AmberYifan commited on
Commit
f57d633
1 Parent(s): 75c3cdf

Training in progress, epoch 2, checkpoint

Browse files
last-checkpoint/global_step626/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6eaa45e9530f3d94283c55295f5df838550f3797754204bd6a943f3ba08f621c
3
+ size 14496051368
last-checkpoint/global_step626/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2601edf9aed3d175e477d9b3d8da64d09732b3c34c11dc4ce1d00f83b390b413
3
+ size 14496051368
last-checkpoint/global_step626/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fc8817173341cb0b47e4e8950086453c26262bc97554145c71baba5385504170
3
+ size 14496051368
last-checkpoint/global_step626/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d2e51c2ee812a2f36c46ba620c91d50fb73a6299cbd982523160da9bc1b2fd5a
3
+ size 14496051368
last-checkpoint/global_step626/zero_pp_rank_0_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e64a9c51ebd628b9bf2608cff3d7237a80d5c468cd8d8b48f515124c642a108c
3
+ size 150629
last-checkpoint/global_step626/zero_pp_rank_1_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7dd73a544ae58c58529ab5f1c8394bd4ca27de8f03d2c08be023cf8d20238c02
3
+ size 150629
last-checkpoint/global_step626/zero_pp_rank_2_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:28d28ca8245f2279317f34b723b619870c155e960ec021b9c34c03a1d87d3b4f
3
+ size 150629
last-checkpoint/global_step626/zero_pp_rank_3_mp_rank_00_model_states.pt ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a1d3e667ee41fda048808b51f84ce0bf14e8f331e4486526d954df9649eeeb8a
3
+ size 150629
last-checkpoint/latest CHANGED
@@ -1 +1 @@
1
- global_step313
 
1
+ global_step626
last-checkpoint/model-00001-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d7bc16b7d14cb1bc85d3ad5586c00e819a8f58211e12d6c5208c133095587746
3
  size 4949453792
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:ba6ef55eea8b65483a8a250fef97347b3f6e02dcba3eedb6bf17f44415c5f732
3
  size 4949453792
last-checkpoint/model-00002-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:d08a9d134f573c80a237f514fbe6fe74e20747ba8a0c78ae3b8f73d6e703d0f0
3
  size 4999819336
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:191c087c6f176fe1c8c5954455cf0c19a6a7eba43b49386055aed3708d046dc0
3
  size 4999819336
last-checkpoint/model-00003-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c5f7441c84f5bb26c9d91046594a17cf738206b5f7da4a20f00ef9db25172e0d
3
  size 4546807800
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:094f61b93fe26129782869ee365a638f652b55856eda586b5a5ead7c68c5e089
3
  size 4546807800
last-checkpoint/rng_state_0.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:f1ce770862b76ceecb3453a5d0b9ef65311eab0f6a60f52997726a89d718de08
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:2307c03867cef25b5028feb9a23f80e784b9af9a615de13ddca560a6a90fb593
3
  size 15024
last-checkpoint/rng_state_1.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2bcae6573442578b752be3e988d4ccb38056b45c31b5c02eb579ec6cebcfa62e
