Training in progress, step 650, 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 97307544
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:bc321b67befde9d32d0f759ae44d1d80ee068c965b067e8f9195d58612be3ea0
|
| 3 |
size 97307544
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 194840426
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0c9e732f66f7d9824044594dce52a639a7c4361adf5450f030c8624a9c1f4741
|
| 3 |
size 194840426
|
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:c8c5cab54e7970b8ccc67a7f8c2fe78f7bf63d36c0111ad86163a4833d9fb630
|
| 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:aa213f8b4f645aefc7b7ed34404a704afe52702d2abc5afd1fcf6555c25785dd
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
-
"best_metric": 2.
|
| 3 |
-
"best_model_checkpoint": "miner_id_24/checkpoint-
|
| 4 |
-
"epoch": 0.
|
| 5 |
"eval_steps": 25,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4407,6 +4407,372 @@
|
|
| 4407 |
"eval_samples_per_second": 7.475,
|
| 4408 |
"eval_steps_per_second": 7.475,
|
| 4409 |
"step": 600
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4410 |
}
|
| 4411 |
],
|
| 4412 |
"logging_steps": 1,
|
|
@@ -4435,7 +4801,7 @@
|
|
| 4435 |
"attributes": {}
|
| 4436 |
}
|
| 4437 |
},
|
| 4438 |
-
"total_flos":
|
| 4439 |
"train_batch_size": 1,
|
| 4440 |
"trial_name": null,
|
| 4441 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_metric": 2.107257604598999,
|
| 3 |
+
"best_model_checkpoint": "miner_id_24/checkpoint-650",
|
| 4 |
+
"epoch": 0.18552379253445123,
|
| 5 |
"eval_steps": 25,
|
| 6 |
+
"global_step": 650,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4407 |
"eval_samples_per_second": 7.475,
|
| 4408 |
"eval_steps_per_second": 7.475,
|
| 4409 |
"step": 600
|
| 4410 |
+
},
|
| 4411 |
+
{
|
| 4412 |
+
"epoch": 0.17153815278954646,
|
| 4413 |
+
"grad_norm": 0.4549245536327362,
|
| 4414 |
+
"learning_rate": 6.576765361421342e-05,
|
| 4415 |
+
"loss": 2.5277,
|
| 4416 |
+
"step": 601
|
| 4417 |
+
},
|
| 4418 |
+
{
|
| 4419 |
+
"epoch": 0.17182357400883022,
|
| 4420 |
+
"grad_norm": 0.4663224220275879,
|
| 4421 |
+
"learning_rate": 6.54792795378143e-05,
|
| 4422 |
+
"loss": 2.3604,
|
| 4423 |
+
"step": 602
|
| 4424 |
+
},
|
| 4425 |
+
{
|
| 4426 |
+
"epoch": 0.172108995228114,
|
| 4427 |
+
"grad_norm": 0.5545827746391296,
|
| 4428 |
+
"learning_rate": 6.519150935911562e-05,
|
| 4429 |
+
"loss": 2.3525,
|
| 4430 |
+
"step": 603
|
| 4431 |
+
},
|
| 4432 |
+
{
|
| 4433 |
+
"epoch": 0.17239441644739775,
|
| 4434 |
+
"grad_norm": 0.557524561882019,
|
| 4435 |
+
"learning_rate": 6.490434698155264e-05,
|
| 4436 |
+
"loss": 2.1753,
|
| 4437 |
+
"step": 604
|
| 4438 |
+
},
|
| 4439 |
+
{
|
| 4440 |
+
"epoch": 0.17267983766668155,
|
| 4441 |
+
"grad_norm": 0.5588095784187317,
|
| 4442 |
