Training in progress, step 600, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 66126768
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1fdd0334d51ecc9ab8aa18db51be282267c4701fa1d13c3de14eb7534664f5eb
|
| 3 |
size 66126768
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 34141829
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:14eb5407563d74dce1f96073fa6908dfe69c57e3d6179c798664936d210508c4
|
| 3 |
size 34141829
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14645
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9445552595536daf5bd8731be4eabb308bd26e76a3f4f0c20c4aa55fcf9ea202
|
| 3 |
size 14645
|
last-checkpoint/scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1383
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:cb7fde5111803012042c93a73aa191336bb6e10b3ad44f6bd1d94fc7008a22b6
|
| 3 |
size 1383
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1465
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1785dec699279cf735b471c940e3c7215708e10021bc4f35a643cbd79b28aacf
|
| 3 |
size 1465
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -2,9 +2,9 @@
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
-
"epoch":
|
| 6 |
"eval_steps": 120,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -3400,6 +3400,854 @@
|
|
| 3400 |
"eval_samples_per_second": 0.461,
|
| 3401 |
"eval_steps_per_second": 0.461,
|
| 3402 |
"step": 480
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3403 |
}
|
| 3404 |
],
|
| 3405 |
"logging_steps": 1,
|
|
@@ -3419,7 +4267,7 @@
|
|
| 3419 |
"attributes": {}
|
| 3420 |
}
|
| 3421 |
},
|
| 3422 |
-
"total_flos": 3.
|
| 3423 |
"train_batch_size": 2,
|
| 3424 |
"trial_name": null,
|
| 3425 |
"trial_params": null
|
|
|
|
| 2 |
"best_global_step": null,
|
| 3 |
"best_metric": null,
|
| 4 |
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 12.774193548387096,
|
| 6 |
"eval_steps": 120,
|
| 7 |
+
"global_step": 600,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 3400 |
"eval_samples_per_second": 0.461,
|
| 3401 |
"eval_steps_per_second": 0.461,
|
| 3402 |
"step": 480
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 10.236559139784946,
|
| 3406 |
+
"grad_norm": 0.6188676953315735,
|
| 3407 |
+
"learning_rate": 0.00012533756949960288,
|
| 3408 |
+
"loss": 0.382239431142807,
|
| 3409 |
+
"step": 481
|
| 3410 |
+
},
|
| 3411 |
+
{
|
| 3412 |
+
"epoch": 10.258064516129032,
|
| 3413 |
+
"grad_norm": 0.6917077302932739,
|
| 3414 |
+
"learning_rate": 0.00012517871326449563,
|
| 3415 |
+
"loss": 0.36738157272338867,
|
| 3416 |
+
"step": 482
|
| 3417 |
+
},
|
| 3418 |
+
{
|
| 3419 |
+
"epoch": 10.279569892473118,
|
| 3420 |
+
"grad_norm": 0.710863471031189,
|
| 3421 |
+
"learning_rate": 0.0001250198570293884,
|
| 3422 |
+
"loss": 0.3310585916042328,
|
| 3423 |
+
"step": 483
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 10.301075268817204,
|
| 3427 |
+
"grad_norm": 0.74619460105896,
|
| 3428 |
+
"learning_rate": 0.00012486100079428118,
|
| 3429 |
+
"loss": 0.5340608358383179,
|
| 3430 |
+
"step": 484
|
| 3431 |
+
},
|
| 3432 |
+
{
|
| 3433 |
+
"epoch": 10.32258064516129,
|
| 3434 |
+
"grad_norm": 0.8272573351860046,
|
| 3435 |
+
"learning_rate": 0.00012470214455917396,
|
| 3436 |
+
"loss": 0.39920294284820557,
|
| 3437 |
+
"step": 485
|
| 3438 |
+
},
|
| 3439 |
+
{
|
| 3440 |
+
"epoch": 10.344086021505376,
|
| 3441 |
+
"grad_norm": 0.8804677724838257,
|
| 3442 |
+
"learning_rate": 0.00012454328832406673,
|
| 3443 |
+
"loss": 0.2775310277938843,
|
| 3444 |
+
"step": 486
|
| 3445 |
+
},
|
| 3446 |
+
{
|
| 3447 |
+
"epoch": 10.365591397849462,
|
| 3448 |
+
"grad_norm": 0.7785136103630066,
|
| 3449 |
+
"learning_rate": 0.00012438443208895948,
|
| 3450 |
+
"loss": 0.32668304443359375,
|
| 3451 |
+
"step": 487
|
| 3452 |
+
},
|
| 3453 |
+
{
|
| 3454 |
+
"epoch": 10.387096774193548,
|
| 3455 |
+
"grad_norm": 0.7213490009307861,
|
| 3456 |
+
"learning_rate": 0.00012422557585385226,