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:50218cfaecdd818354e567b7167c13899e3b42297e7d8f58bd7e732cfa547800
3
  size 15024
last-checkpoint/rng_state_2.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:398ff45072fee2975e88b9e078f915103d75d94bc08753303cf855b915973623
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9814a66b49861f5495b06dae3be12ddf7185b88e2cae1fb808ca9efd99d5807f
3
  size 15024
last-checkpoint/rng_state_3.pth CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:7f4b210bc832d4c648e3bee8c21dca26e5a8b365d6ec90c638062005a052e57b
3
  size 15024
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a7be93040a65e0a29975f6c70b94418e1fdf88423a50c58aa572141d3c92fbfc
3
  size 15024
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c019f2bd3222f739db75f458caa9c28765649010b6fa70739851be6f2b00c68b
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:c3fd160e69e5766b513d913785299b2ff6d181cae9fd47f741adc100cb17d22e
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": 1.0,
5
  "eval_steps": 500,
6
- "global_step": 313,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -503,6 +503,487 @@
503
  "eval_samples_per_second": 12.974,
504
  "eval_steps_per_second": 0.454,
505
  "step": 313
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
506
  }
507
  ],
508
  "logging_steps": 10,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 2.0,
5
  "eval_steps": 500,
6
+ "global_step": 626,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
503
  "eval_samples_per_second": 12.974,
504
  "eval_steps_per_second": 0.454,
505
  "step": 313
506
+ },
507
+ {
508
+ "epoch": 1.0223642172523961,
509
+ "grad_norm": 0.0003356864204729289,
510
+ "learning_rate": 3.662721893491124e-07,
511
+ "logits/chosen": -3.5,
512
+ "logits/rejected": -3.296875,
513
+ "logps/chosen": -169.0,
514
+ "logps/rejected": -478.0,
515
+ "loss": 0.0,
516
+ "rewards/accuracies": 1.0,
517
+ "rewards/chosen": -0.498046875,
518
+ "rewards/margins": 31.125,
519
+ "rewards/rejected": -31.625,
520
+ "step": 320
521
+ },
522
+ {
523
+ "epoch": 1.0543130990415335,
524
+ "grad_norm": 5.8557721184616675e-06,
525
+ "learning_rate": 3.603550295857988e-07,
526
+ "logits/chosen": -3.5,
527
+ "logits/rejected": -3.328125,
528
+ "logps/chosen": -156.0,
529
+ "logps/rejected": -480.0,
530
+ "loss": 0.0,
531
+ "rewards/accuracies": 1.0,
532
+ "rewards/chosen": -0.1435546875,
533
+ "rewards/margins": 32.5,
534
+ "rewards/rejected": -32.75,
535
+ "step": 330
536
+ },
537
+ {
538
+ "epoch": 1.0862619808306708,
539
+ "grad_norm": 3.5777121751233287e-06,
540
+ "learning_rate": 3.544378698224852e-07,
541
+ "logits/chosen": -3.515625,
542
+ "logits/rejected": -3.296875,
543
+ "logps/chosen": -144.0,
544
+ "logps/rejected": -478.0,
545
+ "loss": 0.0,
546
+ "rewards/accuracies": 1.0,
547
+ "rewards/chosen": -0.03955078125,
548
+ "rewards/margins": 30.75,
549
+ "rewards/rejected": -30.75,
550
