Initial commit
Browse files- optimizer.pt +1 -1
- pytorch_model.bin +1 -1
- scheduler.pt +1 -1
- trainer_state.json +843 -3
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 655348487
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:e44596804f6bbe6223dc387b01a078b883d9793f67e4205685fa7e6d1bac155b
|
3 |
size 655348487
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 333975623
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:c8e235124e44f7eac920bcb6bf072b75b2cc2272733920b4b17d82b8d859b967
|
3 |
size 333975623
|
scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 623
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:765811e5c0de8679a19ba5175edb8739e017bd63da1b3261fbc6a9c624ff81ea
|
3 |
size 623
|
trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -3486,11 +3486,851 @@
|
|
3486 |
"learning_rate": 4.6042378673957626e-05,
|
3487 |
"loss": 2.1761,
|
3488 |
"step": 290000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3489 |
}
|
3490 |
],
|
3491 |
"max_steps": 3658000,
|
3492 |
"num_train_epochs": 1000,
|
3493 |
-
"total_flos":
|
3494 |
"trial_name": null,
|
3495 |
"trial_params": null
|
3496 |
}
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 98.41443411700382,
|
5 |
+
"global_step": 360000,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
3486 |
"learning_rate": 4.6042378673957626e-05,
|
3487 |
"loss": 2.1761,
|
3488 |
"step": 290000
|
3489 |
+
},
|
3490 |
+
{
|
3491 |
+
"epoch": 79.41,
|
3492 |
+
"learning_rate": 4.603554340396446e-05,
|
3493 |
+
"loss": 2.1889,
|
3494 |
+
"step": 290500
|
3495 |
+
},
|
3496 |
+
{
|
3497 |
+
"epoch": 79.55,
|
3498 |
+
"learning_rate": 4.6028708133971295e-05,
|
3499 |
+
"loss": 2.1878,
|
3500 |
+
"step": 291000
|
3501 |
+
},
|
3502 |
+
{
|
3503 |
+
"epoch": 79.69,
|
3504 |
+
"learning_rate": 4.602187286397813e-05,
|
3505 |
+
"loss": 2.1939,
|
3506 |
+
"step": 291500
|
3507 |
+
},
|
3508 |
+
{
|
3509 |
+
"epoch": 79.83,
|
3510 |
+
"learning_rate": 4.6015037593984965e-05,
|
3511 |
+
"loss": 2.2074,
|
3512 |
+
"step": 292000
|
3513 |
+
},
|
3514 |
+
{
|
3515 |
+
"epoch": 79.96,
|
3516 |
+
"learning_rate": 4.60082023239918e-05,
|
3517 |
+
"loss": 2.2052,
|
3518 |
+
"step": 292500
|
3519 |
+
},
|
3520 |
+
{
|
3521 |
+
"epoch": 80.1,
|
3522 |
+
"learning_rate": 4.6001367053998634e-05,
|
3523 |
+
"loss": 2.1769,
|
3524 |
+
"step": 293000
|
3525 |
+
},
|
3526 |
+
{
|
3527 |
+
"epoch": 80.24,
|
3528 |
+
"learning_rate": 4.599453178400547e-05,
|
3529 |
+
"loss": 2.1675,
|
3530 |
+
"step": 293500
|
3531 |
+
},
|
3532 |
+
{
|
3533 |
+
"epoch": 80.37,
|
3534 |
+
"learning_rate": 4.5987696514012304e-05,
|
3535 |
+
"loss": 2.1722,
|
3536 |
+
"step": 294000
|
3537 |
+
},
|
3538 |
+
{
|
3539 |
+
"epoch": 80.51,
|
3540 |
+
"learning_rate": 4.598086124401914e-05,
|
3541 |
+
"loss": 2.1814,
|
3542 |
+
"step": 294500
|
3543 |
+
},
|
3544 |
+
{
|
3545 |
+
"epoch": 80.65,
|
3546 |
+
"learning_rate": 4.5974025974025974e-05,
|
3547 |
+
"loss": 2.183,
|
3548 |
+
