Training in progress, step 716, 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 59933632
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5a0be10fbdab56a6574767c96c6cb6d55e2096cfe6ce02be376170cb6591b455
|
3 |
size 59933632
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 31823460
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:34352f87b92d2f74d7d9b5a0472a2c3eddc2fa9e8a4413356e6f5508c4a7188c
|
3 |
size 31823460
|
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 sha256:075e01a41ae15ae31e4905292e60381e146a6e6459533dfec8259d0b29a913d9
|
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 sha256:3b5f024e7c5c8c41cb2db928207237f59d43ea79815d962e48cc1e5a09368626
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0
|
5 |
"eval_steps": 179,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -3790,6 +3790,1267 @@
|
|
3790 |
"eval_samples_per_second": 27.362,
|
3791 |
"eval_steps_per_second": 13.704,
|
3792 |
"step": 537
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3793 |
}
|
3794 |
],
|
3795 |
"logging_steps": 1,
|
@@ -3804,12 +5065,12 @@
|
|
3804 |
"should_evaluate": false,
|
3805 |
"should_log": false,
|
3806 |
"should_save": true,
|
3807 |
-
"should_training_stop":
|
3808 |
},
|
3809 |
"attributes": {}
|
3810 |
}
|
3811 |
},
|
3812 |
-
"total_flos": 1.
|
3813 |
"train_batch_size": 2,
|
3814 |
"trial_name": null,
|
3815 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0,
|
5 |
"eval_steps": 179,
|
6 |
+
"global_step": 716,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3790 |
"eval_samples_per_second": 27.362,
|
3791 |
"eval_steps_per_second": 13.704,
|
3792 |
"step": 537
|
3793 |
+
},
|
3794 |
+
{
|
3795 |
+
"epoch": 0.7513966480446927,
|
3796 |
+
"grad_norm": 0.5159032940864563,
|
3797 |
+
"learning_rate": 2.9762871191687313e-05,
|
3798 |
+
"loss": 0.0998,
|
3799 |
+
"step": 538
|
3800 |
+
},
|
3801 |
+
{
|
3802 |
+
"epoch": 0.7527932960893855,
|
3803 |
+
"grad_norm": 0.5139814615249634,
|
3804 |
+
"learning_rate": 2.9446822653615614e-05,
|
3805 |
+
"loss": 0.106,
|
3806 |
+
"step": 539
|
3807 |
+
},
|
3808 |
+
{
|
3809 |
+
"epoch": 0.7541899441340782,
|
3810 |
+
"grad_norm": 0.6283978819847107,
|
3811 |
+
"learning_rate": 2.913217114705975e-05,
|
3812 |
+
"loss": 0.122,
|
3813 |
+
"step": 540
|
3814 |
+
},
|
3815 |
+
{
|
3816 |
+
"epoch": 0.755586592178771,
|
3817 |
+
"grad_norm": 0.7149221897125244,
|
3818 |
+
"learning_rate": 2.8818922902470135e-05,
|
3819 |
+
"loss": 0.1113,
|
3820 |
+
"step": 541
|
3821 |
+
},
|
3822 |
+
{
|
3823 |
+
"epoch": 0.7569832402234636,
|
3824 |
+
"grad_norm": 0.6622272729873657,
|
3825 |
+
"learning_rate": 2.850708412251103e-05,
|
3826 |
+
"loss": 0.1471,
|
3827 |
+
"step": 542
|
3828 |
+
},
|
3829 |
+
{
|
3830 |
+
"epoch": 0.7583798882681564,
|
3831 |
+
"grad_norm": 0.4957078993320465,
|
3832 |
+
"learning_rate": 2.819666098193764e-05,
|
3833 |
+
"loss": 0.1187,
|
3834 |
+
"step": 543
|
3835 |
+
},
|
3836 |
+
{
|
3837 |
+
"epoch": 0.7597765363128491,
|
3838 |
+
"grad_norm": 0.6351447105407715,
|
3839 |
+
"learning_rate": 2.7887659627474017e-05,
|
3840 |
+
"loss": 0.1239,
|
3841 |
+
"step": 544
|
3842 |
+
},
|
3843 |
+
{
|
3844 |
+
"epoch": 0.7611731843575419,
|
3845 |
+
"grad_norm": 0.7053154706954956,
|
3846 |
+
"learning_rate": 2.758008617769129e-05,
|
3847 |
+
"loss": 0.1463,
|
3848 |
+
"step": 545
|
3849 |
+
},
|
3850 |
+
{
|
3851 |
+
"epoch": 0.7625698324022346,
|
3852 |
+
"grad_norm": 0.6546825766563416,
|
3853 |
+
"learning_rate": 2.7273946722886366e-05,
|
3854 |
+
"loss": 0.1338,
|
3855 |
+
"step": 546
|
3856 |
+
},
|
3857 |
+
{
|
3858 |
+
"epoch": 0.7639664804469274,
|
3859 |
+
"grad_norm": 0.3492923378944397,
|
3860 |
+
"learning_rate": 2.6969247324961555e-05,
|
3861 |
+
"loss": 0.0717,
|
3862 |
+
"step": 547
|
3863 |
+
},
|
3864 |
+
{
|
3865 |
+
"epoch": 0.7653631284916201,
|
3866 |
+
"grad_norm": 0.8152084946632385,
|
3867 |
+
"learning_rate": 2.6665994017304407e-05,
|
3868 |
+
"loss": 0.1597,
|
3869 |
+
"step": 548
|
3870 |
+
},
|
3871 |
+
{
|
3872 |
+
"epoch": 0.7667597765363129,
|
3873 |
+
"grad_norm": 1.4545857906341553,
|
3874 |
+
"learning_rate": 2.636419280466831e-05,
|
3875 |
+
"loss": 0.1784,
|
3876 |
+
"step": 549
|
3877 |
+
},
|
3878 |
+
{
|
3879 |
+
"epoch": 0.7681564245810056,
|
3880 |
+
"grad_norm": 0.9998311400413513,
|
3881 |
+
"learning_rate": 2.6063849663053475e-05,
|
3882 |
+
"loss": 0.1589,
|
3883 |
+
"step": 550
|
3884 |
+
},
|
3885 |
+
{
|
3886 |
+
"epoch": 0.7695530726256983,
|
3887 |
+
"grad_norm": 0.2502938210964203,
|
3888 |
+
"learning_rate": 2.5764970539588674e-05,
|
3889 |
+
"loss": 0.0333,
|
3890 |
+
"step": 551
|
3891 |
+
},
|
3892 |
+
{
|
3893 |
+
"epoch": 0.770949720670391,
|
3894 |
+
"grad_norm": 0.3657321631908417,
|
3895 |
+
"learning_rate": 2.5467561352413648e-05,
|
3896 |
+
"loss": 0.0664,
|
3897 |
+
"step": 552
|
3898 |
+
},
|
3899 |
+
{
|
3900 |
+
"epoch": 0.7723463687150838,
|
3901 |
+
"grad_norm": 0.36316537857055664,
|
3902 |
+
"learning_rate": 2.5171627990561564e-05,
|
3903 |
+
"loss": 0.1044,
|
3904 |
+
"step": 553