+ "step": 340
551
+ },
552
+ {
553
+ "epoch": 1.1182108626198084,
554
+ "grad_norm": 6.936406729828301e-07,
555
+ "learning_rate": 3.485207100591716e-07,
556
+ "logits/chosen": -3.5,
557
+ "logits/rejected": -3.3125,
558
+ "logps/chosen": -154.0,
559
+ "logps/rejected": -492.0,
560
+ "loss": 0.0,
561
+ "rewards/accuracies": 1.0,
562
+ "rewards/chosen": -0.220703125,
563
+ "rewards/margins": 32.75,
564
+ "rewards/rejected": -33.0,
565
+ "step": 350
566
+ },
567
+ {
568
+ "epoch": 1.1501597444089458,
569
+ "grad_norm": 0.009589555434561007,
570
+ "learning_rate": 3.42603550295858e-07,
571
+ "logits/chosen": -3.5,
572
+ "logits/rejected": -3.34375,
573
+ "logps/chosen": -168.0,
574
+ "logps/rejected": -480.0,
575
+ "loss": 0.0,
576
+ "rewards/accuracies": 1.0,
577
+ "rewards/chosen": -0.146484375,
578
+ "rewards/margins": 32.0,
579
+ "rewards/rejected": -32.25,
580
+ "step": 360
581
+ },
582
+ {
583
+ "epoch": 1.182108626198083,
584
+ "grad_norm": 2.378099652849891e-05,
585
+ "learning_rate": 3.366863905325444e-07,
586
+ "logits/chosen": -3.5,
587
+ "logits/rejected": -3.34375,
588
+ "logps/chosen": -137.0,
589
+ "logps/rejected": -472.0,
590
+ "loss": 0.0,
591
+ "rewards/accuracies": 1.0,
592
+ "rewards/chosen": 0.1630859375,
593
+ "rewards/margins": 31.5,
594
+ "rewards/rejected": -31.375,
595
+ "step": 370
596
+ },
597
+ {
598
+ "epoch": 1.2140575079872205,
599
+ "grad_norm": 5.514648329432445e-05,
600
+ "learning_rate": 3.3076923076923075e-07,
601
+ "logits/chosen": -3.5,
602
+ "logits/rejected": -3.390625,
603
+ "logps/chosen": -151.0,
604
+ "logps/rejected": -484.0,
605
+ "loss": 0.0,
606
+ "rewards/accuracies": 1.0,
607
+ "rewards/chosen": -0.03662109375,
608
+ "rewards/margins": 32.75,
609
+ "rewards/rejected": -32.75,
610
+ "step": 380
611
+ },
612
+ {
613
+ "epoch": 1.2460063897763578,
614
+ "grad_norm": 5.280407319654463e-07,
615
+ "learning_rate": 3.2485207100591716e-07,
616
+ "logits/chosen": -3.5,
617
+ "logits/rejected": -3.3125,
618
+ "logps/chosen": -127.0,
619
+ "logps/rejected": -480.0,
620
+ "loss": 0.0,
621
+ "rewards/accuracies": 1.0,
622
+ "rewards/chosen": 0.23046875,
623
+ "rewards/margins": 33.0,
624
+ "rewards/rejected": -32.75,
625
+ "step": 390
626
+ },
627
+ {
628
+ "epoch": 1.2779552715654952,
629
+ "grad_norm": 0.00046164498605506235,
630
+ "learning_rate": 3.189349112426035e-07,
631
+ "logits/chosen": -3.484375,
632
+ "logits/rejected": -3.359375,
633
+ "logps/chosen": -132.0,
634
+ "logps/rejected": -486.0,
635
+ "loss": 0.0,
636
+ "rewards/accuracies": 1.0,
637
+ "rewards/chosen": -0.1435546875,
638
+ "rewards/margins": 33.0,
639
+ "rewards/rejected": -33.0,
640
+ "step": 400
641
+ },
642
+ {
643
+ "epoch": 1.3099041533546325,
644
+ "grad_norm": 0.0001779209961052298,
645
+ "learning_rate": 3.130177514792899e-07,
646
+ "logits/chosen": -3.515625,
647
+ "logits/rejected": -3.265625,
648
+ "logps/chosen": -144.0,
649
+ "logps/rejected": -496.0,
650