|
| 3457 |
+
"loss": 0.3376646041870117,
|
| 3458 |
+
"step": 488
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 10.408602150537634,
|
| 3462 |
+
"grad_norm": 0.7177138328552246,
|
| 3463 |
+
"learning_rate": 0.00012406671961874503,
|
| 3464 |
+
"loss": 0.42496711015701294,
|
| 3465 |
+
"step": 489
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 10.43010752688172,
|
| 3469 |
+
"grad_norm": 0.7309224009513855,
|
| 3470 |
+
"learning_rate": 0.00012390786338363784,
|
| 3471 |
+
"loss": 0.3748842179775238,
|
| 3472 |
+
"step": 490
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 10.451612903225806,
|
| 3476 |
+
"grad_norm": 0.805568277835846,
|
| 3477 |
+
"learning_rate": 0.00012374900714853058,
|
| 3478 |
+
"loss": 0.39560365676879883,
|
| 3479 |
+
"step": 491
|
| 3480 |
+
},
|
| 3481 |
+
{
|
| 3482 |
+
"epoch": 10.473118279569892,
|
| 3483 |
+
"grad_norm": 0.6789179444313049,
|
| 3484 |
+
"learning_rate": 0.00012359015091342336,
|
| 3485 |
+
"loss": 0.3961871862411499,
|
| 3486 |
+
"step": 492
|
| 3487 |
+
},
|
| 3488 |
+
{
|
| 3489 |
+
"epoch": 10.494623655913978,
|
| 3490 |
+
"grad_norm": 0.6529718041419983,
|
| 3491 |
+
"learning_rate": 0.00012343129467831614,
|
| 3492 |
+
"loss": 0.40176600217819214,
|
| 3493 |
+
"step": 493
|
| 3494 |
+
},
|
| 3495 |
+
{
|
| 3496 |
+
"epoch": 10.516129032258064,
|
| 3497 |
+
"grad_norm": 0.6573322415351868,
|
| 3498 |
+
"learning_rate": 0.0001232724384432089,
|
| 3499 |
+
"loss": 0.36849701404571533,
|
| 3500 |
+
"step": 494
|
| 3501 |
+
},
|
| 3502 |
+
{
|
| 3503 |
+
"epoch": 10.53763440860215,
|
| 3504 |
+
"grad_norm": 0.7451475262641907,
|
| 3505 |
+
"learning_rate": 0.0001231135822081017,
|
| 3506 |
+
"loss": 0.39875566959381104,
|
| 3507 |
+
"step": 495
|
| 3508 |
+
},
|
| 3509 |
+
{
|
| 3510 |
+
"epoch": 10.559139784946236,
|
| 3511 |
+
"grad_norm": 0.6852983236312866,
|
| 3512 |
+
"learning_rate": 0.00012295472597299444,
|
| 3513 |
+
"loss": 0.39747345447540283,
|
| 3514 |
+
"step": 496
|
| 3515 |
+
},
|
| 3516 |
+
{
|
| 3517 |
+
"epoch": 10.580645161290322,
|
| 3518 |
+
"grad_norm": 0.7666296362876892,
|
| 3519 |
+
"learning_rate": 0.0001227958697378872,
|
| 3520 |
+
"loss": 0.38431617617607117,
|
| 3521 |
+
"step": 497
|
| 3522 |
+
},
|
| 3523 |
+
{
|
| 3524 |
+
"epoch": 10.602150537634408,
|
| 3525 |
+
"grad_norm": 0.7944223284721375,
|
| 3526 |
+
"learning_rate": 0.00012263701350278,
|
| 3527 |
+
"loss": 0.3355713486671448,
|
| 3528 |
+
"step": 498
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 10.623655913978494,
|
| 3532 |
+
"grad_norm": 0.7261567711830139,
|
| 3533 |
+
"learning_rate": 0.00012247815726767276,
|
| 3534 |
+
"loss": 0.42981234192848206,
|
| 3535 |
+
"step": 499
|
| 3536 |
+
},
|
| 3537 |
+
{
|
| 3538 |
+
"epoch": 10.64516129032258,
|
| 3539 |
+
"grad_norm": 0.7246410250663757,
|
| 3540 |
+
"learning_rate": 0.0001223193010325655,
|
| 3541 |
+
"loss": 0.41270819306373596,
|
| 3542 |
+
"step": 500
|
| 3543 |
+
},
|
| 3544 |
+
{
|
| 3545 |
+
"epoch": 10.666666666666666,
|
| 3546 |
+
"grad_norm": 0.7120943665504456,
|
| 3547 |
+
"learning_rate": 0.0001221604447974583,
|
| 3548 |
+
"loss": 0.3995019197463989,
|
| 3549 |
+
"step": 501
|
| 3550 |
+
},
|
| 3551 |
+
{
|
| 3552 |
+
"epoch": 10.688172043010752,
|
| 3553 |
+
"grad_norm": 0.7080013155937195,
|
| 3554 |
+
"learning_rate": 0.00012200158856235109,
|
| 3555 |
+
"loss": 0.3284502625465393,
|
| 3556 |
+
"step": 502
|
| 3557 |
+
},
|
| 3558 |
+
{
|
| 3559 |
+
"epoch": 10.709677419354838,
|
| 3560 |
+
"grad_norm": 0.7086816430091858,
|
| 3561 |
+
"learning_rate": 0.00012184273232724385,
|
| 3562 |
+
"loss": 0.41737645864486694,
|
| 3563 |
+
"step": 503
|
| 3564 |
+
},
|
| 3565 |
+
{
|
| 3566 |
+
"epoch": 10.731182795698924,
|
| 3567 |
+
"grad_norm": 0.6332603693008423,
|
| 3568 |
+
"learning_rate": 0.00012168387609213663,
|
| 3569 |
+
"loss": 0.31723546981811523,
|
| 3570 |
+
"step": 504
