Training checkpoint at step 16000
Browse files- trainer_state.json +365 -5
trainer_state.json
CHANGED
|
@@ -1,10 +1,10 @@
|
|
| 1 |
{
|
| 2 |
-
"best_global_step":
|
| 3 |
-
"best_metric": 2.
|
| 4 |
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-15000",
|
| 5 |
-
"epoch": 0.
|
| 6 |
"eval_steps": 100,
|
| 7 |
-
"global_step":
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
@@ -5408,6 +5408,366 @@
|
|
| 5408 |
"eval_samples_per_second": 3.216,
|
| 5409 |
"eval_steps_per_second": 1.608,
|
| 5410 |
"step": 15000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5411 |
}
|
| 5412 |
],
|
| 5413 |
"logging_steps": 25,
|
|
@@ -5427,7 +5787,7 @@
|
|
| 5427 |
"attributes": {}
|
| 5428 |
}
|
| 5429 |
},
|
| 5430 |
-
"total_flos":
|
| 5431 |
"train_batch_size": 1,
|
| 5432 |
"trial_name": null,
|
| 5433 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 15900,
|
| 3 |
+
"best_metric": 2.3957200050354004,
|
| 4 |
"best_model_checkpoint": "../runs/karpathy/fineweb-edu-100b-shuffle/meta-llama/Llama-3.2-1B/linear_adamw_wd1e-03_8x1024_mem32_bs64_hf_armt_dmem64/run_21/checkpoint-15000",
|
| 5 |
+
"epoch": 0.32,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 16000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 5408 |
"eval_samples_per_second": 3.216,
|
| 5409 |
"eval_steps_per_second": 1.608,
|
| 5410 |
"step": 15000
|
| 5411 |
+
},
|
| 5412 |
+
{
|
| 5413 |
+
"epoch": 0.3005,
|
| 5414 |
+
"grad_norm": 0.5657448746286691,
|
| 5415 |
+
"learning_rate": 7.772444444444444e-06,
|
| 5416 |
+
"loss": 2.3897,
|
| 5417 |
+
"step": 15025
|
| 5418 |
+
},
|
| 5419 |
+
{
|
| 5420 |
+
"epoch": 0.301,
|
| 5421 |
+
"grad_norm": 0.5523525627604269,
|
| 5422 |
+
"learning_rate": 7.76688888888889e-06,
|
| 5423 |
+
"loss": 2.3795,
|
| 5424 |
+
"step": 15050
|
| 5425 |
+
},
|
| 5426 |
+
{
|
| 5427 |
+
"epoch": 0.3015,
|
| 5428 |
+
"grad_norm": 0.5950789860717867,
|
| 5429 |
+
"learning_rate": 7.761333333333335e-06,
|
| 5430 |
+
"loss": 2.3914,
|
| 5431 |
+
"step": 15075
|
| 5432 |
+
},
|
| 5433 |
+
{
|
| 5434 |
+
"epoch": 0.302,
|
| 5435 |
+
"grad_norm": 0.5999400034143391,
|
| 5436 |
+
"learning_rate": 7.755777777777778e-06,
|
| 5437 |
+
"loss": 2.3769,
|
| 5438 |
+
"step": 15100
|
| 5439 |
+
},
|
| 5440 |
+
{
|
| 5441 |
+
"epoch": 0.302,
|
| 5442 |
+
"eval_loss": 2.396873950958252,
|
| 5443 |
+
"eval_runtime": 31.7696,
|
| 5444 |
+
"eval_samples_per_second": 3.211,
|
| 5445 |
+
"eval_steps_per_second": 1.605,
|
| 5446 |
+
"step": 15100
|
| 5447 |
+
},
|
| 5448 |
+
{
|
| 5449 |
+
"epoch": 0.3025,
|
| 5450 |
+
"grad_norm": 0.558834977842146,
