End of training
Browse files
all_results.json
CHANGED
|
@@ -1,15 +1,15 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_cer": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 45.
|
| 6 |
"eval_samples": 481,
|
| 7 |
-
"eval_samples_per_second": 10.
|
| 8 |
-
"eval_steps_per_second":
|
| 9 |
-
"eval_wer": 0.
|
| 10 |
-
"train_loss": 0.
|
| 11 |
-
"train_runtime":
|
| 12 |
"train_samples": 2528,
|
| 13 |
-
"train_samples_per_second":
|
| 14 |
-
"train_steps_per_second": 2.
|
| 15 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 75.95,
|
| 3 |
+
"eval_cer": 0.1972293657199707,
|
| 4 |
+
"eval_loss": 0.9162325859069824,
|
| 5 |
+
"eval_runtime": 45.3436,
|
| 6 |
"eval_samples": 481,
|
| 7 |
+
"eval_samples_per_second": 10.608,
|
| 8 |
+
"eval_steps_per_second": 0.684,
|
| 9 |
+
"eval_wer": 0.5159447476125512,
|
| 10 |
+
"train_loss": 0.044292491674423215,
|
| 11 |
+
"train_runtime": 2233.4842,
|
| 12 |
"train_samples": 2528,
|
| 13 |
+
"train_samples_per_second": 85.964,
|
| 14 |
+
"train_steps_per_second": 2.686
|
| 15 |
}
|
eval_results.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"eval_cer": 0.
|
| 4 |
-
"eval_loss": 0.
|
| 5 |
-
"eval_runtime": 45.
|
| 6 |
"eval_samples": 481,
|
| 7 |
-
"eval_samples_per_second": 10.
|
| 8 |
-
"eval_steps_per_second":
|
| 9 |
-
"eval_wer": 0.
|
| 10 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 75.95,
|
| 3 |
+
"eval_cer": 0.1972293657199707,
|
| 4 |
+
"eval_loss": 0.9162325859069824,
|
| 5 |
+
"eval_runtime": 45.3436,
|
| 6 |
"eval_samples": 481,
|
| 7 |
+
"eval_samples_per_second": 10.608,
|
| 8 |
+
"eval_steps_per_second": 0.684,
|
| 9 |
+
"eval_wer": 0.5159447476125512
|
| 10 |
}
|
runs/Dec16_21-09-39_129-146-104-29/events.out.tfevents.1671227491.129-146-104-29.146624.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:40319fc688859220e783165235dc6652e723535f892fdb147887467e472c1a18
|
| 3 |
+
size 405
|
train_results.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
-
"epoch":
|
| 3 |
-
"train_loss": 0.
|
| 4 |
-
"train_runtime":
|
| 5 |
"train_samples": 2528,
|
| 6 |
-
"train_samples_per_second":
|
| 7 |
-
"train_steps_per_second": 2.
|
| 8 |
}
|
|
|
|
| 1 |
{
|
| 2 |
+
"epoch": 75.95,
|
| 3 |
+
"train_loss": 0.044292491674423215,
|
| 4 |
+
"train_runtime": 2233.4842,
|
| 5 |
"train_samples": 2528,
|
| 6 |
+
"train_samples_per_second": 85.964,
|
| 7 |
+
"train_steps_per_second": 2.686
|
| 8 |
}
|
trainer_state.json
CHANGED
|
@@ -1,8 +1,8 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.9162458181381226,
|
| 3 |
"best_model_checkpoint": "./checkpoint-4000",
|
| 4 |
-
"epoch":
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
@@ -3418,18 +3418,328 @@
|
|
| 3418 |
"step": 5500
|
| 3419 |
},
|
| 3420 |
{
|
| 3421 |
-
"epoch": 69.
|
| 3422 |
-
"
|
| 3423 |
-
"
|
| 3424 |
-
"
|
| 3425 |
-
|
| 3426 |
-
|
| 3427 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3428 |
}
|
| 3429 |
],
|
| 3430 |
-
"max_steps":
|
| 3431 |
-
"num_train_epochs":
|
| 3432 |
-
"total_flos": 6.