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 10.75268817204301,
|
| 3574 |
+
"grad_norm": 0.6930424571037292,
|
| 3575 |
+
"learning_rate": 0.0001215250198570294,
|
| 3576 |
+
"loss": 0.36203786730766296,
|
| 3577 |
+
"step": 505
|
| 3578 |
+
},
|
| 3579 |
+
{
|
| 3580 |
+
"epoch": 10.774193548387096,
|
| 3581 |
+
"grad_norm": 0.758611798286438,
|
| 3582 |
+
"learning_rate": 0.00012136616362192217,
|
| 3583 |
+
"loss": 0.38453805446624756,
|
| 3584 |
+
"step": 506
|
| 3585 |
+
},
|
| 3586 |
+
{
|
| 3587 |
+
"epoch": 10.795698924731182,
|
| 3588 |
+
"grad_norm": 0.7258947491645813,
|
| 3589 |
+
"learning_rate": 0.00012120730738681494,
|
| 3590 |
+
"loss": 0.3250986635684967,
|
| 3591 |
+
"step": 507
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 10.817204301075268,
|
| 3595 |
+
"grad_norm": 0.8399383425712585,
|
| 3596 |
+
"learning_rate": 0.0001210484511517077,
|
| 3597 |
+
"loss": 0.36206358671188354,
|
| 3598 |
+
"step": 508
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 10.838709677419354,
|
| 3602 |
+
"grad_norm": 0.7296494841575623,
|
| 3603 |
+
"learning_rate": 0.00012088959491660048,
|
| 3604 |
+
"loss": 0.2987769544124603,
|
| 3605 |
+
"step": 509
|
| 3606 |
+
},
|
| 3607 |
+
{
|
| 3608 |
+
"epoch": 10.86021505376344,
|
| 3609 |
+
"grad_norm": 0.724338710308075,
|
| 3610 |
+
"learning_rate": 0.00012073073868149326,
|
| 3611 |
+
"loss": 0.39433813095092773,
|
| 3612 |
+
"step": 510
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 10.881720430107526,
|
| 3616 |
+
"grad_norm": 1.2827945947647095,
|
| 3617 |
+
"learning_rate": 0.00012057188244638602,
|
| 3618 |
+
"loss": 0.42414143681526184,
|
| 3619 |
+
"step": 511
|
| 3620 |
+
},
|
| 3621 |
+
{
|
| 3622 |
+
"epoch": 10.903225806451612,
|
| 3623 |
+
"grad_norm": 0.6901166439056396,
|
| 3624 |
+
"learning_rate": 0.0001204130262112788,
|
| 3625 |
+
"loss": 0.3207138180732727,
|
| 3626 |
+
"step": 512
|
| 3627 |
+
},
|
| 3628 |
+
{
|
| 3629 |
+
"epoch": 10.924731182795698,
|
| 3630 |
+
"grad_norm": 0.7400676012039185,
|
| 3631 |
+
"learning_rate": 0.00012025416997617156,
|
| 3632 |
+
"loss": 0.4030807316303253,
|
| 3633 |
+
"step": 513
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 10.946236559139784,
|
| 3637 |
+
"grad_norm": 0.7329428791999817,
|
| 3638 |
+
"learning_rate": 0.00012009531374106436,
|
| 3639 |
+
"loss": 0.4216151833534241,
|
| 3640 |
+
"step": 514
|
| 3641 |
+
},
|
| 3642 |
+
{
|
| 3643 |
+
"epoch": 10.967741935483872,
|
| 3644 |
+
"grad_norm": 0.7346771359443665,
|
| 3645 |
+
"learning_rate": 0.00011993645750595712,
|
| 3646 |
+
"loss": 0.4015253782272339,
|
| 3647 |
+
"step": 515
|
| 3648 |
+
},
|
| 3649 |
+
{
|
| 3650 |
+
"epoch": 10.989247311827956,
|
| 3651 |
+
"grad_norm": 0.7754795551300049,
|
| 3652 |
+
"learning_rate": 0.0001197776012708499,
|
| 3653 |
+
"loss": 0.3754139542579651,
|
| 3654 |
+
"step": 516
|
| 3655 |
+
},
|
| 3656 |
+
{
|
| 3657 |
+
"epoch": 11.0,
|
| 3658 |
+
"grad_norm": 0.9674854278564453,
|
| 3659 |
+
"learning_rate": 0.00011961874503574266,
|
| 3660 |
+
"loss": 0.4562542736530304,
|
| 3661 |
+
"step": 517
|
| 3662 |
+
},
|
| 3663 |
+
{
|
| 3664 |
+
"epoch": 11.021505376344086,
|
| 3665 |
+
"grad_norm": 0.6914543509483337,
|
| 3666 |
+
"learning_rate": 0.00011945988880063544,
|
| 3667 |
+
"loss": 0.3569665849208832,
|
| 3668 |
+
"step": 518
|
| 3669 |
+
},
|
| 3670 |
+
{
|
| 3671 |
+
"epoch": 11.043010752688172,
|
| 3672 |
+
"grad_norm": 0.7606181502342224,
|
| 3673 |
+
"learning_rate": 0.0001193010325655282,
|
| 3674 |
+
"loss": 0.3248334228992462,
|
| 3675 |
+
"step": 519
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 11.064516129032258,
|
| 3679 |
+
"grad_norm": 0.764230489730835,
|
| 3680 |
+
"learning_rate": 0.00011914217633042097,
|
| 3681 |
+
"loss": 0.30292606353759766,
|
| 3682 |
+
"step": 520
|
| 3683 |
+
},
|
| 3684 |
+
{
|
| 3685 |
+
"epoch": 11.086021505376344,
|
| 3686 |
+