|
| 5451 |
+
"learning_rate": 7.750222222222222e-06,
|
| 5452 |
+
"loss": 2.3854,
|
| 5453 |
+
"step": 15125
|
| 5454 |
+
},
|
| 5455 |
+
{
|
| 5456 |
+
"epoch": 0.303,
|
| 5457 |
+
"grad_norm": 0.5582295283472423,
|
| 5458 |
+
"learning_rate": 7.744666666666667e-06,
|
| 5459 |
+
"loss": 2.3821,
|
| 5460 |
+
"step": 15150
|
| 5461 |
+
},
|
| 5462 |
+
{
|
| 5463 |
+
"epoch": 0.3035,
|
| 5464 |
+
"grad_norm": 0.5632905015995245,
|
| 5465 |
+
"learning_rate": 7.739111111111112e-06,
|
| 5466 |
+
"loss": 2.3798,
|
| 5467 |
+
"step": 15175
|
| 5468 |
+
},
|
| 5469 |
+
{
|
| 5470 |
+
"epoch": 0.304,
|
| 5471 |
+
"grad_norm": 0.5514118333084079,
|
| 5472 |
+
"learning_rate": 7.733555555555556e-06,
|
| 5473 |
+
"loss": 2.3788,
|
| 5474 |
+
"step": 15200
|
| 5475 |
+
},
|
| 5476 |
+
{
|
| 5477 |
+
"epoch": 0.304,
|
| 5478 |
+
"eval_loss": 2.3965888023376465,
|
| 5479 |
+
"eval_runtime": 31.7152,
|
| 5480 |
+
"eval_samples_per_second": 3.216,
|
| 5481 |
+
"eval_steps_per_second": 1.608,
|
| 5482 |
+
"step": 15200
|
| 5483 |
+
},
|
| 5484 |
+
{
|
| 5485 |
+
"epoch": 0.3045,
|
| 5486 |
+
"grad_norm": 0.5649018768322466,
|
| 5487 |
+
"learning_rate": 7.728000000000001e-06,
|
| 5488 |
+
"loss": 2.3912,
|
| 5489 |
+
"step": 15225
|
| 5490 |
+
},
|
| 5491 |
+
{
|
| 5492 |
+
"epoch": 0.305,
|
| 5493 |
+
"grad_norm": 0.581561230195339,
|
| 5494 |
+
"learning_rate": 7.722444444444445e-06,
|
| 5495 |
+
"loss": 2.3766,
|
| 5496 |
+
"step": 15250
|
| 5497 |
+
},
|
| 5498 |
+
{
|
| 5499 |
+
"epoch": 0.3055,
|
| 5500 |
+
"grad_norm": 0.5604985750115082,
|
| 5501 |
+
"learning_rate": 7.71688888888889e-06,
|
| 5502 |
+
"loss": 2.3852,
|
| 5503 |
+
"step": 15275
|
| 5504 |
+
},
|
| 5505 |
+
{
|
| 5506 |
+
"epoch": 0.306,
|
| 5507 |
+
"grad_norm": 0.5602736035393524,
|
| 5508 |
+
"learning_rate": 7.711333333333334e-06,
|
| 5509 |
+
"loss": 2.3867,
|
| 5510 |
+
"step": 15300
|
| 5511 |
+
},
|
| 5512 |
+
{
|
| 5513 |
+
"epoch": 0.306,
|
| 5514 |
+
"eval_loss": 2.3968026638031006,
|
| 5515 |
+
"eval_runtime": 31.8105,
|
| 5516 |
+
"eval_samples_per_second": 3.206,
|
| 5517 |
+
"eval_steps_per_second": 1.603,
|
| 5518 |
+
"step": 15300
|
| 5519 |
+
},
|
| 5520 |
+
{
|
| 5521 |
+
"epoch": 0.3065,
|
| 5522 |
+
"grad_norm": 0.5404472339052024,
|
| 5523 |
+
"learning_rate": 7.705777777777779e-06,
|
| 5524 |
+
"loss": 2.3835,
|
| 5525 |
+
"step": 15325
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 0.307,
|
| 5529 |
+
"grad_norm": 0.5732167481475767,
|
| 5530 |
+
"learning_rate": 7.700222222222224e-06,
|
| 5531 |
+
"loss": 2.386,
|
| 5532 |
