Training in progress, step 800, 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 323014168
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2316cbf7807803c10504f0552a3df41c6728ec9769dcfef615a415ea7c583af0
|
| 3 |
size 323014168
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 165484738
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:833cdbd9a1b69a3478278dfc0e64ccecf27f27d112d125d68306b87027ce427a
|
| 3 |
size 165484738
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e7b315b29c598de44532516d3ac125be85a14a2f551aba8e3d144136258f1dba
|
| 3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:d2c17f8131c6c5ea7f0a3b5c0825855397d8dc22a4373cf0983d7dd31b7657c7
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.3638736605644226,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-650",
|
| 4 |
-
"epoch": 1.
|
| 5 |
"eval_steps": 50,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5385,6 +5385,364 @@
|
|
| 5385 |
"eval_samples_per_second": 2.928,
|
| 5386 |
"eval_steps_per_second": 2.928,
|
| 5387 |
"step": 750
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5388 |
}
|
| 5389 |
],
|
| 5390 |
"logging_steps": 1,
|
|
@@ -5399,7 +5757,7 @@
|
|
| 5399 |
"early_stopping_threshold": 0.0
|
| 5400 |
},
|
| 5401 |
"attributes": {
|
| 5402 |
-
"early_stopping_patience_counter":
|
| 5403 |
}
|
| 5404 |
},
|
| 5405 |
"TrainerControl": {
|
|
@@ -5408,12 +5766,12 @@
|
|
| 5408 |
"should_evaluate": false,
|
| 5409 |
"should_log": false,
|
| 5410 |
"should_save": true,
|
| 5411 |
-
"should_training_stop":
|
| 5412 |
},
|
| 5413 |
"attributes": {}
|
| 5414 |
}
|
| 5415 |
},
|
| 5416 |
-
"total_flos":
|
| 5417 |
"train_batch_size": 1,
|
| 5418 |
"trial_name": null,
|
| 5419 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.3638736605644226,
|
| 3 |
"best_model_checkpoint": "miner_id_24/checkpoint-650",
|
| 4 |
+
"epoch": 1.3579461065138978,
|
| 5 |
"eval_steps": 50,
|
| 6 |
+
"global_step": 800,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5385 |
"eval_samples_per_second": 2.928,
|
| 5386 |
"eval_steps_per_second": 2.928,
|
| 5387 |
"step": 750
|
| 5388 |
+
},
|
| 5389 |
+
{
|
| 5390 |
+
"epoch": 1.2747719074899215,
|
| 5391 |
+
"grad_norm": 0.2629989981651306,
|
| 5392 |
+
"learning_rate": 4.4440055569454936e-05,
|
| 5393 |
+
"loss": 0.5687,
|
| 5394 |
+
"step": 751
|
| 5395 |
+
},
|
| 5396 |
+
{
|
| 5397 |
+
"epoch": 1.2764693401230638,
|
| 5398 |
+
"grad_norm": 0.29386061429977417,
|
| 5399 |
+
"learning_rate": 4.410240728232653e-05,
|
| 5400 |
+
"loss": 0.4559,
|
| 5401 |
+
"step": 752
|
| 5402 |
+
},
|
| 5403 |
+
{
|
| 5404 |
+
"epoch": 1.2781667727562063,
|
| 5405 |
+
"grad_norm": 0.32051095366477966,
|
| 5406 |
+
"learning_rate": 4.376582538278114e-05,
|
| 5407 |
+
"loss": 0.7155,
|
| 5408 |
+
"step": 753
|
| 5409 |
+
},
|
| 5410 |
+
{
|
| 5411 |
+
"epoch": 1.2798642053893485,
|
| 5412 |
+
