Training checkpoint at step 11000
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-9000",
|
| 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,
|
|
@@ -3608,6 +3608,366 @@
|
|
| 3608 |
"eval_samples_per_second": 3.215,
|
| 3609 |
"eval_steps_per_second": 1.607,
|
| 3610 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 3611 |
}
|
| 3612 |
],
|
| 3613 |
"logging_steps": 25,
|
|
@@ -3627,7 +3987,7 @@
|
|
| 3627 |
"attributes": {}
|
| 3628 |
}
|
| 3629 |
},
|
| 3630 |
-
"total_flos": 3.
|
| 3631 |
"train_batch_size": 1,
|
| 3632 |
"trial_name": null,
|
| 3633 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
+
"best_global_step": 10900,
|
| 3 |
+
"best_metric": 2.4055566787719727,
|
| 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-9000",
|
| 5 |
+
"epoch": 0.22,
|
| 6 |
"eval_steps": 100,
|
| 7 |
+
"global_step": 11000,
|
| 8 |
"is_hyper_param_search": false,
|
| 9 |
"is_local_process_zero": true,
|
| 10 |
"is_world_process_zero": true,
|
|
|
|
| 3608 |
"eval_samples_per_second": 3.215,
|
| 3609 |
"eval_steps_per_second": 1.607,
|
| 3610 |
"step": 10000
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 0.2005,
|
| 3614 |
+
"grad_norm": 0.5959223333719136,
|
| 3615 |
+
"learning_rate": 8.883555555555557e-06,
|
| 3616 |
+
"loss": 2.387,
|
| 3617 |
+
"step": 10025
|
| 3618 |
+
},
|
| 3619 |
+
{
|
| 3620 |
+
"epoch": 0.201,
|
| 3621 |
+
"grad_norm": 0.604008744038432,
|
| 3622 |
+
"learning_rate": 8.878e-06,
|
| 3623 |
+
"loss": 2.4016,
|
| 3624 |
+
"step": 10050
|
| 3625 |
+
},
|
| 3626 |
+
{
|
| 3627 |
+
"epoch": 0.2015,
|
| 3628 |
+
"grad_norm": 0.5721419521050413,
|
| 3629 |
+
"learning_rate": 8.872444444444444e-06,
|
| 3630 |
+
"loss": 2.3884,
|
| 3631 |
+
"step": 10075
|
| 3632 |
+
},
|
| 3633 |
+
{
|
| 3634 |
+
"epoch": 0.202,
|
| 3635 |
+
"grad_norm": 0.5986167284289824,
|
| 3636 |
+
"learning_rate": 8.86688888888889e-06,
|
| 3637 |
+
"loss": 2.3945,
|
| 3638 |
+
"step": 10100
|
| 3639 |
+
},
|
| 3640 |
+
{
|
| 3641 |
+
"epoch": 0.202,
|
| 3642 |
+
"eval_loss": 2.4074654579162598,
|
| 3643 |
+
"eval_runtime": 31.8658,
|
| 3644 |
+
"eval_samples_per_second": 3.201,
|
| 3645 |
+
"eval_steps_per_second": 1.6,
|
| 3646 |
+
"step": 10100
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 0.2025,
|
| 3650 |
+
"grad_norm": 0.6046479507995179,
|
| 3651 |
+
"learning_rate": 8.861333333333334e-06,
|
| 3652 |
+
"loss": 2.3858,
|
| 3653 |
+
"step": 10125
|
| 3654 |
+
},
|
| 3655 |
+
{
|
| 3656 |
+
"epoch": 0.203,
|
| 3657 |
+
"grad_norm": 0.5633013817443194,