|
3905 |
+
},
|
3906 |
+
{
|
3907 |
+
"epoch": 0.7737430167597765,
|
3908 |
+
"grad_norm": 0.3267860412597656,
|
3909 |
+
"learning_rate": 2.4877176313842753e-05,
|
3910 |
+
"loss": 0.0955,
|
3911 |
+
"step": 554
|
3912 |
+
},
|
3913 |
+
{
|
3914 |
+
"epoch": 0.7751396648044693,
|
3915 |
+
"grad_norm": 0.5272498726844788,
|
3916 |
+
"learning_rate": 2.4584212152728403e-05,
|
3917 |
+
"loss": 0.1501,
|
3918 |
+
"step": 555
|
3919 |
+
},
|
3920 |
+
{
|
3921 |
+
"epoch": 0.776536312849162,
|
3922 |
+
"grad_norm": 0.4217236340045929,
|
3923 |
+
"learning_rate": 2.4292741308235345e-05,
|
3924 |
+
"loss": 0.1161,
|
3925 |
+
"step": 556
|
3926 |
+
},
|
3927 |
+
{
|
3928 |
+
"epoch": 0.7779329608938548,
|
3929 |
+
"grad_norm": 0.29607003927230835,
|
3930 |
+
"learning_rate": 2.4002769551811033e-05,
|
3931 |
+
"loss": 0.0847,
|
3932 |
+
"step": 557
|
3933 |
+
},
|
3934 |
+
{
|
3935 |
+
"epoch": 0.7793296089385475,
|
3936 |
+
"grad_norm": 0.4450053870677948,
|
3937 |
+
"learning_rate": 2.3714302625219243e-05,
|
3938 |
+
"loss": 0.1128,
|
3939 |
+
"step": 558
|
3940 |
+
},
|
3941 |
+
{
|
3942 |
+
"epoch": 0.7807262569832403,
|
3943 |
+
"grad_norm": 0.8715819120407104,
|
3944 |
+
"learning_rate": 2.3427346240426617e-05,
|
3945 |
+
"loss": 0.1642,
|
3946 |
+
"step": 559
|
3947 |
+
},
|
3948 |
+
{
|
3949 |
+
"epoch": 0.7821229050279329,
|
3950 |
+
"grad_norm": 0.3655119836330414,
|
3951 |
+
"learning_rate": 2.3141906079489183e-05,
|
3952 |
+
"loss": 0.1071,
|
3953 |
+
"step": 560
|
3954 |
+
},
|
3955 |
+
{
|
3956 |
+
"epoch": 0.7835195530726257,
|
3957 |
+
"grad_norm": 0.3583777844905853,
|
3958 |
+
"learning_rate": 2.2857987794440205e-05,
|
3959 |
+
"loss": 0.0747,
|
3960 |
+
"step": 561
|
3961 |
+
},
|
3962 |
+
{
|
3963 |
+
"epoch": 0.7849162011173184,
|
3964 |
+
"grad_norm": 0.34693679213523865,
|
3965 |
+
"learning_rate": 2.2575597007177984e-05,
|
3966 |
+
"loss": 0.0968,
|
3967 |
+
"step": 562
|
3968 |
+
},
|
3969 |
+
{
|
3970 |
+
"epoch": 0.7863128491620112,
|
3971 |
+
"grad_norm": 0.29578697681427,
|
3972 |
+
"learning_rate": 2.229473930935475e-05,
|
3973 |
+
"loss": 0.1034,
|
3974 |
+
"step": 563
|
3975 |
+
},
|
3976 |
+
{
|
3977 |
+
"epoch": 0.7877094972067039,
|
3978 |
+
"grad_norm": 0.45241570472717285,
|
3979 |
+
"learning_rate": 2.2015420262265863e-05,
|
3980 |
+
"loss": 0.1307,
|
3981 |
+
"step": 564
|
3982 |
+
},
|
3983 |
+
{
|
3984 |
+
"epoch": 0.7891061452513967,
|
3985 |
+
"grad_norm": 0.6317943334579468,
|
3986 |
+
"learning_rate": 2.173764539673957e-05,
|
3987 |
+
"loss": 0.1464,
|
3988 |
+
"step": 565
|
3989 |
+
},
|
3990 |
+
{
|
3991 |
+
"epoch": 0.7905027932960894,
|
3992 |
+
"grad_norm": 0.25777265429496765,
|
3993 |
+
"learning_rate": 2.1461420213027772e-05,
|
3994 |
+
"loss": 0.0866,
|
3995 |
+
"step": 566
|
3996 |
+
},
|
3997 |
+
{
|
3998 |
+
"epoch": 0.7918994413407822,
|
3999 |
+
"grad_norm": 0.38986483216285706,
|
4000 |
+
"learning_rate": 2.118675018069679e-05,
|
4001 |
+
"loss": 0.0812,
|
4002 |
+
"step": 567
|
4003 |
+
},
|
4004 |
+
{
|
4005 |
+
"epoch": 0.7932960893854749,
|
4006 |
+
"grad_norm": 0.3201923370361328,
|
4007 |
+
"learning_rate": 2.0913640738519335e-05,
|
4008 |
+
"loss": 0.1062,
|
4009 |
+
"step": 568
|
4010 |
+
},
|
4011 |
+
{
|
4012 |
+
"epoch": 0.7946927374301676,
|
4013 |
+
"grad_norm": 0.34531423449516296,
|
4014 |
+
"learning_rate": 2.0642097294366557e-05,
|
4015 |
+
"loss": 0.0983,
|
4016 |
+
"step": 569
|
4017 |
+
},
|
4018 |
+
{
|
4019 |
+
"epoch": 0.7960893854748603,
|
4020 |
+
"grad_norm": 0.33722755312919617,
|
4021 |
+
"learning_rate": 2.0372125225101234e-05,
|
4022 |
+
"loss": 0.1119,
|
4023 |
+
"step": 570
|
4024 |
+
},
|
4025 |
+
{
|
4026 |
+
"epoch": 0.797486033519553,
|
4027 |
+
"grad_norm": 0.3482882082462311,
|
4028 |
+
"learning_rate": 2.0103729876471145e-05,
|
4029 |
+
"loss": 0.102,
|
4030 |
+
"step": 571
|
4031 |
+
},
|
4032 |
+
{
|
4033 |
+
"epoch": 0.7988826815642458,
|
4034 |
+
"grad_norm": 0.38442498445510864,
|
4035 |
+
"learning_rate": 1.983691656300314e-05,
|
4036 |
+
"loss": 0.1045,
|
4037 |
+
"step": 572
|
4038 |
+
},
|
4039 |
+
{
|
4040 |
+
"epoch": 0.8002793296089385,
|
4041 |
+
"grad_norm": 0.32199081778526306,
|
4042 |
+
"learning_rate": 1.957169056789814e-05,
|
4043 |
+
"loss": 0.1116,
|
4044 |
+
"step": 573
|
4045 |
+
},
|
4046 |
+
{
|
4047 |
+
"epoch": 0.8016759776536313,
|
4048 |
+
"grad_norm": 0.7939237952232361,
|
4049 |
+
"learning_rate": 1.930805714292634e-05,
|
4050 |
+
"loss": 0.1522,
|
4051 |
+
"step": 574
|
4052 |
+
},
|
4053 |
+
{
|
4054 |
+
"epoch": 0.803072625698324,
|
4055 |
+
"grad_norm": 0.32797643542289734,
|
4056 |
+
"learning_rate": 1.9046021508323243e-05,
|
4057 |
+
"loss": 0.0768,
|
4058 |
+
"step": 575
|
4059 |
+
},
|
4060 |
+
{
|
4061 |
+
"epoch": 0.8044692737430168,
|
4062 |
+
"grad_norm": 0.35652273893356323,
|
4063 |
+
"learning_rate": 1.8785588852686376e-05,
|
4064 |
+
"loss": 0.1023,
|
4065 |
+
"step": 576
|
4066 |
+
},
|
4067 |
+
{
|
4068 |
+
"epoch": 0.8058659217877095,
|
4069 |
+
"grad_norm": 0.424141526222229,
|
4070 |
+
