shulijia commited on
Commit
701e917
·
verified ·
1 Parent(s): 0bba35a

Training in progress, step 1000, checkpoint

Browse files
last-checkpoint/model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:450b317279dc110653fc9a73f39508a4d3194f7cfed04a348e348b6c624648db
3
  size 2384234968
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:061dc4d2658410b6acb6b73f235d87b3f5265cef287f6769738bad18b4c0a56d
3
  size 2384234968
last-checkpoint/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b4c8f30f751caea8784934de0edc0324d11e38c766bd5e30d9578144253c7ef0
3
  size 4768663315
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:42fe8edb8b2246c1d5a01ee698f646d3a2bb595570993cb3e2a0c0fd442749b7
3
  size 4768663315
last-checkpoint/rng_state.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:de69a2834426ff9ef8199d077e00892579278af31d8969d77f98235b5cfc010a
3
  size 14645
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2534e434cd5abbb8f7668d3eab0549db0ef95d6a797a3efa86b712e8e32266a7
3
  size 14645
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a5f8ea6759cdb8f0cac09a72bb10eaef8b1793c42711c2d3033d0f1c8eea47e9
3
  size 1465
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:89fc96babd031398dc4f49db529af14042a214d2e774c56f2d58620c95114b55
3
  size 1465
last-checkpoint/trainer_state.json CHANGED
@@ -2,9 +2,9 @@
2
  "best_global_step": null,
3
  "best_metric": null,
4
  "best_model_checkpoint": null,
5
- "epoch": 1.3898364079359555,
6
  "eval_steps": 100,
7
- "global_step": 500,
8
  "is_hyper_param_search": false,
9
  "is_local_process_zero": true,
10
  "is_world_process_zero": true,
@@ -458,6 +458,456 @@
458
  "mean_token_accuracy": 0.9125856146216392,
459
  "num_tokens": 4088832.0,
460
  "step": 500
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
461
  }
462
  ],
463
  "logging_steps": 10,
@@ -477,7 +927,7 @@
477
  "attributes": {}
478
  }
479
  },
480
- "total_flos": 1.0805984574308352e+16,
481
  "train_batch_size": 2,
482
  "trial_name": null,
483
  "trial_params": null
 
2
  "best_global_step": null,
3
  "best_metric": null,
4
  "best_model_checkpoint": null,
5
+ "epoch": 2.779672815871911,
6
  "eval_steps": 100,
7
+ "global_step": 1000,
8
  "is_hyper_param_search": false,
9
  "is_local_process_zero": true,
10
  "is_world_process_zero": true,
 
458
  "mean_token_accuracy": 0.9125856146216392,
459
  "num_tokens": 4088832.0,
460
  "step": 500
461
+ },
462
+ {
463
+ "epoch": 1.4176818656456667,
464
+ "grad_norm": 1.0593403577804565,
465
+ "learning_rate": 5.87448559670782e-06,
466
+ "loss": 0.0926,
467
+ "mean_token_accuracy": 0.9135763190686703,
468
+ "num_tokens": 4170752.0,
469
+ "step": 510
470
+ },
471
+ {
472
+ "epoch": 1.4455273233553776,
473
+ "grad_norm": 0.865151584148407,
474
+ "learning_rate": 5.771604938271605e-06,
475
+ "loss": 0.096,
476
+ "mean_token_accuracy": 0.9089041098952293,
477
+ "num_tokens": 4252672.0,
478
+ "step": 520
479
+ },
480
+ {
481
+ "epoch": 1.4733727810650887,
482
+ "grad_norm": 0.823634684085846,
483
+ "learning_rate": 5.6687242798353915e-06,
484
+ "loss": 0.0826,
485
