End of training
Browse files
all_results.json
CHANGED
|
@@ -1,12 +1,12 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_loss": 0.8710034489631653,
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
"eval_wer": 60.05599273607748,
|
| 8 |
-
"train_loss": 0.
|
| 9 |
-
"train_runtime":
|
| 10 |
-
"train_samples_per_second":
|
| 11 |
-
"train_steps_per_second": 0.
|
| 12 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 25.0,
|
| 3 |
"eval_loss": 0.8710034489631653,
|
| 4 |
+
"eval_runtime": 324.5791,
|
| 5 |
+
"eval_samples_per_second": 1.577,
|
| 6 |
+
"eval_steps_per_second": 0.049,
|
| 7 |
"eval_wer": 60.05599273607748,
|
| 8 |
+
"train_loss": 0.16543263983726503,
|
| 9 |
+
"train_runtime": 2579.5458,
|
| 10 |
+
"train_samples_per_second": 24.811,
|
| 11 |
+
"train_steps_per_second": 0.388
|
| 12 |
}
|
eval_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
"eval_loss": 0.8710034489631653,
|
| 4 |
-
"eval_runtime":
|
| 5 |
-
"eval_samples_per_second":
|
| 6 |
-
"eval_steps_per_second": 0.
|
| 7 |
"eval_wer": 60.05599273607748
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 25.0,
|
| 3 |
"eval_loss": 0.8710034489631653,
|
| 4 |
+
"eval_runtime": 324.5791,
|
| 5 |
+
"eval_samples_per_second": 1.577,
|
| 6 |
+
"eval_steps_per_second": 0.049,
|
| 7 |
"eval_wer": 60.05599273607748
|
| 8 |
}
|
runs/Dec20_18-16-58_129-146-32-172/events.out.tfevents.1671563202.129-146-32-172.137520.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c06e882b57bbb25ce16c01b5e19af5d55ef76cf57fb5e4836eeb3b94df95f92e
|
| 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": 25.0,
|
| 3 |
+
"train_loss": 0.16543263983726503,
|
| 4 |
+
"train_runtime": 2579.5458,
|
| 5 |
+
"train_samples_per_second": 24.811,
|
| 6 |
+
"train_steps_per_second": 0.388
|
| 7 |
}
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.8710034489631653,
|
| 3 |
"best_model_checkpoint": "./checkpoint-400",
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -353,18 +353,363 @@
|
|
| 353 |
"step": 500
|
| 354 |
},
|
| 355 |
{
|
| 356 |
-
"epoch": 12.
|
| 357 |
-
"
|
| 358 |
-
"
|
| 359 |
-
"
|
| 360 |
-
|
| 361 |
-
|
| 362 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 363 |
}
|
| 364 |
],
|
| 365 |
-
"max_steps":
|
| 366 |
-
"num_train_epochs":
|
| 367 |
-
"total_flos":
|
| 368 |
"trial_name": null,
|
| 369 |
"trial_params": null
|
| 370 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.8710034489631653,
|
| 3 |
"best_model_checkpoint": "./checkpoint-400",
|
| 4 |
+
"epoch": 25.0,
|
| 5 |
+
"global_step": 1000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 353 |
"step": 500
|
| 354 |
},
|
| 355 |
{
|
| 356 |
+
"epoch": 12.75,
|
| 357 |
+
"learning_rate": 5.092783505154639e-06,
|
| 358 |
+
"loss": 0.4501,
|
| 359 |
+
"step": 510
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 13.0,
|
| 363 |
+
"learning_rate": 4.989690721649485e-06,
|
| 364 |
+
"loss": 0.4693,
|
| 365 |
+
"step": 520
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 13.25,
|
| 369 |
+
"learning_rate": 4.88659793814433e-06,
|
| 370 |
+
"loss": 0.4261,
|
| 371 |
+
"step": 530
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 13.5,
|
| 375 |
+
"learning_rate": 4.783505154639176e-06,
|
| 376 |
+
"loss": 0.4336,
|
| 377 |
+
"step": 540
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 13.75,
|
| 381 |
+
"learning_rate": 4.680412371134021e-06,
|
| 382 |
+
"loss": 0.4303,
|
| 383 |
+
"step": 550
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 14.0,
|
| 387 |
+
"learning_rate": 4.577319587628866e-06,
|
| 388 |
+
"loss": 0.4231,
|
| 389 |
+
"step": 560
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 14.25,
|
| 393 |
+
"learning_rate": 4.474226804123712e-06,
|
| 394 |
+
"loss": 0.4178,
|
| 395 |
+
"step": 570
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 14.5,
|
| 399 |
+
"learning_rate": 4.371134020618557e-06,
|
| 400 |
+
"loss": 0.4046,
|
| 401 |
+
"step": 580
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 14.75,
|
| 405 |
+
"learning_rate": 4.2680412371134025e-06,
|
| 406 |
+
"loss": 0.3927,
|
| 407 |
+
"step": 590
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 15.0,
|
| 411 |
+
"learning_rate": 4.164948453608248e-06,
|
| 412 |
+
"loss": 0.3882,
|
| 413 |
+
"step": 600
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 15.0,