"learning_rate": 1.8526764332872447e-05,
|
4071 |
+
"loss": 0.1488,
|
4072 |
+
"step": 577
|
4073 |
+
},
|
4074 |
+
{
|
4075 |
+
"epoch": 0.8072625698324022,
|
4076 |
+
"grad_norm": 0.35929641127586365,
|
4077 |
+
"learning_rate": 1.8269553073895375e-05,
|
4078 |
+
"loss": 0.1021,
|
4079 |
+
"step": 578
|
4080 |
+
},
|
4081 |
+
{
|
4082 |
+
"epoch": 0.8086592178770949,
|
4083 |
+
"grad_norm": 0.4345034062862396,
|
4084 |
+
"learning_rate": 1.801396016882456e-05,
|
4085 |
+
"loss": 0.1036,
|
4086 |
+
"step": 579
|
4087 |
+
},
|
4088 |
+
{
|
4089 |
+
"epoch": 0.8100558659217877,
|
4090 |
+
"grad_norm": 0.34810173511505127,
|
4091 |
+
"learning_rate": 1.7759990678684335e-05,
|
4092 |
+
"loss": 0.0864,
|
4093 |
+
"step": 580
|
4094 |
+
},
|
4095 |
+
{
|
4096 |
+
"epoch": 0.8114525139664804,
|
4097 |
+
"grad_norm": 0.40283820033073425,
|
4098 |
+
"learning_rate": 1.7507649632353574e-05,
|
4099 |
+
"loss": 0.1067,
|
4100 |
+
"step": 581
|
4101 |
+
},
|
4102 |
+
{
|
4103 |
+
"epoch": 0.8128491620111732,
|
4104 |
+
"grad_norm": 0.38133350014686584,
|
4105 |
+
"learning_rate": 1.7256942026466072e-05,
|
4106 |
+
"loss": 0.0923,
|
4107 |
+
"step": 582
|
4108 |
+
},
|
4109 |
+
{
|
4110 |
+
"epoch": 0.8142458100558659,
|
4111 |
+
"grad_norm": 0.359263151884079,
|
4112 |
+
"learning_rate": 1.70078728253118e-05,
|
4113 |
+
"loss": 0.102,
|
4114 |
+
"step": 583
|
4115 |
+
},
|
4116 |
+
{
|
4117 |
+
"epoch": 0.8156424581005587,
|
4118 |
+
"grad_norm": 0.43518075346946716,
|
4119 |
+
"learning_rate": 1.6760446960738364e-05,
|
4120 |
+
"loss": 0.1261,
|
4121 |
+
"step": 584
|
4122 |
+
},
|
4123 |
+
{
|
4124 |
+
"epoch": 0.8170391061452514,
|
4125 |
+
"grad_norm": 0.6720304489135742,
|
4126 |
+
"learning_rate": 1.6514669332053634e-05,
|
4127 |
+
"loss": 0.1277,
|
4128 |
+
"step": 585
|
4129 |
+
},
|
4130 |
+
{
|
4131 |
+
"epoch": 0.8184357541899442,
|
4132 |
+
"grad_norm": 0.45705312490463257,
|
4133 |
+
"learning_rate": 1.6270544805928424e-05,
|
4134 |
+
"loss": 0.0853,
|
4135 |
+
"step": 586
|
4136 |
+
},
|
4137 |
+
{
|
4138 |
+
"epoch": 0.8198324022346368,
|
4139 |
+
"grad_norm": 0.5168368220329285,
|
4140 |
+
"learning_rate": 1.6028078216300336e-05,
|
4141 |
+
"loss": 0.1083,
|
4142 |
+
"step": 587
|
4143 |
+
},
|
4144 |
+
{
|
4145 |
+
"epoch": 0.8212290502793296,
|
4146 |
+
"grad_norm": 0.6351551413536072,
|
4147 |
+
"learning_rate": 1.5787274364278004e-05,
|
4148 |
+
"loss": 0.1065,
|
4149 |
+
"step": 588
|
4150 |
+
},
|
4151 |
+
{
|
4152 |
+
"epoch": 0.8226256983240223,
|
4153 |
+
"grad_norm": 0.7593076229095459,
|
4154 |
+
"learning_rate": 1.5548138018045964e-05,
|
4155 |
+
"loss": 0.1282,
|
4156 |
+
"step": 589
|
4157 |
+
},
|
4158 |
+
{
|
4159 |
+
"epoch": 0.8240223463687151,
|
4160 |
+
"grad_norm": 0.6074228286743164,
|
4161 |
+
"learning_rate": 1.5310673912770312e-05,
|
4162 |
+
"loss": 0.1313,
|
4163 |
+
"step": 590
|
4164 |
+
},
|
4165 |
+
{
|
4166 |
+
"epoch": 0.8254189944134078,
|
4167 |
+
"grad_norm": 0.4425796866416931,
|
4168 |
+
"learning_rate": 1.5074886750504846e-05,
|
4169 |
+
"loss": 0.0936,
|
4170 |
+
"step": 591
|
4171 |
+
},
|
4172 |
+
{
|
4173 |
+
"epoch": 0.8268156424581006,
|
4174 |
+
"grad_norm": 0.8903464674949646,
|
4175 |
+
"learning_rate": 1.4840781200098152e-05,
|
4176 |
+
"loss": 0.1461,
|
4177 |
+
"step": 592
|
4178 |
+
},
|
4179 |
+
{
|
4180 |
+
"epoch": 0.8282122905027933,
|
4181 |
+
"grad_norm": 0.4715719223022461,
|
4182 |
+
"learning_rate": 1.4608361897100908e-05,
|
4183 |
+
"loss": 0.1171,
|
4184 |
+
"step": 593
|
4185 |
+
},
|
4186 |
+
{
|
4187 |
+
"epoch": 0.8296089385474861,
|
4188 |
+
"grad_norm": 0.5678585171699524,
|
4189 |
+
"learning_rate": 1.4377633443674233e-05,
|
4190 |
+
"loss": 0.0987,
|
4191 |
+
"step": 594
|
4192 |
+
},
|
4193 |
+
{
|
4194 |
+
"epoch": 0.8310055865921788,
|
4195 |
+
"grad_norm": 0.6416317820549011,
|
4196 |
+
"learning_rate": 1.4148600408498592e-05,
|
4197 |
+
"loss": 0.1381,
|
4198 |
+
"step": 595
|
4199 |
+
},
|
4200 |
+
{
|
4201 |
+
"epoch": 0.8324022346368715,
|
4202 |
+
"grad_norm": 0.5329287648200989,
|
4203 |
+
"learning_rate": 1.392126732668323e-05,
|
4204 |
+
"loss": 0.1136,
|
4205 |
+
"step": 596
|
4206 |
+
},
|
4207 |
+
{
|
4208 |
+
"epoch": 0.8337988826815642,
|
4209 |
+
"grad_norm": 0.6972792744636536,
|
4210 |
+
"learning_rate": 1.3695638699676494e-05,
|
4211 |
+
"loss": 0.1039,
|
4212 |
+
"step": 597
|
4213 |
+
},
|
4214 |
+
{
|
4215 |
+
"epoch": 0.835195530726257,
|
4216 |
+
"grad_norm": 0.7396121025085449,
|
4217 |
+
"learning_rate": 1.3471718995176507e-05,
|
4218 |
+
"loss": 0.1172,
|
4219 |
+
"step": 598
|
4220 |
+
},
|
4221 |
+
{
|
4222 |
+
"epoch": 0.8365921787709497,
|
4223 |
+
"grad_norm": 0.5911096930503845,
|
4224 |
+
"learning_rate": 1.3249512647042917e-05,
|
4225 |
+
"loss": 0.1097,
|
4226 |
+
"step": 599
|
4227 |
+
},
|
4228 |
+
{
|
4229 |
+
"epoch": 0.8379888268156425,
|
4230 |
+
"grad_norm": 0.8591598868370056,
|
4231 |
+
"learning_rate": 1.3029024055209015e-05,
|
4232 |
+
"loss": 0.1149,
|
4233 |
+
"step": 600
|
4234 |
+
},
|
4235 |
+
{
|
4236 |
+
"epoch": 0.8393854748603352,