+ "loss": 0.0,
651
+ "rewards/accuracies": 1.0,
652
+ "rewards/chosen": 0.07177734375,
653
+ "rewards/margins": 35.0,
654
+ "rewards/rejected": -35.0,
655
+ "step": 410
656
+ },
657
+ {
658
+ "epoch": 1.34185303514377,
659
+ "grad_norm": 4.9071818579178655e-05,
660
+ "learning_rate": 3.071005917159763e-07,
661
+ "logits/chosen": -3.515625,
662
+ "logits/rejected": -3.375,
663
+ "logps/chosen": -160.0,
664
+ "logps/rejected": -478.0,
665
+ "loss": 0.0,
666
+ "rewards/accuracies": 1.0,
667
+ "rewards/chosen": -0.12158203125,
668
+ "rewards/margins": 32.5,
669
+ "rewards/rejected": -32.5,
670
+ "step": 420
671
+ },
672
+ {
673
+ "epoch": 1.3738019169329074,
674
+ "grad_norm": 4.25683016680881e-06,
675
+ "learning_rate": 3.011834319526627e-07,
676
+ "logits/chosen": -3.5,
677
+ "logits/rejected": -3.3125,
678
+ "logps/chosen": -149.0,
679
+ "logps/rejected": -490.0,
680
+ "loss": 0.0,
681
+ "rewards/accuracies": 1.0,
682
+ "rewards/chosen": -0.1220703125,
683
+ "rewards/margins": 33.75,
684
+ "rewards/rejected": -33.75,
685
+ "step": 430
686
+ },
687
+ {
688
+ "epoch": 1.4057507987220448,
689
+ "grad_norm": 0.0009887420156956786,
690
+ "learning_rate": 2.952662721893491e-07,
691
+ "logits/chosen": -3.5,
692
+ "logits/rejected": -3.28125,
693
+ "logps/chosen": -157.0,
694
+ "logps/rejected": -482.0,
695
+ "loss": 0.0,
696
+ "rewards/accuracies": 1.0,
697
+ "rewards/chosen": -0.130859375,
698
+ "rewards/margins": 32.5,
699
+ "rewards/rejected": -32.5,
700
+ "step": 440
701
+ },
702
+ {
703
+ "epoch": 1.4376996805111821,
704
+ "grad_norm": 1.4837415298689954e-05,
705
+ "learning_rate": 2.893491124260355e-07,
706
+ "logits/chosen": -3.5,
707
+ "logits/rejected": -3.34375,
708
+ "logps/chosen": -156.0,
709
+ "logps/rejected": -488.0,
710
+ "loss": 0.0,
711
+ "rewards/accuracies": 1.0,
712
+ "rewards/chosen": 0.12890625,
713
+ "rewards/margins": 34.75,
714
+ "rewards/rejected": -34.5,
715
+ "step": 450
716
+ },
717
+ {
718
+ "epoch": 1.4696485623003195,
719
+ "grad_norm": 8.248177987361765e-06,
720
+ "learning_rate": 2.834319526627219e-07,
721
+ "logits/chosen": -3.5,
722
+ "logits/rejected": -3.375,
723
+ "logps/chosen": -148.0,
724
+ "logps/rejected": -498.0,
725
+ "loss": 0.0,
726
+ "rewards/accuracies": 1.0,
727
+ "rewards/chosen": -0.150390625,
728
+ "rewards/margins": 34.5,
729
+ "rewards/rejected": -34.5,
730
+ "step": 460
731
+ },
732
+ {
733
+ "epoch": 1.5015974440894568,
734
+ "grad_norm": 1.9806936842810454e-05,
735
+ "learning_rate": 2.7751479289940824e-07,
736
+ "logits/chosen": -3.515625,
737
+ "logits/rejected": -3.375,
738
+ "logps/chosen": -157.0,
739
+ "logps/rejected": -478.0,
740
+ "loss": 0.0,
741
+ "rewards/accuracies": 1.0,
742
+ "rewards/chosen": -0.396484375,
743
+ "rewards/margins": 32.25,
744
+ "rewards/rejected": -32.5,
745
+ "step": 470
746
+ },
747
+ {
748
+ "epoch": 1.5335463258785942,
749
+ "grad_norm": 5.907801581150698e-07,
750
+ "learning_rate": 2.715976331360947e-07,
