Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +354 -4
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -6657,6 +6657,356 @@
|
|
| 6657 |
"learning_rate": 6.819348298638839e-07,
|
| 6658 |
"loss": 0.0077,
|
| 6659 |
"step": 9500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 6660 |
}
|
| 6661 |
],
|
| 6662 |
"logging_steps": 10,
|
|
@@ -6671,12 +7021,12 @@
|
|
| 6671 |
"should_evaluate": false,
|
| 6672 |
"should_log": false,
|
| 6673 |
"should_save": true,
|
| 6674 |
-
"should_training_stop":
|
| 6675 |
},
|
| 6676 |
"attributes": {}
|
| 6677 |
}
|
| 6678 |
},
|
| 6679 |
-
"total_flos": 3.
|
| 6680 |
"train_batch_size": 16,
|
| 6681 |
"trial_name": null,
|
| 6682 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 20.70393374741201,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 10000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 6657 |
"learning_rate": 6.819348298638839e-07,
|
| 6658 |
"loss": 0.0077,
|
| 6659 |
"step": 9500
|
| 6660 |
+
},
|
| 6661 |
+
{
|
| 6662 |
+
"epoch": 19.68944099378882,
|
| 6663 |
+
"grad_norm": 0.20381984114646912,
|
| 6664 |
+
"learning_rate": 6.549893279788277e-07,
|
| 6665 |
+
"loss": 0.0079,
|
| 6666 |
+
"step": 9510
|
| 6667 |
+
},
|
| 6668 |
+
{
|
| 6669 |
+
"epoch": 19.71014492753623,
|
| 6670 |
+
"grad_norm": 0.15530520677566528,
|
| 6671 |
+
"learning_rate": 6.285834552247128e-07,
|
| 6672 |
+
"loss": 0.0062,
|
| 6673 |
+
"step": 9520
|
| 6674 |
+
},
|
| 6675 |
+
{
|
| 6676 |
+
"epoch": 19.730848861283643,
|
| 6677 |
+
"grad_norm": 0.05255560576915741,
|
| 6678 |
+
"learning_rate": 6.027175003719354e-07,
|
| 6679 |
+
"loss": 0.0047,
|
| 6680 |
+
"step": 9530
|
| 6681 |
+
},
|
| 6682 |
+
{
|
| 6683 |
+
"epoch": 19.751552795031056,
|
| 6684 |
+
"grad_norm": 0.06348340958356857,
|
| 6685 |
+
"learning_rate": 5.773917462864264e-07,
|
| 6686 |
+
"loss": 0.0062,
|
| 6687 |
+
"step": 9540
|
| 6688 |
+
},
|
| 6689 |
+
{
|
| 6690 |
+
"epoch": 19.77225672877847,
|
| 6691 |
+
"grad_norm": 0.12964807450771332,
|
| 6692 |
+
"learning_rate": 5.526064699265753e-07,
|
| 6693 |
+
"loss": 0.0127,
|
| 6694 |
+
"step": 9550
|
| 6695 |
+
},
|
| 6696 |
+
{
|
| 6697 |
+
"epoch": 19.79296066252588,
|
| 6698 |
+
"grad_norm": 0.1979799121618271,
|
| 6699 |
+
"learning_rate": 5.283619423401998e-07,
|
| 6700 |
+
"loss": 0.0048,
|
| 6701 |
+
"step": 9560
|
| 6702 |
+
},
|
| 6703 |
+
{
|
| 6704 |
+
"epoch": 19.81366459627329,
|
| 6705 |
+
"grad_norm": 0.14146266877651215,
|
| 6706 |
+
"learning_rate": 5.046584286615697e-07,
|
| 6707 |
+
"loss": 0.0089,
|
| 6708 |
+