"grad_norm": 0.2592981159687042,
|
| 5413 |
+
"learning_rate": 4.3430313260194697e-05,
|
| 5414 |
+
"loss": 0.4972,
|
| 5415 |
+
"step": 754
|
| 5416 |
+
},
|
| 5417 |
+
{
|
| 5418 |
+
"epoch": 1.281561638022491,
|
| 5419 |
+
"grad_norm": 0.23039335012435913,
|
| 5420 |
+
"learning_rate": 4.309587429317061e-05,
|
| 5421 |
+
"loss": 0.4049,
|
| 5422 |
+
"step": 755
|
| 5423 |
+
},
|
| 5424 |
+
{
|
| 5425 |
+
"epoch": 1.2832590706556333,
|
| 5426 |
+
"grad_norm": 0.2293672114610672,
|
| 5427 |
+
"learning_rate": 4.2762511849505476e-05,
|
| 5428 |
+
"loss": 0.4085,
|
| 5429 |
+
"step": 756
|
| 5430 |
+
},
|
| 5431 |
+
{
|
| 5432 |
+
"epoch": 1.2849565032887758,
|
| 5433 |
+
"grad_norm": 0.2180144041776657,
|
| 5434 |
+
"learning_rate": 4.2430229286155484e-05,
|
| 5435 |
+
"loss": 0.3829,
|
| 5436 |
+
"step": 757
|
| 5437 |
+
},
|
| 5438 |
+
{
|
| 5439 |
+
"epoch": 1.286653935921918,
|
| 5440 |
+
"grad_norm": 0.22328434884548187,
|
| 5441 |
+
"learning_rate": 4.209902994920235e-05,
|
| 5442 |
+
"loss": 0.381,
|
| 5443 |
+
"step": 758
|
| 5444 |
+
},
|
| 5445 |
+
{
|
| 5446 |
+
"epoch": 1.2883513685550605,
|
| 5447 |
+
"grad_norm": 0.19186857342720032,
|
| 5448 |
+
"learning_rate": 4.176891717381967e-05,
|
| 5449 |
+
"loss": 0.2676,
|
| 5450 |
+
"step": 759
|
| 5451 |
+
},
|
| 5452 |
+
{
|
| 5453 |
+
"epoch": 1.290048801188203,
|
| 5454 |
+
"grad_norm": 0.19644580781459808,
|
| 5455 |
+
"learning_rate": 4.143989428423947e-05,
|
| 5456 |
+
"loss": 0.2604,
|
| 5457 |
+
"step": 760
|
| 5458 |
+
},
|
| 5459 |
+
{
|
| 5460 |
+
"epoch": 1.2917462338213452,
|
| 5461 |
+
"grad_norm": 0.1470167487859726,
|
| 5462 |
+
"learning_rate": 4.111196459371862e-05,
|
| 5463 |
+
"loss": 0.1697,
|
| 5464 |
+
"step": 761
|
| 5465 |
+
},
|
| 5466 |
+
{
|
| 5467 |
+
"epoch": 1.2934436664544875,
|
| 5468 |
+
"grad_norm": 0.10344719886779785,
|
| 5469 |
+
"learning_rate": 4.0785131404505376e-05,
|
| 5470 |
+
"loss": 0.0742,
|
| 5471 |
+
"step": 762
|
| 5472 |
+
},
|
| 5473 |
+
{
|
| 5474 |
+
"epoch": 1.29514109908763,
|
| 5475 |
+
"grad_norm": 0.08508791774511337,
|
| 5476 |
+
"learning_rate": 4.045939800780639e-05,
|
| 5477 |
+
"loss": 0.0492,
|
| 5478 |
+
"step": 763
|
| 5479 |
+
},
|
| 5480 |
+
{
|
| 5481 |
+
"epoch": 1.2968385317207725,
|
| 5482 |
+
"grad_norm": 0.08410584181547165,
|
| 5483 |
+
"learning_rate": 4.0134767683753385e-05,
|
| 5484 |
+
"loss": 0.049,
|
| 5485 |
+
"step": 764
|
| 5486 |
+
},
|
| 5487 |
+
{
|
| 5488 |
+
"epoch": 1.2985359643539147,
|
| 5489 |
+
"grad_norm": 0.001906770863570273,
|
| 5490 |
+
"learning_rate": 3.981124370137001e-05,
|
| 5491 |
+
"loss": 0.0001,
|
| 5492 |
+
"step": 765
|
| 5493 |
+
},
|
| 5494 |
+
{
|
| 5495 |
+
"epoch": 1.300233396987057,
|
| 5496 |
+
"grad_norm": 0.14595600962638855,
|
| 5497 |
+
"learning_rate": 3.948882931853924e-05,
|
| 5498 |
+
"loss": 0.119,
|
| 5499 |
+
"step": 766
|
| 5500 |