+
"learning_rate": 6.46177963003161e-05,
|
| 4443 |
+
"loss": 2.3271,
|
| 4444 |
+
"step": 605
|
| 4445 |
+
},
|
| 4446 |
+
{
|
| 4447 |
+
"epoch": 0.1729652588859653,
|
| 4448 |
+
"grad_norm": 0.5639928579330444,
|
| 4449 |
+
"learning_rate": 6.43318612022995e-05,
|
| 4450 |
+
"loss": 2.3426,
|
| 4451 |
+
"step": 606
|
| 4452 |
+
},
|
| 4453 |
+
{
|
| 4454 |
+
"epoch": 0.17325068010524908,
|
| 4455 |
+
"grad_norm": 0.5503498315811157,
|
| 4456 |
+
"learning_rate": 6.404654556604627e-05,
|
| 4457 |
+
"loss": 2.1217,
|
| 4458 |
+
"step": 607
|
| 4459 |
+
},
|
| 4460 |
+
{
|
| 4461 |
+
"epoch": 0.17353610132453284,
|
| 4462 |
+
"grad_norm": 0.5461228489875793,
|
| 4463 |
+
"learning_rate": 6.376185326169721e-05,
|
| 4464 |
+
"loss": 1.8886,
|
| 4465 |
+
"step": 608
|
| 4466 |
+
},
|
| 4467 |
+
{
|
| 4468 |
+
"epoch": 0.1738215225438166,
|
| 4469 |
+
"grad_norm": 0.4937599003314972,
|
| 4470 |
+
"learning_rate": 6.347778815093796e-05,
|
| 4471 |
+
"loss": 2.1941,
|
| 4472 |
+
"step": 609
|
| 4473 |
+
},
|
| 4474 |
+
{
|
| 4475 |
+
"epoch": 0.17410694376310037,
|
| 4476 |
+
"grad_norm": 0.5086988806724548,
|
| 4477 |
+
"learning_rate": 6.31943540869467e-05,
|
| 4478 |
+
"loss": 1.9679,
|
| 4479 |
+
"step": 610
|
| 4480 |
+
},
|
| 4481 |
+
{
|
| 4482 |
+
"epoch": 0.17439236498238417,
|
| 4483 |
+
"grad_norm": 0.5381971001625061,
|
| 4484 |
+
"learning_rate": 6.291155491434175e-05,
|
| 4485 |
+
"loss": 2.0835,
|
| 4486 |
+
"step": 611
|
| 4487 |
+
},
|
| 4488 |
+
{
|
| 4489 |
+
"epoch": 0.17467778620166793,
|
| 4490 |
+
"grad_norm": 0.554766833782196,
|
| 4491 |
+
"learning_rate": 6.262939446912959e-05,
|
| 4492 |
+
"loss": 1.8391,
|
| 4493 |
+
"step": 612
|
| 4494 |
+
},
|
| 4495 |
+
{
|
| 4496 |
+
"epoch": 0.1749632074209517,
|
| 4497 |
+
"grad_norm": 0.5361195206642151,
|
| 4498 |
+
"learning_rate": 6.234787657865266e-05,
|
| 4499 |
+
"loss": 2.1932,
|
| 4500 |
+
"step": 613
|
| 4501 |
+
},
|
| 4502 |
+
{
|
| 4503 |
+
"epoch": 0.17524862864023547,
|
| 4504 |
+
"grad_norm": 0.5431385040283203,
|
| 4505 |
+
"learning_rate": 6.206700506153755e-05,
|
| 4506 |
+
"loss": 2.089,
|
| 4507 |
+
"step": 614
|
| 4508 |
+
},
|
| 4509 |
+
{
|
| 4510 |
+
"epoch": 0.17553404985951923,
|
| 4511 |
+
"grad_norm": 0.5752652287483215,
|
| 4512 |
+
"learning_rate": 6.178678372764315e-05,
|
| 4513 |
+
"loss": 2.0846,
|
| 4514 |
+
"step": 615
|
| 4515 |
+
},
|
| 4516 |
+
{
|
| 4517 |
+
"epoch": 0.17581947107880302,
|
| 4518 |
+
"grad_norm": 0.5880393981933594,
|
| 4519 |
+
"learning_rate": 6.150721637800901e-05,
|
| 4520 |
+
"loss": 2.0762,
|
| 4521 |
+
"step": 616
|
| 4522 |
+
},
|
| 4523 |
+
{
|
| 4524 |
+
"epoch": 0.1761048922980868,
|
| 4525 |
+
"grad_norm": 0.5371225476264954,
|
| 4526 |
+
"learning_rate": 6.122830680480383e-05,
|
| 4527 |
+