"step": 295000
|
3549 |
+
},
|
3550 |
+
{
|
3551 |
+
"epoch": 80.78,
|
3552 |
+
"learning_rate": 4.5967190704032815e-05,
|
3553 |
+
"loss": 2.1974,
|
3554 |
+
"step": 295500
|
3555 |
+
},
|
3556 |
+
{
|
3557 |
+
"epoch": 80.92,
|
3558 |
+
"learning_rate": 4.596035543403965e-05,
|
3559 |
+
"loss": 2.1974,
|
3560 |
+
"step": 296000
|
3561 |
+
},
|
3562 |
+
{
|
3563 |
+
"epoch": 81.06,
|
3564 |
+
"learning_rate": 4.5953520164046485e-05,
|
3565 |
+
"loss": 2.1853,
|
3566 |
+
"step": 296500
|
3567 |
+
},
|
3568 |
+
{
|
3569 |
+
"epoch": 81.19,
|
3570 |
+
"learning_rate": 4.594668489405331e-05,
|
3571 |
+
"loss": 2.1593,
|
3572 |
+
"step": 297000
|
3573 |
+
},
|
3574 |
+
{
|
3575 |
+
"epoch": 81.33,
|
3576 |
+
"learning_rate": 4.5939849624060154e-05,
|
3577 |
+
"loss": 2.1672,
|
3578 |
+
"step": 297500
|
3579 |
+
},
|
3580 |
+
{
|
3581 |
+
"epoch": 81.47,
|
3582 |
+
"learning_rate": 4.593301435406699e-05,
|
3583 |
+
"loss": 2.1748,
|
3584 |
+
"step": 298000
|
3585 |
+
},
|
3586 |
+
{
|
3587 |
+
"epoch": 81.6,
|
3588 |
+
"learning_rate": 4.5926179084073824e-05,
|
3589 |
+
"loss": 2.1794,
|
3590 |
+
"step": 298500
|
3591 |
+
},
|
3592 |
+
{
|
3593 |
+
"epoch": 81.74,
|
3594 |
+
"learning_rate": 4.591934381408066e-05,
|
3595 |
+
"loss": 2.1888,
|
3596 |
+
"step": 299000
|
3597 |
+
},
|
3598 |
+
{
|
3599 |
+
"epoch": 81.88,
|
3600 |
+
"learning_rate": 4.5912508544087494e-05,
|
3601 |
+
"loss": 2.1932,
|
3602 |
+
"step": 299500
|
3603 |
+
},
|
3604 |
+
{
|
3605 |
+
"epoch": 82.01,
|
3606 |
+
"learning_rate": 4.590567327409433e-05,
|
3607 |
+
"loss": 2.1958,
|
3608 |
+
"step": 300000
|
3609 |
+
},
|
3610 |
+
{
|
3611 |
+
"epoch": 82.15,
|
3612 |
+
"learning_rate": 4.589883800410116e-05,
|
3613 |
+
"loss": 2.1528,
|
3614 |
+
"step": 300500
|
3615 |
+
},
|
3616 |
+
{
|
3617 |
+
"epoch": 82.29,
|
3618 |
+
"learning_rate": 4.5892002734108e-05,
|
3619 |
+
"loss": 2.1538,
|
3620 |
+
"step": 301000
|
3621 |
+
},
|
3622 |
+
{
|
3623 |
+
"epoch": 82.42,
|
3624 |
+
"learning_rate": 4.588516746411484e-05,
|
3625 |
+
"loss": 2.1697,
|
3626 |
+
"step": 301500
|
3627 |
+
},
|
3628 |
+
{
|
3629 |
+
"epoch": 82.56,
|
3630 |
+
"learning_rate": 4.587833219412167e-05,
|
3631 |
+
"loss": 2.1753,
|
3632 |
+
"step": 302000
|
3633 |
+
},
|
3634 |
+
{
|
3635 |
+
"epoch": 82.7,
|
3636 |
+
"learning_rate": 4.58714969241285e-05,
|
3637 |
+
"loss": 2.1771,
|
3638 |
+
"step": 302500
|
3639 |
+
},
|
3640 |
+
{
|
3641 |
+
"epoch": 82.83,
|
3642 |
+
"learning_rate": 4.586466165413534e-05,
|
3643 |
+
"loss": 2.1842,
|
3644 |
+
"step": 303000
|
3645 |
+
},
|
3646 |
+
{
|
3647 |
+
"epoch": 82.97,
|
3648 |
+
"learning_rate": 4.585782638414218e-05,
|
3649 |
+
"loss": 2.1916,
|
3650 |
+
"step": 303500
|
3651 |
+
},
|
3652 |
+
{
|
3653 |
+
"epoch": 83.11,
|
3654 |
+
"learning_rate": 4.5850991114149013e-05,
|
3655 |
+
"loss": 2.1436,
|
3656 |
+
"step": 304000
|
3657 |
+
},
|
3658 |
+
{
|
3659 |
+
"epoch": 83.24,
|
3660 |
+
"learning_rate": 4.584415584415585e-05,
|
3661 |