|
| 417 |
+
"eval_loss": 0.8819407224655151,
|
| 418 |
+
"eval_runtime": 211.3554,
|
| 419 |
+
"eval_samples_per_second": 2.422,
|
| 420 |
+
"eval_steps_per_second": 0.076,
|
| 421 |
+
"eval_wer": 59.06476997578692,
|
| 422 |
+
"step": 600
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 15.25,
|
| 426 |
+
"learning_rate": 4.061855670103093e-06,
|
| 427 |
+
"loss": 0.372,
|
| 428 |
+
"step": 610
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 15.5,
|
| 432 |
+
"learning_rate": 3.958762886597938e-06,
|
| 433 |
+
"loss": 0.3832,
|
| 434 |
+
"step": 620
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 15.75,
|
| 438 |
+
"learning_rate": 3.855670103092784e-06,
|
| 439 |
+
"loss": 0.3756,
|
| 440 |
+
"step": 630
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 16.0,
|
| 444 |
+
"learning_rate": 3.752577319587629e-06,
|
| 445 |
+
"loss": 0.3818,
|
| 446 |
+
"step": 640
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 16.25,
|
| 450 |
+
"learning_rate": 3.6494845360824745e-06,
|
| 451 |
+
"loss": 0.3669,
|
| 452 |
+
"step": 650
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 16.5,
|
| 456 |
+
"learning_rate": 3.54639175257732e-06,
|
| 457 |
+
"loss": 0.3481,
|
| 458 |
+
"step": 660
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 16.75,
|
| 462 |
+
"learning_rate": 3.443298969072165e-06,
|
| 463 |
+
"loss": 0.3568,
|
| 464 |
+
"step": 670
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 17.0,
|
| 468 |
+
"learning_rate": 3.3402061855670103e-06,
|
| 469 |
+
"loss": 0.3567,
|
| 470 |
+
"step": 680
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 17.25,
|
| 474 |
+
"learning_rate": 3.2371134020618563e-06,
|
| 475 |
+
"loss": 0.3513,
|
| 476 |
+
"step": 690
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 17.5,
|
| 480 |
+
"learning_rate": 3.1340206185567014e-06,
|
| 481 |
+
"loss": 0.3346,
|
| 482 |
+
"step": 700
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 17.5,
|
| 486 |
+
"eval_loss": 0.9031797647476196,
|
| 487 |
+
"eval_runtime": 183.8672,
|
| 488 |
+
"eval_samples_per_second": 2.785,
|
| 489 |
+
"eval_steps_per_second": 0.087,
|
| 490 |
+
"eval_wer": 59.480932203389834,
|
| 491 |
+
"step": 700
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 17.75,
|
| 495 |
+
"learning_rate": 3.0309278350515465e-06,
|
| 496 |
+
"loss": 0.3322,
|
| 497 |
+
"step": 710
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 18.0,
|
| 501 |
+
"learning_rate": 2.927835051546392e-06,
|
| 502 |
+
"loss": 0.3114,
|
| 503 |
+
"step": 720
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 18.25,
|
| 507 |
+
"learning_rate": 2.8247422680412372e-06,
|
| 508 |
+
"loss": 0.3155,
|
| 509 |
+
"step": 730
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 18.5,
|
| 513 |
+
"learning_rate": 2.7216494845360823e-06,
|
| 514 |
+
"loss": 0.3295,
|
| 515 |
+
"step": 740
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 18.75,
|
| 519 |
+
"learning_rate": 2.6185567010309283e-06,
|
| 520 |
+
"loss": 0.314,
|
| 521 |
+
"step": 750
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 19.0,
|
| 525 |
+
"learning_rate": 2.5154639175257734e-06,
|
| 526 |
+
"loss": 0.3068,
|
| 527 |
+
"step": 760
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 19.25,
|
| 531 |
+
"learning_rate": 2.412371134020619e-06,
|
| 532 |
+
"loss": 0.312,
|
| 533 |
+
"step": 770
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 19.5,
|
| 537 |
+
"learning_rate": 2.309278350515464e-06,
|
| 538 |
+
"loss": 0.2938,
|
| 539 |
+
"step": 780
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 19.75,
|
| 543 |
+
"learning_rate": 2.2061855670103092e-06,
|
| 544 |
+
"loss": 0.3052,
|
| 545 |
+
"step": 790
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 20.0,
|
| 549 |
+
"learning_rate": 2.1030927835051548e-06,
|
| 550 |
+
"loss": 0.2947,
|
| 551 |
+
"step": 800
|
| 552 |
+
},
|
| 553 |
+
{
|
| 554 |
+
"epoch": 20.0,
|
| 555 |
+
"eval_loss": 0.9144465923309326,
|
| 556 |
+
"eval_runtime": 185.6121,
|
| 557 |
+
"eval_samples_per_second": 2.758,
|
| 558 |
+
"eval_steps_per_second": 0.086,
|
| 559 |
+
"eval_wer": 59.76846246973365,
|
| 560 |
+
"step": 800
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 20.25,
|
| 564 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 565 |
+
