| 1058 |
+
"step": 4225
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 249.35,
|
| 1062 |
+
"learning_rate": 1.6803529411764705e-06,
|
| 1063 |
+
"loss": 0.0003,
|
| 1064 |
+
"step": 4250
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 250.82,
|
| 1068 |
+
"learning_rate": 1.6715294117647059e-06,
|
| 1069 |
+
"loss": 0.0003,
|
| 1070 |
+
"step": 4275
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 252.29,
|
| 1074 |
+
"learning_rate": 1.6627058823529412e-06,
|
| 1075 |
+
"loss": 0.0004,
|
| 1076 |
+
"step": 4300
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 253.76,
|
| 1080 |
+
"learning_rate": 1.6538823529411765e-06,
|
| 1081 |
+
"loss": 0.0004,
|
| 1082 |
+
"step": 4325
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 255.24,
|
| 1086 |
+
"learning_rate": 1.6450588235294119e-06,
|
| 1087 |
+
"loss": 0.0003,
|
| 1088 |
+
"step": 4350
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 256.71,
|
| 1092 |
+
"learning_rate": 1.6362352941176472e-06,
|
| 1093 |
+
"loss": 0.0003,
|
| 1094 |
+
"step": 4375
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 258.18,
|
| 1098 |
+
"learning_rate": 1.6274117647058825e-06,
|
| 1099 |
+
"loss": 0.0003,
|
| 1100 |
+
"step": 4400
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 259.65,
|
| 1104 |
+
"learning_rate": 1.6185882352941178e-06,
|
| 1105 |
+
"loss": 0.0007,
|
| 1106 |
+
"step": 4425
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 261.12,
|
| 1110 |
+
"learning_rate": 1.6097647058823532e-06,
|
| 1111 |
+
"loss": 0.0003,
|
| 1112 |
+
"step": 4450
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 262.59,
|
| 1116 |
+
"learning_rate": 1.600941176470588e-06,
|
| 1117 |
+
"loss": 0.0004,
|
| 1118 |
+
"step": 4475
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 264.06,
|
| 1122 |
+
"learning_rate": 1.5921176470588234e-06,
|
| 1123 |
+
"loss": 0.0003,
|
| 1124 |
+
"step": 4500
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 265.53,
|
| 1128 |
+
"learning_rate": 1.5832941176470587e-06,
|
| 1129 |
+
"loss": 0.0004,
|
| 1130 |
+
"step": 4525
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 267.0,
|
| 1134 |
+
"learning_rate": 1.574470588235294e-06,
|
| 1135 |
+
"loss": 0.0002,
|
| 1136 |
+
"step": 4550
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 268.47,
|
| 1140 |
+
"learning_rate": 1.5656470588235294e-06,
|
| 1141 |
+
"loss": 0.0004,
|
| 1142 |
+
"step": 4575
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 269.94,
|
| 1146 |
+
"learning_rate": 1.5568235294117647e-06,
|
| 1147 |
+
"loss": 0.0003,
|
| 1148 |
+
"step": 4600
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 271.41,
|
| 1152 |
+
"learning_rate": 1.548e-06,
|
| 1153 |
+
"loss": 0.0002,
|
| 1154 |
+
"step": 4625
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 272.88,
|
| 1158 |
+
"learning_rate": 1.5391764705882354e-06,
|
| 1159 |
+
"loss": 0.0002,
|
| 1160 |
+
"step": 4650
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 274.35,
|
| 1164 |
+
"learning_rate": 1.5303529411764707e-06,
|
| 1165 |
+
"loss": 0.0003,
|
| 1166 |
+
"step": 4675
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 275.82,
|
| 1170 |
+
"learning_rate": 1.521529411764706e-06,
|
| 1171 |
+
"loss": 0.0003,
|
| 1172 |
+
"step": 4700
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 277.29,
|
| 1176 |
+
"learning_rate": 1.5127058823529411e-06,
|
| 1177 |
+
"loss": 0.0003,
|
| 1178 |
+
"step": 4725
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 278.76,
|
| 1182 |
+
"learning_rate": 1.5038823529411765e-06,
|
| 1183 |
+
"loss": 0.0002,
|
| 1184 |
+
"step": 4750
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 280.24,
|
| 1188 |
+
"learning_rate": 1.4950588235294118e-06,
|
| 1189 |
+
"loss": 0.0002,
|
| 1190 |
+
"step": 4775
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 281.71,
|
| 1194 |
+
"learning_rate": 1.486235294117647e-06,
|
| 1195 |
+
"loss": 0.0003,
|
| 1196 |
+
"step": 4800
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 283.18,
|
| 1200 |
+
"learning_rate": 1.4774117647058823e-06,
|
| 1201 |
+
"loss": 0.0003,
|
| 1202 |
+
"step": 4825
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 284.65,
|
| 1206 |
+
"learning_rate": 1.4685882352941176e-06,
|
| 1207 |
+
"loss": 0.0004,
|
| 1208 |
+
"step": 4850
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 286.12,