+
"loss": 2.148,
|
3662 |
+
"step": 304500
|
3663 |
+
},
|
3664 |
+
{
|
3665 |
+
"epoch": 83.38,
|
3666 |
+
"learning_rate": 4.583732057416268e-05,
|
3667 |
+
"loss": 2.1559,
|
3668 |
+
"step": 305000
|
3669 |
+
},
|
3670 |
+
{
|
3671 |
+
"epoch": 83.52,
|
3672 |
+
"learning_rate": 4.583048530416952e-05,
|
3673 |
+
"loss": 2.1686,
|
3674 |
+
"step": 305500
|
3675 |
+
},
|
3676 |
+
{
|
3677 |
+
"epoch": 83.65,
|
3678 |
+
"learning_rate": 4.582365003417635e-05,
|
3679 |
+
"loss": 2.1744,
|
3680 |
+
"step": 306000
|
3681 |
+
},
|
3682 |
+
{
|
3683 |
+
"epoch": 83.79,
|
3684 |
+
"learning_rate": 4.581681476418319e-05,
|
3685 |
+
"loss": 2.1742,
|
3686 |
+
"step": 306500
|
3687 |
+
},
|
3688 |
+
{
|
3689 |
+
"epoch": 83.93,
|
3690 |
+
"learning_rate": 4.580997949419002e-05,
|
3691 |
+
"loss": 2.1881,
|
3692 |
+
"step": 307000
|
3693 |
+
},
|
3694 |
+
{
|
3695 |
+
"epoch": 84.06,
|
3696 |
+
"learning_rate": 4.580314422419686e-05,
|
3697 |
+
"loss": 2.1643,
|
3698 |
+
"step": 307500
|
3699 |
+
},
|
3700 |
+
{
|
3701 |
+
"epoch": 84.2,
|
3702 |
+
"learning_rate": 4.579630895420369e-05,
|
3703 |
+
"loss": 2.1451,
|
3704 |
+
"step": 308000
|
3705 |
+
},
|
3706 |
+
{
|
3707 |
+
"epoch": 84.34,
|
3708 |
+
"learning_rate": 4.5789473684210527e-05,
|
3709 |
+
"loss": 2.1479,
|
3710 |
+
"step": 308500
|
3711 |
+
},
|
3712 |
+
{
|
3713 |
+
"epoch": 84.47,
|
3714 |
+
"learning_rate": 4.578263841421737e-05,
|
3715 |
+
"loss": 2.1579,
|
3716 |
+
"step": 309000
|
3717 |
+
},
|
3718 |
+
{
|
3719 |
+
"epoch": 84.61,
|
3720 |
+
"learning_rate": 4.57758031442242e-05,
|
3721 |
+
"loss": 2.1608,
|
3722 |
+
"step": 309500
|
3723 |
+
},
|
3724 |
+
{
|
3725 |
+
"epoch": 84.75,
|
3726 |
+
"learning_rate": 4.576896787423103e-05,
|
3727 |
+
"loss": 2.1697,
|
3728 |
+
"step": 310000
|
3729 |
+
},
|
3730 |
+
{
|
3731 |
+
"epoch": 84.88,
|
3732 |
+
"learning_rate": 4.5762132604237866e-05,
|
3733 |
+
"loss": 2.1809,
|
3734 |
+
"step": 310500
|
3735 |
+
},
|
3736 |
+
{
|
3737 |
+
"epoch": 85.02,
|
3738 |
+
"learning_rate": 4.575529733424471e-05,
|
3739 |
+
"loss": 2.1723,
|
3740 |
+
"step": 311000
|
3741 |
+
},
|
3742 |
+
{
|
3743 |
+
"epoch": 85.16,
|
3744 |
+
"learning_rate": 4.574846206425154e-05,
|
3745 |
+
"loss": 2.1286,
|
3746 |
+
"step": 311500
|
3747 |
+
},
|
3748 |
+
{
|
3749 |
+
"epoch": 85.29,
|
3750 |
+
"learning_rate": 4.574162679425838e-05,
|
3751 |
+
"loss": 2.1458,
|
3752 |
+
"step": 312000
|
3753 |
+
},
|
3754 |
+
{
|
3755 |
+
"epoch": 85.43,
|
3756 |
+
"learning_rate": 4.5734791524265205e-05,
|
3757 |
+
"loss": 2.1496,
|
3758 |
+
"step": 312500
|
3759 |
+
},
|
3760 |
+
{
|
3761 |
+
"epoch": 85.57,
|
3762 |
+
"learning_rate": 4.5727956254272047e-05,
|
3763 |
+
"loss": 2.1572,
|
3764 |
+
"step": 313000
|
3765 |
+
},
|
3766 |
+
{
|
3767 |
+
"epoch": 85.7,
|
3768 |
+
"learning_rate": 4.572112098427888e-05,
|
3769 |
+
"loss": 2.1648,
|
3770 |
+
"step": 313500
|
3771 |
+
},
|
3772 |
+
{
|
3773 |
+
"epoch": 85.84,
|
3774 |
+