+
},
|
| 5501 |
+
{
|
| 5502 |
+
"epoch": 1.3019308296201995,
|
| 5503 |
+
"grad_norm": 0.00104467140045017,
|
| 5504 |
+
"learning_rate": 3.916752778197039e-05,
|
| 5505 |
+
"loss": 0.0001,
|
| 5506 |
+
"step": 767
|
| 5507 |
+
},
|
| 5508 |
+
{
|
| 5509 |
+
"epoch": 1.303628262253342,
|
| 5510 |
+
"grad_norm": 0.00029723646002821624,
|
| 5511 |
+
"learning_rate": 3.8847342327166244e-05,
|
| 5512 |
+
"loss": 0.0,
|
| 5513 |
+
"step": 768
|
| 5514 |
+
},
|
| 5515 |
+
{
|
| 5516 |
+
"epoch": 1.3053256948864842,
|
| 5517 |
+
"grad_norm": 0.07592643052339554,
|
| 5518 |
+
"learning_rate": 3.852827617839084e-05,
|
| 5519 |
+
"loss": 0.026,
|
| 5520 |
+
"step": 769
|
| 5521 |
+
},
|
| 5522 |
+
{
|
| 5523 |
+
"epoch": 1.3070231275196265,
|
| 5524 |
+
"grad_norm": 0.00041246655746363103,
|
| 5525 |
+
"learning_rate": 3.8210332548636796e-05,
|
| 5526 |
+
"loss": 0.0,
|
| 5527 |
+
"step": 770
|
| 5528 |
+
},
|
| 5529 |
+
{
|
| 5530 |
+
"epoch": 1.308720560152769,
|
| 5531 |
+
"grad_norm": 0.002659448655322194,
|
| 5532 |
+
"learning_rate": 3.7893514639592895e-05,
|
| 5533 |
+
"loss": 0.0001,
|
| 5534 |
+
"step": 771
|
| 5535 |
+
},
|
| 5536 |
+
{
|
| 5537 |
+
"epoch": 1.3104179927859114,
|
| 5538 |
+
"grad_norm": 0.00021023498266004026,
|
| 5539 |
+
"learning_rate": 3.757782564161191e-05,
|
| 5540 |
+
"loss": 0.0,
|
| 5541 |
+
"step": 772
|
| 5542 |
+
},
|
| 5543 |
+
{
|
| 5544 |
+
"epoch": 1.3121154254190537,
|
| 5545 |
+
"grad_norm": 0.0010800276650115848,
|
| 5546 |
+
"learning_rate": 3.7263268733678606e-05,
|
| 5547 |
+
"loss": 0.0,
|
| 5548 |
+
"step": 773
|
| 5549 |
+
},
|
| 5550 |
+
{
|
| 5551 |
+
"epoch": 1.313812858052196,
|
| 5552 |
+
"grad_norm": 0.0006105787470005453,
|
| 5553 |
+
"learning_rate": 3.694984708337756e-05,
|
| 5554 |
+
"loss": 0.0,
|
| 5555 |
+
"step": 774
|
| 5556 |
+
},
|
| 5557 |
+
{
|
| 5558 |
+
"epoch": 1.3155102906853384,
|
| 5559 |
+
"grad_norm": 0.0007423445931635797,
|
| 5560 |
+
"learning_rate": 3.663756384686127e-05,
|
| 5561 |
+
"loss": 0.0,
|
| 5562 |
+
"step": 775
|
| 5563 |
+
},
|
| 5564 |
+
{
|
| 5565 |
+
"epoch": 1.317207723318481,
|
| 5566 |
+
"grad_norm": 0.0007892303401604295,
|
| 5567 |
+
"learning_rate": 3.632642216881847e-05,
|
| 5568 |
+
"loss": 0.0,
|
| 5569 |
+
"step": 776
|
| 5570 |
+
},
|
| 5571 |
+
{
|
| 5572 |
+
"epoch": 1.3189051559516232,
|
| 5573 |
+
"grad_norm": 0.0015005484456196427,
|
| 5574 |
+
"learning_rate": 3.601642518244247e-05,
|
| 5575 |
+
"loss": 0.0,
|
| 5576 |
+
"step": 777
|
| 5577 |
+
},
|
| 5578 |
+
{
|
| 5579 |
+
"epoch": 1.3206025885847654,
|
| 5580 |
+
"grad_norm": 0.0008817919879220426,
|
| 5581 |
+
"learning_rate": 3.570757600939939e-05,
|
| 5582 |
+
"loss": 0.0,
|
| 5583 |
+
"step": 778
|
| 5584 |
+
},
|
| 5585 |
+
{
|
| 5586 |
+
"epoch": 1.322300021217908,
|
| 5587 |
+
"grad_norm": 0.0003715140337590128,
|
| 5588 |
+
"learning_rate": 3.5399877759797e-05,