|
| 1212 |
+
"learning_rate": 1.459764705882353e-06,
|
| 1213 |
+
"loss": 0.0002,
|
| 1214 |
+
"step": 4875
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 287.59,
|
| 1218 |
+
"learning_rate": 1.4509411764705882e-06,
|
| 1219 |
+
"loss": 0.0002,
|
| 1220 |
+
"step": 4900
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 289.06,
|
| 1224 |
+
"learning_rate": 1.4421176470588236e-06,
|
| 1225 |
+
"loss": 0.0012,
|
| 1226 |
+
"step": 4925
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 290.53,
|
| 1230 |
+
"learning_rate": 1.433294117647059e-06,
|
| 1231 |
+
"loss": 0.0003,
|
| 1232 |
+
"step": 4950
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 292.0,
|
| 1236 |
+
"learning_rate": 1.4244705882352942e-06,
|
| 1237 |
+
"loss": 0.0002,
|
| 1238 |
+
"step": 4975
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 293.47,
|
| 1242 |
+
"learning_rate": 1.4156470588235296e-06,
|
| 1243 |
+
"loss": 0.0004,
|
| 1244 |
+
"step": 5000
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 293.47,
|
| 1248 |
+
"eval_loss": 0.55322265625,
|
| 1249 |
+
"eval_runtime": 164.7504,
|
| 1250 |
+
"eval_samples_per_second": 1.651,
|
| 1251 |
+
"eval_steps_per_second": 0.103,
|
| 1252 |
+
"eval_wer": 10.104011887072808,
|
| 1253 |
+
"step": 5000
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 294.94,
|
| 1257 |
+
"learning_rate": 1.407529411764706e-06,
|
| 1258 |
+
"loss": 0.0002,
|
| 1259 |
+
"step": 5025
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 296.41,
|
| 1263 |
+
"learning_rate": 1.3987058823529413e-06,
|
| 1264 |
+
"loss": 0.0004,
|
| 1265 |
+
"step": 5050
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 297.88,
|
| 1269 |
+
"learning_rate": 1.3898823529411764e-06,
|
| 1270 |
+
"loss": 0.0003,
|
| 1271 |
+
"step": 5075
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 299.35,
|
| 1275 |
+
"learning_rate": 1.3810588235294117e-06,
|
| 1276 |
+
"loss": 0.0003,
|
| 1277 |
+
"step": 5100
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 300.82,
|
| 1281 |
+
"learning_rate": 1.372235294117647e-06,
|
| 1282 |
+
"loss": 0.0004,
|
| 1283 |
+
"step": 5125
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 302.29,
|
| 1287 |
+
"learning_rate": 1.3634117647058824e-06,
|
| 1288 |
+
"loss": 0.0004,
|
| 1289 |
+
"step": 5150
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 303.76,
|
| 1293 |
+
"learning_rate": 1.3545882352941177e-06,
|
| 1294 |
+
"loss": 0.0002,
|
| 1295 |
+
"step": 5175
|
| 1296 |
+
},
|
| 1297 |
+
{
|
| 1298 |
+
"epoch": 305.24,
|
| 1299 |
+
"learning_rate": 1.3457647058823528e-06,
|
| 1300 |
+
"loss": 0.0003,
|
| 1301 |
+
"step": 5200
|
| 1302 |
+
},
|
| 1303 |
+
{
|
| 1304 |
+
"epoch": 306.71,
|
| 1305 |
+
"learning_rate": 1.3369411764705881e-06,
|
| 1306 |
+
"loss": 0.0005,
|
| 1307 |
+
"step": 5225
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 308.18,
|
| 1311 |
+
"learning_rate": 1.3281176470588235e-06,
|
| 1312 |
+
"loss": 0.0002,
|
| 1313 |
+
"step": 5250
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 309.65,
|
| 1317 |
+
"learning_rate": 1.3192941176470588e-06,
|
| 1318 |
+
"loss": 0.0002,
|
| 1319 |
+
"step": 5275
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 311.12,
|
| 1323 |
+
"learning_rate": 1.3104705882352941e-06,
|
| 1324 |
+
"loss": 0.0002,
|
| 1325 |
+
"step": 5300
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 312.59,
|
| 1329 |
+
"learning_rate": 1.3016470588235295e-06,
|
| 1330 |
+
"loss": 0.0003,
|
| 1331 |
+
"step": 5325
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 314.06,
|
| 1335 |
+
"learning_rate": 1.2928235294117648e-06,
|
| 1336 |
+
"loss": 0.0003,
|
| 1337 |
+
"step": 5350
|
| 1338 |
+
},
|
| 1339 |
+
{
|
| 1340 |
+
"epoch": 315.53,
|
| 1341 |
+
"learning_rate": 1.284e-06,
|
| 1342 |
+
"loss": 0.0003,
|
| 1343 |
+
"step": 5375
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 317.0,
|
| 1347 |
+
"learning_rate": 1.2751764705882352e-06,
|
| 1348 |
+
"loss": 0.0003,
|
| 1349 |
+
"step": 5400
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 318.47,
|
| 1353 |
+
"learning_rate": 1.2663529411764706e-06,
|
| 1354 |
+
"loss": 0.0003,
|
| 1355 |
+
"step": 5425
|
| 1356 |
+
},
|
| 1357 |
+
{
|
| 1358 |
+
"epoch": 319.94,
|
| 1359 |
+
"learning_rate": 1.2575294117647059e-06,
|
| 1360 |
+