+ "mean_token_accuracy": 0.9167196683585643,
486
+ "num_tokens": 4334592.0,
487
+ "step": 530
488
+ },
489
+ {
490
+ "epoch": 1.5012182387747999,
491
+ "grad_norm": 1.3224678039550781,
492
+ "learning_rate": 5.565843621399178e-06,
493
+ "loss": 0.0948,
494
+ "mean_token_accuracy": 0.909222112596035,
495
+ "num_tokens": 4416512.0,
496
+ "step": 540
497
+ },
498
+ {
499
+ "epoch": 1.529063696484511,
500
+ "grad_norm": 1.1047239303588867,
501
+ "learning_rate": 5.462962962962963e-06,
502
+ "loss": 0.0908,
503
+ "mean_token_accuracy": 0.911081214249134,
504
+ "num_tokens": 4498432.0,
505
+ "step": 550
506
+ },
507
+ {
508
+ "epoch": 1.556909154194222,
509
+ "grad_norm": 1.124988317489624,
510
+ "learning_rate": 5.360082304526749e-06,
511
+ "loss": 0.0944,
512
+ "mean_token_accuracy": 0.9126834630966186,
513
+ "num_tokens": 4580352.0,
514
+ "step": 560
515
+ },
516
+ {
517
+ "epoch": 1.584754611903933,
518
+ "grad_norm": 1.1450597047805786,
519
+ "learning_rate": 5.257201646090535e-06,
520
+ "loss": 0.09,
521
+ "mean_token_accuracy": 0.9106775917112827,
522
+ "num_tokens": 4662272.0,
523
+ "step": 570
524
+ },
525
+ {
526
+ "epoch": 1.6126000696136442,
527
+ "grad_norm": 0.9272586107254028,
528
+ "learning_rate": 5.154320987654321e-06,
529
+ "loss": 0.0864,
530
+ "mean_token_accuracy": 0.9094911932945251,
531
+ "num_tokens": 4744192.0,
532
+ "step": 580
533
+ },
534
+ {
535
+ "epoch": 1.6404455273233554,
536
+ "grad_norm": 1.3119956254959106,
537
+ "learning_rate": 5.051440329218107e-06,
538
+ "loss": 0.0887,
539
+ "mean_token_accuracy": 0.9084026426076889,
540
+ "num_tokens": 4826112.0,
541
+ "step": 590
542
+ },
543
+ {
544
+ "epoch": 1.6682909850330665,
545
+ "grad_norm": 0.9722729325294495,
546
+ "learning_rate": 4.9485596707818935e-06,
547
+ "loss": 0.0863,
548
+ "mean_token_accuracy": 0.9236545950174332,
549
+ "num_tokens": 4908032.0,
550
+ "step": 600
551
+ },
552
+ {
553
+ "epoch": 1.6961364427427776,
554
+ "grad_norm": 0.9531897902488708,
555
+ "learning_rate": 4.845679012345679e-06,
556
+ "loss": 0.0851,
557
+ "mean_token_accuracy": 0.9154476508498192,
558
+ "num_tokens": 4989952.0,
559
+ "step": 610
560
+ },
561
+ {
562
+ "epoch": 1.7239819004524888,
563
+ "grad_norm": 1.1761149168014526,
564
+ "learning_rate": 4.742798353909465e-06,
565
+ "loss": 0.0889,
566
+ "mean_token_accuracy": 0.9128913849592208,
567
+ "num_tokens": 5071872.0,
568
+ "step": 620
569
+ },
570
+ {
571
+ "epoch": 1.7518273581622,
572
+ "grad_norm": 0.8723099231719971,
573
+ "learning_rate": 4.6399176954732515e-06,
574
+ "loss": 0.0923,
575
+ "mean_token_accuracy": 0.9085738733410835,
576
+ "num_tokens": 5153792.0,
577
+ "step": 630
578
+ },
579
+ {
580
+ "epoch": 1.7796728158719108,
581
+ "grad_norm": 1.3783780336380005,
582
+ "learning_rate": 4.537037037037038e-06,
583
+ "loss": 0.0921,
584
+ "mean_token_accuracy": 0.9185420729219913,
585
+ "num_tokens": 5235712.0,
586
+ "step": 640
587
+ },
588
+ {
589
+ "epoch": 1.807518273581622,
590
+ "grad_norm": 0.9514628648757935,
591
+ "learning_rate": 4.434156378600823e-06,
592
+ "loss": 0.0914,
593