|
4237 |
+
"grad_norm": 0.2031138688325882,
|
4238 |
+
"learning_rate": 1.281025758559451e-05,
|
4239 |
+
"loss": 0.0357,
|
4240 |
+
"step": 601
|
4241 |
+
},
|
4242 |
+
{
|
4243 |
+
"epoch": 0.840782122905028,
|
4244 |
+
"grad_norm": 0.41612380743026733,
|
4245 |
+
"learning_rate": 1.2593217570019267e-05,
|
4246 |
+
"loss": 0.088,
|
4247 |
+
"step": 602
|
4248 |
+
},
|
4249 |
+
{
|
4250 |
+
"epoch": 0.8421787709497207,
|
4251 |
+
"grad_norm": 0.544515073299408,
|
4252 |
+
"learning_rate": 1.2377908306117391e-05,
|
4253 |
+
"loss": 0.1182,
|
4254 |
+
"step": 603
|
4255 |
+
},
|
4256 |
+
{
|
4257 |
+
"epoch": 0.8435754189944135,
|
4258 |
+
"grad_norm": 0.39283642172813416,
|
4259 |
+
"learning_rate": 1.2164334057252203e-05,
|
4260 |
+
"loss": 0.1298,
|
4261 |
+
"step": 604
|
4262 |
+
},
|
4263 |
+
{
|
4264 |
+
"epoch": 0.8449720670391061,
|
4265 |
+
"grad_norm": 0.29914194345474243,
|
4266 |
+
"learning_rate": 1.1952499052431753e-05,
|
4267 |
+
"loss": 0.1081,
|
4268 |
+
"step": 605
|
4269 |
+
},
|
4270 |
+
{
|
4271 |
+
"epoch": 0.8463687150837989,
|
4272 |
+
"grad_norm": 0.32783523201942444,
|
4273 |
+
"learning_rate": 1.174240748622516e-05,
|
4274 |
+
"loss": 0.0814,
|
4275 |
+
"step": 606
|
4276 |
+
},
|
4277 |
+
{
|
4278 |
+
"epoch": 0.8477653631284916,
|
4279 |
+
"grad_norm": 0.3241506516933441,
|
4280 |
+
"learning_rate": 1.1534063518679516e-05,
|
4281 |
+
"loss": 0.1111,
|
4282 |
+
"step": 607
|
4283 |
+
},
|
4284 |
+
{
|
4285 |
+
"epoch": 0.8491620111731844,
|
4286 |
+
"grad_norm": 0.41864722967147827,
|
4287 |
+
"learning_rate": 1.1327471275237456e-05,
|
4288 |
+
"loss": 0.1169,
|
4289 |
+
"step": 608
|
4290 |
+
},
|
4291 |
+
{
|
4292 |
+
"epoch": 0.8505586592178771,
|
4293 |
+
"grad_norm": 0.31106308102607727,
|
4294 |
+
"learning_rate": 1.11226348466556e-05,
|
4295 |
+
"loss": 0.0782,
|
4296 |
+
"step": 609
|
4297 |
+
},
|
4298 |
+
{
|
4299 |
+
"epoch": 0.8519553072625698,
|
4300 |
+
"grad_norm": 0.3926340341567993,
|
4301 |
+
"learning_rate": 1.0919558288923426e-05,
|
4302 |
+
"loss": 0.1118,
|
4303 |
+
"step": 610
|
4304 |
+
},
|
4305 |
+
{
|
4306 |
+
"epoch": 0.8533519553072626,
|
4307 |
+
"grad_norm": 0.3361690640449524,
|
4308 |
+
"learning_rate": 1.0718245623183066e-05,
|
4309 |
+
"loss": 0.1096,
|
4310 |
+
"step": 611
|
4311 |
+
},
|
4312 |
+
{
|
4313 |
+
"epoch": 0.8547486033519553,
|
4314 |
+
"grad_norm": 0.3871351480484009,
|
4315 |
+
"learning_rate": 1.0518700835649553e-05,
|
4316 |
+
"loss": 0.1092,
|
4317 |
+
"step": 612
|
4318 |
+
},
|
4319 |
+
{
|
4320 |
+
"epoch": 0.8561452513966481,
|
4321 |
+
"grad_norm": 0.24205343425273895,
|
4322 |
+
"learning_rate": 1.0320927877531971e-05,
|
4323 |
+
"loss": 0.0836,
|
4324 |
+
"step": 613
|
4325 |
+
},
|
4326 |
+
{
|
4327 |
+
"epoch": 0.8575418994413407,
|
4328 |
+
"grad_norm": 0.31774112582206726,
|
4329 |
+
"learning_rate": 1.0124930664955301e-05,
|
4330 |
+
"loss": 0.098,
|
4331 |
+
"step": 614
|
4332 |
+
},
|
4333 |
+
{
|
4334 |
+
"epoch": 0.8589385474860335,
|
4335 |
+
"grad_norm": 0.34544655680656433,
|
4336 |
+
"learning_rate": 9.930713078882659e-06,
|
4337 |
+
"loss": 0.1347,
|
4338 |
+
"step": 615
|
4339 |
+
},
|
4340 |
+
{
|
4341 |
+
"epoch": 0.8603351955307262,
|
4342 |
+
"grad_norm": 0.24583694338798523,
|
4343 |
+
"learning_rate": 9.73827896503865e-06,
|
4344 |
+
"loss": 0.0752,
|
4345 |
+
"step": 616
|
4346 |
+
},
|
4347 |
+
{
|
4348 |
+
"epoch": 0.861731843575419,
|
4349 |
+
"grad_norm": 0.3795362114906311,
|
4350 |
+
"learning_rate": 9.54763213383314e-06,
|
4351 |
+
"loss": 0.1214,
|
4352 |
+
"step": 617
|
4353 |
+
},
|
4354 |
+
{
|
4355 |
+
"epoch": 0.8631284916201117,
|
4356 |
+
"grad_norm": 0.3082280158996582,
|
4357 |
+
"learning_rate": 9.358776360285759e-06,
|
4358 |
+
"loss": 0.1001,
|
4359 |
+
"step": 618
|
4360 |
+
},
|
4361 |
+
{
|
4362 |
+
"epoch": 0.8645251396648045,
|
4363 |
+
"grad_norm": 0.4244152307510376,
|
4364 |
+
"learning_rate": 9.171715383951251e-06,
|
4365 |
+
"loss": 0.1184,
|
4366 |
+
"step": 619
|
4367 |
+
},
|
4368 |
+
{
|
4369 |
+
"epoch": 0.8659217877094972,
|
4370 |
+
"grad_norm": 0.315733402967453,
|
4371 |
+
"learning_rate": 8.986452908845322e-06,
|
4372 |
+
"loss": 0.0912,
|
4373 |
+
"step": 620
|
4374 |
+
},
|
4375 |
+
{
|
4376 |
+
"epoch": 0.86731843575419,
|
4377 |
+
"grad_norm": 0.2573086619377136,
|
4378 |
+
"learning_rate": 8.80299260337144e-06,
|
4379 |
+
"loss": 0.0754,
|
4380 |
+
"step": 621
|
4381 |
+
},
|
4382 |
+
{
|
4383 |
+
"epoch": 0.8687150837988827,
|
4384 |
+
"grad_norm": 0.3076595067977905,
|
4385 |
+
"learning_rate": 8.621338100247988e-06,
|
4386 |
+
"loss": 0.122,
|
4387 |
+
"step": 622
|
4388 |
+
},
|
4389 |
+
{
|
4390 |
+
"epoch": 0.8701117318435754,
|
4391 |
+
"grad_norm": 0.3168458044528961,
|
4392 |
+
"learning_rate": 8.441492996436573e-06,
|
4393 |
+
"loss": 0.0862,
|
4394 |
+
"step": 623
|
4395 |
+
},
|
4396 |
+
{
|
4397 |
+
"epoch": 0.8715083798882681,
|
4398 |
+
"grad_norm": 0.292837917804718,
|
4399 |
+
"learning_rate": 8.26346085307057e-06,
|
4400 |
+
"loss": 0.122,
|
4401 |
+
"step": 624