|
| 3433 |
"trial_name": null,
|
| 3434 |
"trial_params": null
|
| 3435 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.9162458181381226,
|
| 3 |
"best_model_checkpoint": "./checkpoint-4000",
|
| 4 |
+
"epoch": 75.9493670886076,
|
| 5 |
+
"global_step": 6000,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
|
|
|
| 3418 |
"step": 5500
|
| 3419 |
},
|
| 3420 |
{
|
| 3421 |
+
"epoch": 69.75,
|
| 3422 |
+
"learning_rate": 7.38e-06,
|
| 3423 |
+
"loss": 0.525,
|
| 3424 |
+
"step": 5510
|
| 3425 |
+
},
|
| 3426 |
+
{
|
| 3427 |
+
"epoch": 69.87,
|
| 3428 |
+
"learning_rate": 7.229999999999999e-06,
|
| 3429 |
+
"loss": 0.5232,
|
| 3430 |
+
"step": 5520
|
| 3431 |
+
},
|
| 3432 |
+
{
|
| 3433 |
+
"epoch": 70.0,
|
| 3434 |
+
"learning_rate": 7.079999999999999e-06,
|
| 3435 |
+
"loss": 0.5318,
|
| 3436 |
+
"step": 5530
|
| 3437 |
+
},
|
| 3438 |
+
{
|
| 3439 |
+
"epoch": 70.13,
|
| 3440 |
+
"learning_rate": 6.929999999999999e-06,
|
| 3441 |
+
"loss": 0.562,
|
| 3442 |
+
"step": 5540
|
| 3443 |
+
},
|
| 3444 |
+
{
|
| 3445 |
+
"epoch": 70.25,
|
| 3446 |
+
"learning_rate": 6.779999999999999e-06,
|
| 3447 |
+
"loss": 0.494,
|
| 3448 |
+
"step": 5550
|
| 3449 |
+
},
|
| 3450 |
+
{
|
| 3451 |
+
"epoch": 70.38,
|
| 3452 |
+
"learning_rate": 6.63e-06,
|
| 3453 |
+
"loss": 0.5314,
|
| 3454 |
+
"step": 5560
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 70.51,
|
| 3458 |
+
"learning_rate": 6.48e-06,
|
| 3459 |
+
"loss": 0.5332,
|
| 3460 |
+
"step": 5570
|
| 3461 |
+
},
|
| 3462 |
+
{
|
| 3463 |
+
"epoch": 70.63,
|
| 3464 |
+
"learning_rate": 6.3299999999999995e-06,
|
| 3465 |
+
"loss": 0.552,
|
| 3466 |
+
"step": 5580
|
| 3467 |
+
},
|
| 3468 |
+
{
|
| 3469 |
+
"epoch": 70.76,
|
| 3470 |
+
"learning_rate": 6.179999999999999e-06,
|
| 3471 |
+
"loss": 0.5538,
|
| 3472 |
+
"step": 5590
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 70.89,
|
| 3476 |
+
"learning_rate": 6.029999999999999e-06,
|
| 3477 |
+
"loss": 0.5507,
|
| 3478 |
+
"step": 5600
|
| 3479 |
+
},
|
| 3480 |
+
{
|
| 3481 |
+
"epoch": 71.01,
|
| 3482 |
+
"learning_rate": 5.88e-06,
|
| 3483 |
+
"loss": 0.5207,
|
| 3484 |
+
"step": 5610
|
| 3485 |
+
},
|
| 3486 |
+
{
|
| 3487 |
+
"epoch": 71.14,
|
| 3488 |
+
"learning_rate": 5.729999999999999e-06,
|
| 3489 |
+
"loss": 0.5613,
|
| 3490 |
+
"step": 5620
|
| 3491 |
+
},
|
| 3492 |
+
{
|
| 3493 |
+
"epoch": 71.27,
|
| 3494 |
+
"learning_rate": 5.579999999999999e-06,
|
| 3495 |
+
"loss": 0.5263,
|
| 3496 |
+
"step": 5630
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 71.39,
|
| 3500 |
+
"learning_rate": 5.43e-06,
|
| 3501 |
+
"loss": 0.5138,
|
| 3502 |
+
"step": 5640
|
| 3503 |
+
},
|
| 3504 |
+
{
|
| 3505 |
+
"epoch": 71.52,
|
| 3506 |
+
"learning_rate": 5.28e-06,
|
| 3507 |
+
"loss": 0.5268,
|
| 3508 |
+
"step": 5650
|
| 3509 |
+
},
|
| 3510 |
+
{
|
| 3511 |
+
"epoch": 71.65,
|
| 3512 |
+
"learning_rate": 5.13e-06,
|
| 3513 |
+
"loss": 0.5285,
|
| 3514 |
+
"step": 5660
|
| 3515 |
+
},
|
| 3516 |
+
{
|
| 3517 |
+
"epoch": 71.77,
|
| 3518 |
+
"learning_rate": 4.98e-06,
|
| 3519 |
+
"loss": 0.539,
|
| 3520 |
+
"step": 5670
|
| 3521 |
+
},
|
| 3522 |
+
{