+ "mean_token_accuracy": 0.918224073201418,
594
+ "num_tokens": 5317632.0,
595
+ "step": 650
596
+ },
597
+ {
598
+ "epoch": 1.8353637312913331,
599
+ "grad_norm": 1.5927624702453613,
600
+ "learning_rate": 4.331275720164609e-06,
601
+ "loss": 0.0948,
602
+ "mean_token_accuracy": 0.9134784743189812,
603
+ "num_tokens": 5399552.0,
604
+ "step": 660
605
+ },
606
+ {
607
+ "epoch": 1.863209189001044,
608
+ "grad_norm": 1.1536865234375,
609
+ "learning_rate": 4.228395061728396e-06,
610
+ "loss": 0.0963,
611
+ "mean_token_accuracy": 0.9069471605122089,
612
+ "num_tokens": 5481472.0,
613
+ "step": 670
614
+ },
615
+ {
616
+ "epoch": 1.8910546467107552,
617
+ "grad_norm": 0.9462175965309143,
618
+ "learning_rate": 4.125514403292181e-06,
619
+ "loss": 0.0898,
620
+ "mean_token_accuracy": 0.9125489234924317,
621
+ "num_tokens": 5563392.0,
622
+ "step": 680
623
+ },
624
+ {
625
+ "epoch": 1.9189001044204663,
626
+ "grad_norm": 0.940758228302002,
627
+ "learning_rate": 4.022633744855967e-06,
628
+ "loss": 0.0949,
629
+ "mean_token_accuracy": 0.9125366877764464,
630
+ "num_tokens": 5645312.0,
631
+ "step": 690
632
+ },
633
+ {
634
+ "epoch": 1.9467455621301775,
635
+ "grad_norm": 0.9007344245910645,
636
+ "learning_rate": 3.9197530864197535e-06,
637
+ "loss": 0.082,
638
+ "mean_token_accuracy": 0.9042441304773092,
639
+ "num_tokens": 5727232.0,
640
+ "step": 700
641
+ },
642
+ {
643
+ "epoch": 1.9745910198398886,
644
+ "grad_norm": 0.880010187625885,
645
+ "learning_rate": 3.81687242798354e-06,
646
+ "loss": 0.0882,
647
+ "mean_token_accuracy": 0.9144691735506058,
648
+ "num_tokens": 5809152.0,
649
+ "step": 710
650
+ },
651
+ {
652
+ "epoch": 2.0,
653
+ "grad_norm": 3.2822415828704834,
654
+ "learning_rate": 3.7139917695473256e-06,
655
+ "loss": 0.0894,
656
+ "mean_token_accuracy": 0.9187732877796644,
657
+ "num_tokens": 5883904.0,
658
+ "step": 720
659
+ },
660
+ {
661
+ "epoch": 2.027845457709711,
662
+ "grad_norm": 0.8666273951530457,
663
+ "learning_rate": 3.6111111111111115e-06,
664
+ "loss": 0.0783,
665
+ "mean_token_accuracy": 0.9194716215133667,
666
+ "num_tokens": 5965824.0,
667
+ "step": 730
668
+ },
669
+ {
670
+ "epoch": 2.0556909154194223,
671
+ "grad_norm": 1.257120132446289,
672
+ "learning_rate": 3.5082304526748973e-06,
673
+ "loss": 0.0792,
674
+ "mean_token_accuracy": 0.9152030311524868,
675
+ "num_tokens": 6047744.0,
676
+ "step": 740
677
+ },
678
+ {
679
+ "epoch": 2.0835363731291334,
680
+ "grad_norm": 1.344621181488037,
681
+ "learning_rate": 3.405349794238683e-06,
682
+ "loss": 0.0868,
683
+ "mean_token_accuracy": 0.9004647728055716,
684
+ "num_tokens": 6129664.0,
685
+ "step": 750
686
+ },
687
+ {
688
+ "epoch": 2.1113818308388446,
689
+ "grad_norm": 1.0627198219299316,
690
+ "learning_rate": 3.30246913580247e-06,
691
+ "loss": 0.0804,
692
+ "mean_token_accuracy": 0.915875731408596,
693
+ "num_tokens": 6211584.0,
694
+ "step": 760
695
+ },
696
+ {
697
+ "epoch": 2.1392272885485557,
698
+ "grad_norm": 1.065942406654358,
699
+ "learning_rate": 3.1995884773662556e-06,
700
+ "loss": 0.0769,
701
+ "mean_token_accuracy": 0.9183219157159328,
