End of training
Browse files
all_results.json
CHANGED
@@ -1,12 +1,12 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"eval_loss": 1.
|
4 |
-
"eval_runtime":
|
5 |
-
"eval_samples_per_second":
|
6 |
-
"eval_steps_per_second": 0.
|
7 |
"eval_wer": 56.651029055690074,
|
8 |
-
"train_loss": 0.
|
9 |
-
"train_runtime":
|
10 |
-
"train_samples_per_second":
|
11 |
-
"train_steps_per_second": 0.
|
12 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 114.29,
|
3 |
+
"eval_loss": 1.227304458618164,
|
4 |
+
"eval_runtime": 520.721,
|
5 |
+
"eval_samples_per_second": 0.983,
|
6 |
+
"eval_steps_per_second": 0.031,
|
7 |
"eval_wer": 56.651029055690074,
|
8 |
+
"train_loss": 0.0002984794275835156,
|
9 |
+
"train_runtime": 870.3377,
|
10 |
+
"train_samples_per_second": 58.828,
|
11 |
+
"train_steps_per_second": 0.919
|
12 |
}
|
eval_results.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"epoch":
|
3 |
-
"eval_loss": 1.
|
4 |
-
"eval_runtime":
|
5 |
-
"eval_samples_per_second":
|
6 |
-
"eval_steps_per_second": 0.
|
7 |
"eval_wer": 56.651029055690074
|
8 |
}
|
|
|
1 |
{
|
2 |
+
"epoch": 114.29,
|
3 |
+
"eval_loss": 1.227304458618164,
|
4 |
+
"eval_runtime": 520.721,
|
5 |
+
"eval_samples_per_second": 0.983,
|
6 |
+
"eval_steps_per_second": 0.031,
|
7 |
"eval_wer": 56.651029055690074
|
8 |
}
|
runs/Dec16_18-49-35_129-146-104-29/events.out.tfevents.1671218042.129-146-104-29.141275.2
ADDED
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
1 |
+
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f788da650b08ffb9209415d679779a83d79ba0758e68bcd848249eb8b2bd3fcc
|
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": 114.29,
|
3 |
+
"train_loss": 0.0002984794275835156,
|
4 |
+
"train_runtime": 870.3377,
|
5 |
+
"train_samples_per_second": 58.828,
|
6 |
+
"train_steps_per_second": 0.919
|
7 |
}
|
trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"best_metric": 56.651029055690074,
|
3 |
"best_model_checkpoint": "./checkpoint-600",
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -383,18 +383,87 @@
|
|
383 |
"step": 700
|
384 |
},
|
385 |
{
|
386 |
-
"epoch":
|
387 |
-
"
|
388 |
-
"
|
389 |
-
"
|
390 |
-
|
391 |
-
|
392 |
-
"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
393 |
}
|
394 |
],
|
395 |
-
"max_steps":
|
396 |
-
"num_train_epochs":
|
397 |
-
"total_flos":
|
398 |
"trial_name": null,
|
399 |
"trial_params": null
|
400 |
}
|
|
|
1 |
{
|
2 |
"best_metric": 56.651029055690074,
|
3 |
"best_model_checkpoint": "./checkpoint-600",
|
4 |
+
"epoch": 114.28571428571429,
|
5 |
+
"global_step": 800,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
383 |
"step": 700
|
384 |
},
|
385 |
{
|
386 |
+
"epoch": 101.43,
|
387 |
+
"learning_rate": 3.133333333333334e-06,
|
388 |
+
"loss": 0.0026,
|
389 |
+
"step": 710
|
390 |
+
},
|
391 |
+
{
|
392 |
+
"epoch": 102.86,
|
393 |
+
"learning_rate": 2.8000000000000003e-06,
|
394 |
+
"loss": 0.0025,
|
395 |
+
"step": 720
|
396 |
+
},
|
397 |
+
{
|
398 |
+
"epoch": 104.29,
|
399 |
+
"learning_rate": 2.466666666666667e-06,
|
400 |
+
"loss": 0.0025,
|
401 |
+
"step": 730
|
402 |
+
},
|
403 |
+
{
|
404 |
+
"epoch": 105.71,
|
405 |
+
"learning_rate": 2.133333333333334e-06,
|
406 |
+
"loss": 0.0024,
|
407 |
+
"step": 740
|
408 |
+
},
|
409 |
+
{
|
410 |
+
"epoch": 107.14,
|
411 |
+
"learning_rate": 1.8000000000000001e-06,
|
412 |
+
"loss": 0.0024,
|
413 |
+
"step": 750
|
414 |
+
},
|
415 |
+
{
|
416 |
+
"epoch": 108.57,
|
417 |
+
"learning_rate": 1.4666666666666669e-06,
|
418 |
+
"loss": 0.0023,
|
419 |
+
"step": 760
|
420 |
+
},
|
421 |
+
{
|
422 |
+
"epoch": 110.0,
|
423 |
+
"learning_rate": 1.1333333333333334e-06,
|
424 |
+
"loss": 0.0023,
|
425 |
+
"step": 770
|
426 |
+
},
|
427 |
+
{
|
428 |
+
"epoch": 111.43,
|
429 |
+
"learning_rate": 8.000000000000001e-07,
|
430 |
+
"loss": 0.0023,
|
431 |
+
"step": 780
|
432 |
+
},
|
433 |
+
{
|
434 |
+
"epoch": 112.86,
|
435 |
+
"learning_rate": 4.666666666666667e-07,
|
436 |
+
"loss": 0.0023,
|
437 |
+
"step": 790
|
438 |
+
},
|
439 |
+
{
|
440 |
+
"epoch": 114.29,
|
441 |
+
"learning_rate": 1.3333333333333336e-07,
|
442 |
+
"loss": 0.0023,
|
443 |
+
"step": 800
|
444 |
+
},
|
445 |
+
{
|
446 |
+
"epoch": 114.29,
|
447 |
+
"eval_loss": 1.2710366249084473,
|
448 |
+
"eval_runtime": 511.2588,
|
449 |
+
"eval_samples_per_second": 1.001,
|
450 |
+
"eval_steps_per_second": 0.031,
|
451 |
+
"eval_wer": 56.75696125907991,
|
452 |
+
"step": 800
|
453 |
+
},
|
454 |
+
{
|
455 |
+
"epoch": 114.29,
|
456 |
+
"step": 800,
|
457 |
+
"total_flos": 6.3156914860032e+18,
|
458 |
+
"train_loss": 0.0002984794275835156,
|
459 |
+
"train_runtime": 870.3377,
|
460 |
+
"train_samples_per_second": 58.828,
|
461 |
+
"train_steps_per_second": 0.919
|
462 |
}
|
463 |
],
|
464 |
+
"max_steps": 800,
|
465 |
+
"num_train_epochs": 115,
|
466 |
+
"total_flos": 6.3156914860032e+18,
|
467 |
"trial_name": null,
|
468 |
"trial_params": null
|
469 |
}
|