"step": 9570
|
| 6709 |
+
},
|
| 6710 |
+
{
|
| 6711 |
+
"epoch": 19.834368530020704,
|
| 6712 |
+
"grad_norm": 0.1563977748155594,
|
| 6713 |
+
"learning_rate": 4.814961881085045e-07,
|
| 6714 |
+
"loss": 0.0042,
|
| 6715 |
+
"step": 9580
|
| 6716 |
+
},
|
| 6717 |
+
{
|
| 6718 |
+
"epoch": 19.855072463768117,
|
| 6719 |
+
"grad_norm": 0.04129205644130707,
|
| 6720 |
+
"learning_rate": 4.5887547397955864e-07,
|
| 6721 |
+
"loss": 0.0047,
|
| 6722 |
+
"step": 9590
|
| 6723 |
+
},
|
| 6724 |
+
{
|
| 6725 |
+
"epoch": 19.875776397515526,
|
| 6726 |
+
"grad_norm": 0.2608759105205536,
|
| 6727 |
+
"learning_rate": 4.367965336512403e-07,
|
| 6728 |
+
"loss": 0.0134,
|
| 6729 |
+
"step": 9600
|
| 6730 |
+
},
|
| 6731 |
+
{
|
| 6732 |
+
"epoch": 19.89648033126294,
|
| 6733 |
+
"grad_norm": 0.16297097504138947,
|
| 6734 |
+
"learning_rate": 4.1525960857530243e-07,
|
| 6735 |
+
"loss": 0.0035,
|
| 6736 |
+
"step": 9610
|
| 6737 |
+
},
|
| 6738 |
+
{
|
| 6739 |
+
"epoch": 19.917184265010352,
|
| 6740 |
+
"grad_norm": 0.14169001579284668,
|
| 6741 |
+
"learning_rate": 3.9426493427611177e-07,
|
| 6742 |
+
"loss": 0.0076,
|
| 6743 |
+
"step": 9620
|
| 6744 |
+
},
|
| 6745 |
+
{
|
| 6746 |
+
"epoch": 19.937888198757765,
|
| 6747 |
+
"grad_norm": 0.141464963555336,
|
| 6748 |
+
"learning_rate": 3.738127403480507e-07,
|
| 6749 |
+
"loss": 0.0052,
|
| 6750 |
+
"step": 9630
|
| 6751 |
+
},
|
| 6752 |
+
{
|
| 6753 |
+
"epoch": 19.958592132505174,
|
| 6754 |
+
"grad_norm": 0.08023510873317719,
|
| 6755 |
+
"learning_rate": 3.5390325045304706e-07,
|
| 6756 |
+
"loss": 0.0055,
|
| 6757 |
+
"step": 9640
|
| 6758 |
+
},
|
| 6759 |
+
{
|
| 6760 |
+
"epoch": 19.979296066252587,
|
| 6761 |
+
"grad_norm": 0.09788880497217178,
|
| 6762 |
+
"learning_rate": 3.3453668231809286e-07,
|
| 6763 |
+
"loss": 0.0086,
|
| 6764 |
+
"step": 9650
|
| 6765 |
+
},
|
| 6766 |
+
{
|
| 6767 |
+
"epoch": 20.0,
|
| 6768 |
+
"grad_norm": 0.2991919219493866,
|
| 6769 |
+
"learning_rate": 3.157132477328628e-07,
|
| 6770 |
+
"loss": 0.0117,
|
| 6771 |
+
"step": 9660
|
| 6772 |
+
},
|
| 6773 |
+
{
|
| 6774 |
+
"epoch": 20.020703933747413,
|
| 6775 |
+
"grad_norm": 0.152107372879982,
|
| 6776 |
+
"learning_rate": 2.9743315254743833e-07,
|
| 6777 |
+
"loss": 0.0212,
|
| 6778 |
+
"step": 9670
|
| 6779 |
+
},
|
| 6780 |
+
{
|
| 6781 |
+
"epoch": 20.041407867494826,
|
| 6782 |
+
"grad_norm": 0.1049988642334938,
|
| 6783 |
+
"learning_rate": 2.796965966699927e-07,
|
| 6784 |
+
"loss": 0.012,
|
| 6785 |
+
"step": 9680
|
| 6786 |
+
},
|
| 6787 |
+
{
|
| 6788 |
+