"loss": 2.1387,
|
| 4528 |
+
"step": 617
|
| 4529 |
+
},
|
| 4530 |
+
{
|
| 4531 |
+
"epoch": 0.17639031351737056,
|
| 4532 |
+
"grad_norm": 0.5906757712364197,
|
| 4533 |
+
"learning_rate": 6.095005879127379e-05,
|
| 4534 |
+
"loss": 2.2487,
|
| 4535 |
+
"step": 618
|
| 4536 |
+
},
|
| 4537 |
+
{
|
| 4538 |
+
"epoch": 0.17667573473665432,
|
| 4539 |
+
"grad_norm": 0.5833765268325806,
|
| 4540 |
+
"learning_rate": 6.067247611169153e-05,
|
| 4541 |
+
"loss": 2.0569,
|
| 4542 |
+
"step": 619
|
| 4543 |
+
},
|
| 4544 |
+
{
|
| 4545 |
+
"epoch": 0.1769611559559381,
|
| 4546 |
+
"grad_norm": 0.5555745959281921,
|
| 4547 |
+
"learning_rate": 6.039556253130481e-05,
|
| 4548 |
+
"loss": 2.1929,
|
| 4549 |
+
"step": 620
|
| 4550 |
+
},
|
| 4551 |
+
{
|
| 4552 |
+
"epoch": 0.17724657717522188,
|
| 4553 |
+
"grad_norm": 0.5676312446594238,
|
| 4554 |
+
"learning_rate": 6.0119321806285366e-05,
|
| 4555 |
+
"loss": 2.0863,
|
| 4556 |
+
"step": 621
|
| 4557 |
+
},
|
| 4558 |
+
{
|
| 4559 |
+
"epoch": 0.17753199839450565,
|
| 4560 |
+
"grad_norm": 0.5911099910736084,
|
| 4561 |
+
"learning_rate": 5.984375768367816e-05,
|
| 4562 |
+
"loss": 1.9681,
|
| 4563 |
+
"step": 622
|
| 4564 |
+
},
|
| 4565 |
+
{
|
| 4566 |
+
"epoch": 0.1778174196137894,
|
| 4567 |
+
"grad_norm": 0.6491994857788086,
|
| 4568 |
+
"learning_rate": 5.9568873901350295e-05,
|
| 4569 |
+
"loss": 2.1629,
|
| 4570 |
+
"step": 623
|
| 4571 |
+
},
|
| 4572 |
+
{
|
| 4573 |
+
"epoch": 0.17810284083307318,
|
| 4574 |
+
"grad_norm": 0.5989968776702881,
|
| 4575 |
+
"learning_rate": 5.9294674187940635e-05,
|
| 4576 |
+
"loss": 1.817,
|
| 4577 |
+
"step": 624
|
| 4578 |
+
},
|
| 4579 |
+
{
|
| 4580 |
+
"epoch": 0.17838826205235694,
|
| 4581 |
+
"grad_norm": 0.5644077658653259,
|
| 4582 |
+
"learning_rate": 5.902116226280887e-05,
|
| 4583 |
+
"loss": 2.2051,
|
| 4584 |
+
"step": 625
|
| 4585 |
+
},
|
| 4586 |
+
{
|
| 4587 |
+
"epoch": 0.17838826205235694,
|
| 4588 |
+
"eval_loss": 2.110799789428711,
|
| 4589 |
+
"eval_runtime": 6.6942,
|
| 4590 |
+
"eval_samples_per_second": 7.469,
|
| 4591 |
+
"eval_steps_per_second": 7.469,
|
| 4592 |
+
"step": 625
|
| 4593 |
+
},
|
| 4594 |
+
{
|
| 4595 |
+
"epoch": 0.17867368327164074,
|
| 4596 |
+
"grad_norm": 0.5663304924964905,
|
| 4597 |
+
"learning_rate": 5.874834183598541e-05,
|
| 4598 |
+
"loss": 2.2308,
|
| 4599 |
+
"step": 626
|
| 4600 |
+
},
|
| 4601 |
+
{
|
| 4602 |
+
"epoch": 0.1789591044909245,
|
| 4603 |
+
"grad_norm": 0.6335240602493286,
|
| 4604 |
+
"learning_rate": 5.847621660812067e-05,
|
| 4605 |
+
"loss": 2.3482,
|
| 4606 |
+
"step": 627
|
| 4607 |
+
},
|
| 4608 |
+
{
|
| 4609 |
+
"epoch": 0.17924452571020827,
|
| 4610 |
+
"grad_norm": 0.5667268633842468,
|
| 4611 |
+
"learning_rate": 5.820479027043531e-05,
|
| 4612 |
+
"loss": 2.3666,