"learning_rate": 4.5714285714285716e-05,
|
3775 |
+
"loss": 2.1736,
|
3776 |
+
"step": 314000
|
3777 |
+
},
|
3778 |
+
{
|
3779 |
+
"epoch": 85.98,
|
3780 |
+
"learning_rate": 4.570745044429256e-05,
|
3781 |
+
"loss": 2.1776,
|
3782 |
+
"step": 314500
|
3783 |
+
},
|
3784 |
+
{
|
3785 |
+
"epoch": 86.11,
|
3786 |
+
"learning_rate": 4.5700615174299386e-05,
|
3787 |
+
"loss": 2.1304,
|
3788 |
+
"step": 315000
|
3789 |
+
},
|
3790 |
+
{
|
3791 |
+
"epoch": 86.25,
|
3792 |
+
"learning_rate": 4.569377990430622e-05,
|
3793 |
+
"loss": 2.1302,
|
3794 |
+
"step": 315500
|
3795 |
+
},
|
3796 |
+
{
|
3797 |
+
"epoch": 86.39,
|
3798 |
+
"learning_rate": 4.5686944634313055e-05,
|
3799 |
+
"loss": 2.1441,
|
3800 |
+
"step": 316000
|
3801 |
+
},
|
3802 |
+
{
|
3803 |
+
"epoch": 86.52,
|
3804 |
+
"learning_rate": 4.56801093643199e-05,
|
3805 |
+
"loss": 2.1551,
|
3806 |
+
"step": 316500
|
3807 |
+
},
|
3808 |
+
{
|
3809 |
+
"epoch": 86.66,
|
3810 |
+
"learning_rate": 4.567327409432673e-05,
|
3811 |
+
"loss": 2.1582,
|
3812 |
+
"step": 317000
|
3813 |
+
},
|
3814 |
+
{
|
3815 |
+
"epoch": 86.8,
|
3816 |
+
"learning_rate": 4.566643882433356e-05,
|
3817 |
+
"loss": 2.1666,
|
3818 |
+
"step": 317500
|
3819 |
+
},
|
3820 |
+
{
|
3821 |
+
"epoch": 86.93,
|
3822 |
+
"learning_rate": 4.5659603554340394e-05,
|
3823 |
+
"loss": 2.1599,
|
3824 |
+
"step": 318000
|
3825 |
+
},
|
3826 |
+
{
|
3827 |
+
"epoch": 87.07,
|
3828 |
+
"learning_rate": 4.5652768284347236e-05,
|
3829 |
+
"loss": 2.1394,
|
3830 |
+
"step": 318500
|
3831 |
+
},
|
3832 |
+
{
|
3833 |
+
"epoch": 87.21,
|
3834 |
+
"learning_rate": 4.564593301435407e-05,
|
3835 |
+
"loss": 2.1233,
|
3836 |
+
"step": 319000
|
3837 |
+
},
|
3838 |
+
{
|
3839 |
+
"epoch": 87.34,
|
3840 |
+
"learning_rate": 4.5639097744360906e-05,
|
3841 |
+
"loss": 2.1365,
|
3842 |
+
"step": 319500
|
3843 |
+
},
|
3844 |
+
{
|
3845 |
+
"epoch": 87.48,
|
3846 |
+
"learning_rate": 4.563226247436774e-05,
|
3847 |
+
"loss": 2.1441,
|
3848 |
+
"step": 320000
|
3849 |
+
},
|
3850 |
+
{
|
3851 |
+
"epoch": 87.62,
|
3852 |
+
"learning_rate": 4.5625427204374575e-05,
|
3853 |
+
"loss": 2.1482,
|
3854 |
+
"step": 320500
|
3855 |
+
},
|
3856 |
+
{
|
3857 |
+
"epoch": 87.75,
|
3858 |
+
"learning_rate": 4.561859193438141e-05,
|
3859 |
+
"loss": 2.1599,
|
3860 |
+
"step": 321000
|
3861 |
+
},
|
3862 |
+
{
|
3863 |
+
"epoch": 87.89,
|
3864 |
+
"learning_rate": 4.5611756664388245e-05,
|
3865 |
+
"loss": 2.1559,
|
3866 |
+
"step": 321500
|
3867 |
+
},
|
3868 |
+
{
|
3869 |
+
"epoch": 88.03,
|
3870 |
+
"learning_rate": 4.560492139439508e-05,
|
3871 |
+
"loss": 2.1513,
|
3872 |
+
"step": 322000
|
3873 |
+
},
|
3874 |
+
{
|
3875 |
+
"epoch": 88.16,
|
3876 |
+
"learning_rate": 4.5598086124401914e-05,
|
3877 |
+
"loss": 2.115,
|
3878 |
+
"step": 322500
|
3879 |
+
},
|
3880 |
+
{
|
3881 |
+
"epoch": 88.3,
|
3882 |
+
"learning_rate": 4.559125085440875e-05,
|
3883 |
+
"loss": 2.1336,
|
3884 |
+
"step": 323000
|
3885 |
+
},
|
3886 |
+
{
|
3887 |
+