|
| 3658 |
+
"learning_rate": 8.855777777777778e-06,
|
| 3659 |
+
"loss": 2.3879,
|
| 3660 |
+
"step": 10150
|
| 3661 |
+
},
|
| 3662 |
+
{
|
| 3663 |
+
"epoch": 0.2035,
|
| 3664 |
+
"grad_norm": 0.5953174401982892,
|
| 3665 |
+
"learning_rate": 8.850222222222223e-06,
|
| 3666 |
+
"loss": 2.3967,
|
| 3667 |
+
"step": 10175
|
| 3668 |
+
},
|
| 3669 |
+
{
|
| 3670 |
+
"epoch": 0.204,
|
| 3671 |
+
"grad_norm": 0.6306212647705982,
|
| 3672 |
+
"learning_rate": 8.844666666666667e-06,
|
| 3673 |
+
"loss": 2.3927,
|
| 3674 |
+
"step": 10200
|
| 3675 |
+
},
|
| 3676 |
+
{
|
| 3677 |
+
"epoch": 0.204,
|
| 3678 |
+
"eval_loss": 2.407031297683716,
|
| 3679 |
+
"eval_runtime": 31.7801,
|
| 3680 |
+
"eval_samples_per_second": 3.21,
|
| 3681 |
+
"eval_steps_per_second": 1.605,
|
| 3682 |
+
"step": 10200
|
| 3683 |
+
},
|
| 3684 |
+
{
|
| 3685 |
+
"epoch": 0.2045,
|
| 3686 |
+
"grad_norm": 0.5605617492602121,
|
| 3687 |
+
"learning_rate": 8.839111111111112e-06,
|
| 3688 |
+
"loss": 2.4081,
|
| 3689 |
+
"step": 10225
|
| 3690 |
+
},
|
| 3691 |
+
{
|
| 3692 |
+
"epoch": 0.205,
|
| 3693 |
+
"grad_norm": 0.5739246143474902,
|
| 3694 |
+
"learning_rate": 8.833555555555556e-06,
|
| 3695 |
+
"loss": 2.3841,
|
| 3696 |
+
"step": 10250
|
| 3697 |
+
},
|
| 3698 |
+
{
|
| 3699 |
+
"epoch": 0.2055,
|
| 3700 |
+
"grad_norm": 0.5938549959471341,
|
| 3701 |
+
"learning_rate": 8.828000000000001e-06,
|
| 3702 |
+
"loss": 2.3902,
|
| 3703 |
+
"step": 10275
|
| 3704 |
+
},
|
| 3705 |
+
{
|
| 3706 |
+
"epoch": 0.206,
|
| 3707 |
+
"grad_norm": 0.5902936931354175,
|
| 3708 |
+
"learning_rate": 8.822444444444446e-06,
|
| 3709 |
+
"loss": 2.3905,
|
| 3710 |
+
"step": 10300
|
| 3711 |
+
},
|
| 3712 |
+
{
|
| 3713 |
+
"epoch": 0.206,
|
| 3714 |
+
"eval_loss": 2.4066004753112793,
|
| 3715 |
+
"eval_runtime": 31.7707,
|
| 3716 |
+
"eval_samples_per_second": 3.211,
|
| 3717 |
+
"eval_steps_per_second": 1.605,
|
| 3718 |
+
"step": 10300
|
| 3719 |
+
},
|
| 3720 |
+
{
|
| 3721 |
+
"epoch": 0.2065,
|
| 3722 |
+
"grad_norm": 0.5697435057211838,
|
| 3723 |
+
"learning_rate": 8.81688888888889e-06,
|
| 3724 |
+
"loss": 2.3854,
|
| 3725 |
+
"step": 10325
|
| 3726 |
+
},
|
| 3727 |
+
{
|
| 3728 |
+
"epoch": 0.207,
|
| 3729 |
+
"grad_norm": 0.5879126074250441,
|
| 3730 |
+
"learning_rate": 8.811333333333333e-06,
|
| 3731 |
+
"loss": 2.3917,
|
| 3732 |
+
"step": 10350
|
| 3733 |
+
},
|
| 3734 |
+
{
|
| 3735 |
+
"epoch": 0.2075,
|
| 3736 |
+
"grad_norm": 0.5800642153182343,
|
| 3737 |
+
"learning_rate": 8.805777777777778e-06,