|
| 3523 |
+
"epoch": 71.9,
|
| 3524 |
+
"learning_rate": 4.8299999999999995e-06,
|
| 3525 |
+
"loss": 0.5518,
|
| 3526 |
+
"step": 5680
|
| 3527 |
+
},
|
| 3528 |
+
{
|
| 3529 |
+
"epoch": 72.03,
|
| 3530 |
+
"learning_rate": 4.679999999999999e-06,
|
| 3531 |
+
"loss": 0.5392,
|
| 3532 |
+
"step": 5690
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 72.15,
|
| 3536 |
+
"learning_rate": 4.53e-06,
|
| 3537 |
+
"loss": 0.5341,
|
| 3538 |
+
"step": 5700
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 72.28,
|
| 3542 |
+
"learning_rate": 4.3799999999999996e-06,
|
| 3543 |
+
"loss": 0.528,
|
| 3544 |
+
"step": 5710
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 72.41,
|
| 3548 |
+
"learning_rate": 4.229999999999999e-06,
|
| 3549 |
+
"loss": 0.5285,
|
| 3550 |
+
"step": 5720
|
| 3551 |
+
},
|
| 3552 |
+
{
|
| 3553 |
+
"epoch": 72.53,
|
| 3554 |
+
"learning_rate": 4.079999999999999e-06,
|
| 3555 |
+
"loss": 0.5291,
|
| 3556 |
+
"step": 5730
|
| 3557 |
+
},
|
| 3558 |
+
{
|
| 3559 |
+
"epoch": 72.66,
|
| 3560 |
+
"learning_rate": 3.93e-06,
|
| 3561 |
+
"loss": 0.56,
|
| 3562 |
+
"step": 5740
|
| 3563 |
+
},
|
| 3564 |
+
{
|
| 3565 |
+
"epoch": 72.78,
|
| 3566 |
+
"learning_rate": 3.78e-06,
|
| 3567 |
+
"loss": 0.5638,
|
| 3568 |
+
"step": 5750
|
| 3569 |
+
},
|
| 3570 |
+
{
|
| 3571 |
+
"epoch": 72.91,
|
| 3572 |
+
"learning_rate": 3.6299999999999995e-06,
|
| 3573 |
+
"loss": 0.5111,
|
| 3574 |
+
"step": 5760
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 73.04,
|
| 3578 |
+
"learning_rate": 3.4799999999999993e-06,
|
| 3579 |
+
"loss": 0.5239,
|
| 3580 |
+
"step": 5770
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 73.16,
|
| 3584 |
+
"learning_rate": 3.33e-06,
|
| 3585 |
+
"loss": 0.5386,
|
| 3586 |
+
"step": 5780
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 73.29,
|
| 3590 |
+
"learning_rate": 3.1799999999999996e-06,
|
| 3591 |
+
"loss": 0.525,
|
| 3592 |
+
"step": 5790
|
| 3593 |
+
},
|
| 3594 |
+
{
|
| 3595 |
+
"epoch": 73.42,
|
| 3596 |
+
"learning_rate": 3.03e-06,
|
| 3597 |
+
"loss": 0.5051,
|
| 3598 |
+
"step": 5800
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 73.54,
|
| 3602 |
+
"learning_rate": 2.8799999999999995e-06,
|
| 3603 |
+
"loss": 0.5119,
|
| 3604 |
+
"step": 5810
|
| 3605 |
+
},
|
| 3606 |
+
{
|
| 3607 |
+
"epoch": 73.67,
|
| 3608 |
+
"learning_rate": 2.7299999999999997e-06,
|
| 3609 |
+
"loss": 0.5209,
|
| 3610 |
+
"step": 5820
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 73.8,
|
| 3614 |
+
"learning_rate": 2.58e-06,
|
| 3615 |
+
"loss": 0.5659,
|
| 3616 |
+
"step": 5830
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 73.92,
|
| 3620 |
+
"learning_rate": 2.4299999999999996e-06,
|
| 3621 |
+
"loss": 0.5178,
|
| 3622 |
+
"step": 5840
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 74.05,
|
| 3626 |
+
"learning_rate": 2.2799999999999998e-06,
|
| 3627 |
+
"loss": 0.5523,
|
| 3628 |
+
"step": 5850
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 74.18,
|
| 3632 |
+
"learning_rate": 2.13e-06,
|
| 3633 |
+
"loss": 0.5048,
|
| 3634 |
+
"step": 5860
|
| 3635 |
+
},
|
| 3636 |
+
{
|
| 3637 |