"epoch": 88.44,
|
3888 |
+
"learning_rate": 4.5584415584415584e-05,
|
3889 |
+
"loss": 2.1394,
|
3890 |
+
"step": 323500
|
3891 |
+
},
|
3892 |
+
{
|
3893 |
+
"epoch": 88.57,
|
3894 |
+
"learning_rate": 4.5577580314422426e-05,
|
3895 |
+
"loss": 2.1366,
|
3896 |
+
"step": 324000
|
3897 |
+
},
|
3898 |
+
{
|
3899 |
+
"epoch": 88.71,
|
3900 |
+
"learning_rate": 4.557074504442926e-05,
|
3901 |
+
"loss": 2.1424,
|
3902 |
+
"step": 324500
|
3903 |
+
},
|
3904 |
+
{
|
3905 |
+
"epoch": 88.85,
|
3906 |
+
"learning_rate": 4.5563909774436095e-05,
|
3907 |
+
"loss": 2.151,
|
3908 |
+
"step": 325000
|
3909 |
+
},
|
3910 |
+
{
|
3911 |
+
"epoch": 88.98,
|
3912 |
+
"learning_rate": 4.555707450444292e-05,
|
3913 |
+
"loss": 2.1611,
|
3914 |
+
"step": 325500
|
3915 |
+
},
|
3916 |
+
{
|
3917 |
+
"epoch": 89.12,
|
3918 |
+
"learning_rate": 4.5550239234449765e-05,
|
3919 |
+
"loss": 2.1216,
|
3920 |
+
"step": 326000
|
3921 |
+
},
|
3922 |
+
{
|
3923 |
+
"epoch": 89.26,
|
3924 |
+
"learning_rate": 4.55434039644566e-05,
|
3925 |
+
"loss": 2.1112,
|
3926 |
+
"step": 326500
|
3927 |
+
},
|
3928 |
+
{
|
3929 |
+
"epoch": 89.39,
|
3930 |
+
"learning_rate": 4.5536568694463434e-05,
|
3931 |
+
"loss": 2.1246,
|
3932 |
+
"step": 327000
|
3933 |
+
},
|
3934 |
+
{
|
3935 |
+
"epoch": 89.53,
|
3936 |
+
"learning_rate": 4.552973342447027e-05,
|
3937 |
+
"loss": 2.1361,
|
3938 |
+
"step": 327500
|
3939 |
+
},
|
3940 |
+
{
|
3941 |
+
"epoch": 89.67,
|
3942 |
+
"learning_rate": 4.5522898154477104e-05,
|
3943 |
+
"loss": 2.1375,
|
3944 |
+
"step": 328000
|
3945 |
+
},
|
3946 |
+
{
|
3947 |
+
"epoch": 89.8,
|
3948 |
+
"learning_rate": 4.551606288448394e-05,
|
3949 |
+
"loss": 2.1438,
|
3950 |
+
"step": 328500
|
3951 |
+
},
|
3952 |
+
{
|
3953 |
+
"epoch": 89.94,
|
3954 |
+
"learning_rate": 4.5509227614490773e-05,
|
3955 |
+
"loss": 2.1525,
|
3956 |
+
"step": 329000
|
3957 |
+
},
|
3958 |
+
{
|
3959 |
+
"epoch": 90.08,
|
3960 |
+
"learning_rate": 4.550239234449761e-05,
|
3961 |
+
"loss": 2.127,
|
3962 |
+
"step": 329500
|
3963 |
+
},
|
3964 |
+
{
|
3965 |
+
"epoch": 90.21,
|
3966 |
+
"learning_rate": 4.549555707450445e-05,
|
3967 |
+
"loss": 2.1043,
|
3968 |
+
"step": 330000
|
3969 |
+
},
|
3970 |
+
{
|
3971 |
+
"epoch": 90.35,
|
3972 |
+
"learning_rate": 4.548872180451128e-05,
|
3973 |
+
"loss": 2.1148,
|
3974 |
+
"step": 330500
|
3975 |
+
},
|
3976 |
+
{
|
3977 |
+
"epoch": 90.49,
|
3978 |
+
"learning_rate": 4.548188653451811e-05,
|
3979 |
+
"loss": 2.1282,
|
3980 |
+
"step": 331000
|
3981 |
+
},
|
3982 |
+
{
|
3983 |
+
"epoch": 90.62,
|
3984 |
+
"learning_rate": 4.547505126452495e-05,
|
3985 |
+
"loss": 2.1347,
|
3986 |
+
"step": 331500
|
3987 |
+
},
|
3988 |
+
{
|
3989 |
+
"epoch": 90.76,
|
3990 |
+
"learning_rate": 4.546821599453179e-05,
|
3991 |
+
"loss": 2.1465,
|
3992 |
+
"step": 332000
|
3993 |
+
},
|
3994 |
+
{
|
3995 |
+
"epoch": 90.9,
|
3996 |
+
"learning_rate": 4.5461380724538624e-05,
|
3997 |
+
"loss": 2.1457,
|
3998 |
+
"step": 332500
|