|
| 3738 |
+
"loss": 2.3929,
|
| 3739 |
+
"step": 10375
|
| 3740 |
+
},
|
| 3741 |
+
{
|
| 3742 |
+
"epoch": 0.208,
|
| 3743 |
+
"grad_norm": 0.5794546973922929,
|
| 3744 |
+
"learning_rate": 8.800222222222224e-06,
|
| 3745 |
+
"loss": 2.3912,
|
| 3746 |
+
"step": 10400
|
| 3747 |
+
},
|
| 3748 |
+
{
|
| 3749 |
+
"epoch": 0.208,
|
| 3750 |
+
"eval_loss": 2.4065024852752686,
|
| 3751 |
+
"eval_runtime": 31.7191,
|
| 3752 |
+
"eval_samples_per_second": 3.216,
|
| 3753 |
+
"eval_steps_per_second": 1.608,
|
| 3754 |
+
"step": 10400
|
| 3755 |
+
},
|
| 3756 |
+
{
|
| 3757 |
+
"epoch": 0.2085,
|
| 3758 |
+
"grad_norm": 0.5776454190712899,
|
| 3759 |
+
"learning_rate": 8.794666666666667e-06,
|
| 3760 |
+
"loss": 2.386,
|
| 3761 |
+
"step": 10425
|
| 3762 |
+
},
|
| 3763 |
+
{
|
| 3764 |
+
"epoch": 0.209,
|
| 3765 |
+
"grad_norm": 0.5578455228918948,
|
| 3766 |
+
"learning_rate": 8.78911111111111e-06,
|
| 3767 |
+
"loss": 2.3869,
|
| 3768 |
+
"step": 10450
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 0.2095,
|
| 3772 |
+
"grad_norm": 0.5721674793656858,
|
| 3773 |
+
"learning_rate": 8.783555555555556e-06,
|
| 3774 |
+
"loss": 2.3779,
|
| 3775 |
+
"step": 10475
|
| 3776 |
+
},
|
| 3777 |
+
{
|
| 3778 |
+
"epoch": 0.21,
|
| 3779 |
+
"grad_norm": 0.5950633442730316,
|
| 3780 |
+
"learning_rate": 8.778000000000001e-06,
|
| 3781 |
+
"loss": 2.3845,
|
| 3782 |
+
"step": 10500
|
| 3783 |
+
},
|
| 3784 |
+
{
|
| 3785 |
+
"epoch": 0.21,
|
| 3786 |
+
"eval_loss": 2.4065566062927246,
|
| 3787 |
+
"eval_runtime": 31.8091,
|
| 3788 |
+
"eval_samples_per_second": 3.207,
|
| 3789 |
+
"eval_steps_per_second": 1.603,
|
| 3790 |
+
"step": 10500
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 0.2105,
|
| 3794 |
+
"grad_norm": 0.605078293663896,
|
| 3795 |
+
"learning_rate": 8.772444444444445e-06,
|
| 3796 |
+
"loss": 2.3913,
|
| 3797 |
+
"step": 10525
|
| 3798 |
+
},
|
| 3799 |
+
{
|
| 3800 |
+
"epoch": 0.211,
|
| 3801 |
+
"grad_norm": 0.567849892850204,
|
| 3802 |
+
"learning_rate": 8.766888888888888e-06,
|
| 3803 |
+
"loss": 2.3966,
|
| 3804 |
+
"step": 10550
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 0.2115,
|
| 3808 |
+
"grad_norm": 0.6876645024191659,
|
| 3809 |
+
"learning_rate": 8.761333333333334e-06,
|
| 3810 |
+
"loss": 2.3993,
|
| 3811 |
+
"step": 10575
|
| 3812 |
+
},
|
| 3813 |
+
{
|
| 3814 |
+
"epoch": 0.212,
|
| 3815 |
+
"grad_norm": 0.5841938304908528,
|
| 3816 |
+
"learning_rate": 8.755777777777779e-06,
|
| 3817 |
+
"loss": 2.3916,
|