"epoch": 20.062111801242235,
|
| 6789 |
+
"grad_norm": 0.28902608156204224,
|
| 6790 |
+
"learning_rate": 2.625037740646763e-07,
|
| 6791 |
+
"loss": 0.0103,
|
| 6792 |
+
"step": 9690
|
| 6793 |
+
},
|
| 6794 |
+
{
|
| 6795 |
+
"epoch": 20.082815734989648,
|
| 6796 |
+
"grad_norm": 0.2800842821598053,
|
| 6797 |
+
"learning_rate": 2.458548727494292e-07,
|
| 6798 |
+
"loss": 0.0067,
|
| 6799 |
+
"step": 9700
|
| 6800 |
+
},
|
| 6801 |
+
{
|
| 6802 |
+
"epoch": 20.10351966873706,
|
| 6803 |
+
"grad_norm": 0.18260431289672852,
|
| 6804 |
+
"learning_rate": 2.2975007479397738e-07,
|
| 6805 |
+
"loss": 0.0049,
|
| 6806 |
+
"step": 9710
|
| 6807 |
+
},
|
| 6808 |
+
{
|
| 6809 |
+
"epoch": 20.124223602484474,
|
| 6810 |
+
"grad_norm": 0.19015970826148987,
|
| 6811 |
+
"learning_rate": 2.1418955631781202e-07,
|
| 6812 |
+
"loss": 0.0117,
|
| 6813 |
+
"step": 9720
|
| 6814 |
+
},
|
| 6815 |
+
{
|
| 6816 |
+
"epoch": 20.144927536231883,
|
| 6817 |
+
"grad_norm": 0.1346769630908966,
|
| 6818 |
+
"learning_rate": 1.9917348748826335e-07,
|
| 6819 |
+
"loss": 0.0065,
|
| 6820 |
+
"step": 9730
|
| 6821 |
+
},
|
| 6822 |
+
{
|
| 6823 |
+
"epoch": 20.165631469979296,
|
| 6824 |
+
"grad_norm": 0.12038490176200867,
|
| 6825 |
+
"learning_rate": 1.847020325186577e-07,
|
| 6826 |
+
"loss": 0.0078,
|
| 6827 |
+
"step": 9740
|
| 6828 |
+
},
|
| 6829 |
+
{
|
| 6830 |
+
"epoch": 20.18633540372671,
|
| 6831 |
+
"grad_norm": 0.2004089653491974,
|
| 6832 |
+
"learning_rate": 1.7077534966650766e-07,
|
| 6833 |
+
"loss": 0.0096,
|
| 6834 |
+
"step": 9750
|
| 6835 |
+
},
|
| 6836 |
+
{
|
| 6837 |
+
"epoch": 20.20703933747412,
|
| 6838 |
+
"grad_norm": 0.19906413555145264,
|
| 6839 |
+
"learning_rate": 1.5739359123178587e-07,
|
| 6840 |
+
"loss": 0.0078,
|
| 6841 |
+
"step": 9760
|
| 6842 |
+
},
|
| 6843 |
+
{
|
| 6844 |
+
"epoch": 20.22774327122153,
|
| 6845 |
+
"grad_norm": 0.15226063132286072,
|
| 6846 |
+
"learning_rate": 1.4455690355525964e-07,
|
| 6847 |
+
"loss": 0.0049,
|
| 6848 |
+
"step": 9770
|
| 6849 |
+
},
|
| 6850 |
+
{
|
| 6851 |
+
"epoch": 20.248447204968944,
|
| 6852 |
+
"grad_norm": 0.12389522045850754,
|
| 6853 |
+
"learning_rate": 1.3226542701689215e-07,
|
| 6854 |
+
"loss": 0.0062,
|
| 6855 |
+
"step": 9780
|
| 6856 |
+
},
|
| 6857 |
+
{
|
| 6858 |
+
"epoch": 20.269151138716357,
|
| 6859 |
+
"grad_norm": 0.10260294377803802,
|
| 6860 |
+
"learning_rate": 1.2051929603428825e-07,
|
| 6861 |
+
"loss": 0.0131,
|
| 6862 |
+
"step": 9790
|
| 6863 |
+
},
|
| 6864 |
+
{
|
| 6865 |
+
"epoch": 20.28985507246377,
|
| 6866 |
+