"loss": 0.0002,
|
| 1361 |
+
"step": 5450
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 321.41,
|
| 1365 |
+
"learning_rate": 1.2487058823529412e-06,
|
| 1366 |
+
"loss": 0.0002,
|
| 1367 |
+
"step": 5475
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 322.88,
|
| 1371 |
+
"learning_rate": 1.2398823529411765e-06,
|
| 1372 |
+
"loss": 0.0002,
|
| 1373 |
+
"step": 5500
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 324.35,
|
| 1377 |
+
"learning_rate": 1.2310588235294119e-06,
|
| 1378 |
+
"loss": 0.0001,
|
| 1379 |
+
"step": 5525
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 325.82,
|
| 1383 |
+
"learning_rate": 1.2222352941176472e-06,
|
| 1384 |
+
"loss": 0.0002,
|
| 1385 |
+
"step": 5550
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 327.29,
|
| 1389 |
+
"learning_rate": 1.2134117647058825e-06,
|
| 1390 |
+
"loss": 0.0002,
|
| 1391 |
+
"step": 5575
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 328.76,
|
| 1395 |
+
"learning_rate": 1.2045882352941177e-06,
|
| 1396 |
+
"loss": 0.0003,
|
| 1397 |
+
"step": 5600
|
| 1398 |
+
},
|
| 1399 |
+
{
|
| 1400 |
+
"epoch": 330.24,
|
| 1401 |
+
"learning_rate": 1.195764705882353e-06,
|
| 1402 |
+
"loss": 0.0003,
|
| 1403 |
+
"step": 5625
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 331.71,
|
| 1407 |
+
"learning_rate": 1.1869411764705883e-06,
|
| 1408 |
+
"loss": 0.0003,
|
| 1409 |
+
"step": 5650
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 333.18,
|
| 1413 |
+
"learning_rate": 1.1781176470588236e-06,
|
| 1414 |
+
"loss": 0.0003,
|
| 1415 |
+
"step": 5675
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 334.65,
|
| 1419 |
+
"learning_rate": 1.169294117647059e-06,
|
| 1420 |
+
"loss": 0.0004,
|
| 1421 |
+
"step": 5700
|
| 1422 |
+
},
|
| 1423 |
+
{
|
| 1424 |
+
"epoch": 336.12,
|
| 1425 |
+
"learning_rate": 1.1604705882352943e-06,
|
| 1426 |
+
"loss": 0.0002,
|
| 1427 |
+
"step": 5725
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 337.59,
|
| 1431 |
+
"learning_rate": 1.1516470588235294e-06,
|
| 1432 |
+
"loss": 0.0003,
|
| 1433 |
+
"step": 5750
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 339.06,
|
| 1437 |
+
"learning_rate": 1.1428235294117647e-06,
|
| 1438 |
+
"loss": 0.0002,
|
| 1439 |
+
"step": 5775
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 340.53,
|
| 1443 |
+
"learning_rate": 1.134e-06,
|
| 1444 |
+
"loss": 0.0002,
|
| 1445 |
+
"step": 5800
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 342.0,
|
| 1449 |
+
"learning_rate": 1.1251764705882354e-06,
|
| 1450 |
+
"loss": 0.0003,
|
| 1451 |
+
"step": 5825
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 343.47,
|
| 1455 |
+
"learning_rate": 1.1163529411764707e-06,
|
| 1456 |
+
"loss": 0.0002,
|
| 1457 |
+
"step": 5850
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 344.94,
|
| 1461 |
+
"learning_rate": 1.1075294117647058e-06,
|
| 1462 |
+
"loss": 0.0002,
|
| 1463 |
+
"step": 5875
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 346.41,
|
| 1467 |
+
"learning_rate": 1.0987058823529412e-06,
|
| 1468 |
+
"loss": 0.0002,
|
| 1469 |
+
"step": 5900
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 347.88,
|
| 1473 |
+
"learning_rate": 1.0898823529411765e-06,
|
| 1474 |
+
"loss": 0.0002,
|
| 1475 |
+
"step": 5925
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 349.35,
|
| 1479 |
+
"learning_rate": 1.0810588235294118e-06,
|
| 1480 |
+
"loss": 0.0003,
|
| 1481 |
+
"step": 5950
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 350.82,
|
| 1485 |
+
"learning_rate": 1.0722352941176472e-06,
|
| 1486 |
+
"loss": 0.0002,
|
| 1487 |
+
"step": 5975
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 352.29,
|
| 1491 |
+
"learning_rate": 1.0634117647058823e-06,
|
| 1492 |
+
"loss": 0.0013,
|
| 1493 |
+
"step": 6000
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 352.29,
|
| 1497 |
+
"eval_loss": 0.564453125,
|
| 1498 |
+
"eval_runtime": 167.4651,
|
| 1499 |
+
"eval_samples_per_second": 1.624,
|
| 1500 |
+
"eval_steps_per_second": 0.102,
|
| 1501 |
+
"eval_wer": 10.085438335809807,
|
| 1502 |
+
"step": 6000
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 353.76,
|
| 1506 |
+
"learning_rate": 1.0552941176470589e-06,
|
| 1507 |
+
"loss": 0.0002,
|
| 1508 |
+
"step": 6025
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 355.24,