751
+ "logits/chosen": -3.515625,
752
+ "logits/rejected": -3.390625,
753
+ "logps/chosen": -153.0,
754
+ "logps/rejected": -460.0,
755
+ "loss": 0.0,
756
+ "rewards/accuracies": 1.0,
757
+ "rewards/chosen": -0.17578125,
758
+ "rewards/margins": 30.5,
759
+ "rewards/rejected": -30.75,
760
+ "step": 480
761
+ },
762
+ {
763
+ "epoch": 1.5654952076677318,
764
+ "grad_norm": 0.00014451950487479103,
765
+ "learning_rate": 2.6568047337278106e-07,
766
+ "logits/chosen": -3.515625,
767
+ "logits/rejected": -3.34375,
768
+ "logps/chosen": -158.0,
769
+ "logps/rejected": -470.0,
770
+ "loss": 0.0,
771
+ "rewards/accuracies": 1.0,
772
+ "rewards/chosen": -0.0057373046875,
773
+ "rewards/margins": 32.25,
774
+ "rewards/rejected": -32.25,
775
+ "step": 490
776
+ },
777
+ {
778
+ "epoch": 1.5974440894568689,
779
+ "grad_norm": 0.0012272802134721346,
780
+ "learning_rate": 2.5976331360946746e-07,
781
+ "logits/chosen": -3.5,
782
+ "logits/rejected": -3.3125,
783
+ "logps/chosen": -148.0,
784
+ "logps/rejected": -482.0,
785
+ "loss": 0.0,
786
+ "rewards/accuracies": 1.0,
787
+ "rewards/chosen": 0.19921875,
788
+ "rewards/margins": 33.0,
789
+ "rewards/rejected": -32.75,
790
+ "step": 500
791
+ },
792
+ {
793
+ "epoch": 1.6293929712460065,
794
+ "grad_norm": 2.9833293563020383e-05,
795
+ "learning_rate": 2.538461538461538e-07,
796
+ "logits/chosen": -3.5,
797
+ "logits/rejected": -3.28125,
798
+ "logps/chosen": -150.0,
799
+ "logps/rejected": -494.0,
800
+ "loss": 0.0,
801
+ "rewards/accuracies": 1.0,
802
+ "rewards/chosen": 0.1962890625,
803
+ "rewards/margins": 34.75,
804
+ "rewards/rejected": -34.75,
805
+ "step": 510
806
+ },
807
+ {
808
+ "epoch": 1.6613418530351438,
809
+ "grad_norm": 0.001356611209656493,
810
+ "learning_rate": 2.479289940828402e-07,
811
+ "logits/chosen": -3.484375,
812
+ "logits/rejected": -3.34375,
813
+ "logps/chosen": -135.0,
814
+ "logps/rejected": -500.0,
815
+ "loss": 0.0,
816
+ "rewards/accuracies": 1.0,
817
+ "rewards/chosen": 0.035400390625,
818
+ "rewards/margins": 34.75,
819
+ "rewards/rejected": -34.75,
820
+ "step": 520
821
+ },
822
+ {
823
+ "epoch": 1.6932907348242812,
824
+ "grad_norm": 0.00022456450452624014,
825
+ "learning_rate": 2.420118343195266e-07,
826
+ "logits/chosen": -3.53125,
827
+ "logits/rejected": -3.359375,
828
+ "logps/chosen": -166.0,
829
+ "logps/rejected": -484.0,
830
+ "loss": 0.0,
831
+ "rewards/accuracies": 1.0,
832
+ "rewards/chosen": -0.35546875,
833
+ "rewards/margins": 31.5,
834
+ "rewards/rejected": -31.75,
835
+ "step": 530
836
+ },
837
+ {
838
+ "epoch": 1.7252396166134185,
839
+ "grad_norm": 6.820094977898695e-06,
840
+ "learning_rate": 2.36094674556213e-07,
841
+ "logits/chosen": -3.46875,
842
+ "logits/rejected": -3.34375,
843
+ "logps/chosen": -169.0,
844
+ "logps/rejected": -478.0,
845
+ "loss": 0.0,
846
+ "rewards/accuracies": 1.0,
847
+ "rewards/chosen": -0.298828125,
848
+ "rewards/margins": 33.25,
849