"grad_norm": 0.059663962572813034,
|
| 6867 |
+
"learning_rate": 1.0931863906127327e-07,
|
| 6868 |
+
"loss": 0.0059,
|
| 6869 |
+
"step": 9800
|
| 6870 |
+
},
|
| 6871 |
+
{
|
| 6872 |
+
"epoch": 20.31055900621118,
|
| 6873 |
+
"grad_norm": 0.05520065873861313,
|
| 6874 |
+
"learning_rate": 9.866357858642205e-08,
|
| 6875 |
+
"loss": 0.0074,
|
| 6876 |
+
"step": 9810
|
| 6877 |
+
},
|
| 6878 |
+
{
|
| 6879 |
+
"epoch": 20.33126293995859,
|
| 6880 |
+
"grad_norm": 0.21085630357265472,
|
| 6881 |
+
"learning_rate": 8.855423113177664e-08,
|
| 6882 |
+
"loss": 0.0072,
|
| 6883 |
+
"step": 9820
|
| 6884 |
+
},
|
| 6885 |
+
{
|
| 6886 |
+
"epoch": 20.351966873706004,
|
| 6887 |
+
"grad_norm": 0.1336776465177536,
|
| 6888 |
+
"learning_rate": 7.899070725153613e-08,
|
| 6889 |
+
"loss": 0.0054,
|
| 6890 |
+
"step": 9830
|
| 6891 |
+
},
|
| 6892 |
+
{
|
| 6893 |
+
"epoch": 20.372670807453417,
|
| 6894 |
+
"grad_norm": 0.19630035758018494,
|
| 6895 |
+
"learning_rate": 6.997311153086883e-08,
|
| 6896 |
+
"loss": 0.0046,
|
| 6897 |
+
"step": 9840
|
| 6898 |
+
},
|
| 6899 |
+
{
|
| 6900 |
+
"epoch": 20.393374741200827,
|
| 6901 |
+
"grad_norm": 0.20599442720413208,
|
| 6902 |
+
"learning_rate": 6.150154258476315e-08,
|
| 6903 |
+
"loss": 0.0081,
|
| 6904 |
+
"step": 9850
|
| 6905 |
+
},
|
| 6906 |
+
{
|
| 6907 |
+
"epoch": 20.41407867494824,
|
| 6908 |
+
"grad_norm": 0.22885958850383759,
|
| 6909 |
+
"learning_rate": 5.3576093056922906e-08,
|
| 6910 |
+
"loss": 0.0076,
|
| 6911 |
+
"step": 9860
|
| 6912 |
+
},
|
| 6913 |
+
{
|
| 6914 |
+
"epoch": 20.434782608695652,
|
| 6915 |
+
"grad_norm": 0.07992696017026901,
|
| 6916 |
+
"learning_rate": 4.619684961881254e-08,
|
| 6917 |
+
"loss": 0.0071,
|
| 6918 |
+
"step": 9870
|
| 6919 |
+
},
|
| 6920 |
+
{
|
| 6921 |
+
"epoch": 20.455486542443065,
|
| 6922 |
+
"grad_norm": 0.2755813002586365,
|
| 6923 |
+
"learning_rate": 3.936389296864129e-08,
|
| 6924 |
+
"loss": 0.0064,
|
| 6925 |
+
"step": 9880
|
| 6926 |
+
},
|
| 6927 |
+
{
|
| 6928 |
+
"epoch": 20.476190476190474,
|
| 6929 |
+
"grad_norm": 0.1255924105644226,
|
| 6930 |
+
"learning_rate": 3.3077297830541584e-08,
|
| 6931 |
+
"loss": 0.0075,
|
| 6932 |
+
"step": 9890
|
| 6933 |
+
},
|
| 6934 |
+
{
|
| 6935 |
+
"epoch": 20.496894409937887,
|
| 6936 |
+
"grad_norm": 0.10394856333732605,
|
| 6937 |
+
"learning_rate": 2.7337132953697554e-08,
|
| 6938 |
+
"loss": 0.0072,
|
| 6939 |
+
"step": 9900
|
| 6940 |
+
},
|
| 6941 |
+
{
|
| 6942 |
+
"epoch": 20.5175983436853,
|
| 6943 |
+
"grad_norm": 0.11971770226955414,
|
| 6944 |
+