|
| 1512 |
+
"learning_rate": 1.0464705882352942e-06,
|
| 1513 |
+
"loss": 0.0002,
|
| 1514 |
+
"step": 6050
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 356.71,
|
| 1518 |
+
"learning_rate": 1.0376470588235295e-06,
|
| 1519 |
+
"loss": 0.0002,
|
| 1520 |
+
"step": 6075
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 358.18,
|
| 1524 |
+
"learning_rate": 1.0288235294117649e-06,
|
| 1525 |
+
"loss": 0.0002,
|
| 1526 |
+
"step": 6100
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 359.65,
|
| 1530 |
+
"learning_rate": 1.0200000000000002e-06,
|
| 1531 |
+
"loss": 0.0002,
|
| 1532 |
+
"step": 6125
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 361.12,
|
| 1536 |
+
"learning_rate": 1.0111764705882353e-06,
|
| 1537 |
+
"loss": 0.0002,
|
| 1538 |
+
"step": 6150
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 362.59,
|
| 1542 |
+
"learning_rate": 1.0023529411764706e-06,
|
| 1543 |
+
"loss": 0.0002,
|
| 1544 |
+
"step": 6175
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 364.06,
|
| 1548 |
+
"learning_rate": 9.93529411764706e-07,
|
| 1549 |
+
"loss": 0.0004,
|
| 1550 |
+
"step": 6200
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 365.53,
|
| 1554 |
+
"learning_rate": 9.847058823529413e-07,
|
| 1555 |
+
"loss": 0.0002,
|
| 1556 |
+
"step": 6225
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 367.0,
|
| 1560 |
+
"learning_rate": 9.758823529411766e-07,
|
| 1561 |
+
"loss": 0.0002,
|
| 1562 |
+
"step": 6250
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 368.47,
|
| 1566 |
+
"learning_rate": 9.670588235294117e-07,
|
| 1567 |
+
"loss": 0.0002,
|
| 1568 |
+
"step": 6275
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 369.94,
|
| 1572 |
+
"learning_rate": 9.58235294117647e-07,
|
| 1573 |
+
"loss": 0.0002,
|
| 1574 |
+
"step": 6300
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 371.41,
|
| 1578 |
+
"learning_rate": 9.494117647058824e-07,
|
| 1579 |
+
"loss": 0.0002,
|
| 1580 |
+
"step": 6325
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 372.88,
|
| 1584 |
+
"learning_rate": 9.405882352941177e-07,
|
| 1585 |
+
"loss": 0.0003,
|
| 1586 |
+
"step": 6350
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 374.35,
|
| 1590 |
+
"learning_rate": 9.31764705882353e-07,
|
| 1591 |
+
"loss": 0.0002,
|
| 1592 |
+
"step": 6375
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 375.82,
|
| 1596 |
+
"learning_rate": 9.229411764705882e-07,
|
| 1597 |
+
"loss": 0.0002,
|
| 1598 |
+
"step": 6400
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 377.29,
|
| 1602 |
+
"learning_rate": 9.141176470588235e-07,
|
| 1603 |
+
"loss": 0.0002,
|
| 1604 |
+
"step": 6425
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 378.76,
|
| 1608 |
+
"learning_rate": 9.052941176470588e-07,
|
| 1609 |
+
"loss": 0.0002,
|
| 1610 |
+
"step": 6450
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 380.24,
|
| 1614 |
+
"learning_rate": 8.964705882352942e-07,
|
| 1615 |
+
"loss": 0.0002,
|
| 1616 |
+
"step": 6475
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 381.71,
|
| 1620 |
+
"learning_rate": 8.876470588235295e-07,
|
| 1621 |
+
"loss": 0.0002,
|
| 1622 |
+
"step": 6500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 383.18,
|
| 1626 |
+
"learning_rate": 8.788235294117648e-07,
|
| 1627 |
+
"loss": 0.0002,
|
| 1628 |
+
"step": 6525
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 384.65,
|
| 1632 |
+
"learning_rate": 8.699999999999999e-07,
|
| 1633 |
+
"loss": 0.0003,
|
| 1634 |
+
"step": 6550
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 386.12,
|
| 1638 |
+
"learning_rate": 8.611764705882353e-07,
|
| 1639 |
+
"loss": 0.0003,
|
| 1640 |
+
"step": 6575
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 387.59,
|
| 1644 |
+
"learning_rate": 8.523529411764706e-07,
|
| 1645 |
+
"loss": 0.0002,
|
| 1646 |
+
"step": 6600
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 389.06,
|
| 1650 |
+
"learning_rate": 8.435294117647059e-07,
|
| 1651 |
+
"loss": 0.0002,
|
| 1652 |
+
"step": 6625
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 390.53,
|
| 1656 |
+
"learning_rate": 8.347058823529412e-07,
|
| 1657 |
+
"loss": 0.0001,
|
| 1658 |
+
"step": 6650
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 392.0,
|
| 1662 |
+
"learning_rate": 8.258823529411765e-07,
|