"grad_norm": 0.8754153847694397,
|
| 3687 |
+
"learning_rate": 0.00011898332009531375,
|
| 3688 |
+
"loss": 0.32286104559898376,
|
| 3689 |
+
"step": 521
|
| 3690 |
+
},
|
| 3691 |
+
{
|
| 3692 |
+
"epoch": 11.10752688172043,
|
| 3693 |
+
"grad_norm": 0.9021672606468201,
|
| 3694 |
+
"learning_rate": 0.00011882446386020651,
|
| 3695 |
+
"loss": 0.37493667006492615,
|
| 3696 |
+
"step": 522
|
| 3697 |
+
},
|
| 3698 |
+
{
|
| 3699 |
+
"epoch": 11.129032258064516,
|
| 3700 |
+
"grad_norm": 0.7618964910507202,
|
| 3701 |
+
"learning_rate": 0.00011866560762509929,
|
| 3702 |
+
"loss": 0.2827898859977722,
|
| 3703 |
+
"step": 523
|
| 3704 |
+
},
|
| 3705 |
+
{
|
| 3706 |
+
"epoch": 11.150537634408602,
|
| 3707 |
+
"grad_norm": 0.7415357828140259,
|
| 3708 |
+
"learning_rate": 0.00011850675138999205,
|
| 3709 |
+
"loss": 0.3536837697029114,
|
| 3710 |
+
"step": 524
|
| 3711 |
+
},
|
| 3712 |
+
{
|
| 3713 |
+
"epoch": 11.172043010752688,
|
| 3714 |
+
"grad_norm": 0.7170859575271606,
|
| 3715 |
+
"learning_rate": 0.00011834789515488482,
|
| 3716 |
+
"loss": 0.3306007385253906,
|
| 3717 |
+
"step": 525
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 11.193548387096774,
|
| 3721 |
+
"grad_norm": 0.7188425660133362,
|
| 3722 |
+
"learning_rate": 0.00011818903891977761,
|
| 3723 |
+
"loss": 0.312721312046051,
|
| 3724 |
+
"step": 526
|
| 3725 |
+
},
|
| 3726 |
+
{
|
| 3727 |
+
"epoch": 11.21505376344086,
|
| 3728 |
+
"grad_norm": 0.7460401058197021,
|
| 3729 |
+
"learning_rate": 0.00011803018268467039,
|
| 3730 |
+
"loss": 0.34166038036346436,
|
| 3731 |
+
"step": 527
|
| 3732 |
+
},
|
| 3733 |
+
{
|
| 3734 |
+
"epoch": 11.236559139784946,
|
| 3735 |
+
"grad_norm": 0.7806810736656189,
|
| 3736 |
+
"learning_rate": 0.00011787132644956315,
|
| 3737 |
+
"loss": 0.28128740191459656,
|
| 3738 |
+
"step": 528
|
| 3739 |
+
},
|
| 3740 |
+
{
|
| 3741 |
+
"epoch": 11.258064516129032,
|
| 3742 |
+
"grad_norm": 0.7604184746742249,
|
| 3743 |
+
"learning_rate": 0.00011771247021445593,
|
| 3744 |
+
"loss": 0.25918617844581604,
|
| 3745 |
+
"step": 529
|
| 3746 |
+
},
|
| 3747 |
+
{
|
| 3748 |
+
"epoch": 11.279569892473118,
|
| 3749 |
+
"grad_norm": 0.9017161726951599,
|
| 3750 |
+
"learning_rate": 0.0001175536139793487,
|
| 3751 |
+
"loss": 0.3212778866291046,
|
| 3752 |
+
"step": 530
|
| 3753 |
+
},
|
| 3754 |
+
{
|
| 3755 |
+
"epoch": 11.301075268817204,
|
| 3756 |
+
"grad_norm": 0.7170535326004028,
|
| 3757 |
+
"learning_rate": 0.00011739475774424147,
|
| 3758 |
+
"loss": 0.20179462432861328,
|
| 3759 |
+
"step": 531
|
| 3760 |
+
},
|
| 3761 |
+
{
|
| 3762 |
+
"epoch": 11.32258064516129,
|
| 3763 |
+
"grad_norm": 0.7809444069862366,
|
| 3764 |
+
"learning_rate": 0.00011723590150913424,
|
| 3765 |
+
"loss": 0.2983512282371521,
|
| 3766 |
+
"step": 532
|
| 3767 |
+
},
|
| 3768 |
+
{
|
| 3769 |
+
"epoch": 11.344086021505376,
|
| 3770 |
+
"grad_norm": 0.7415927052497864,
|
| 3771 |
+
"learning_rate": 0.000117077045274027,
|
| 3772 |
+
"loss": 0.2575104534626007,
|
| 3773 |
+
"step": 533
|
| 3774 |
+
},
|
| 3775 |
+
{
|
| 3776 |
+
"epoch": 11.365591397849462,
|
| 3777 |
+
"grad_norm": 0.7183315753936768,
|
| 3778 |
+
"learning_rate": 0.00011691818903891978,
|
| 3779 |
+
"loss": 0.27599287033081055,
|
| 3780 |
+
"step": 534
|
| 3781 |
+
},
|
| 3782 |
+
{
|
| 3783 |
+
"epoch": 11.387096774193548,
|
| 3784 |
+
"grad_norm": 1.682211995124817,
|
| 3785 |
+
"learning_rate": 0.00011675933280381256,
|
| 3786 |
+
"loss": 0.3844273090362549,
|
| 3787 |
+
"step": 535
|
| 3788 |
+
},
|
| 3789 |
+
{
|
| 3790 |
+
"epoch": 11.408602150537634,
|
| 3791 |
+
"grad_norm": 0.707069993019104,
|
| 3792 |
+
"learning_rate": 0.00011660047656870532,
|
| 3793 |
+
"loss": 0.28967398405075073,
|
| 3794 |
+
"step": 536
|
| 3795 |
+
},
|
| 3796 |
+
{
|
| 3797 |
+
"epoch": 11.43010752688172,
|
| 3798 |
+
"grad_norm": 0.7777815461158752,
|
| 3799 |
+
"learning_rate": 0.0001164416203335981,
|
| 3800 |
+