3999 |
+
},
|
4000 |
+
{
|
4001 |
+
"epoch": 91.03,
|
4002 |
+
"learning_rate": 4.545454545454546e-05,
|
4003 |
+
"loss": 2.1369,
|
4004 |
+
"step": 333000
|
4005 |
+
},
|
4006 |
+
{
|
4007 |
+
"epoch": 91.17,
|
4008 |
+
"learning_rate": 4.544771018455229e-05,
|
4009 |
+
"loss": 2.0965,
|
4010 |
+
"step": 333500
|
4011 |
+
},
|
4012 |
+
{
|
4013 |
+
"epoch": 91.31,
|
4014 |
+
"learning_rate": 4.544087491455913e-05,
|
4015 |
+
"loss": 2.1165,
|
4016 |
+
"step": 334000
|
4017 |
+
},
|
4018 |
+
{
|
4019 |
+
"epoch": 91.44,
|
4020 |
+
"learning_rate": 4.543403964456596e-05,
|
4021 |
+
"loss": 2.1213,
|
4022 |
+
"step": 334500
|
4023 |
+
},
|
4024 |
+
{
|
4025 |
+
"epoch": 91.58,
|
4026 |
+
"learning_rate": 4.54272043745728e-05,
|
4027 |
+
"loss": 2.1286,
|
4028 |
+
"step": 335000
|
4029 |
+
},
|
4030 |
+
{
|
4031 |
+
"epoch": 91.72,
|
4032 |
+
"learning_rate": 4.542036910457963e-05,
|
4033 |
+
"loss": 2.1284,
|
4034 |
+
"step": 335500
|
4035 |
+
},
|
4036 |
+
{
|
4037 |
+
"epoch": 91.85,
|
4038 |
+
"learning_rate": 4.541353383458647e-05,
|
4039 |
+
"loss": 2.1407,
|
4040 |
+
"step": 336000
|
4041 |
+
},
|
4042 |
+
{
|
4043 |
+
"epoch": 91.99,
|
4044 |
+
"learning_rate": 4.54066985645933e-05,
|
4045 |
+
"loss": 2.1417,
|
4046 |
+
"step": 336500
|
4047 |
+
},
|
4048 |
+
{
|
4049 |
+
"epoch": 92.13,
|
4050 |
+
"learning_rate": 4.539986329460014e-05,
|
4051 |
+
"loss": 2.0955,
|
4052 |
+
"step": 337000
|
4053 |
+
},
|
4054 |
+
{
|
4055 |
+
"epoch": 92.26,
|
4056 |
+
"learning_rate": 4.539302802460698e-05,
|
4057 |
+
"loss": 2.0994,
|
4058 |
+
"step": 337500
|
4059 |
+
},
|
4060 |
+
{
|
4061 |
+
"epoch": 92.4,
|
4062 |
+
"learning_rate": 4.538619275461381e-05,
|
4063 |
+
"loss": 2.1162,
|
4064 |
+
"step": 338000
|
4065 |
+
},
|
4066 |
+
{
|
4067 |
+
"epoch": 92.54,
|
4068 |
+
"learning_rate": 4.537935748462064e-05,
|
4069 |
+
"loss": 2.1158,
|
4070 |
+
"step": 338500
|
4071 |
+
},
|
4072 |
+
{
|
4073 |
+
"epoch": 92.67,
|
4074 |
+
"learning_rate": 4.5372522214627476e-05,
|
4075 |
+
"loss": 2.1289,
|
4076 |
+
"step": 339000
|
4077 |
+
},
|
4078 |
+
{
|
4079 |
+
"epoch": 92.81,
|
4080 |
+
"learning_rate": 4.536568694463432e-05,
|
4081 |
+
"loss": 2.1288,
|
4082 |
+
"step": 339500
|
4083 |
+
},
|
4084 |
+
{
|
4085 |
+
"epoch": 92.95,
|
4086 |
+
"learning_rate": 4.535885167464115e-05,
|
4087 |
+
"loss": 2.1417,
|
4088 |
+
"step": 340000
|
4089 |
+
},
|
4090 |
+
{
|
4091 |
+
"epoch": 93.08,
|
4092 |
+
"learning_rate": 4.535201640464799e-05,
|
4093 |
+
"loss": 2.1089,
|
4094 |
+
"step": 340500
|
4095 |
+
},
|
4096 |
+
{
|
4097 |
+
"epoch": 93.22,
|
4098 |
+
"learning_rate": 4.5345181134654815e-05,
|
4099 |
+
"loss": 2.0924,
|
4100 |
+
"step": 341000
|
4101 |
+
},
|
4102 |
+
{
|
4103 |
+
"epoch": 93.36,
|
4104 |
+
"learning_rate": 4.533834586466166e-05,
|
4105 |
+
"loss": 2.1037,
|
4106 |
+
"step": 341500
|
4107 |
+
},
|
4108 |
+
{
|
4109 |
+
"epoch": 93.49,
|
4110 |
+
"learning_rate": 4.533151059466849e-05,
|
4111 |
+
"loss": 2.1097,