+
"epoch": 74.3,
|
| 3638 |
+
"learning_rate": 1.9799999999999997e-06,
|
| 3639 |
+
"loss": 0.5109,
|
| 3640 |
+
"step": 5870
|
| 3641 |
+
},
|
| 3642 |
+
{
|
| 3643 |
+
"epoch": 74.43,
|
| 3644 |
+
"learning_rate": 1.83e-06,
|
| 3645 |
+
"loss": 0.5092,
|
| 3646 |
+
"step": 5880
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 74.56,
|
| 3650 |
+
"learning_rate": 1.6799999999999998e-06,
|
| 3651 |
+
"loss": 0.5439,
|
| 3652 |
+
"step": 5890
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 74.68,
|
| 3656 |
+
"learning_rate": 1.53e-06,
|
| 3657 |
+
"loss": 0.5501,
|
| 3658 |
+
"step": 5900
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 74.81,
|
| 3662 |
+
"learning_rate": 1.38e-06,
|
| 3663 |
+
"loss": 0.5628,
|
| 3664 |
+
"step": 5910
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 74.94,
|
| 3668 |
+
"learning_rate": 1.23e-06,
|
| 3669 |
+
"loss": 0.5097,
|
| 3670 |
+
"step": 5920
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 75.06,
|
| 3674 |
+
"learning_rate": 1.0799999999999998e-06,
|
| 3675 |
+
"loss": 0.5363,
|
| 3676 |
+
"step": 5930
|
| 3677 |
+
},
|
| 3678 |
+
{
|
| 3679 |
+
"epoch": 75.19,
|
| 3680 |
+
"learning_rate": 9.299999999999999e-07,
|
| 3681 |
+
"loss": 0.5304,
|
| 3682 |
+
"step": 5940
|
| 3683 |
+
},
|
| 3684 |
+
{
|
| 3685 |
+
"epoch": 75.32,
|
| 3686 |
+
"learning_rate": 7.799999999999999e-07,
|
| 3687 |
+
"loss": 0.5358,
|
| 3688 |
+
"step": 5950
|
| 3689 |
+
},
|
| 3690 |
+
{
|
| 3691 |
+
"epoch": 75.44,
|
| 3692 |
+
"learning_rate": 6.299999999999999e-07,
|
| 3693 |
+
"loss": 0.5262,
|
| 3694 |
+
"step": 5960
|
| 3695 |
+
},
|
| 3696 |
+
{
|
| 3697 |
+
"epoch": 75.57,
|
| 3698 |
+
"learning_rate": 4.8e-07,
|
| 3699 |
+
"loss": 0.5258,
|
| 3700 |
+
"step": 5970
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 75.7,
|
| 3704 |
+
"learning_rate": 3.3e-07,
|
| 3705 |
+
"loss": 0.4952,
|
| 3706 |
+
"step": 5980
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 75.82,
|
| 3710 |
+
"learning_rate": 1.7999999999999997e-07,
|
| 3711 |
+
"loss": 0.5285,
|
| 3712 |
+
"step": 5990
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 75.95,
|
| 3716 |
+
"learning_rate": 3e-08,
|
| 3717 |
+
"loss": 0.5252,
|
| 3718 |
+
"step": 6000
|
| 3719 |
+
},
|
| 3720 |
+
{
|
| 3721 |
+
"epoch": 75.95,
|
| 3722 |
+
"eval_cer": 0.19692902596061795,
|
| 3723 |
+
"eval_loss": 0.9505288004875183,
|
| 3724 |
+
"eval_runtime": 45.9888,
|
| 3725 |
+
"eval_samples_per_second": 10.459,
|
| 3726 |
+
"eval_steps_per_second": 0.674,
|
| 3727 |
+
"eval_wer": 0.5117667121418826,
|
| 3728 |
+
"step": 6000
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 75.95,
|
| 3732 |
+
"step": 6000,
|
| 3733 |
+
"total_flos": 6.910110276723645e+19,
|
| 3734 |
+
"train_loss": 0.044292491674423215,
|
| 3735 |
+
"train_runtime": 2233.4842,
|
| 3736 |
+
"train_samples_per_second": 85.964,
|
| 3737 |
+
"train_steps_per_second": 2.686
|
| 3738 |
}
|
| 3739 |
],
|
| 3740 |
+
"max_steps": 6000,
|
| 3741 |
+
"num_train_epochs": 76,
|
| 3742 |
+
"total_flos": 6.910110276723645e+19,
|
| 3743 |
"trial_name": null,
|
| 3744 |
"trial_params": null
|
| 3745 |
}
|