"loss": 0.36004626750946045,
|
| 3801 |
+
"step": 537
|
| 3802 |
+
},
|
| 3803 |
+
{
|
| 3804 |
+
"epoch": 11.451612903225806,
|
| 3805 |
+
"grad_norm": 0.8384584188461304,
|
| 3806 |
+
"learning_rate": 0.00011628276409849088,
|
| 3807 |
+
"loss": 0.26809942722320557,
|
| 3808 |
+
"step": 538
|
| 3809 |
+
},
|
| 3810 |
+
{
|
| 3811 |
+
"epoch": 11.473118279569892,
|
| 3812 |
+
"grad_norm": 0.8009209036827087,
|
| 3813 |
+
"learning_rate": 0.00011612390786338366,
|
| 3814 |
+
"loss": 0.31199246644973755,
|
| 3815 |
+
"step": 539
|
| 3816 |
+
},
|
| 3817 |
+
{
|
| 3818 |
+
"epoch": 11.494623655913978,
|
| 3819 |
+
"grad_norm": 0.7819718718528748,
|
| 3820 |
+
"learning_rate": 0.00011596505162827642,
|
| 3821 |
+
"loss": 0.371052622795105,
|
| 3822 |
+
"step": 540
|
| 3823 |
+
},
|
| 3824 |
+
{
|
| 3825 |
+
"epoch": 11.516129032258064,
|
| 3826 |
+
"grad_norm": 0.7900601029396057,
|
| 3827 |
+
"learning_rate": 0.0001158061953931692,
|
| 3828 |
+
"loss": 0.29752612113952637,
|
| 3829 |
+
"step": 541
|
| 3830 |
+
},
|
| 3831 |
+
{
|
| 3832 |
+
"epoch": 11.53763440860215,
|
| 3833 |
+
"grad_norm": 0.7695387005805969,
|
| 3834 |
+
"learning_rate": 0.00011564733915806196,
|
| 3835 |
+
"loss": 0.2954059839248657,
|
| 3836 |
+
"step": 542
|
| 3837 |
+
},
|
| 3838 |
+
{
|
| 3839 |
+
"epoch": 11.559139784946236,
|
| 3840 |
+
"grad_norm": 0.7167935371398926,
|
| 3841 |
+
"learning_rate": 0.00011548848292295473,
|
| 3842 |
+
"loss": 0.3413902223110199,
|
| 3843 |
+
"step": 543
|
| 3844 |
+
},
|
| 3845 |
+
{
|
| 3846 |
+
"epoch": 11.580645161290322,
|
| 3847 |
+
"grad_norm": 0.9060370922088623,
|
| 3848 |
+
"learning_rate": 0.0001153296266878475,
|
| 3849 |
+
"loss": 0.3013741075992584,
|
| 3850 |
+
"step": 544
|
| 3851 |
+
},
|
| 3852 |
+
{
|
| 3853 |
+
"epoch": 11.602150537634408,
|
| 3854 |
+
"grad_norm": 0.7279766798019409,
|
| 3855 |
+
"learning_rate": 0.00011517077045274027,
|
| 3856 |
+
"loss": 0.26471853256225586,
|
| 3857 |
+
"step": 545
|
| 3858 |
+
},
|
| 3859 |
+
{
|
| 3860 |
+
"epoch": 11.623655913978494,
|
| 3861 |
+
"grad_norm": 0.7126449942588806,
|
| 3862 |
+
"learning_rate": 0.00011501191421763305,
|
| 3863 |
+
"loss": 0.3059503138065338,
|
| 3864 |
+
"step": 546
|
| 3865 |
+
},
|
| 3866 |
+
{
|
| 3867 |
+
"epoch": 11.64516129032258,
|
| 3868 |
+
"grad_norm": 0.8045353293418884,
|
| 3869 |
+
"learning_rate": 0.00011485305798252581,
|
| 3870 |
+
"loss": 0.22076934576034546,
|
| 3871 |
+
"step": 547
|
| 3872 |
+
},
|
| 3873 |
+
{
|
| 3874 |
+
"epoch": 11.666666666666666,
|
| 3875 |
+
"grad_norm": 0.8084357380867004,
|
| 3876 |
+
"learning_rate": 0.00011469420174741859,
|
| 3877 |
+
"loss": 0.3207700252532959,
|
| 3878 |
+
"step": 548
|
| 3879 |
+
},
|
| 3880 |
+
{
|
| 3881 |
+
"epoch": 11.688172043010752,
|
| 3882 |
+
"grad_norm": 0.7768396139144897,
|
| 3883 |
+
"learning_rate": 0.00011453534551231135,
|
| 3884 |
+
"loss": 0.36251944303512573,
|
| 3885 |
+
"step": 549
|
| 3886 |
+
},
|
| 3887 |
+
{
|
| 3888 |
+
"epoch": 11.709677419354838,
|
| 3889 |
+
"grad_norm": 0.8002042770385742,
|
| 3890 |
+
"learning_rate": 0.00011437648927720415,
|
| 3891 |
+
"loss": 0.3310307264328003,
|
| 3892 |
+
"step": 550
|
| 3893 |
+
},
|
| 3894 |
+
{
|
| 3895 |
+
"epoch": 11.731182795698924,
|
| 3896 |
+
"grad_norm": 0.8118063807487488,
|
| 3897 |
+
"learning_rate": 0.00011421763304209691,
|
| 3898 |
+
"loss": 0.4163286089897156,
|
| 3899 |
+
"step": 551
|
| 3900 |
+
},
|
| 3901 |
+
{
|
| 3902 |
+
"epoch": 11.75268817204301,
|
| 3903 |
+
"grad_norm": 0.8167386054992676,
|
| 3904 |
+
"learning_rate": 0.00011405877680698969,
|
| 3905 |
+
"loss": 0.3315795660018921,
|
| 3906 |
+
"step": 552
|
| 3907 |
+
},
|
| 3908 |
+
{
|
| 3909 |
+
"epoch": 11.774193548387096,
|
| 3910 |
+
"grad_norm": 0.7709652185440063,
|
| 3911 |
+
"learning_rate": 0.00011389992057188245,
|
| 3912 |
+
"loss": 0.3589479327201843,
|
| 3913 |
+
"step": 553
|
| 3914 |
+
},
|
| 3915 |
+