| 1663 |
+
"loss": 0.0002,
|
| 1664 |
+
"step": 6675
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 393.47,
|
| 1668 |
+
"learning_rate": 8.170588235294118e-07,
|
| 1669 |
+
"loss": 0.0001,
|
| 1670 |
+
"step": 6700
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 394.94,
|
| 1674 |
+
"learning_rate": 8.082352941176471e-07,
|
| 1675 |
+
"loss": 0.0001,
|
| 1676 |
+
"step": 6725
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 396.41,
|
| 1680 |
+
"learning_rate": 7.994117647058823e-07,
|
| 1681 |
+
"loss": 0.0002,
|
| 1682 |
+
"step": 6750
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 397.88,
|
| 1686 |
+
"learning_rate": 7.905882352941177e-07,
|
| 1687 |
+
"loss": 0.0003,
|
| 1688 |
+
"step": 6775
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 399.35,
|
| 1692 |
+
"learning_rate": 7.817647058823529e-07,
|
| 1693 |
+
"loss": 0.0004,
|
| 1694 |
+
"step": 6800
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 400.82,
|
| 1698 |
+
"learning_rate": 7.729411764705882e-07,
|
| 1699 |
+
"loss": 0.0003,
|
| 1700 |
+
"step": 6825
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 402.29,
|
| 1704 |
+
"learning_rate": 7.641176470588236e-07,
|
| 1705 |
+
"loss": 0.0002,
|
| 1706 |
+
"step": 6850
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 403.76,
|
| 1710 |
+
"learning_rate": 7.552941176470589e-07,
|
| 1711 |
+
"loss": 0.0002,
|
| 1712 |
+
"step": 6875
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 405.24,
|
| 1716 |
+
"learning_rate": 7.464705882352941e-07,
|
| 1717 |
+
"loss": 0.0002,
|
| 1718 |
+
"step": 6900
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 406.71,
|
| 1722 |
+
"learning_rate": 7.376470588235294e-07,
|
| 1723 |
+
"loss": 0.0002,
|
| 1724 |
+
"step": 6925
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 408.18,
|
| 1728 |
+
"learning_rate": 7.288235294117648e-07,
|
| 1729 |
+
"loss": 0.0001,
|
| 1730 |
+
"step": 6950
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 409.65,
|
| 1734 |
+
"learning_rate": 7.2e-07,
|
| 1735 |
+
"loss": 0.0001,
|
| 1736 |
+
"step": 6975
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 411.12,
|
| 1740 |
+
"learning_rate": 7.111764705882353e-07,
|
| 1741 |
+
"loss": 0.0002,
|
| 1742 |
+
"step": 7000
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 411.12,
|
| 1746 |
+
"eval_loss": 0.56689453125,
|
| 1747 |
+
"eval_runtime": 168.4232,
|
| 1748 |
+
"eval_samples_per_second": 1.615,
|
| 1749 |
+
"eval_steps_per_second": 0.101,
|
| 1750 |
+
"eval_wer": 10.11329866270431,
|
| 1751 |
+
"step": 7000
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 412.59,
|
| 1755 |
+
"learning_rate": 7.030588235294118e-07,
|
| 1756 |
+
"loss": 0.0002,
|
| 1757 |
+
"step": 7025
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 414.06,
|
| 1761 |
+
"learning_rate": 6.94235294117647e-07,
|
| 1762 |
+
"loss": 0.0003,
|
| 1763 |
+
"step": 7050
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 415.53,
|
| 1767 |
+
"learning_rate": 6.854117647058824e-07,
|
| 1768 |
+
"loss": 0.0002,
|
| 1769 |
+
"step": 7075
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 417.0,
|
| 1773 |
+
"learning_rate": 6.765882352941177e-07,
|
| 1774 |
+
"loss": 0.0001,
|
| 1775 |
+
"step": 7100
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 418.47,
|
| 1779 |
+
"learning_rate": 6.677647058823529e-07,
|
| 1780 |
+
"loss": 0.0003,
|
| 1781 |
+
"step": 7125
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 419.94,
|
| 1785 |
+
"learning_rate": 6.589411764705882e-07,
|
| 1786 |
+
"loss": 0.0002,
|
| 1787 |
+
"step": 7150
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 421.41,
|
| 1791 |
+
"learning_rate": 6.501176470588235e-07,
|
| 1792 |
+
"loss": 0.0003,
|
| 1793 |
+
"step": 7175
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 422.88,
|
| 1797 |
+
"learning_rate": 6.412941176470588e-07,
|
| 1798 |
+
"loss": 0.0003,
|
| 1799 |
+
"step": 7200
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 424.35,
|
| 1803 |
+
"learning_rate": 6.324705882352941e-07,
|
| 1804 |
+
"loss": 0.0003,
|
| 1805 |
+
"step": 7225
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 425.82,
|
| 1809 |
+
"learning_rate": 6.236470588235294e-07,
|
| 1810 |
+
"loss": 0.0002,
|
| 1811 |
+
"step": 7250
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 427.29,
|