"learning_rate": 2.214346111164556e-08,
|
| 6945 |
+
"loss": 0.007,
|
| 6946 |
+
"step": 9910
|
| 6947 |
+
},
|
| 6948 |
+
{
|
| 6949 |
+
"epoch": 20.538302277432713,
|
| 6950 |
+
"grad_norm": 0.20464111864566803,
|
| 6951 |
+
"learning_rate": 1.749633910153592e-08,
|
| 6952 |
+
"loss": 0.0046,
|
| 6953 |
+
"step": 9920
|
| 6954 |
+
},
|
| 6955 |
+
{
|
| 6956 |
+
"epoch": 20.559006211180126,
|
| 6957 |
+
"grad_norm": 0.12219670414924622,
|
| 6958 |
+
"learning_rate": 1.3395817743561134e-08,
|
| 6959 |
+
"loss": 0.0082,
|
| 6960 |
+
"step": 9930
|
| 6961 |
+
},
|
| 6962 |
+
{
|
| 6963 |
+
"epoch": 20.579710144927535,
|
| 6964 |
+
"grad_norm": 0.2536000907421112,
|
| 6965 |
+
"learning_rate": 9.841941880361916e-09,
|
| 6966 |
+
"loss": 0.0152,
|
| 6967 |
+
"step": 9940
|
| 6968 |
+
},
|
| 6969 |
+
{
|
| 6970 |
+
"epoch": 20.600414078674948,
|
| 6971 |
+
"grad_norm": 0.16146942973136902,
|
| 6972 |
+
"learning_rate": 6.834750376549792e-09,
|
| 6973 |
+
"loss": 0.0055,
|
| 6974 |
+
"step": 9950
|
| 6975 |
+
},
|
| 6976 |
+
{
|
| 6977 |
+
"epoch": 20.62111801242236,
|
| 6978 |
+
"grad_norm": 0.1295255571603775,
|
| 6979 |
+
"learning_rate": 4.3742761183018784e-09,
|
| 6980 |
+
"loss": 0.0118,
|
| 6981 |
+
"step": 9960
|
| 6982 |
+
},
|
| 6983 |
+
{
|
| 6984 |
+
"epoch": 20.641821946169774,
|
| 6985 |
+
"grad_norm": 0.19239592552185059,
|
| 6986 |
+
"learning_rate": 2.4605460129556445e-09,
|
| 6987 |
+
"loss": 0.0065,
|
| 6988 |
+
"step": 9970
|
| 6989 |
+
},
|
| 6990 |
+
{
|
| 6991 |
+
"epoch": 20.662525879917183,
|
| 6992 |
+
"grad_norm": 0.21993553638458252,
|
| 6993 |
+
"learning_rate": 1.0935809887702154e-09,
|
| 6994 |
+
"loss": 0.0065,
|
| 6995 |
+
"step": 9980
|
| 6996 |
+
},
|
| 6997 |
+
{
|
| 6998 |
+
"epoch": 20.683229813664596,
|
| 6999 |
+
"grad_norm": 0.15056583285331726,
|
| 7000 |
+
"learning_rate": 2.7339599464326627e-10,
|
| 7001 |
+
"loss": 0.0062,
|
| 7002 |
+
"step": 9990
|
| 7003 |
+
},
|
| 7004 |
+
{
|
| 7005 |
+
"epoch": 20.70393374741201,
|
| 7006 |
+
"grad_norm": 0.24346260726451874,
|
| 7007 |
+
"learning_rate": 0.0,
|
| 7008 |
+
"loss": 0.0084,
|
| 7009 |
+
"step": 10000
|
| 7010 |
}
|
| 7011 |
],
|
| 7012 |
"logging_steps": 10,
|
|
|
|
| 7021 |
"should_evaluate": false,
|
| 7022 |
"should_log": false,
|
| 7023 |
"should_save": true,
|
| 7024 |
+
"should_training_stop": true
|
| 7025 |
},
|
| 7026 |
"attributes": {}
|
| 7027 |
}
|
| 7028 |
},
|
| 7029 |
+
"total_flos": 3.6267190177586125e+17,
|
| 7030 |
"train_batch_size": 16,
|
| 7031 |
"trial_name": null,
|
| 7032 |
"trial_params": null
|