|
4112 |
+
"step": 342000
|
4113 |
+
},
|
4114 |
+
{
|
4115 |
+
"epoch": 93.63,
|
4116 |
+
"learning_rate": 4.5324675324675326e-05,
|
4117 |
+
"loss": 2.1164,
|
4118 |
+
"step": 342500
|
4119 |
+
},
|
4120 |
+
{
|
4121 |
+
"epoch": 93.77,
|
4122 |
+
"learning_rate": 4.531784005468217e-05,
|
4123 |
+
"loss": 2.1277,
|
4124 |
+
"step": 343000
|
4125 |
+
},
|
4126 |
+
{
|
4127 |
+
"epoch": 93.9,
|
4128 |
+
"learning_rate": 4.5311004784688996e-05,
|
4129 |
+
"loss": 2.1281,
|
4130 |
+
"step": 343500
|
4131 |
+
},
|
4132 |
+
{
|
4133 |
+
"epoch": 94.04,
|
4134 |
+
"learning_rate": 4.530416951469583e-05,
|
4135 |
+
"loss": 2.1238,
|
4136 |
+
"step": 344000
|
4137 |
+
},
|
4138 |
+
{
|
4139 |
+
"epoch": 94.18,
|
4140 |
+
"learning_rate": 4.5297334244702666e-05,
|
4141 |
+
"loss": 2.0879,
|
4142 |
+
"step": 344500
|
4143 |
+
},
|
4144 |
+
{
|
4145 |
+
"epoch": 94.31,
|
4146 |
+
"learning_rate": 4.529049897470951e-05,
|
4147 |
+
"loss": 2.0903,
|
4148 |
+
"step": 345000
|
4149 |
+
},
|
4150 |
+
{
|
4151 |
+
"epoch": 94.45,
|
4152 |
+
"learning_rate": 4.528366370471634e-05,
|
4153 |
+
"loss": 2.1058,
|
4154 |
+
"step": 345500
|
4155 |
+
},
|
4156 |
+
{
|
4157 |
+
"epoch": 94.59,
|
4158 |
+
"learning_rate": 4.527682843472317e-05,
|
4159 |
+
"loss": 2.1162,
|
4160 |
+
"step": 346000
|
4161 |
+
},
|
4162 |
+
{
|
4163 |
+
"epoch": 94.72,
|
4164 |
+
"learning_rate": 4.5269993164730005e-05,
|
4165 |
+
"loss": 2.1162,
|
4166 |
+
"step": 346500
|
4167 |
+
},
|
4168 |
+
{
|
4169 |
+
"epoch": 94.86,
|
4170 |
+
"learning_rate": 4.5263157894736846e-05,
|
4171 |
+
"loss": 2.1276,
|
4172 |
+
"step": 347000
|
4173 |
+
},
|
4174 |
+
{
|
4175 |
+
"epoch": 95.0,
|
4176 |
+
"learning_rate": 4.525632262474368e-05,
|
4177 |
+
"loss": 2.1259,
|
4178 |
+
"step": 347500
|
4179 |
+
},
|
4180 |
+
{
|
4181 |
+
"epoch": 95.13,
|
4182 |
+
"learning_rate": 4.5249487354750516e-05,
|
4183 |
+
"loss": 2.0818,
|
4184 |
+
"step": 348000
|
4185 |
+
},
|
4186 |
+
{
|
4187 |
+
"epoch": 95.27,
|
4188 |
+
"learning_rate": 4.524265208475735e-05,
|
4189 |
+
"loss": 2.0907,
|
4190 |
+
"step": 348500
|
4191 |
+
},
|
4192 |
+
{
|
4193 |
+
"epoch": 95.41,
|
4194 |
+
"learning_rate": 4.5235816814764186e-05,
|
4195 |
+
"loss": 2.0926,
|
4196 |
+
"step": 349000
|
4197 |
+
},
|
4198 |
+
{
|
4199 |
+
"epoch": 95.54,
|
4200 |
+
"learning_rate": 4.522898154477102e-05,
|
4201 |
+
"loss": 2.1014,
|
4202 |
+
"step": 349500
|
4203 |
+
},
|
4204 |
+
{
|
4205 |
+
"epoch": 95.68,
|
4206 |
+
"learning_rate": 4.5222146274777855e-05,
|
4207 |
+
"loss": 2.1133,
|
4208 |
+
"step": 350000
|
4209 |
+
},
|
4210 |
+
{
|
4211 |
+
"epoch": 95.82,
|
4212 |
+
"learning_rate": 4.521531100478469e-05,
|
4213 |
+
"loss": 2.1189,
|
4214 |
+
"step": 350500
|
4215 |
+
},
|
4216 |
+
{
|
4217 |
+
"epoch": 95.95,
|
4218 |
+
"learning_rate": 4.5208475734791525e-05,
|
4219 |
+
"loss": 2.1202,
|
4220 |
+
"step": 351000
|
4221 |
+
},
|
4222 |
+
{
|
4223 |
+
"epoch": 96.09,
|
4224 |
+
"learning_rate": 4.520164046479836e-05,