{
|
| 3916 |
+
"epoch": 11.795698924731182,
|
| 3917 |
+
"grad_norm": 0.7389516830444336,
|
| 3918 |
+
"learning_rate": 0.00011374106433677523,
|
| 3919 |
+
"loss": 0.3652232587337494,
|
| 3920 |
+
"step": 554
|
| 3921 |
+
},
|
| 3922 |
+
{
|
| 3923 |
+
"epoch": 11.817204301075268,
|
| 3924 |
+
"grad_norm": 0.7422162294387817,
|
| 3925 |
+
"learning_rate": 0.000113582208101668,
|
| 3926 |
+
"loss": 0.3981531262397766,
|
| 3927 |
+
"step": 555
|
| 3928 |
+
},
|
| 3929 |
+
{
|
| 3930 |
+
"epoch": 11.838709677419354,
|
| 3931 |
+
"grad_norm": 0.7577189803123474,
|
| 3932 |
+
"learning_rate": 0.00011342335186656076,
|
| 3933 |
+
"loss": 0.2992894649505615,
|
| 3934 |
+
"step": 556
|
| 3935 |
+
},
|
| 3936 |
+
{
|
| 3937 |
+
"epoch": 11.86021505376344,
|
| 3938 |
+
"grad_norm": 0.7978619337081909,
|
| 3939 |
+
"learning_rate": 0.00011326449563145354,
|
| 3940 |
+
"loss": 0.30227142572402954,
|
| 3941 |
+
"step": 557
|
| 3942 |
+
},
|
| 3943 |
+
{
|
| 3944 |
+
"epoch": 11.881720430107526,
|
| 3945 |
+
"grad_norm": 0.7575592398643494,
|
| 3946 |
+
"learning_rate": 0.0001131056393963463,
|
| 3947 |
+
"loss": 0.3341229557991028,
|
| 3948 |
+
"step": 558
|
| 3949 |
+
},
|
| 3950 |
+
{
|
| 3951 |
+
"epoch": 11.903225806451612,
|
| 3952 |
+
"grad_norm": 0.6892173886299133,
|
| 3953 |
+
"learning_rate": 0.00011294678316123908,
|
| 3954 |
+
"loss": 0.36578071117401123,
|
| 3955 |
+
"step": 559
|
| 3956 |
+
},
|
| 3957 |
+
{
|
| 3958 |
+
"epoch": 11.924731182795698,
|
| 3959 |
+
"grad_norm": 0.796484649181366,
|
| 3960 |
+
"learning_rate": 0.00011278792692613185,
|
| 3961 |
+
"loss": 0.35607850551605225,
|
| 3962 |
+
"step": 560
|
| 3963 |
+
},
|
| 3964 |
+
{
|
| 3965 |
+
"epoch": 11.946236559139784,
|
| 3966 |
+
"grad_norm": 0.7589250206947327,
|
| 3967 |
+
"learning_rate": 0.00011262907069102462,
|
| 3968 |
+
"loss": 0.3695809841156006,
|
| 3969 |
+
"step": 561
|
| 3970 |
+
},
|
| 3971 |
+
{
|
| 3972 |
+
"epoch": 11.967741935483872,
|
| 3973 |
+
"grad_norm": 0.8040224313735962,
|
| 3974 |
+
"learning_rate": 0.0001124702144559174,
|
| 3975 |
+
"loss": 0.3417878746986389,
|
| 3976 |
+
"step": 562
|
| 3977 |
+
},
|
| 3978 |
+
{
|
| 3979 |
+
"epoch": 11.989247311827956,
|
| 3980 |
+
"grad_norm": 0.808239758014679,
|
| 3981 |
+
"learning_rate": 0.00011231135822081018,
|
| 3982 |
+
"loss": 0.2793850898742676,
|
| 3983 |
+
"step": 563
|
| 3984 |
+
},
|
| 3985 |
+
{
|
| 3986 |
+
"epoch": 12.0,
|
| 3987 |
+
"grad_norm": 0.9436281323432922,
|
| 3988 |
+
"learning_rate": 0.00011215250198570296,
|
| 3989 |
+
"loss": 0.3655521273612976,
|
| 3990 |
+
"step": 564
|
| 3991 |
+
},
|
| 3992 |
+
{
|
| 3993 |
+
"epoch": 12.021505376344086,
|
| 3994 |
+
"grad_norm": 0.7834585905075073,
|
| 3995 |
+
"learning_rate": 0.00011199364575059572,
|
| 3996 |
+
"loss": 0.30738168954849243,
|
| 3997 |
+
"step": 565
|
| 3998 |
+
},
|
| 3999 |
+
{
|
| 4000 |
+
"epoch": 12.043010752688172,
|
| 4001 |
+
"grad_norm": 0.795142412185669,
|
| 4002 |
+
"learning_rate": 0.0001118347895154885,
|
| 4003 |
+
"loss": 0.2081073522567749,
|
| 4004 |
+
"step": 566
|
| 4005 |
+
},
|
| 4006 |
+
{
|
| 4007 |
+
"epoch": 12.064516129032258,
|
| 4008 |
+
"grad_norm": 0.6989514827728271,
|
| 4009 |
+
"learning_rate": 0.00011167593328038126,
|
| 4010 |
+
"loss": 0.23969483375549316,
|
| 4011 |
+
"step": 567
|
| 4012 |
+
},
|
| 4013 |
+
{
|
| 4014 |
+
"epoch": 12.086021505376344,
|
| 4015 |
+
"grad_norm": 0.9184768199920654,
|
| 4016 |
+
"learning_rate": 0.00011151707704527403,
|
| 4017 |
+
"loss": 0.23692801594734192,
|
| 4018 |
+
"step": 568
|
| 4019 |
+
},
|
| 4020 |
+
{
|
| 4021 |
+
"epoch": 12.10752688172043,
|
| 4022 |
+
"grad_norm": 0.8651571273803711,
|
| 4023 |
+
"learning_rate": 0.0001113582208101668,
|
| 4024 |
+
"loss": 0.2645634412765503,
|
| 4025 |
+
"step": 569
|
| 4026 |
+
},
|
| 4027 |
+
{
|
| 4028 |
+
"epoch": 12.129032258064516,
|
| 4029 |
+
"grad_norm": 1.062126874923706,
|
| 4030 |
+