|
4402 |
+
},
|
4403 |
+
{
|
4404 |
+
"epoch": 0.8729050279329609,
|
4405 |
+
"grad_norm": 0.3604884445667267,
|
4406 |
+
"learning_rate": 8.087245195384774e-06,
|
4407 |
+
"loss": 0.1143,
|
4408 |
+
"step": 625
|
4409 |
+
},
|
4410 |
+
{
|
4411 |
+
"epoch": 0.8743016759776536,
|
4412 |
+
"grad_norm": 0.3604539632797241,
|
4413 |
+
"learning_rate": 7.91284951264557e-06,
|
4414 |
+
"loss": 0.1162,
|
4415 |
+
"step": 626
|
4416 |
+
},
|
4417 |
+
{
|
4418 |
+
"epoch": 0.8756983240223464,
|
4419 |
+
"grad_norm": 0.33363595604896545,
|
4420 |
+
"learning_rate": 7.740277258081696e-06,
|
4421 |
+
"loss": 0.0852,
|
4422 |
+
"step": 627
|
4423 |
+
},
|
4424 |
+
{
|
4425 |
+
"epoch": 0.8770949720670391,
|
4426 |
+
"grad_norm": 0.6355361342430115,
|
4427 |
+
"learning_rate": 7.569531848816147e-06,
|
4428 |
+
"loss": 0.1154,
|
4429 |
+
"step": 628
|
4430 |
+
},
|
4431 |
+
{
|
4432 |
+
"epoch": 0.8784916201117319,
|
4433 |
+
"grad_norm": 0.3323967754840851,
|
4434 |
+
"learning_rate": 7.400616665798199e-06,
|
4435 |
+
"loss": 0.1003,
|
4436 |
+
"step": 629
|
4437 |
+
},
|
4438 |
+
{
|
4439 |
+
"epoch": 0.8798882681564246,
|
4440 |
+
"grad_norm": 0.516983151435852,
|
4441 |
+
"learning_rate": 7.233535053736706e-06,
|
4442 |
+
"loss": 0.0863,
|
4443 |
+
"step": 630
|
4444 |
+
},
|
4445 |
+
{
|
4446 |
+
"epoch": 0.8812849162011173,
|
4447 |
+
"grad_norm": 0.45355257391929626,
|
4448 |
+
"learning_rate": 7.068290321033688e-06,
|
4449 |
+
"loss": 0.1229,
|
4450 |
+
"step": 631
|
4451 |
+
},
|
4452 |
+
{
|
4453 |
+
"epoch": 0.88268156424581,
|
4454 |
+
"grad_norm": 0.37939712405204773,
|
4455 |
+
"learning_rate": 6.90488573971898e-06,
|
4456 |
+
"loss": 0.1081,
|
4457 |
+
"step": 632
|
4458 |
+
},
|
4459 |
+
{
|
4460 |
+
"epoch": 0.8840782122905028,
|
4461 |
+
"grad_norm": 0.45642518997192383,
|
4462 |
+
"learning_rate": 6.743324545385354e-06,
|
4463 |
+
"loss": 0.091,
|
4464 |
+
"step": 633
|
4465 |
+
},
|
4466 |
+
{
|
4467 |
+
"epoch": 0.8854748603351955,
|
4468 |
+
"grad_norm": 0.365570068359375,
|
4469 |
+
"learning_rate": 6.583609937124435e-06,
|
4470 |
+
"loss": 0.1067,
|
4471 |
+
"step": 634
|
4472 |
+
},
|
4473 |
+
{
|
4474 |
+
"epoch": 0.8868715083798883,
|
4475 |
+
"grad_norm": 0.38732656836509705,
|
4476 |
+
"learning_rate": 6.425745077463408e-06,
|
4477 |
+
"loss": 0.105,
|
4478 |
+
"step": 635
|
4479 |
+
},
|
4480 |
+
{
|
4481 |
+
"epoch": 0.888268156424581,
|
4482 |
+
"grad_norm": 2.0591869354248047,
|
4483 |
+
"learning_rate": 6.269733092302399e-06,
|
4484 |
+
"loss": 0.1407,
|
4485 |
+
"step": 636
|
4486 |
+
},
|
4487 |
+
{
|
4488 |
+
"epoch": 0.8896648044692738,
|
4489 |
+
"grad_norm": 0.48740479350090027,
|
4490 |
+
"learning_rate": 6.115577070852507e-06,
|
4491 |
+
"loss": 0.1058,
|
4492 |
+
"step": 637
|
4493 |
+
},
|
4494 |
+
{
|
4495 |
+
"epoch": 0.8910614525139665,
|
4496 |
+
"grad_norm": 3.680542469024658,
|
4497 |
+
"learning_rate": 5.963280065574694e-06,
|
4498 |
+
"loss": 0.1259,
|
4499 |
+
"step": 638
|
4500 |
+
},
|
4501 |
+
{
|
4502 |
+
"epoch": 0.8924581005586593,
|
4503 |
+
"grad_norm": 0.5749529004096985,
|
4504 |
+
"learning_rate": 5.8128450921193674e-06,
|
4505 |
+
"loss": 0.1333,
|
4506 |
+
"step": 639
|
4507 |
+
},
|
4508 |
+
{
|
4509 |
+
"epoch": 0.8938547486033519,
|
4510 |
+
"grad_norm": 0.3572239875793457,
|
4511 |
+
"learning_rate": 5.664275129266605e-06,
|
4512 |
+
"loss": 0.104,
|
4513 |
+
"step": 640
|
4514 |
+
},
|
4515 |
+
{
|
4516 |
+
"epoch": 0.8952513966480447,
|
4517 |
+
"grad_norm": 0.49053096771240234,
|
4518 |
+
"learning_rate": 5.51757311886717e-06,
|
4519 |
+
"loss": 0.1293,
|
4520 |
+
"step": 641
|
4521 |
+
},
|
4522 |
+
{
|
4523 |
+
"epoch": 0.8966480446927374,
|
4524 |
+
"grad_norm": 0.8410274386405945,
|
4525 |
+
"learning_rate": 5.372741965784323e-06,
|
4526 |
+
"loss": 0.1495,
|
4527 |
+
"step": 642
|
4528 |
+
},
|
4529 |
+
{
|
4530 |
+
"epoch": 0.8980446927374302,
|
4531 |
+
"grad_norm": 0.436309278011322,
|
4532 |
+
"learning_rate": 5.2297845378362795e-06,
|
4533 |
+
"loss": 0.1046,
|
4534 |
+
"step": 643
|
4535 |
+
},
|
4536 |
+
{
|
4537 |
+
"epoch": 0.8994413407821229,
|
4538 |
+
"grad_norm": 0.6086000800132751,
|
4539 |
+
"learning_rate": 5.088703665739336e-06,
|
4540 |
+
"loss": 0.1314,
|
4541 |
+
"step": 644
|
4542 |
+
},
|
4543 |
+
{
|
4544 |
+
"epoch": 0.9008379888268156,
|
4545 |
+
"grad_norm": 0.5451857447624207,
|
4546 |
+
"learning_rate": 4.949502143051976e-06,
|
4547 |
+
"loss": 0.1268,
|
4548 |
+
"step": 645
|
4549 |
+
},
|
4550 |
+
{
|
4551 |
+
"epoch": 0.9022346368715084,
|
4552 |
+
"grad_norm": 0.55224609375,
|
4553 |
+
"learning_rate": 4.812182726119397e-06,
|
4554 |
+
"loss": 0.074,
|
4555 |
+
"step": 646
|
4556 |
+
},
|
4557 |
+
{
|
4558 |
+
"epoch": 0.9036312849162011,
|
4559 |
+
"grad_norm": 0.5905503630638123,
|
4560 |
+
"learning_rate": 4.676748134019105e-06,
|
4561 |
+
"loss": 0.1371,
|
4562 |
+
"step": 647
|
4563 |
+
},
|
4564 |
+
{
|
4565 |
+
"epoch": 0.9050279329608939,
|
4566 |
+
"grad_norm": 0.3724439740180969,
|
4567 |
+
"learning_rate": 4.543201048506851e-06,
|