| 3818 |
+
"step": 10600
|
| 3819 |
+
},
|
| 3820 |
+
{
|
| 3821 |
+
"epoch": 0.212,
|
| 3822 |
+
"eval_loss": 2.4061877727508545,
|
| 3823 |
+
"eval_runtime": 31.8484,
|
| 3824 |
+
"eval_samples_per_second": 3.203,
|
| 3825 |
+
"eval_steps_per_second": 1.601,
|
| 3826 |
+
"step": 10600
|
| 3827 |
+
},
|
| 3828 |
+
{
|
| 3829 |
+
"epoch": 0.2125,
|
| 3830 |
+
"grad_norm": 0.5649004204666818,
|
| 3831 |
+
"learning_rate": 8.750222222222223e-06,
|
| 3832 |
+
"loss": 2.381,
|
| 3833 |
+
"step": 10625
|
| 3834 |
+
},
|
| 3835 |
+
{
|
| 3836 |
+
"epoch": 0.213,
|
| 3837 |
+
"grad_norm": 0.5678489376050115,
|
| 3838 |
+
"learning_rate": 8.744666666666666e-06,
|
| 3839 |
+
"loss": 2.3995,
|
| 3840 |
+
"step": 10650
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 0.2135,
|
| 3844 |
+
"grad_norm": 0.5712733595317334,
|
| 3845 |
+
"learning_rate": 8.739111111111111e-06,
|
| 3846 |
+
"loss": 2.3954,
|
| 3847 |
+
"step": 10675
|
| 3848 |
+
},
|
| 3849 |
+
{
|
| 3850 |
+
"epoch": 0.214,
|
| 3851 |
+
"grad_norm": 0.573353636066434,
|
| 3852 |
+
"learning_rate": 8.733555555555557e-06,
|
| 3853 |
+
"loss": 2.379,
|
| 3854 |
+
"step": 10700
|
| 3855 |
+
},
|
| 3856 |
+
{
|
| 3857 |
+
"epoch": 0.214,
|
| 3858 |
+
"eval_loss": 2.4055771827697754,
|
| 3859 |
+
"eval_runtime": 31.8192,
|
| 3860 |
+
"eval_samples_per_second": 3.206,
|
| 3861 |
+
"eval_steps_per_second": 1.603,
|
| 3862 |
+
"step": 10700
|
| 3863 |
+
},
|
| 3864 |
+
{
|
| 3865 |
+
"epoch": 0.2145,
|
| 3866 |
+
"grad_norm": 0.6133309651928519,
|
| 3867 |
+
"learning_rate": 8.728e-06,
|
| 3868 |
+
"loss": 2.3946,
|
| 3869 |
+
"step": 10725
|
| 3870 |
+
},
|
| 3871 |
+
{
|
| 3872 |
+
"epoch": 0.215,
|
| 3873 |
+
"grad_norm": 0.6033931866035528,
|
| 3874 |
+
"learning_rate": 8.722444444444445e-06,
|
| 3875 |
+
"loss": 2.3935,
|
| 3876 |
+
"step": 10750
|
| 3877 |
+
},
|
| 3878 |
+
{
|
| 3879 |
+
"epoch": 0.2155,
|
| 3880 |
+
"grad_norm": 0.6008672136487845,
|
| 3881 |
+
"learning_rate": 8.716888888888889e-06,
|
| 3882 |
+
"loss": 2.3872,
|
| 3883 |
+
"step": 10775
|
| 3884 |
+
},
|
| 3885 |
+
{
|
| 3886 |
+
"epoch": 0.216,
|
| 3887 |
+
"grad_norm": 0.5728704483928734,
|
| 3888 |
+
"learning_rate": 8.711333333333334e-06,
|
| 3889 |
+
"loss": 2.3917,
|
| 3890 |
+
"step": 10800
|
| 3891 |
+
},
|
| 3892 |
+
{
|
| 3893 |
+
"epoch": 0.216,
|
| 3894 |
+
"eval_loss": 2.4059016704559326,
|
| 3895 |
+
"eval_runtime": 31.7995,
|
| 3896 |
+
"eval_samples_per_second": 3.208,
|
| 3897 |
+
"eval_steps_per_second": 1.604,
|
| 3898 |
+