"learning_rate": 0.00011119936457505957,
|
| 4031 |
+
"loss": 0.2957659065723419,
|
| 4032 |
+
"step": 570
|
| 4033 |
+
},
|
| 4034 |
+
{
|
| 4035 |
+
"epoch": 12.150537634408602,
|
| 4036 |
+
"grad_norm": 0.7741546630859375,
|
| 4037 |
+
"learning_rate": 0.00011104050833995235,
|
| 4038 |
+
"loss": 0.22269803285598755,
|
| 4039 |
+
"step": 571
|
| 4040 |
+
},
|
| 4041 |
+
{
|
| 4042 |
+
"epoch": 12.172043010752688,
|
| 4043 |
+
"grad_norm": 0.6743382215499878,
|
| 4044 |
+
"learning_rate": 0.00011088165210484511,
|
| 4045 |
+
"loss": 0.24519725143909454,
|
| 4046 |
+
"step": 572
|
| 4047 |
+
},
|
| 4048 |
+
{
|
| 4049 |
+
"epoch": 12.193548387096774,
|
| 4050 |
+
"grad_norm": 0.7190341949462891,
|
| 4051 |
+
"learning_rate": 0.00011072279586973789,
|
| 4052 |
+
"loss": 0.21414649486541748,
|
| 4053 |
+
"step": 573
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 12.21505376344086,
|
| 4057 |
+
"grad_norm": 1.0120435953140259,
|
| 4058 |
+
"learning_rate": 0.00011056393963463067,
|
| 4059 |
+
"loss": 0.2711429297924042,
|
| 4060 |
+
"step": 574
|
| 4061 |
+
},
|
| 4062 |
+
{
|
| 4063 |
+
"epoch": 12.236559139784946,
|
| 4064 |
+
"grad_norm": 0.7273455858230591,
|
| 4065 |
+
"learning_rate": 0.00011040508339952345,
|
| 4066 |
+
"loss": 0.26170241832733154,
|
| 4067 |
+
"step": 575
|
| 4068 |
+
},
|
| 4069 |
+
{
|
| 4070 |
+
"epoch": 12.258064516129032,
|
| 4071 |
+
"grad_norm": 0.7552404999732971,
|
| 4072 |
+
"learning_rate": 0.00011024622716441621,
|
| 4073 |
+
"loss": 0.3006550669670105,
|
| 4074 |
+
"step": 576
|
| 4075 |
+
},
|
| 4076 |
+
{
|
| 4077 |
+
"epoch": 12.279569892473118,
|
| 4078 |
+
"grad_norm": 0.9425163865089417,
|
| 4079 |
+
"learning_rate": 0.00011008737092930899,
|
| 4080 |
+
"loss": 0.1508064717054367,
|
| 4081 |
+
"step": 577
|
| 4082 |
+
},
|
| 4083 |
+
{
|
| 4084 |
+
"epoch": 12.301075268817204,
|
| 4085 |
+
"grad_norm": 0.9021785855293274,
|
| 4086 |
+
"learning_rate": 0.00010992851469420175,
|
| 4087 |
+
"loss": 0.3232599198818207,
|
| 4088 |
+
"step": 578
|
| 4089 |
+
},
|
| 4090 |
+
{
|
| 4091 |
+
"epoch": 12.32258064516129,
|
| 4092 |
+
"grad_norm": 0.7773718237876892,
|
| 4093 |
+
"learning_rate": 0.00010976965845909453,
|
| 4094 |
+
"loss": 0.19899800419807434,
|
| 4095 |
+
"step": 579
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 12.344086021505376,
|
| 4099 |
+
"grad_norm": 0.8534666895866394,
|
| 4100 |
+
"learning_rate": 0.0001096108022239873,
|
| 4101 |
+
"loss": 0.2873285412788391,
|
| 4102 |
+
"step": 580
|
| 4103 |
+
},
|
| 4104 |
+
{
|
| 4105 |
+
"epoch": 12.365591397849462,
|
| 4106 |
+
"grad_norm": 0.9202722907066345,
|
| 4107 |
+
"learning_rate": 0.00010945194598888006,
|
| 4108 |
+
"loss": 0.3199685513973236,
|
| 4109 |
+
"step": 581
|
| 4110 |
+
},
|
| 4111 |
+
{
|
| 4112 |
+
"epoch": 12.387096774193548,
|
| 4113 |
+
"grad_norm": 0.8233513832092285,
|
| 4114 |
+
"learning_rate": 0.00010929308975377284,
|
| 4115 |
+
"loss": 0.3001951575279236,
|
| 4116 |
+
"step": 582
|
| 4117 |
+
},
|
| 4118 |
+
{
|
| 4119 |
+
"epoch": 12.408602150537634,
|
| 4120 |
+
"grad_norm": 0.8201817870140076,
|
| 4121 |
+
"learning_rate": 0.0001091342335186656,
|
| 4122 |
+
"loss": 0.23249655961990356,
|
| 4123 |
+
"step": 583
|
| 4124 |
+
},
|
| 4125 |
+
{
|
| 4126 |
+
"epoch": 12.43010752688172,
|
| 4127 |
+
"grad_norm": 0.7741047739982605,
|
| 4128 |
+
"learning_rate": 0.00010897537728355838,
|
| 4129 |
+
"loss": 0.1689053624868393,
|
| 4130 |
+
"step": 584
|
| 4131 |
+
},
|
| 4132 |
+
{
|
| 4133 |
+
"epoch": 12.451612903225806,
|
| 4134 |
+
"grad_norm": 0.8471128940582275,
|
| 4135 |
+
"learning_rate": 0.00010881652104845115,
|
| 4136 |
+
"loss": 0.2790966033935547,
|
| 4137 |
+
"step": 585
|
| 4138 |
+
},
|
| 4139 |
+
{
|
| 4140 |
+
"epoch": 12.473118279569892,
|
| 4141 |
+
"grad_norm": 0.9554393291473389,
|
| 4142 |
+
"learning_rate": 0.00010865766481334392,
|
| 4143 |
+
"loss": 0.27663490176200867,