+
"step": 15350
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 0.3075,
|
| 5536 |
+
"grad_norm": 0.5668975128857069,
|
| 5537 |
+
"learning_rate": 7.694666666666668e-06,
|
| 5538 |
+
"loss": 2.3838,
|
| 5539 |
+
"step": 15375
|
| 5540 |
+
},
|
| 5541 |
+
{
|
| 5542 |
+
"epoch": 0.308,
|
| 5543 |
+
"grad_norm": 0.5478312505357384,
|
| 5544 |
+
"learning_rate": 7.689111111111111e-06,
|
| 5545 |
+
"loss": 2.4068,
|
| 5546 |
+
"step": 15400
|
| 5547 |
+
},
|
| 5548 |
+
{
|
| 5549 |
+
"epoch": 0.308,
|
| 5550 |
+
"eval_loss": 2.39662766456604,
|
| 5551 |
+
"eval_runtime": 31.4625,
|
| 5552 |
+
"eval_samples_per_second": 3.242,
|
| 5553 |
+
"eval_steps_per_second": 1.621,
|
| 5554 |
+
"step": 15400
|
| 5555 |
+
},
|
| 5556 |
+
{
|
| 5557 |
+
"epoch": 0.3085,
|
| 5558 |
+
"grad_norm": 0.5853236703412803,
|
| 5559 |
+
"learning_rate": 7.683555555555556e-06,
|
| 5560 |
+
"loss": 2.3781,
|
| 5561 |
+
"step": 15425
|
| 5562 |
+
},
|
| 5563 |
+
{
|
| 5564 |
+
"epoch": 0.309,
|
| 5565 |
+
"grad_norm": 0.566498029803985,
|
| 5566 |
+
"learning_rate": 7.678000000000002e-06,
|
| 5567 |
+
"loss": 2.3825,
|
| 5568 |
+
"step": 15450
|
| 5569 |
+
},
|
| 5570 |
+
{
|
| 5571 |
+
"epoch": 0.3095,
|
| 5572 |
+
"grad_norm": 0.5876295223419085,
|
| 5573 |
+
"learning_rate": 7.672444444444445e-06,
|
| 5574 |
+
"loss": 2.3821,
|
| 5575 |
+
"step": 15475
|
| 5576 |
+
},
|
| 5577 |
+
{
|
| 5578 |
+
"epoch": 0.31,
|
| 5579 |
+
"grad_norm": 0.5308633915785282,
|
| 5580 |
+
"learning_rate": 7.666888888888889e-06,
|
| 5581 |
+
"loss": 2.3762,
|
| 5582 |
+
"step": 15500
|
| 5583 |
+
},
|
| 5584 |
+
{
|
| 5585 |
+
"epoch": 0.31,
|
| 5586 |
+
"eval_loss": 2.39650559425354,
|
| 5587 |
+
"eval_runtime": 31.6255,
|
| 5588 |
+
"eval_samples_per_second": 3.225,
|
| 5589 |
+
"eval_steps_per_second": 1.613,
|
| 5590 |
+
"step": 15500
|
| 5591 |
+
},
|
| 5592 |
+
{
|
| 5593 |
+
"epoch": 0.3105,
|
| 5594 |
+
"grad_norm": 1.090575647217174,
|
| 5595 |
+
"learning_rate": 7.661333333333334e-06,
|
| 5596 |
+
"loss": 2.3854,
|
| 5597 |
+
"step": 15525
|
| 5598 |
+
},
|
| 5599 |
+
{
|
| 5600 |
+
"epoch": 0.311,
|
| 5601 |
+
"grad_norm": 0.5608565584872227,
|
| 5602 |
+
"learning_rate": 7.65577777777778e-06,
|
| 5603 |
+
"loss": 2.3909,
|
| 5604 |
+
"step": 15550
|
| 5605 |
+
},
|
| 5606 |
+
{
|
| 5607 |
+
"epoch": 0.3115,
|
| 5608 |
+
"grad_norm": 0.5664910219445479,
|
| 5609 |
+
"learning_rate": 7.650222222222223e-06,
|
| 5610 |
+
"loss": 2.3876,
|
| 5611 |
+
"step": 15575
|
| 5612 |
+
},
|
| 5613 |
+
{
|
| 5614 |
+
"epoch": 0.312,
|
| 5615 |
+
"grad_norm": 0.5743138998726522,