4568 |
+
"loss": 0.0814,
|
4569 |
+
"step": 648
|
4570 |
+
},
|
4571 |
+
{
|
4572 |
+
"epoch": 0.9064245810055865,
|
4573 |
+
"grad_norm": 0.5553332567214966,
|
4574 |
+
"learning_rate": 4.41154411396375e-06,
|
4575 |
+
"loss": 0.1375,
|
4576 |
+
"step": 649
|
4577 |
+
},
|
4578 |
+
{
|
4579 |
+
"epoch": 0.9078212290502793,
|
4580 |
+
"grad_norm": 0.5004715919494629,
|
4581 |
+
"learning_rate": 4.2817799373437994e-06,
|
4582 |
+
"loss": 0.0615,
|
4583 |
+
"step": 650
|
4584 |
+
},
|
4585 |
+
{
|
4586 |
+
"epoch": 0.909217877094972,
|
4587 |
+
"grad_norm": 0.22518612444400787,
|
4588 |
+
"learning_rate": 4.153911088122231e-06,
|
4589 |
+
"loss": 0.0299,
|
4590 |
+
"step": 651
|
4591 |
+
},
|
4592 |
+
{
|
4593 |
+
"epoch": 0.9106145251396648,
|
4594 |
+
"grad_norm": 0.3304058313369751,
|
4595 |
+
"learning_rate": 4.027940098244753e-06,
|
4596 |
+
"loss": 0.0849,
|
4597 |
+
"step": 652
|
4598 |
+
},
|
4599 |
+
{
|
4600 |
+
"epoch": 0.9120111731843575,
|
4601 |
+
"grad_norm": 0.413650780916214,
|
4602 |
+
"learning_rate": 3.90386946207727e-06,
|
4603 |
+
"loss": 0.1193,
|
4604 |
+
"step": 653
|
4605 |
+
},
|
4606 |
+
{
|
4607 |
+
"epoch": 0.9134078212290503,
|
4608 |
+
"grad_norm": 0.3433137834072113,
|
4609 |
+
"learning_rate": 3.7817016363566493e-06,
|
4610 |
+
"loss": 0.101,
|
4611 |
+
"step": 654
|
4612 |
+
},
|
4613 |
+
{
|
4614 |
+
"epoch": 0.914804469273743,
|
4615 |
+
"grad_norm": 0.30257925391197205,
|
4616 |
+
"learning_rate": 3.6614390401419453e-06,
|
4617 |
+
"loss": 0.1137,
|
4618 |
+
"step": 655
|
4619 |
+
},
|
4620 |
+
{
|
4621 |
+
"epoch": 0.9162011173184358,
|
4622 |
+
"grad_norm": 0.23929886519908905,
|
4623 |
+
"learning_rate": 3.54308405476651e-06,
|
4624 |
+
"loss": 0.1059,
|
4625 |
+
"step": 656
|
4626 |
+
},
|
4627 |
+
{
|
4628 |
+
"epoch": 0.9175977653631285,
|
4629 |
+
"grad_norm": 0.33015841245651245,
|
4630 |
+
"learning_rate": 3.4266390237909676e-06,
|
4631 |
+
"loss": 0.1267,
|
4632 |
+
"step": 657
|
4633 |
+
},
|
4634 |
+
{
|
4635 |
+
"epoch": 0.9189944134078212,
|
4636 |
+
"grad_norm": 0.3994545638561249,
|
4637 |
+
"learning_rate": 3.312106252956626e-06,
|
4638 |
+
"loss": 0.1451,
|
4639 |
+
"step": 658
|
4640 |
+
},
|
4641 |
+
{
|
4642 |
+
"epoch": 0.9203910614525139,
|
4643 |
+
"grad_norm": 0.27575787901878357,
|
4644 |
+
"learning_rate": 3.1994880101399726e-06,
|
4645 |
+
"loss": 0.088,
|
4646 |
+
"step": 659
|
4647 |
+
},
|
4648 |
+
{
|
4649 |
+
"epoch": 0.9217877094972067,
|
4650 |
+
"grad_norm": 0.37436044216156006,
|
4651 |
+
"learning_rate": 3.0887865253076632e-06,
|
4652 |
+
"loss": 0.1151,
|
4653 |
+
"step": 660
|
4654 |
+
},
|
4655 |
+
{
|
4656 |
+
"epoch": 0.9231843575418994,
|
4657 |
+
"grad_norm": 0.27707546949386597,
|
4658 |
+
"learning_rate": 2.9800039904724463e-06,
|
4659 |
+
"loss": 0.0835,
|
4660 |
+
"step": 661
|
4661 |
+
},
|
4662 |
+
{
|
4663 |
+
"epoch": 0.9245810055865922,
|
4664 |
+
"grad_norm": 0.4059697687625885,
|
4665 |
+
"learning_rate": 2.873142559649722e-06,
|
4666 |
+
"loss": 0.1027,
|
4667 |
+
"step": 662
|
4668 |
+
},
|
4669 |
+
{
|
4670 |
+
"epoch": 0.9259776536312849,
|
4671 |
+
"grad_norm": 0.22227145731449127,
|
4672 |
+
"learning_rate": 2.7682043488148513e-06,
|
4673 |
+
"loss": 0.1016,
|
4674 |
+
"step": 663
|
4675 |
+
},
|
4676 |
+
{
|
4677 |
+
"epoch": 0.9273743016759777,
|
4678 |
+
"grad_norm": 0.38108816742897034,
|
4679 |
+
"learning_rate": 2.6651914358613252e-06,
|
4680 |
+
"loss": 0.1187,
|
4681 |
+
"step": 664
|
4682 |
+
},
|
4683 |
+
{
|
4684 |
+
"epoch": 0.9287709497206704,
|
4685 |
+
"grad_norm": 0.29107677936553955,
|
4686 |
+
"learning_rate": 2.564105860559607e-06,
|
4687 |
+
"loss": 0.0908,
|
4688 |
+
"step": 665
|
4689 |
+
},
|
4690 |
+
{
|
4691 |
+
"epoch": 0.9301675977653632,
|
4692 |
+
"grad_norm": 0.3450527787208557,
|
4693 |
+
"learning_rate": 2.464949624516688e-06,
|
4694 |
+
"loss": 0.1126,
|
4695 |
+
"step": 666
|
4696 |
+
},
|
4697 |
+
{
|
4698 |
+
"epoch": 0.9315642458100558,
|
4699 |
+
"grad_norm": 0.3469128906726837,
|
4700 |
+
"learning_rate": 2.3677246911365304e-06,
|
4701 |
+
"loss": 0.1213,
|
4702 |
+
"step": 667
|
4703 |
+
},
|
4704 |
+
{
|
4705 |
+
"epoch": 0.9329608938547486,
|
4706 |
+
"grad_norm": 0.4438486099243164,
|
4707 |
+
"learning_rate": 2.272432985581119e-06,
|
4708 |
+
"loss": 0.1135,
|
4709 |
+
"step": 668
|
4710 |
+
},
|
4711 |
+
{
|
4712 |
+
"epoch": 0.9343575418994413,
|
4713 |
+
"grad_norm": 0.28915104269981384,
|
4714 |
+
"learning_rate": 2.1790763947324046e-06,
|
4715 |
+
"loss": 0.1028,
|
4716 |
+
"step": 669
|
4717 |
+
},
|
4718 |
+
{
|
4719 |
+
"epoch": 0.9357541899441341,
|
4720 |
+
"grad_norm": 0.25452756881713867,
|
4721 |
+
"learning_rate": 2.0876567671548773e-06,
|
4722 |
+
"loss": 0.1019,
|
4723 |
+
"step": 670
|
4724 |
+
},
|
4725 |
+
{
|
4726 |
+
"epoch": 0.9371508379888268,
|
4727 |
+
"grad_norm": 0.3689977526664734,
|
4728 |
+
"learning_rate": 1.9981759130590305e-06,
|
4729 |
+
"loss": 0.0743,
|
4730 |
+
"step": 671
|
4731 |
+
},
|
4732 |
+
{
|
4733 |
+
"epoch": 0.9385474860335196,
|
4734 |
+