702
+ "num_tokens": 6293504.0,
703
+ "step": 770
704
+ },
705
+ {
706
+ "epoch": 2.167072746258267,
707
+ "grad_norm": 1.218607783317566,
708
+ "learning_rate": 3.0967078189300415e-06,
709
+ "loss": 0.0847,
710
+ "mean_token_accuracy": 0.9176369860768319,
711
+ "num_tokens": 6375424.0,
712
+ "step": 780
713
+ },
714
+ {
715
+ "epoch": 2.1949182039679775,
716
+ "grad_norm": 0.8840998411178589,
717
+ "learning_rate": 2.9938271604938273e-06,
718
+ "loss": 0.076,
719
+ "mean_token_accuracy": 0.9088307194411754,
720
+ "num_tokens": 6457344.0,
721
+ "step": 790
722
+ },
723
+ {
724
+ "epoch": 2.2227636616776887,
725
+ "grad_norm": 1.3679556846618652,
726
+ "learning_rate": 2.890946502057613e-06,
727
+ "loss": 0.0804,
728
+ "mean_token_accuracy": 0.9142979431897402,
729
+ "num_tokens": 6539264.0,
730
+ "step": 800
731
+ },
732
+ {
733
+ "epoch": 2.2506091193874,
734
+ "grad_norm": 1.7247203588485718,
735
+ "learning_rate": 2.7880658436213994e-06,
736
+ "loss": 0.085,
737
+ "mean_token_accuracy": 0.9157534211874008,
738
+ "num_tokens": 6621184.0,
739
+ "step": 810
740
+ },
741
+ {
742
+ "epoch": 2.278454577097111,
743
+ "grad_norm": 1.2949681282043457,
744
+ "learning_rate": 2.6851851851851856e-06,
745
+ "loss": 0.0792,
746
+ "mean_token_accuracy": 0.9105797454714775,
747
+ "num_tokens": 6703104.0,
748
+ "step": 820
749
+ },
750
+ {
751
+ "epoch": 2.306300034806822,
752
+ "grad_norm": 1.1340421438217163,
753
+ "learning_rate": 2.5823045267489715e-06,
754
+ "loss": 0.0909,
755
+ "mean_token_accuracy": 0.9104696653783322,
756
+ "num_tokens": 6785024.0,
757
+ "step": 830
758
+ },
759
+ {
760
+ "epoch": 2.3341454925165332,
761
+ "grad_norm": 1.4311587810516357,
762
+ "learning_rate": 2.4794238683127577e-06,
763
+ "loss": 0.07,
764
+ "mean_token_accuracy": 0.9179427556693553,
765
+ "num_tokens": 6866944.0,
766
+ "step": 840
767
+ },
768
+ {
769
+ "epoch": 2.3619909502262444,
770
+ "grad_norm": 1.2658636569976807,
771
+ "learning_rate": 2.3765432098765435e-06,
772
+ "loss": 0.078,
773
+ "mean_token_accuracy": 0.9194593898952007,
774
+ "num_tokens": 6948864.0,
775
+ "step": 850
776
+ },
777
+ {
778
+ "epoch": 2.3898364079359555,
779
+ "grad_norm": 1.1735376119613647,
780
+ "learning_rate": 2.2736625514403294e-06,
781
+ "loss": 0.0771,
782
+ "mean_token_accuracy": 0.9146893344819546,
783
+ "num_tokens": 7030784.0,
784
+ "step": 860
785
+ },
786
+ {
787
+ "epoch": 2.4176818656456667,
788
+ "grad_norm": 1.3436040878295898,
789
+ "learning_rate": 2.1707818930041156e-06,
790
+ "loss": 0.0761,
791
+ "mean_token_accuracy": 0.9215264186263085,
792
+ "num_tokens": 7112704.0,
793
+ "step": 870
794
+ },
795
+ {
796
+ "epoch": 2.445527323355378,
797
+ "grad_norm": 1.7530394792556763,
798
+ "learning_rate": 2.0679012345679015e-06,
799
+ "loss": 0.0809,
800
+ "mean_token_accuracy": 0.9206335641443729,
801
+ "num_tokens": 7194624.0,
802
+ "step": 880
803
+ },
804
+ {
805
+ "epoch": 2.4733727810650885,
806
+ "grad_norm": 1.2825167179107666,
807
+ "learning_rate": 1.9650205761316873e-06,
808
+ "loss": 0.0763,
809
+ "mean_token_accuracy": 0.915472112596035,