|
| 5616 |
+
"learning_rate": 7.644666666666666e-06,
|
| 5617 |
+
"loss": 2.3891,
|
| 5618 |
+
"step": 15600
|
| 5619 |
+
},
|
| 5620 |
+
{
|
| 5621 |
+
"epoch": 0.312,
|
| 5622 |
+
"eval_loss": 2.395846128463745,
|
| 5623 |
+
"eval_runtime": 31.422,
|
| 5624 |
+
"eval_samples_per_second": 3.246,
|
| 5625 |
+
"eval_steps_per_second": 1.623,
|
| 5626 |
+
"step": 15600
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.3125,
|
| 5630 |
+
"grad_norm": 0.5838966503811626,
|
| 5631 |
+
"learning_rate": 7.639111111111112e-06,
|
| 5632 |
+
"loss": 2.3744,
|
| 5633 |
+
"step": 15625
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.313,
|
| 5637 |
+
"grad_norm": 0.5861982665217826,
|
| 5638 |
+
"learning_rate": 7.633555555555557e-06,
|
| 5639 |
+
"loss": 2.386,
|
| 5640 |
+
"step": 15650
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.3135,
|
| 5644 |
+
"grad_norm": 0.5623110973377239,
|
| 5645 |
+
"learning_rate": 7.628000000000001e-06,
|
| 5646 |
+
"loss": 2.3729,
|
| 5647 |
+
"step": 15675
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.314,
|
| 5651 |
+
"grad_norm": 0.5546807091447383,
|
| 5652 |
+
"learning_rate": 7.622444444444445e-06,
|
| 5653 |
+
"loss": 2.3758,
|
| 5654 |
+
"step": 15700
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.314,
|
| 5658 |
+
"eval_loss": 2.396050453186035,
|
| 5659 |
+
"eval_runtime": 31.4839,
|
| 5660 |
+
"eval_samples_per_second": 3.24,
|
| 5661 |
+
"eval_steps_per_second": 1.62,
|
| 5662 |
+
"step": 15700
|
| 5663 |
+
},
|
| 5664 |
+
{
|
| 5665 |
+
"epoch": 0.3145,
|
| 5666 |
+
"grad_norm": 0.566357543453858,
|
| 5667 |
+
"learning_rate": 7.616888888888889e-06,
|
| 5668 |
+
"loss": 2.3814,
|
| 5669 |
+
"step": 15725
|
| 5670 |
+
},
|
| 5671 |
+
{
|
| 5672 |
+
"epoch": 0.315,
|
| 5673 |
+
"grad_norm": 0.5863021742964364,
|
| 5674 |
+
"learning_rate": 7.611333333333334e-06,
|
| 5675 |
+
"loss": 2.3912,
|
| 5676 |
+
"step": 15750
|
| 5677 |
+
},
|
| 5678 |
+
{
|
| 5679 |
+
"epoch": 0.3155,
|
| 5680 |
+
"grad_norm": 0.5448091994015362,
|
| 5681 |
+
"learning_rate": 7.605777777777779e-06,
|
| 5682 |
+
"loss": 2.3949,
|
| 5683 |
+
"step": 15775
|
| 5684 |
+
},
|
| 5685 |
+
{
|
| 5686 |
+
"epoch": 0.316,
|
| 5687 |
+
"grad_norm": 0.5571622234957405,
|
| 5688 |
+
"learning_rate": 7.600222222222223e-06,
|
| 5689 |
+
"loss": 2.3893,
|
| 5690 |
+
"step": 15800
|
| 5691 |
+
},
|
| 5692 |
+
{
|
| 5693 |
+
"epoch": 0.316,
|
| 5694 |
+
"eval_loss": 2.3957884311676025,
|
| 5695 |
+
"eval_runtime": 31.4676,
|
| 5696 |
+
"eval_samples_per_second": 3.241,
|
| 5697 |
+