"grad_norm": 0.27568790316581726,
|
4735 |
+
"learning_rate": 1.910635604265465e-06,
|
4736 |
+
"loss": 0.0833,
|
4737 |
+
"step": 672
|
4738 |
+
},
|
4739 |
+
{
|
4740 |
+
"epoch": 0.9399441340782123,
|
4741 |
+
"grad_norm": 0.3405630588531494,
|
4742 |
+
"learning_rate": 1.82503757416983e-06,
|
4743 |
+
"loss": 0.092,
|
4744 |
+
"step": 673
|
4745 |
+
},
|
4746 |
+
{
|
4747 |
+
"epoch": 0.9413407821229051,
|
4748 |
+
"grad_norm": 0.28969278931617737,
|
4749 |
+
"learning_rate": 1.7413835177084835e-06,
|
4750 |
+
"loss": 0.084,
|
4751 |
+
"step": 674
|
4752 |
+
},
|
4753 |
+
{
|
4754 |
+
"epoch": 0.9427374301675978,
|
4755 |
+
"grad_norm": 0.38578617572784424,
|
4756 |
+
"learning_rate": 1.6596750913249304e-06,
|
4757 |
+
"loss": 0.108,
|
4758 |
+
"step": 675
|
4759 |
+
},
|
4760 |
+
{
|
4761 |
+
"epoch": 0.9441340782122905,
|
4762 |
+
"grad_norm": 0.3050103187561035,
|
4763 |
+
"learning_rate": 1.5799139129370588e-06,
|
4764 |
+
"loss": 0.0972,
|
4765 |
+
"step": 676
|
4766 |
+
},
|
4767 |
+
{
|
4768 |
+
"epoch": 0.9455307262569832,
|
4769 |
+
"grad_norm": 0.3887476325035095,
|
4770 |
+
"learning_rate": 1.502101561905067e-06,
|
4771 |
+
"loss": 0.0775,
|
4772 |
+
"step": 677
|
4773 |
+
},
|
4774 |
+
{
|
4775 |
+
"epoch": 0.946927374301676,
|
4776 |
+
"grad_norm": 0.39781996607780457,
|
4777 |
+
"learning_rate": 1.4262395790001881e-06,
|
4778 |
+
"loss": 0.0953,
|
4779 |
+
"step": 678
|
4780 |
+
},
|
4781 |
+
{
|
4782 |
+
"epoch": 0.9483240223463687,
|
4783 |
+
"grad_norm": 0.2917478382587433,
|
4784 |
+
"learning_rate": 1.3523294663742025e-06,
|
4785 |
+
"loss": 0.096,
|
4786 |
+
"step": 679
|
4787 |
+
},
|
4788 |
+
{
|
4789 |
+
"epoch": 0.9497206703910615,
|
4790 |
+
"grad_norm": 0.36107248067855835,
|
4791 |
+
"learning_rate": 1.2803726875296963e-06,
|
4792 |
+
"loss": 0.1049,
|
4793 |
+
"step": 680
|
4794 |
+
},
|
4795 |
+
{
|
4796 |
+
"epoch": 0.9511173184357542,
|
4797 |
+
"grad_norm": 0.49688655138015747,
|
4798 |
+
"learning_rate": 1.2103706672910497e-06,
|
4799 |
+
"loss": 0.1086,
|
4800 |
+
"step": 681
|
4801 |
+
},
|
4802 |
+
{
|
4803 |
+
"epoch": 0.952513966480447,
|
4804 |
+
"grad_norm": 0.39250099658966064,
|
4805 |
+
"learning_rate": 1.142324791776239e-06,
|
4806 |
+
"loss": 0.0881,
|
4807 |
+
"step": 682
|
4808 |
+
},
|
4809 |
+
{
|
4810 |
+
"epoch": 0.9539106145251397,
|
4811 |
+
"grad_norm": 0.41053643822669983,
|
4812 |
+
"learning_rate": 1.0762364083694464e-06,
|
4813 |
+
"loss": 0.098,
|
4814 |
+
"step": 683
|
4815 |
+
},
|
4816 |
+
{
|
4817 |
+
"epoch": 0.9553072625698324,
|
4818 |
+
"grad_norm": 0.34778672456741333,
|
4819 |
+
"learning_rate": 1.01210682569427e-06,
|
4820 |
+
"loss": 0.1198,
|
4821 |
+
"step": 684
|
4822 |
+
},
|
4823 |
+
{
|
4824 |
+
"epoch": 0.9567039106145251,
|
4825 |
+
"grad_norm": 0.47372210025787354,
|
4826 |
+
"learning_rate": 9.499373135879008e-07,
|
4827 |
+
"loss": 0.1078,
|
4828 |
+
"step": 685
|
4829 |
+
},
|
4830 |
+
{
|
4831 |
+
"epoch": 0.9581005586592178,
|
4832 |
+
"grad_norm": 0.5244646668434143,
|
4833 |
+
"learning_rate": 8.897291030759314e-07,
|
4834 |
+
"loss": 0.1052,
|
4835 |
+
"step": 686
|
4836 |
+
},
|
4837 |
+
{
|
4838 |
+
"epoch": 0.9594972067039106,
|
4839 |
+
"grad_norm": 0.4699241816997528,
|
4840 |
+
"learning_rate": 8.314833863480198e-07,
|
4841 |
+
"loss": 0.1193,
|
4842 |
+
"step": 687
|
4843 |
+
},
|
4844 |
+
{
|
4845 |
+
"epoch": 0.9608938547486033,
|
4846 |
+
"grad_norm": 0.33639252185821533,
|
4847 |
+
"learning_rate": 7.752013167342531e-07,
|
4848 |
+
"loss": 0.0922,
|
4849 |
+
"step": 688
|
4850 |
+
},
|
4851 |
+
{
|
4852 |
+
"epoch": 0.9622905027932961,
|
4853 |
+
"grad_norm": 0.46195724606513977,
|
4854 |
+
"learning_rate": 7.208840086822988e-07,
|
4855 |
+
"loss": 0.1021,
|
4856 |
+
"step": 689
|
4857 |
+
},
|
4858 |
+
{
|
4859 |
+
"epoch": 0.9636871508379888,
|
4860 |
+
"grad_norm": 0.4881128668785095,
|
4861 |
+
"learning_rate": 6.68532537735389e-07,
|
4862 |
+
"loss": 0.1044,
|
4863 |
+
"step": 690
|
4864 |
+
},
|
4865 |
+
{
|
4866 |
+
"epoch": 0.9650837988826816,
|
4867 |
+
"grad_norm": 0.5272928476333618,
|
4868 |
+
"learning_rate": 6.181479405109602e-07,
|
4869 |
+
"loss": 0.1259,
|
4870 |
+
"step": 691
|
4871 |
+
},
|
4872 |
+
{
|
4873 |
+
"epoch": 0.9664804469273743,
|
4874 |
+
"grad_norm": 0.6659584641456604,
|
4875 |
+
"learning_rate": 5.697312146801915e-07,
|
4876 |
+
"loss": 0.1545,
|
4877 |
+
"step": 692
|
4878 |
+
},
|
4879 |
+
{
|
4880 |
+
"epoch": 0.9678770949720671,
|
4881 |
+
"grad_norm": 0.49742087721824646,
|
4882 |
+
"learning_rate": 5.232833189481645e-07,
|
4883 |
+
"loss": 0.0908,
|
4884 |
+
"step": 693
|
4885 |
+
},
|
4886 |
+
{
|
4887 |
+
"epoch": 0.9692737430167597,
|
4888 |
+
"grad_norm": 0.593102753162384,
|
4889 |
+
"learning_rate": 4.788051730349907e-07,
|
4890 |
+
"loss": 0.155,
|
4891 |
+
"step": 694
|
4892 |
+
},
|
4893 |
+
{
|
4894 |
+
"epoch": 0.9706703910614525,
|
4895 |
+
"grad_norm": 0.9965668320655823,
|
4896 |
+
"learning_rate": 4.3629765765751396e-07,
|
4897 |
+
"loss": 0.168,
|
4898 |
+
"step": 695
|
4899 |
+
},
|
4900 |
+
{
|