|
| 5589 |
+
"loss": 0.0,
|
| 5590 |
+
"step": 779
|
| 5591 |
+
},
|
| 5592 |
+
{
|
| 5593 |
+
"epoch": 1.3239974538510504,
|
| 5594 |
+
"grad_norm": 0.001832145731896162,
|
| 5595 |
+
"learning_rate": 3.509333353215331e-05,
|
| 5596 |
+
"loss": 0.0001,
|
| 5597 |
+
"step": 780
|
| 5598 |
+
},
|
| 5599 |
+
{
|
| 5600 |
+
"epoch": 1.3256948864841926,
|
| 5601 |
+
"grad_norm": 0.0006237781490199268,
|
| 5602 |
+
"learning_rate": 3.47879464133652e-05,
|
| 5603 |
+
"loss": 0.0,
|
| 5604 |
+
"step": 781
|
| 5605 |
+
},
|
| 5606 |
+
{
|
| 5607 |
+
"epoch": 1.327392319117335,
|
| 5608 |
+
"grad_norm": 0.0015418545808643103,
|
| 5609 |
+
"learning_rate": 3.448371947867763e-05,
|
| 5610 |
+
"loss": 0.0001,
|
| 5611 |
+
"step": 782
|
| 5612 |
+
},
|
| 5613 |
+
{
|
| 5614 |
+
"epoch": 1.3290897517504774,
|
| 5615 |
+
"grad_norm": 0.0035702355671674013,
|
| 5616 |
+
"learning_rate": 3.4180655791652476e-05,
|
| 5617 |
+
"loss": 0.0001,
|
| 5618 |
+
"step": 783
|
| 5619 |
+
},
|
| 5620 |
+
{
|
| 5621 |
+
"epoch": 1.3307871843836199,
|
| 5622 |
+
"grad_norm": 0.0013561249943450093,
|
| 5623 |
+
"learning_rate": 3.3878758404137624e-05,
|
| 5624 |
+
"loss": 0.0001,
|
| 5625 |
+
"step": 784
|
| 5626 |
+
},
|
| 5627 |
+
{
|
| 5628 |
+
"epoch": 1.3324846170167621,
|
| 5629 |
+
"grad_norm": 0.001489428337663412,
|
| 5630 |
+
"learning_rate": 3.3578030356236455e-05,
|
| 5631 |
+
"loss": 0.0001,
|
| 5632 |
+
"step": 785
|
| 5633 |
+
},
|
| 5634 |
+
{
|
| 5635 |
+
"epoch": 1.3341820496499044,
|
| 5636 |
+
"grad_norm": 0.0011818065540865064,
|
| 5637 |
+
"learning_rate": 3.3278474676277114e-05,
|
| 5638 |
+
"loss": 0.0,
|
| 5639 |
+
"step": 786
|
| 5640 |
+
},
|
| 5641 |
+
{
|
| 5642 |
+
"epoch": 1.3358794822830469,
|
| 5643 |
+
"grad_norm": 0.0076867276802659035,
|
| 5644 |
+
"learning_rate": 3.298009438078194e-05,
|
| 5645 |
+
"loss": 0.0003,
|
| 5646 |
+
"step": 787
|
| 5647 |
+
},
|
| 5648 |
+
{
|
| 5649 |
+
"epoch": 1.3375769149161894,
|
| 5650 |
+
"grad_norm": 0.002071813913062215,
|
| 5651 |
+
"learning_rate": 3.268289247443713e-05,
|
| 5652 |
+
"loss": 0.0001,
|
| 5653 |
+
"step": 788
|
| 5654 |
+
},
|
| 5655 |
+
{
|
| 5656 |
+
"epoch": 1.3392743475493316,
|
| 5657 |
+
"grad_norm": 0.002192431129515171,
|
| 5658 |
+
"learning_rate": 3.238687195006264e-05,
|
| 5659 |
+
"loss": 0.0001,
|
| 5660 |
+
"step": 789
|
| 5661 |
+
},
|
| 5662 |
+
{
|
| 5663 |
+
"epoch": 1.3409717801824739,
|
| 5664 |
+
"grad_norm": 0.24951210618019104,
|
| 5665 |
+
"learning_rate": 3.209203578858191e-05,
|
| 5666 |
+
"loss": 0.6115,
|
| 5667 |
+
"step": 790
|
| 5668 |
+
},
|
| 5669 |
+
{
|
| 5670 |
+
"epoch": 1.3426692128156164,
|
| 5671 |
+
"grad_norm": 0.29264548420906067,
|
| 5672 |
+
"learning_rate": 3.1798386958991714e-05,
|
| 5673 |
+
"loss": 0.71,
|
| 5674 |
+
"step": 791
|
| 5675 |
+
},
|
| 5676 |
+
{
|
| 5677 |
+
"epoch": 1.3443666454487588,