"loss": 0.2897,
|
| 566 |
+
"step": 810
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 20.5,
|
| 570 |
+
"learning_rate": 1.8969072164948455e-06,
|
| 571 |
+
"loss": 0.2856,
|
| 572 |
+
"step": 820
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 20.75,
|
| 576 |
+
"learning_rate": 1.7938144329896908e-06,
|
| 577 |
+
"loss": 0.3027,
|
| 578 |
+
"step": 830
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 21.0,
|
| 582 |
+
"learning_rate": 1.6907216494845363e-06,
|
| 583 |
+
"loss": 0.3446,
|
| 584 |
+
"step": 840
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 21.25,
|
| 588 |
+
"learning_rate": 1.5876288659793815e-06,
|
| 589 |
+
"loss": 0.2766,
|
| 590 |
+
"step": 850
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 21.5,
|
| 594 |
+
"learning_rate": 1.484536082474227e-06,
|
| 595 |
+
"loss": 0.2816,
|
| 596 |
+
"step": 860
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 21.75,
|
| 600 |
+
"learning_rate": 1.3814432989690724e-06,
|
| 601 |
+
"loss": 0.2845,
|
| 602 |
+
"step": 870
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 22.0,
|
| 606 |
+
"learning_rate": 1.2783505154639175e-06,
|
| 607 |
+
"loss": 0.2828,
|
| 608 |
+
"step": 880
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 22.25,
|
| 612 |
+
"learning_rate": 1.175257731958763e-06,
|
| 613 |
+
"loss": 0.2766,
|
| 614 |
+
"step": 890
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 22.5,
|
| 618 |
+
"learning_rate": 1.0721649484536084e-06,
|
| 619 |
+
"loss": 0.2724,
|
| 620 |
+
"step": 900
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 22.5,
|
| 624 |
+
"eval_loss": 0.9288888573646545,
|
| 625 |
+
"eval_runtime": 275.793,
|
| 626 |
+
"eval_samples_per_second": 1.856,
|
| 627 |
+
"eval_steps_per_second": 0.058,
|
| 628 |
+
"eval_wer": 58.98153753026635,
|
| 629 |
+
"step": 900
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 22.75,
|
| 633 |
+
"learning_rate": 9.690721649484537e-07,
|
| 634 |
+
"loss": 0.267,
|
| 635 |
+
"step": 910
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 23.0,
|
| 639 |
+
"learning_rate": 8.65979381443299e-07,
|
| 640 |
+
"loss": 0.2602,
|
| 641 |
+
"step": 920
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 23.25,
|
| 645 |
+
"learning_rate": 7.628865979381445e-07,
|
| 646 |
+
"loss": 0.2692,
|
| 647 |
+
"step": 930
|
| 648 |
+
},
|
| 649 |
+
{
|
| 650 |
+
"epoch": 23.5,
|
| 651 |
+
"learning_rate": 6.597938144329897e-07,
|
| 652 |
+
"loss": 0.2623,
|
| 653 |
+
"step": 940
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 23.75,
|
| 657 |
+
"learning_rate": 5.567010309278352e-07,
|
| 658 |
+
"loss": 0.2729,
|
| 659 |
+
"step": 950
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 24.0,
|
| 663 |
+
"learning_rate": 4.5360824742268044e-07,
|
| 664 |
+
"loss": 0.263,
|
| 665 |
+
"step": 960
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 24.25,
|
| 669 |
+
"learning_rate": 3.505154639175258e-07,
|
| 670 |
+
"loss": 0.2656,
|
| 671 |
+
"step": 970
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 24.5,
|
| 675 |
+
"learning_rate": 2.474226804123711e-07,
|
| 676 |
+
"loss": 0.2633,
|
| 677 |
+
"step": 980
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 24.75,
|
| 681 |
+
"learning_rate": 1.443298969072165e-07,
|
| 682 |
+
"loss": 0.2661,
|
| 683 |
+
"step": 990
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 25.0,
|
| 687 |
+
"learning_rate": 4.123711340206186e-08,
|
| 688 |
+
"loss": 0.2785,
|
| 689 |
+
"step": 1000
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 25.0,
|
| 693 |
+
"eval_loss": 0.933917224407196,
|
| 694 |
+
"eval_runtime": 339.9994,
|
| 695 |
+
"eval_samples_per_second": 1.506,
|
| 696 |
+
"eval_steps_per_second": 0.047,
|
| 697 |
+
"eval_wer": 59.200968523002416,
|
| 698 |
+
"step": 1000
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 25.0,
|
| 702 |
+
"step": 1000,
|
| 703 |
+
"total_flos": 4.048885260288e+18,
|
| 704 |
+
"train_loss": 0.16543263983726503,
|
| 705 |
+
"train_runtime": 2579.5458,
|
| 706 |
+
"train_samples_per_second": 24.811,
|
| 707 |
+
"train_steps_per_second": 0.388
|
| 708 |
}
|
| 709 |
],
|
| 710 |
+
"max_steps": 1000,
|
| 711 |
+
"num_train_epochs": 25,
|
| 712 |
+
"total_flos": 4.048885260288e+18,
|
| 713 |
"trial_name": null,
|
| 714 |
"trial_params": null
|
| 715 |
}
|