| 1815 |
+
"learning_rate": 6.148235294117648e-07,
|
| 1816 |
+
"loss": 0.0002,
|
| 1817 |
+
"step": 7275
|
| 1818 |
+
},
|
| 1819 |
+
{
|
| 1820 |
+
"epoch": 428.76,
|
| 1821 |
+
"learning_rate": 6.060000000000001e-07,
|
| 1822 |
+
"loss": 0.0003,
|
| 1823 |
+
"step": 7300
|
| 1824 |
+
},
|
| 1825 |
+
{
|
| 1826 |
+
"epoch": 430.24,
|
| 1827 |
+
"learning_rate": 5.971764705882353e-07,
|
| 1828 |
+
"loss": 0.0002,
|
| 1829 |
+
"step": 7325
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 431.71,
|
| 1833 |
+
"learning_rate": 5.883529411764707e-07,
|
| 1834 |
+
"loss": 0.0003,
|
| 1835 |
+
"step": 7350
|
| 1836 |
+
},
|
| 1837 |
+
{
|
| 1838 |
+
"epoch": 433.18,
|
| 1839 |
+
"learning_rate": 5.795294117647059e-07,
|
| 1840 |
+
"loss": 0.0004,
|
| 1841 |
+
"step": 7375
|
| 1842 |
+
},
|
| 1843 |
+
{
|
| 1844 |
+
"epoch": 434.65,
|
| 1845 |
+
"learning_rate": 5.707058823529412e-07,
|
| 1846 |
+
"loss": 0.0002,
|
| 1847 |
+
"step": 7400
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 436.12,
|
| 1851 |
+
"learning_rate": 5.618823529411765e-07,
|
| 1852 |
+
"loss": 0.0001,
|
| 1853 |
+
"step": 7425
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 437.59,
|
| 1857 |
+
"learning_rate": 5.530588235294118e-07,
|
| 1858 |
+
"loss": 0.0002,
|
| 1859 |
+
"step": 7450
|
| 1860 |
+
},
|
| 1861 |
+
{
|
| 1862 |
+
"epoch": 439.06,
|
| 1863 |
+
"learning_rate": 5.442352941176471e-07,
|
| 1864 |
+
"loss": 0.0002,
|
| 1865 |
+
"step": 7475
|
| 1866 |
+
},
|
| 1867 |
+
{
|
| 1868 |
+
"epoch": 440.53,
|
| 1869 |
+
"learning_rate": 5.354117647058824e-07,
|
| 1870 |
+
"loss": 0.0001,
|
| 1871 |
+
"step": 7500
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 442.0,
|
| 1875 |
+
"learning_rate": 5.265882352941176e-07,
|
| 1876 |
+
"loss": 0.0003,
|
| 1877 |
+
"step": 7525
|
| 1878 |
+
},
|
| 1879 |
+
{
|
| 1880 |
+
"epoch": 443.47,
|
| 1881 |
+
"learning_rate": 5.17764705882353e-07,
|
| 1882 |
+
"loss": 0.0008,
|
| 1883 |
+
"step": 7550
|
| 1884 |
+
},
|
| 1885 |
+
{
|
| 1886 |
+
"epoch": 444.94,
|
| 1887 |
+
"learning_rate": 5.089411764705882e-07,
|
| 1888 |
+
"loss": 0.0002,
|
| 1889 |
+
"step": 7575
|
| 1890 |
+
},
|
| 1891 |
+
{
|
| 1892 |
+
"epoch": 446.41,
|
| 1893 |
+
"learning_rate": 5.001176470588235e-07,
|
| 1894 |
+
"loss": 0.0001,
|
| 1895 |
+
"step": 7600
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 447.88,
|
| 1899 |
+
"learning_rate": 4.912941176470589e-07,
|
| 1900 |
+
"loss": 0.0001,
|
| 1901 |
+
"step": 7625
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 449.35,
|
| 1905 |
+
"learning_rate": 4.824705882352941e-07,
|
| 1906 |
+
"loss": 0.0001,
|
| 1907 |
+
"step": 7650
|
| 1908 |
+
},
|
| 1909 |
+
{
|
| 1910 |
+
"epoch": 450.82,
|
| 1911 |
+
"learning_rate": 4.7364705882352946e-07,
|
| 1912 |
+
"loss": 0.0001,
|
| 1913 |
+
"step": 7675
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 452.29,
|
| 1917 |
+
"learning_rate": 4.648235294117648e-07,
|
| 1918 |
+
"loss": 0.0001,
|
| 1919 |
+
"step": 7700
|
| 1920 |
+
},
|
| 1921 |
+
{
|
| 1922 |
+
"epoch": 453.76,
|
| 1923 |
+
"learning_rate": 4.56e-07,
|
| 1924 |
+
"loss": 0.0003,
|
| 1925 |
+
"step": 7725
|
| 1926 |
+
},
|
| 1927 |
+
{
|
| 1928 |
+
"epoch": 455.24,
|
| 1929 |
+
"learning_rate": 4.4717647058823534e-07,
|
| 1930 |
+
"loss": 0.0001,
|
| 1931 |
+
"step": 7750
|
| 1932 |
+
},
|
| 1933 |
+
{
|
| 1934 |
+
"epoch": 456.71,
|
| 1935 |
+
"learning_rate": 4.3835294117647056e-07,
|
| 1936 |
+
"loss": 0.0003,
|
| 1937 |
+
"step": 7775
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 458.18,
|
| 1941 |
+
"learning_rate": 4.295294117647059e-07,
|
| 1942 |
+
"loss": 0.0001,
|
| 1943 |
+
"step": 7800
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 459.65,
|
| 1947 |
+
"learning_rate": 4.207058823529412e-07,
|
| 1948 |
+
"loss": 0.0002,
|
| 1949 |
+
"step": 7825
|
| 1950 |
+
},
|
| 1951 |
+
{
|
| 1952 |
+
"epoch": 461.12,
|
| 1953 |
+
"learning_rate": 4.1188235294117644e-07,
|
| 1954 |
+
"loss": 0.0002,
|
| 1955 |
+
"step": 7850
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 462.59,
|
| 1959 |
+
"learning_rate": 4.0305882352941177e-07,
|
| 1960 |
+
"loss": 0.0001,
|
| 1961 |
+
"step": 7875
|
| 1962 |
+
},
|
| 1963 |
+
{
|
| 1964 |
+
"epoch": 464.06,
|
| 1965 |
+
"learning_rate": 3.942352941176471e-07,
|
| 1966 |
+
"loss": 0.0002,
|
| 1967 |