"eval_steps_per_second": 1.621,
|
| 5698 |
+
"step": 15800
|
| 5699 |
+
},
|
| 5700 |
+
{
|
| 5701 |
+
"epoch": 0.3165,
|
| 5702 |
+
"grad_norm": 0.6175149611764096,
|
| 5703 |
+
"learning_rate": 7.594666666666667e-06,
|
| 5704 |
+
"loss": 2.3858,
|
| 5705 |
+
"step": 15825
|
| 5706 |
+
},
|
| 5707 |
+
{
|
| 5708 |
+
"epoch": 0.317,
|
| 5709 |
+
"grad_norm": 0.5811416818392343,
|
| 5710 |
+
"learning_rate": 7.589111111111111e-06,
|
| 5711 |
+
"loss": 2.3893,
|
| 5712 |
+
"step": 15850
|
| 5713 |
+
},
|
| 5714 |
+
{
|
| 5715 |
+
"epoch": 0.3175,
|
| 5716 |
+
"grad_norm": 0.5685262674194088,
|
| 5717 |
+
"learning_rate": 7.5835555555555566e-06,
|
| 5718 |
+
"loss": 2.3895,
|
| 5719 |
+
"step": 15875
|
| 5720 |
+
},
|
| 5721 |
+
{
|
| 5722 |
+
"epoch": 0.318,
|
| 5723 |
+
"grad_norm": 0.5726231388910242,
|
| 5724 |
+
"learning_rate": 7.578000000000001e-06,
|
| 5725 |
+
"loss": 2.3924,
|
| 5726 |
+
"step": 15900
|
| 5727 |
+
},
|
| 5728 |
+
{
|
| 5729 |
+
"epoch": 0.318,
|
| 5730 |
+
"eval_loss": 2.3957200050354004,
|
| 5731 |
+
"eval_runtime": 31.6833,
|
| 5732 |
+
"eval_samples_per_second": 3.219,
|
| 5733 |
+
"eval_steps_per_second": 1.61,
|
| 5734 |
+
"step": 15900
|
| 5735 |
+
},
|
| 5736 |
+
{
|
| 5737 |
+
"epoch": 0.3185,
|
| 5738 |
+
"grad_norm": 0.5881014617899262,
|
| 5739 |
+
"learning_rate": 7.572444444444445e-06,
|
| 5740 |
+
"loss": 2.3719,
|
| 5741 |
+
"step": 15925
|
| 5742 |
+
},
|
| 5743 |
+
{
|
| 5744 |
+
"epoch": 0.319,
|
| 5745 |
+
"grad_norm": 0.5635459036409981,
|
| 5746 |
+
"learning_rate": 7.566888888888889e-06,
|
| 5747 |
+
"loss": 2.378,
|
| 5748 |
+
"step": 15950
|
| 5749 |
+
},
|
| 5750 |
+
{
|
| 5751 |
+
"epoch": 0.3195,
|
| 5752 |
+
"grad_norm": 0.5604907919572244,
|
| 5753 |
+
"learning_rate": 7.561333333333334e-06,
|
| 5754 |
+
"loss": 2.3744,
|
| 5755 |
+
"step": 15975
|
| 5756 |
+
},
|
| 5757 |
+
{
|
| 5758 |
+
"epoch": 0.32,
|
| 5759 |
+
"grad_norm": 0.5743956921241223,
|
| 5760 |
+
"learning_rate": 7.555777777777779e-06,
|
| 5761 |
+
"loss": 2.3872,
|
| 5762 |
+
"step": 16000
|
| 5763 |
+
},
|
| 5764 |
+
{
|
| 5765 |
+
"epoch": 0.32,
|
| 5766 |
+
"eval_loss": 2.3958442211151123,
|
| 5767 |
+
"eval_runtime": 31.9703,
|
| 5768 |
+
"eval_samples_per_second": 3.19,
|
| 5769 |
+
"eval_steps_per_second": 1.595,
|
| 5770 |
+
"step": 16000
|
| 5771 |
}
|
| 5772 |
],
|
| 5773 |
"logging_steps": 25,
|
|
|
|
| 5787 |
"attributes": {}
|
| 5788 |
}
|
| 5789 |
},
|
| 5790 |
+
"total_flos": 5.093123677323526e+19,
|
| 5791 |
"train_batch_size": 1,
|
| 5792 |
"trial_name": null,
|
| 5793 |
"trial_params": null
|