|
| 5678 |
+
"grad_norm": 0.2815876603126526,
|
| 5679 |
+
"learning_rate": 3.1505928418332574e-05,
|
| 5680 |
+
"loss": 0.6586,
|
| 5681 |
+
"step": 792
|
| 5682 |
+
},
|
| 5683 |
+
{
|
| 5684 |
+
"epoch": 1.346064078081901,
|
| 5685 |
+
"grad_norm": 0.25446027517318726,
|
| 5686 |
+
"learning_rate": 3.121466311165875e-05,
|
| 5687 |
+
"loss": 0.5623,
|
| 5688 |
+
"step": 793
|
| 5689 |
+
},
|
| 5690 |
+
{
|
| 5691 |
+
"epoch": 1.3477615107150436,
|
| 5692 |
+
"grad_norm": 0.2643551230430603,
|
| 5693 |
+
"learning_rate": 3.092459397200861e-05,
|
| 5694 |
+
"loss": 0.6298,
|
| 5695 |
+
"step": 794
|
| 5696 |
+
},
|
| 5697 |
+
{
|
| 5698 |
+
"epoch": 1.3494589433481858,
|
| 5699 |
+
"grad_norm": 0.24555166065692902,
|
| 5700 |
+
"learning_rate": 3.0635723920375164e-05,
|
| 5701 |
+
"loss": 0.5237,
|
| 5702 |
+
"step": 795
|
| 5703 |
+
},
|
| 5704 |
+
{
|
| 5705 |
+
"epoch": 1.3511563759813283,
|
| 5706 |
+
"grad_norm": 0.2268795222043991,
|
| 5707 |
+
"learning_rate": 3.0348055865676707e-05,
|
| 5708 |
+
"loss": 0.3814,
|
| 5709 |
+
"step": 796
|
| 5710 |
+
},
|
| 5711 |
+
{
|
| 5712 |
+
"epoch": 1.3528538086144706,
|
| 5713 |
+
"grad_norm": 0.25621020793914795,
|
| 5714 |
+
"learning_rate": 3.0061592704727257e-05,
|
| 5715 |
+
"loss": 0.5183,
|
| 5716 |
+
"step": 797
|
| 5717 |
+
},
|
| 5718 |
+
{
|
| 5719 |
+
"epoch": 1.354551241247613,
|
| 5720 |
+
"grad_norm": 0.2662016451358795,
|
| 5721 |
+
"learning_rate": 2.9776337322207687e-05,
|
| 5722 |
+
"loss": 0.5679,
|
| 5723 |
+
"step": 798
|
| 5724 |
+
},
|
| 5725 |
+
{
|
| 5726 |
+
"epoch": 1.3562486738807553,
|
| 5727 |
+
"grad_norm": 0.20577247440814972,
|
| 5728 |
+
"learning_rate": 2.9492292590636613e-05,
|
| 5729 |
+
"loss": 0.3231,
|
| 5730 |
+
"step": 799
|
| 5731 |
+
},
|
| 5732 |
+
{
|
| 5733 |
+
"epoch": 1.3579461065138978,
|
| 5734 |
+
"grad_norm": 0.23013556003570557,
|
| 5735 |
+
"learning_rate": 2.9209461370341204e-05,
|
| 5736 |
+
"loss": 0.4158,
|
| 5737 |
+
"step": 800
|
| 5738 |
+
},
|
| 5739 |
+
{
|
| 5740 |
+
"epoch": 1.3579461065138978,
|
| 5741 |
+
"eval_loss": 0.36441436409950256,
|
| 5742 |
+
"eval_runtime": 65.9028,
|
| 5743 |
+
"eval_samples_per_second": 2.929,
|
| 5744 |
+
"eval_steps_per_second": 2.929,
|
| 5745 |
+
"step": 800
|
| 5746 |
}
|
| 5747 |
],
|
| 5748 |
"logging_steps": 1,
|
|
|
|
| 5757 |
"early_stopping_threshold": 0.0
|
| 5758 |
},
|
| 5759 |
"attributes": {
|
| 5760 |
+
"early_stopping_patience_counter": 3
|
| 5761 |
}
|
| 5762 |
},
|
| 5763 |
"TrainerControl": {
|
|
|
|
| 5766 |
"should_evaluate": false,
|
| 5767 |
"should_log": false,
|
| 5768 |
"should_save": true,
|
| 5769 |
+
"should_training_stop": true
|
| 5770 |
},
|
| 5771 |
"attributes": {}
|
| 5772 |
}
|
| 5773 |
},
|
| 5774 |
+
"total_flos": 5.185996556504924e+17,
|
| 5775 |
"train_batch_size": 1,
|
| 5776 |
"trial_name": null,
|
| 5777 |
"trial_params": null
|