4901 |
+
"epoch": 0.9720670391061452,
|
4902 |
+
"grad_norm": 0.4818421006202698,
|
4903 |
+
"learning_rate": 3.9576161451186923e-07,
|
4904 |
+
"loss": 0.1031,
|
4905 |
+
"step": 696
|
4906 |
+
},
|
4907 |
+
{
|
4908 |
+
"epoch": 0.973463687150838,
|
4909 |
+
"grad_norm": 0.6472490429878235,
|
4910 |
+
"learning_rate": 3.571978462568959e-07,
|
4911 |
+
"loss": 0.1141,
|
4912 |
+
"step": 697
|
4913 |
+
},
|
4914 |
+
{
|
4915 |
+
"epoch": 0.9748603351955307,
|
4916 |
+
"grad_norm": 0.4191659688949585,
|
4917 |
+
"learning_rate": 3.2060711649817277e-07,
|
4918 |
+
"loss": 0.0783,
|
4919 |
+
"step": 698
|
4920 |
+
},
|
4921 |
+
{
|
4922 |
+
"epoch": 0.9762569832402235,
|
4923 |
+
"grad_norm": 0.5093485713005066,
|
4924 |
+
"learning_rate": 2.8599014977289675e-07,
|
4925 |
+
"loss": 0.0947,
|
4926 |
+
"step": 699
|
4927 |
+
},
|
4928 |
+
{
|
4929 |
+
"epoch": 0.9776536312849162,
|
4930 |
+
"grad_norm": 1.8124933242797852,
|
4931 |
+
"learning_rate": 2.5334763153559424e-07,
|
4932 |
+
"loss": 0.1765,
|
4933 |
+
"step": 700
|
4934 |
+
},
|
4935 |
+
{
|
4936 |
+
"epoch": 0.979050279329609,
|
4937 |
+
"grad_norm": 0.30719417333602905,
|
4938 |
+
"learning_rate": 2.2268020814447676e-07,
|
4939 |
+
"loss": 0.0636,
|
4940 |
+
"step": 701
|
4941 |
+
},
|
4942 |
+
{
|
4943 |
+
"epoch": 0.9804469273743017,
|
4944 |
+
"grad_norm": 1.8330652713775635,
|
4945 |
+
"learning_rate": 1.939884868487174e-07,
|
4946 |
+
"loss": 0.1424,
|
4947 |
+
"step": 702
|
4948 |
+
},
|
4949 |
+
{
|
4950 |
+
"epoch": 0.9818435754189944,
|
4951 |
+
"grad_norm": 0.6687912344932556,
|
4952 |
+
"learning_rate": 1.6727303577633858e-07,
|
4953 |
+
"loss": 0.1007,
|
4954 |
+
"step": 703
|
4955 |
+
},
|
4956 |
+
{
|
4957 |
+
"epoch": 0.9832402234636871,
|
4958 |
+
"grad_norm": 0.29703205823898315,
|
4959 |
+
"learning_rate": 1.4253438392304307e-07,
|
4960 |
+
"loss": 0.0762,
|
4961 |
+
"step": 704
|
4962 |
+
},
|
4963 |
+
{
|
4964 |
+
"epoch": 0.9846368715083799,
|
4965 |
+
"grad_norm": 0.30622580647468567,
|
4966 |
+
"learning_rate": 1.197730211417114e-07,
|
4967 |
+
"loss": 0.1054,
|
4968 |
+
"step": 705
|
4969 |
+
},
|
4970 |
+
{
|
4971 |
+
"epoch": 0.9860335195530726,
|
4972 |
+
"grad_norm": 0.2646753489971161,
|
4973 |
+
"learning_rate": 9.898939813267616e-08,
|
4974 |
+
"loss": 0.0877,
|
4975 |
+
"step": 706
|
4976 |
+
},
|
4977 |
+
{
|
4978 |
+
"epoch": 0.9874301675977654,
|
4979 |
+
"grad_norm": 0.36262211203575134,
|
4980 |
+
"learning_rate": 8.018392643484029e-08,
|
4981 |
+
"loss": 0.0927,
|
4982 |
+
"step": 707
|
4983 |
+
},
|
4984 |
+
{
|
4985 |
+
"epoch": 0.9888268156424581,
|
4986 |
+
"grad_norm": 0.24466976523399353,
|
4987 |
+
"learning_rate": 6.335697841747257e-08,
|
4988 |
+
"loss": 0.084,
|
4989 |
+
"step": 708
|
4990 |
+
},
|
4991 |
+
{
|
4992 |
+
"epoch": 0.9902234636871509,
|
4993 |
+
"grad_norm": 0.4240047335624695,
|
4994 |
+
"learning_rate": 4.850888727290226e-08,
|
4995 |
+
"loss": 0.106,
|
4996 |
+
"step": 709
|
4997 |
+
},
|
4998 |
+
{
|
4999 |
+
"epoch": 0.9916201117318436,
|
5000 |
+
"grad_norm": 0.47176429629325867,
|
5001 |
+
"learning_rate": 3.563994700988005e-08,
|
5002 |
+
"loss": 0.1121,
|
5003 |
+
"step": 710
|
5004 |
+
},
|
5005 |
+
{
|
5006 |
+
"epoch": 0.9930167597765364,
|
5007 |
+
"grad_norm": 0.35240668058395386,
|
5008 |
+
"learning_rate": 2.4750412447749337e-08,
|
5009 |
+
"loss": 0.0961,
|
5010 |
+
"step": 711
|
5011 |
+
},
|
5012 |
+
{
|
5013 |
+
"epoch": 0.994413407821229,
|
5014 |
+
"grad_norm": 0.5831834673881531,
|
5015 |
+
"learning_rate": 1.5840499211439152e-08,
|
5016 |
+
"loss": 0.1229,
|
5017 |
+
"step": 712
|
5018 |
+
},
|
5019 |
+
{
|
5020 |
+
"epoch": 0.9958100558659218,
|
5021 |
+
"grad_norm": 0.5493488907814026,
|
5022 |
+
"learning_rate": 8.910383727156468e-09,
|
5023 |
+
"loss": 0.1237,
|
5024 |
+
"step": 713
|
5025 |
+
},
|
5026 |
+
{
|
5027 |
+
"epoch": 0.9972067039106145,
|
5028 |
+
"grad_norm": 0.4525875747203827,
|
5029 |
+
"learning_rate": 3.960203218911218e-09,
|
5030 |
+
"loss": 0.1178,
|
5031 |
+
"step": 714
|
5032 |
+
},
|
5033 |
+
{
|
5034 |
+
"epoch": 0.9986033519553073,
|
5035 |
+
"grad_norm": 0.7450011968612671,
|
5036 |
+
"learning_rate": 9.9005570577404e-10,
|
5037 |
+
"loss": 0.1197,
|
5038 |
+
"step": 715
|
5039 |
+
},
|
5040 |
+
{
|
5041 |
+
"epoch": 1.0,
|
5042 |
+
"grad_norm": 0.7382752299308777,
|
5043 |
+
"learning_rate": 0.0,
|
5044 |
+
"loss": 0.1041,
|
5045 |
+
"step": 716
|
5046 |
+
},
|
5047 |
+
{
|
5048 |
+
"epoch": 1.0,
|
5049 |
+
"eval_loss": 0.10793300718069077,
|
5050 |
+
"eval_runtime": 22.0207,
|
5051 |
+
"eval_samples_per_second": 27.383,
|
5052 |
+
"eval_steps_per_second": 13.714,
|
5053 |
+
"step": 716
|
5054 |
}
|
5055 |
],
|
5056 |
"logging_steps": 1,
|
|
|
5065 |
"should_evaluate": false,
|
5066 |
"should_log": false,
|
5067 |
"should_save": true,
|
5068 |
+
"should_training_stop": true
|
5069 |
},
|
5070 |
"attributes": {}
|
5071 |
}
|
5072 |
},
|
5073 |
+
"total_flos": 1.963577405771612e+17,
|
5074 |
"train_batch_size": 2,
|
5075 |
"trial_name": null,
|
5076 |
"trial_params": null
|