+
"step": 7900
|
| 1968 |
+
},
|
| 1969 |
+
{
|
| 1970 |
+
"epoch": 465.53,
|
| 1971 |
+
"learning_rate": 3.8541176470588237e-07,
|
| 1972 |
+
"loss": 0.0002,
|
| 1973 |
+
"step": 7925
|
| 1974 |
+
},
|
| 1975 |
+
{
|
| 1976 |
+
"epoch": 467.0,
|
| 1977 |
+
"learning_rate": 3.765882352941177e-07,
|
| 1978 |
+
"loss": 0.0002,
|
| 1979 |
+
"step": 7950
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 468.47,
|
| 1983 |
+
"learning_rate": 3.67764705882353e-07,
|
| 1984 |
+
"loss": 0.0003,
|
| 1985 |
+
"step": 7975
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 469.94,
|
| 1989 |
+
"learning_rate": 3.5894117647058825e-07,
|
| 1990 |
+
"loss": 0.0001,
|
| 1991 |
+
"step": 8000
|
| 1992 |
+
},
|
| 1993 |
+
{
|
| 1994 |
+
"epoch": 469.94,
|
| 1995 |
+
"eval_loss": 0.56689453125,
|
| 1996 |
+
"eval_runtime": 163.3464,
|
| 1997 |
+
"eval_samples_per_second": 1.665,
|
| 1998 |
+
"eval_steps_per_second": 0.104,
|
| 1999 |
+
"eval_wer": 9.899702823179792,
|
| 2000 |
+
"step": 8000
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 471.41,
|
| 2004 |
+
"learning_rate": 3.508235294117647e-07,
|
| 2005 |
+
"loss": 0.0002,
|
| 2006 |
+
"step": 8025
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 472.88,
|
| 2010 |
+
"learning_rate": 3.42e-07,
|
| 2011 |
+
"loss": 0.0001,
|
| 2012 |
+
"step": 8050
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 474.35,
|
| 2016 |
+
"learning_rate": 3.331764705882353e-07,
|
| 2017 |
+
"loss": 0.0001,
|
| 2018 |
+
"step": 8075
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 475.82,
|
| 2022 |
+
"learning_rate": 3.2435294117647057e-07,
|
| 2023 |
+
"loss": 0.0002,
|
| 2024 |
+
"step": 8100
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 477.29,
|
| 2028 |
+
"learning_rate": 3.1552941176470585e-07,
|
| 2029 |
+
"loss": 0.0002,
|
| 2030 |
+
"step": 8125
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 478.76,
|
| 2034 |
+
"learning_rate": 3.067058823529412e-07,
|
| 2035 |
+
"loss": 0.0001,
|
| 2036 |
+
"step": 8150
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 480.24,
|
| 2040 |
+
"learning_rate": 2.978823529411765e-07,
|
| 2041 |
+
"loss": 0.0001,
|
| 2042 |
+
"step": 8175
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 481.71,
|
| 2046 |
+
"learning_rate": 2.890588235294118e-07,
|
| 2047 |
+
"loss": 0.0002,
|
| 2048 |
+
"step": 8200
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 483.18,
|
| 2052 |
+
"learning_rate": 2.8023529411764706e-07,
|
| 2053 |
+
"loss": 0.0002,
|
| 2054 |
+
"step": 8225
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 484.65,
|
| 2058 |
+
"learning_rate": 2.714117647058824e-07,
|
| 2059 |
+
"loss": 0.0001,
|
| 2060 |
+
"step": 8250
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 486.12,
|
| 2064 |
+
"learning_rate": 2.6258823529411766e-07,
|
| 2065 |
+
"loss": 0.0003,
|
| 2066 |
+
"step": 8275
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 487.59,
|
| 2070 |
+
"learning_rate": 2.5376470588235294e-07,
|
| 2071 |
+
"loss": 0.0001,
|
| 2072 |
+
"step": 8300
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 489.06,
|
| 2076 |
+
"learning_rate": 2.449411764705882e-07,
|
| 2077 |
+
"loss": 0.0002,
|
| 2078 |
+
"step": 8325
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 490.53,
|
| 2082 |
+
"learning_rate": 2.3611764705882354e-07,
|
| 2083 |
+
"loss": 0.0001,
|
| 2084 |
+
"step": 8350
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 492.0,
|
| 2088 |
+
"learning_rate": 2.2729411764705884e-07,
|
| 2089 |
+
"loss": 0.0002,
|
| 2090 |
+
"step": 8375
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 493.47,
|
| 2094 |
+
"learning_rate": 2.1847058823529412e-07,
|
| 2095 |
+
"loss": 0.0002,
|
| 2096 |
+
"step": 8400
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 494.94,
|
| 2100 |
+
"learning_rate": 2.096470588235294e-07,
|
| 2101 |
+
"loss": 0.0002,
|
| 2102 |
+
"step": 8425
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 496.41,
|
| 2106 |
+
"learning_rate": 2.0082352941176472e-07,
|
| 2107 |
+
"loss": 0.0003,
|
| 2108 |
+
"step": 8450
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 497.88,
|
| 2112 |
+
"learning_rate": 1.92e-07,
|
| 2113 |
+
"loss": 0.0002,
|
| 2114 |
+
"step": 8475
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 499.35,
|
| 2118 |
+
"learning_rate": 1.831764705882353e-07,