|
4225 |
+
"loss": 2.0911,
|
4226 |
+
"step": 351500
|
4227 |
+
},
|
4228 |
+
{
|
4229 |
+
"epoch": 96.23,
|
4230 |
+
"learning_rate": 4.5194805194805194e-05,
|
4231 |
+
"loss": 2.0824,
|
4232 |
+
"step": 352000
|
4233 |
+
},
|
4234 |
+
{
|
4235 |
+
"epoch": 96.36,
|
4236 |
+
"learning_rate": 4.5187969924812036e-05,
|
4237 |
+
"loss": 2.0898,
|
4238 |
+
"step": 352500
|
4239 |
+
},
|
4240 |
+
{
|
4241 |
+
"epoch": 96.5,
|
4242 |
+
"learning_rate": 4.518113465481887e-05,
|
4243 |
+
"loss": 2.0946,
|
4244 |
+
"step": 353000
|
4245 |
+
},
|
4246 |
+
{
|
4247 |
+
"epoch": 96.64,
|
4248 |
+
"learning_rate": 4.5174299384825705e-05,
|
4249 |
+
"loss": 2.1039,
|
4250 |
+
"step": 353500
|
4251 |
+
},
|
4252 |
+
{
|
4253 |
+
"epoch": 96.77,
|
4254 |
+
"learning_rate": 4.5167464114832533e-05,
|
4255 |
+
"loss": 2.1076,
|
4256 |
+
"step": 354000
|
4257 |
+
},
|
4258 |
+
{
|
4259 |
+
"epoch": 96.91,
|
4260 |
+
"learning_rate": 4.5160628844839375e-05,
|
4261 |
+
"loss": 2.1183,
|
4262 |
+
"step": 354500
|
4263 |
+
},
|
4264 |
+
{
|
4265 |
+
"epoch": 97.05,
|
4266 |
+
"learning_rate": 4.515379357484621e-05,
|
4267 |
+
"loss": 2.1028,
|
4268 |
+
"step": 355000
|
4269 |
+
},
|
4270 |
+
{
|
4271 |
+
"epoch": 97.18,
|
4272 |
+
"learning_rate": 4.5146958304853045e-05,
|
4273 |
+
"loss": 2.0748,
|
4274 |
+
"step": 355500
|
4275 |
+
},
|
4276 |
+
{
|
4277 |
+
"epoch": 97.32,
|
4278 |
+
"learning_rate": 4.514012303485988e-05,
|
4279 |
+
"loss": 2.082,
|
4280 |
+
"step": 356000
|
4281 |
+
},
|
4282 |
+
{
|
4283 |
+
"epoch": 97.46,
|
4284 |
+
"learning_rate": 4.5133287764866714e-05,
|
4285 |
+
"loss": 2.0929,
|
4286 |
+
"step": 356500
|
4287 |
+
},
|
4288 |
+
{
|
4289 |
+
"epoch": 97.59,
|
4290 |
+
"learning_rate": 4.512645249487355e-05,
|
4291 |
+
"loss": 2.0978,
|
4292 |
+
"step": 357000
|
4293 |
+
},
|
4294 |
+
{
|
4295 |
+
"epoch": 97.73,
|
4296 |
+
"learning_rate": 4.5119617224880384e-05,
|
4297 |
+
"loss": 2.1015,
|
4298 |
+
"step": 357500
|
4299 |
+
},
|
4300 |
+
{
|
4301 |
+
"epoch": 97.87,
|
4302 |
+
"learning_rate": 4.511278195488722e-05,
|
4303 |
+
"loss": 2.1031,
|
4304 |
+
"step": 358000
|
4305 |
+
},
|
4306 |
+
{
|
4307 |
+
"epoch": 98.0,
|
4308 |
+
"learning_rate": 4.510594668489406e-05,
|
4309 |
+
"loss": 2.1172,
|
4310 |
+
"step": 358500
|
4311 |
+
},
|
4312 |
+
{
|
4313 |
+
"epoch": 98.14,
|
4314 |
+
"learning_rate": 4.509911141490089e-05,
|
4315 |
+
"loss": 2.0695,
|
4316 |
+
"step": 359000
|
4317 |
+
},
|
4318 |
+
{
|
4319 |
+
"epoch": 98.28,
|
4320 |
+
"learning_rate": 4.509227614490772e-05,
|
4321 |
+
"loss": 2.0733,
|
4322 |
+
"step": 359500
|
4323 |
+
},
|
4324 |
+
{
|
4325 |
+
"epoch": 98.41,
|
4326 |
+
"learning_rate": 4.5085440874914565e-05,
|
4327 |
+
"loss": 2.0832,
|
4328 |
+
"step": 360000
|
4329 |
}
|
4330 |
],
|
4331 |
"max_steps": 3658000,
|
4332 |
"num_train_epochs": 1000,
|
4333 |
+
"total_flos": 724673057990639616,
|
4334 |
"trial_name": null,
|
4335 |
"trial_params": null
|
4336 |
}
|