"step": 10800
|
| 3899 |
+
},
|
| 3900 |
+
{
|
| 3901 |
+
"epoch": 0.2165,
|
| 3902 |
+
"grad_norm": 0.5888944153423502,
|
| 3903 |
+
"learning_rate": 8.705777777777778e-06,
|
| 3904 |
+
"loss": 2.3946,
|
| 3905 |
+
"step": 10825
|
| 3906 |
+
},
|
| 3907 |
+
{
|
| 3908 |
+
"epoch": 0.217,
|
| 3909 |
+
"grad_norm": 0.5947880979306366,
|
| 3910 |
+
"learning_rate": 8.700222222222223e-06,
|
| 3911 |
+
"loss": 2.3736,
|
| 3912 |
+
"step": 10850
|
| 3913 |
+
},
|
| 3914 |
+
{
|
| 3915 |
+
"epoch": 0.2175,
|
| 3916 |
+
"grad_norm": 0.6163696606959644,
|
| 3917 |
+
"learning_rate": 8.694666666666668e-06,
|
| 3918 |
+
"loss": 2.3838,
|
| 3919 |
+
"step": 10875
|
| 3920 |
+
},
|
| 3921 |
+
{
|
| 3922 |
+
"epoch": 0.218,
|
| 3923 |
+
"grad_norm": 0.6004092938812543,
|
| 3924 |
+
"learning_rate": 8.689111111111112e-06,
|
| 3925 |
+
"loss": 2.3942,
|
| 3926 |
+
"step": 10900
|
| 3927 |
+
},
|
| 3928 |
+
{
|
| 3929 |
+
"epoch": 0.218,
|
| 3930 |
+
"eval_loss": 2.4055566787719727,
|
| 3931 |
+
"eval_runtime": 31.7386,
|
| 3932 |
+
"eval_samples_per_second": 3.214,
|
| 3933 |
+
"eval_steps_per_second": 1.607,
|
| 3934 |
+
"step": 10900
|
| 3935 |
+
},
|
| 3936 |
+
{
|
| 3937 |
+
"epoch": 0.2185,
|
| 3938 |
+
"grad_norm": 0.5786273641598245,
|
| 3939 |
+
"learning_rate": 8.683555555555555e-06,
|
| 3940 |
+
"loss": 2.3938,
|
| 3941 |
+
"step": 10925
|
| 3942 |
+
},
|
| 3943 |
+
{
|
| 3944 |
+
"epoch": 0.219,
|
| 3945 |
+
"grad_norm": 0.5764162885826465,
|
| 3946 |
+
"learning_rate": 8.678e-06,
|
| 3947 |
+
"loss": 2.3939,
|
| 3948 |
+
"step": 10950
|
| 3949 |
+
},
|
| 3950 |
+
{
|
| 3951 |
+
"epoch": 0.2195,
|
| 3952 |
+
"grad_norm": 0.5923291223123188,
|
| 3953 |
+
"learning_rate": 8.672444444444446e-06,
|
| 3954 |
+
"loss": 2.3847,
|
| 3955 |
+
"step": 10975
|
| 3956 |
+
},
|
| 3957 |
+
{
|
| 3958 |
+
"epoch": 0.22,
|
| 3959 |
+
"grad_norm": 0.6102815146568634,
|
| 3960 |
+
"learning_rate": 8.66688888888889e-06,
|
| 3961 |
+
"loss": 2.3901,
|
| 3962 |
+
"step": 11000
|
| 3963 |
+
},
|
| 3964 |
+
{
|
| 3965 |
+
"epoch": 0.22,
|
| 3966 |
+
"eval_loss": 2.405616044998169,
|
| 3967 |
+
"eval_runtime": 31.7048,
|
| 3968 |
+
"eval_samples_per_second": 3.217,
|
| 3969 |
+
"eval_steps_per_second": 1.609,
|
| 3970 |
+
"step": 11000
|
| 3971 |
}
|
| 3972 |
],
|
| 3973 |
"logging_steps": 25,
|
|
|
|
| 3987 |
"attributes": {}
|
| 3988 |
}
|
| 3989 |
},
|
| 3990 |
+
"total_flos": 3.5015225281599242e+19,
|
| 3991 |
"train_batch_size": 1,
|
| 3992 |
"trial_name": null,
|
| 3993 |
"trial_params": null
|