|
| 4144 |
+
"step": 586
|
| 4145 |
+
},
|
| 4146 |
+
{
|
| 4147 |
+
"epoch": 12.494623655913978,
|
| 4148 |
+
"grad_norm": 0.7771924734115601,
|
| 4149 |
+
"learning_rate": 0.0001084988085782367,
|
| 4150 |
+
"loss": 0.29540884494781494,
|
| 4151 |
+
"step": 587
|
| 4152 |
+
},
|
| 4153 |
+
{
|
| 4154 |
+
"epoch": 12.516129032258064,
|
| 4155 |
+
"grad_norm": 0.980816125869751,
|
| 4156 |
+
"learning_rate": 0.00010833995234312948,
|
| 4157 |
+
"loss": 0.3429156243801117,
|
| 4158 |
+
"step": 588
|
| 4159 |
+
},
|
| 4160 |
+
{
|
| 4161 |
+
"epoch": 12.53763440860215,
|
| 4162 |
+
"grad_norm": 0.8538633584976196,
|
| 4163 |
+
"learning_rate": 0.00010818109610802226,
|
| 4164 |
+
"loss": 0.2047094702720642,
|
| 4165 |
+
"step": 589
|
| 4166 |
+
},
|
| 4167 |
+
{
|
| 4168 |
+
"epoch": 12.559139784946236,
|
| 4169 |
+
"grad_norm": 0.8534285426139832,
|
| 4170 |
+
"learning_rate": 0.00010802223987291502,
|
| 4171 |
+
"loss": 0.3163786828517914,
|
| 4172 |
+
"step": 590
|
| 4173 |
+
},
|
| 4174 |
+
{
|
| 4175 |
+
"epoch": 12.580645161290322,
|
| 4176 |
+
"grad_norm": 0.7636980414390564,
|
| 4177 |
+
"learning_rate": 0.0001078633836378078,
|
| 4178 |
+
"loss": 0.22881919145584106,
|
| 4179 |
+
"step": 591
|
| 4180 |
+
},
|
| 4181 |
+
{
|
| 4182 |
+
"epoch": 12.602150537634408,
|
| 4183 |
+
"grad_norm": 0.808996856212616,
|
| 4184 |
+
"learning_rate": 0.00010770452740270056,
|
| 4185 |
+
"loss": 0.235946387052536,
|
| 4186 |
+
"step": 592
|
| 4187 |
+
},
|
| 4188 |
+
{
|
| 4189 |
+
"epoch": 12.623655913978494,
|
| 4190 |
+
"grad_norm": 0.8072691559791565,
|
| 4191 |
+
"learning_rate": 0.00010754567116759333,
|
| 4192 |
+
"loss": 0.2396862506866455,
|
| 4193 |
+
"step": 593
|
| 4194 |
+
},
|
| 4195 |
+
{
|
| 4196 |
+
"epoch": 12.64516129032258,
|
| 4197 |
+
"grad_norm": 0.829767644405365,
|
| 4198 |
+
"learning_rate": 0.00010738681493248611,
|
| 4199 |
+
"loss": 0.3337879776954651,
|
| 4200 |
+
"step": 594
|
| 4201 |
+
},
|
| 4202 |
+
{
|
| 4203 |
+
"epoch": 12.666666666666666,
|
| 4204 |
+
"grad_norm": 0.8298752307891846,
|
| 4205 |
+
"learning_rate": 0.00010722795869737887,
|
| 4206 |
+
"loss": 0.3443082571029663,
|
| 4207 |
+
"step": 595
|
| 4208 |
+
},
|
| 4209 |
+
{
|
| 4210 |
+
"epoch": 12.688172043010752,
|
| 4211 |
+
"grad_norm": 0.8143765330314636,
|
| 4212 |
+
"learning_rate": 0.00010706910246227165,
|
| 4213 |
+
"loss": 0.2124541848897934,
|
| 4214 |
+
"step": 596
|
| 4215 |
+
},
|
| 4216 |
+
{
|
| 4217 |
+
"epoch": 12.709677419354838,
|
| 4218 |
+
"grad_norm": 0.8767410516738892,
|
| 4219 |
+
"learning_rate": 0.00010691024622716441,
|
| 4220 |
+
"loss": 0.33744725584983826,
|
| 4221 |
+
"step": 597
|
| 4222 |
+
},
|
| 4223 |
+
{
|
| 4224 |
+
"epoch": 12.731182795698924,
|
| 4225 |
+
"grad_norm": 0.7700532674789429,
|
| 4226 |
+
"learning_rate": 0.00010675138999205718,
|
| 4227 |
+
"loss": 0.28435879945755005,
|
| 4228 |
+
"step": 598
|
| 4229 |
+
},
|
| 4230 |
+
{
|
| 4231 |
+
"epoch": 12.75268817204301,
|
| 4232 |
+
"grad_norm": 0.8594262003898621,
|
| 4233 |
+
"learning_rate": 0.00010659253375694997,
|
| 4234 |
+
"loss": 0.21120105683803558,
|
| 4235 |
+
"step": 599
|
| 4236 |
+
},
|
| 4237 |
+
{
|
| 4238 |
+
"epoch": 12.774193548387096,
|
| 4239 |
+
"grad_norm": 0.7381999492645264,
|
| 4240 |
+
"learning_rate": 0.00010643367752184275,
|
| 4241 |
+
"loss": 0.1675238013267517,
|
| 4242 |
+
"step": 600
|
| 4243 |
+
},
|
| 4244 |
+
{
|
| 4245 |
+
"epoch": 12.774193548387096,
|
| 4246 |
+
"eval_loss": 0.9527648091316223,
|
| 4247 |
+
"eval_runtime": 91.226,
|
| 4248 |
+
"eval_samples_per_second": 0.46,
|
| 4249 |
+
"eval_steps_per_second": 0.46,
|
| 4250 |
+
"step": 600
|
| 4251 |
}
|
| 4252 |
],
|
| 4253 |
"logging_steps": 1,
|
|
|
|
| 4267 |
"attributes": {}
|
| 4268 |
}
|
| 4269 |
},
|
| 4270 |
+
"total_flos": 3.9436180253965824e+17,
|
| 4271 |
"train_batch_size": 2,
|
| 4272 |
"trial_name": null,
|
| 4273 |
"trial_params": null
|