|
| 2119 |
+
"loss": 0.0002,
|
| 2120 |
+
"step": 8500
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 500.82,
|
| 2124 |
+
"learning_rate": 1.743529411764706e-07,
|
| 2125 |
+
"loss": 0.0001,
|
| 2126 |
+
"step": 8525
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 502.29,
|
| 2130 |
+
"learning_rate": 1.6552941176470588e-07,
|
| 2131 |
+
"loss": 0.0002,
|
| 2132 |
+
"step": 8550
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 503.76,
|
| 2136 |
+
"learning_rate": 1.5670588235294118e-07,
|
| 2137 |
+
"loss": 0.0001,
|
| 2138 |
+
"step": 8575
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 505.24,
|
| 2142 |
+
"learning_rate": 1.4788235294117646e-07,
|
| 2143 |
+
"loss": 0.0001,
|
| 2144 |
+
"step": 8600
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 506.71,
|
| 2148 |
+
"learning_rate": 1.3905882352941179e-07,
|
| 2149 |
+
"loss": 0.0002,
|
| 2150 |
+
"step": 8625
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 508.18,
|
| 2154 |
+
"learning_rate": 1.3023529411764706e-07,
|
| 2155 |
+
"loss": 0.0002,
|
| 2156 |
+
"step": 8650
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 509.65,
|
| 2160 |
+
"learning_rate": 1.2141176470588236e-07,
|
| 2161 |
+
"loss": 0.0001,
|
| 2162 |
+
"step": 8675
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 511.12,
|
| 2166 |
+
"learning_rate": 1.1258823529411764e-07,
|
| 2167 |
+
"loss": 0.0001,
|
| 2168 |
+
"step": 8700
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 512.59,
|
| 2172 |
+
"learning_rate": 1.0376470588235296e-07,
|
| 2173 |
+
"loss": 0.0001,
|
| 2174 |
+
"step": 8725
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 514.06,
|
| 2178 |
+
"learning_rate": 9.494117647058823e-08,
|
| 2179 |
+
"loss": 0.0001,
|
| 2180 |
+
"step": 8750
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 515.53,
|
| 2184 |
+
"learning_rate": 8.611764705882352e-08,
|
| 2185 |
+
"loss": 0.0001,
|
| 2186 |
+
"step": 8775
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 517.0,
|
| 2190 |
+
"learning_rate": 7.729411764705882e-08,
|
| 2191 |
+
"loss": 0.0001,
|
| 2192 |
+
"step": 8800
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 518.47,
|
| 2196 |
+
"learning_rate": 6.847058823529411e-08,
|
| 2197 |
+
"loss": 0.0001,
|
| 2198 |
+
"step": 8825
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 519.94,
|
| 2202 |
+
"learning_rate": 5.964705882352941e-08,
|
| 2203 |
+
"loss": 0.0001,
|
| 2204 |
+
"step": 8850
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 521.41,
|
| 2208 |
+
"learning_rate": 5.08235294117647e-08,
|
| 2209 |
+
"loss": 0.0001,
|
| 2210 |
+
"step": 8875
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 522.88,
|
| 2214 |
+
"learning_rate": 4.2e-08,
|
| 2215 |
+
"loss": 0.0001,
|
| 2216 |
+
"step": 8900
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 524.35,
|
| 2220 |
+
"learning_rate": 3.3176470588235295e-08,
|
| 2221 |
+
"loss": 0.0001,
|
| 2222 |
+
"step": 8925
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 525.82,
|
| 2226 |
+
"learning_rate": 2.435294117647059e-08,
|
| 2227 |
+
"loss": 0.0001,
|
| 2228 |
+
"step": 8950
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 527.29,
|
| 2232 |
+
"learning_rate": 1.5529411764705883e-08,
|
| 2233 |
+
"loss": 0.0001,
|
| 2234 |
+
"step": 8975
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 528.76,
|
| 2238 |
+
"learning_rate": 6.705882352941177e-09,
|
| 2239 |
+
"loss": 0.0001,
|
| 2240 |
+
"step": 9000
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 528.76,
|
| 2244 |
+
"eval_loss": 0.564453125,
|
| 2245 |
+
"eval_runtime": 154.1953,
|
| 2246 |
+
"eval_samples_per_second": 1.764,
|
| 2247 |
+
"eval_steps_per_second": 0.11,
|
| 2248 |
+
"eval_wer": 9.927563150074294,
|
| 2249 |
+
"step": 9000
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 528.76,
|
| 2253 |
+
"step": 9000,
|
| 2254 |
+
"total_flos": 2.7722894647818505e+20,
|
| 2255 |
+
"train_loss": 0.00025936167769961886,
|
| 2256 |
+
"train_runtime": 28092.3517,
|
| 2257 |
+
"train_samples_per_second": 10.252,
|
| 2258 |
+
"train_steps_per_second": 0.32
|
| 2259 |
}
|
| 2260 |
],
|
| 2261 |
+
"max_steps": 9000,
|
| 2262 |
+
"num_train_epochs": 530,
|
| 2263 |
+
"total_flos": 2.7722894647818505e+20,
|
| 2264 |
"trial_name": null,
|
| 2265 |
"trial_params": null
|
| 2266 |
}
|