|
| 4613 |
+
"step": 628
|
| 4614 |
+
},
|
| 4615 |
+
{
|
| 4616 |
+
"epoch": 0.17952994692949203,
|
| 4617 |
+
"grad_norm": 0.5804107189178467,
|
| 4618 |
+
"learning_rate": 5.793406650466982e-05,
|
| 4619 |
+
"loss": 1.9951,
|
| 4620 |
+
"step": 629
|
| 4621 |
+
},
|
| 4622 |
+
{
|
| 4623 |
+
"epoch": 0.1798153681487758,
|
| 4624 |
+
"grad_norm": 0.6026704907417297,
|
| 4625 |
+
"learning_rate": 5.7664048983034746e-05,
|
| 4626 |
+
"loss": 1.9337,
|
| 4627 |
+
"step": 630
|
| 4628 |
+
},
|
| 4629 |
+
{
|
| 4630 |
+
"epoch": 0.1801007893680596,
|
| 4631 |
+
"grad_norm": 0.6197707653045654,
|
| 4632 |
+
"learning_rate": 5.739474136816092e-05,
|
| 4633 |
+
"loss": 2.1738,
|
| 4634 |
+
"step": 631
|
| 4635 |
+
},
|
| 4636 |
+
{
|
| 4637 |
+
"epoch": 0.18038621058734336,
|
| 4638 |
+
"grad_norm": 0.60870361328125,
|
| 4639 |
+
"learning_rate": 5.712614731304959e-05,
|
| 4640 |
+
"loss": 2.0636,
|
| 4641 |
+
"step": 632
|
| 4642 |
+
},
|
| 4643 |
+
{
|
| 4644 |
+
"epoch": 0.18067163180662713,
|
| 4645 |
+
"grad_norm": 0.6289341449737549,
|
| 4646 |
+
"learning_rate": 5.6858270461023034e-05,
|
| 4647 |
+
"loss": 2.2225,
|
| 4648 |
+
"step": 633
|
| 4649 |
+
},
|
| 4650 |
+
{
|
| 4651 |
+
"epoch": 0.1809570530259109,
|
| 4652 |
+
"grad_norm": 0.6064093708992004,
|
| 4653 |
+
"learning_rate": 5.6591114445675065e-05,
|
| 4654 |
+
"loss": 2.014,
|
| 4655 |
+
"step": 634
|
| 4656 |
+
},
|
| 4657 |
+
{
|
| 4658 |
+
"epoch": 0.18124247424519466,
|
| 4659 |
+
"grad_norm": 0.6756201386451721,
|
| 4660 |
+
"learning_rate": 5.632468289082184e-05,
|
| 4661 |
+
"loss": 2.0107,
|
| 4662 |
+
"step": 635
|
| 4663 |
+
},
|
| 4664 |
+
{
|
| 4665 |
+
"epoch": 0.18152789546447845,
|
| 4666 |
+
"grad_norm": 0.6307650804519653,
|
| 4667 |
+
"learning_rate": 5.6058979410452436e-05,
|
| 4668 |
+
"loss": 1.9966,
|
| 4669 |
+
"step": 636
|
| 4670 |
+
},
|
| 4671 |
+
{
|
| 4672 |
+
"epoch": 0.18181331668376222,
|
| 4673 |
+
"grad_norm": 0.6745278239250183,
|
| 4674 |
+
"learning_rate": 5.5794007608680235e-05,
|
| 4675 |
+
"loss": 2.075,
|
| 4676 |
+
"step": 637
|
| 4677 |
+
},
|
| 4678 |
+
{
|
| 4679 |
+
"epoch": 0.18209873790304598,
|
| 4680 |
+
"grad_norm": 0.7290084958076477,
|
| 4681 |
+
"learning_rate": 5.552977107969375e-05,
|
| 4682 |
+
"loss": 2.2774,
|
| 4683 |
+
"step": 638
|
| 4684 |
+
},
|
| 4685 |
+
{
|
| 4686 |
+
"epoch": 0.18238415912232975,
|
| 4687 |
+
"grad_norm": 0.649732232093811,
|
| 4688 |
+
"learning_rate": 5.5266273407707934e-05,
|
| 4689 |
+
"loss": 2.2732,
|
| 4690 |
+
"step": 639
|
| 4691 |
+
},
|
| 4692 |
+
{
|
| 4693 |
+
"epoch": 0.1826695803416135,
|
| 4694 |
+
"grad_norm": 0.7042523622512817,
|
| 4695 |
+
"learning_rate": 5.500351816691556e-05,
|
| 4696 |
+
"loss": 2.2515,
|
| 4697 |
+
"step": 640
|
| 4698 |
+
},
|
| 4699 |