810
+ "num_tokens": 7276544.0,
811
+ "step": 890
812
+ },
813
+ {
814
+ "epoch": 2.5012182387747997,
815
+ "grad_norm": 1.0933923721313477,
816
+ "learning_rate": 1.8621399176954735e-06,
817
+ "loss": 0.0824,
818
+ "mean_token_accuracy": 0.91106898188591,
819
+ "num_tokens": 7358464.0,
820
+ "step": 900
821
+ },
822
+ {
823
+ "epoch": 2.529063696484511,
824
+ "grad_norm": 1.5624420642852783,
825
+ "learning_rate": 1.7592592592592594e-06,
826
+ "loss": 0.0788,
827
+ "mean_token_accuracy": 0.9117661438882351,
828
+ "num_tokens": 7440384.0,
829
+ "step": 910
830
+ },
831
+ {
832
+ "epoch": 2.556909154194222,
833
+ "grad_norm": 0.8869456052780151,
834
+ "learning_rate": 1.6563786008230454e-06,
835
+ "loss": 0.0744,
836
+ "mean_token_accuracy": 0.9128913849592208,
837
+ "num_tokens": 7522304.0,
838
+ "step": 920
839
+ },
840
+ {
841
+ "epoch": 2.584754611903933,
842
+ "grad_norm": 1.7279531955718994,
843
+ "learning_rate": 1.5534979423868312e-06,
844
+ "loss": 0.0879,
845
+ "mean_token_accuracy": 0.9147627178579569,
846
+ "num_tokens": 7604224.0,
847
+ "step": 930
848
+ },
849
+ {
850
+ "epoch": 2.612600069613644,
851
+ "grad_norm": 0.9950515031814575,
852
+ "learning_rate": 1.4506172839506175e-06,
853
+ "loss": 0.0785,
854
+ "mean_token_accuracy": 0.9133194699883461,
855
+ "num_tokens": 7686144.0,
856
+ "step": 940
857
+ },
858
+ {
859
+ "epoch": 2.6404455273233554,
860
+ "grad_norm": 1.2250232696533203,
861
+ "learning_rate": 1.3477366255144033e-06,
862
+ "loss": 0.0781,
863
+ "mean_token_accuracy": 0.9132705468684434,
864
+ "num_tokens": 7768064.0,
865
+ "step": 950
866
+ },
867
+ {
868
+ "epoch": 2.6682909850330665,
869
+ "grad_norm": 1.1228382587432861,
870
+ "learning_rate": 1.2448559670781894e-06,
871
+ "loss": 0.0858,
872
+ "mean_token_accuracy": 0.9010029338300228,
873
+ "num_tokens": 7849984.0,
874
+ "step": 960
875
+ },
876
+ {
877
+ "epoch": 2.6961364427427776,
878
+ "grad_norm": 1.256042718887329,
879
+ "learning_rate": 1.1419753086419754e-06,
880
+ "loss": 0.079,
881
+ "mean_token_accuracy": 0.9115582153201103,
882
+ "num_tokens": 7931904.0,
883
+ "step": 970
884
+ },
885
+ {
886
+ "epoch": 2.723981900452489,
887
+ "grad_norm": 0.9604980945587158,
888
+ "learning_rate": 1.0390946502057615e-06,
889
+ "loss": 0.0757,
890
+ "mean_token_accuracy": 0.9120229929685593,
891
+ "num_tokens": 8013824.0,
892
+ "step": 980
893
+ },
894
+ {
895
+ "epoch": 2.7518273581622,
896
+ "grad_norm": 1.2086925506591797,
897
+ "learning_rate": 9.362139917695474e-07,
898
+ "loss": 0.082,
899
+ "mean_token_accuracy": 0.9099559649825096,
900
+ "num_tokens": 8095744.0,
901
+ "step": 990
902
+ },
903
+ {
904
+ "epoch": 2.779672815871911,
905
+ "grad_norm": 0.9486561417579651,
906
+ "learning_rate": 8.333333333333333e-07,
907
+ "loss": 0.0888,
908
+ "mean_token_accuracy": 0.910922210663557,
909
+ "num_tokens": 8177664.0,
910
+ "step": 1000
911
  }
912
  ],
913
  "logging_steps": 10,
 
927
  "attributes": {}
928
  }
929
  },
930
+ "total_flos": 2.1611969148616704e+16,
931
  "train_batch_size": 2,
932
  "trial_name": null,
933
  "trial_params": null