+
{
|
| 4700 |
+
"epoch": 0.1829550015608973,
|
| 4701 |
+
"grad_norm": 0.6616380214691162,
|
| 4702 |
+
"learning_rate": 5.474150892143879e-05,
|
| 4703 |
+
"loss": 1.8062,
|
| 4704 |
+
"step": 641
|
| 4705 |
+
},
|
| 4706 |
+
{
|
| 4707 |
+
"epoch": 0.18324042278018107,
|
| 4708 |
+
"grad_norm": 0.6693776845932007,
|
| 4709 |
+
"learning_rate": 5.448024922528079e-05,
|
| 4710 |
+
"loss": 1.988,
|
| 4711 |
+
"step": 642
|
| 4712 |
+
},
|
| 4713 |
+
{
|
| 4714 |
+
"epoch": 0.18352584399946484,
|
| 4715 |
+
"grad_norm": 0.658955991268158,
|
| 4716 |
+
"learning_rate": 5.421974262227747e-05,
|
| 4717 |
+
"loss": 2.1457,
|
| 4718 |
+
"step": 643
|
| 4719 |
+
},
|
| 4720 |
+
{
|
| 4721 |
+
"epoch": 0.1838112652187486,
|
| 4722 |
+
"grad_norm": 0.7057983875274658,
|
| 4723 |
+
"learning_rate": 5.395999264604956e-05,
|
| 4724 |
+
"loss": 2.051,
|
| 4725 |
+
"step": 644
|
| 4726 |
+
},
|
| 4727 |
+
{
|
| 4728 |
+
"epoch": 0.18409668643803237,
|
| 4729 |
+
"grad_norm": 0.6857619881629944,
|
| 4730 |
+
"learning_rate": 5.370100281995445e-05,
|
| 4731 |
+
"loss": 1.8961,
|
| 4732 |
+
"step": 645
|
| 4733 |
+
},
|
| 4734 |
+
{
|
| 4735 |
+
"epoch": 0.18438210765731614,
|
| 4736 |
+
"grad_norm": 0.7401403188705444,
|
| 4737 |
+
"learning_rate": 5.344277665703872e-05,
|
| 4738 |
+
"loss": 2.1315,
|
| 4739 |
+
"step": 646
|
| 4740 |
+
},
|
| 4741 |
+
{
|
| 4742 |
+
"epoch": 0.18466752887659993,
|
| 4743 |
+
"grad_norm": 0.7094218134880066,
|
| 4744 |
+
"learning_rate": 5.318531765999015e-05,
|
| 4745 |
+
"loss": 2.0304,
|
| 4746 |
+
"step": 647
|
| 4747 |
+
},
|
| 4748 |
+
{
|
| 4749 |
+
"epoch": 0.1849529500958837,
|
| 4750 |
+
"grad_norm": 1.1581752300262451,
|
| 4751 |
+
"learning_rate": 5.292862932109044e-05,
|
| 4752 |
+
"loss": 2.3675,
|
| 4753 |
+
"step": 648
|
| 4754 |
+
},
|
| 4755 |
+
{
|
| 4756 |
+
"epoch": 0.18523837131516746,
|
| 4757 |
+
"grad_norm": 0.8512890934944153,
|
| 4758 |
+
"learning_rate": 5.267271512216774e-05,
|
| 4759 |
+
"loss": 2.1611,
|
| 4760 |
+
"step": 649
|
| 4761 |
+
},
|
| 4762 |
+
{
|
| 4763 |
+
"epoch": 0.18552379253445123,
|
| 4764 |
+
"grad_norm": 1.1400418281555176,
|
| 4765 |
+
"learning_rate": 5.2417578534549406e-05,
|
| 4766 |
+
"loss": 2.0556,
|
| 4767 |
+
"step": 650
|
| 4768 |
+
},
|
| 4769 |
+
{
|
| 4770 |
+
"epoch": 0.18552379253445123,
|
| 4771 |
+
"eval_loss": 2.107257604598999,
|
| 4772 |
+
"eval_runtime": 6.7052,
|
| 4773 |
+
"eval_samples_per_second": 7.457,
|
| 4774 |
+
"eval_steps_per_second": 7.457,
|
| 4775 |
+
"step": 650
|
| 4776 |
}
|
| 4777 |
],
|
| 4778 |
"logging_steps": 1,
|
|
|
|
| 4801 |
"attributes": {}
|
| 4802 |
}
|
| 4803 |
},
|
| 4804 |
+
"total_flos": 7.266940971374346e+17,
|
| 4805 |
"train_batch_size": 1,
|
| 4806 |
"trial_name": null,
|
| 4807 |
"trial_params": null
|