+ "rewards/rejected": -33.5,
850
+ "step": 540
851
+ },
852
+ {
853
+ "epoch": 1.7571884984025559,
854
+ "grad_norm": 0.001992382751935317,
855
+ "learning_rate": 2.301775147928994e-07,
856
+ "logits/chosen": -3.515625,
857
+ "logits/rejected": -3.296875,
858
+ "logps/chosen": -171.0,
859
+ "logps/rejected": -492.0,
860
+ "loss": 0.0,
861
+ "rewards/accuracies": 1.0,
862
+ "rewards/chosen": 0.10107421875,
863
+ "rewards/margins": 33.5,
864
+ "rewards/rejected": -33.5,
865
+ "step": 550
866
+ },
867
+ {
868
+ "epoch": 1.7891373801916934,
869
+ "grad_norm": 1.115393663415116e-05,
870
+ "learning_rate": 2.242603550295858e-07,
871
+ "logits/chosen": -3.484375,
872
+ "logits/rejected": -3.375,
873
+ "logps/chosen": -142.0,
874
+ "logps/rejected": -512.0,
875
+ "loss": 0.0,
876
+ "rewards/accuracies": 1.0,
877
+ "rewards/chosen": 0.16796875,
878
+ "rewards/margins": 35.5,
879
+ "rewards/rejected": -35.5,
880
+ "step": 560
881
+ },
882
+ {
883
+ "epoch": 1.8210862619808306,
884
+ "grad_norm": 1.489072794590662e-05,
885
+ "learning_rate": 2.1834319526627217e-07,
886
+ "logits/chosen": -3.484375,
887
+ "logits/rejected": -3.359375,
888
+ "logps/chosen": -154.0,
889
+ "logps/rejected": -476.0,
890
+ "loss": 0.0,
891
+ "rewards/accuracies": 1.0,
892
+ "rewards/chosen": -0.06640625,
893
+ "rewards/margins": 32.0,
894
+ "rewards/rejected": -32.0,
895
+ "step": 570
896
+ },
897
+ {
898
+ "epoch": 1.8530351437699681,
899
+ "grad_norm": 8.145269082773688e-06,
900
+ "learning_rate": 2.1242603550295858e-07,
901
+ "logits/chosen": -3.5,
902
+ "logits/rejected": -3.359375,
903
+ "logps/chosen": -154.0,
904
+ "logps/rejected": -472.0,
905
+ "loss": 0.0,
906
+ "rewards/accuracies": 1.0,
907
+ "rewards/chosen": -0.10400390625,
908
+ "rewards/margins": 32.25,
909
+ "rewards/rejected": -32.25,
910
+ "step": 580
911
+ },
912
+ {
913
+ "epoch": 1.8849840255591053,
914
+ "grad_norm": 0.0005030582745028686,
915
+ "learning_rate": 2.0650887573964496e-07,
916
+ "logits/chosen": -3.515625,
917
+ "logits/rejected": -3.328125,
918
+ "logps/chosen": -154.0,
919
+ "logps/rejected": -488.0,
920
+ "loss": 0.0,
921
+ "rewards/accuracies": 1.0,
922
+ "rewards/chosen": -0.003448486328125,
923
+ "rewards/margins": 32.0,
924
+ "rewards/rejected": -32.0,
925
+ "step": 590
926
+ },
927
+ {
928
+ "epoch": 1.9169329073482428,
929
+ "grad_norm": 9.704683387564733e-06,
930
+ "learning_rate": 2.0059171597633133e-07,
931
+ "logits/chosen": -3.484375,
932
+ "logits/rejected": -3.34375,
933
+ "logps/chosen": -150.0,
934
+ "logps/rejected": -474.0,
935
+ "loss": 0.0,
936
+ "rewards/accuracies": 1.0,
937
+ "rewards/chosen": -0.1357421875,
938
+ "rewards/margins": 31.375,
939
+ "rewards/rejected": -31.5,
940
+ "step": 600
941
+ },
942
+ {
943
+ "epoch": 1.9488817891373802,
944
+ "grad_norm": 1.7369899748244282e-05,
945
+ "learning_rate": 1.9467455621301774e-07,
946
+ "logits/chosen": -3.484375,
947
+ "logits/rejected": -3.328125,
948
+ "logps/chosen": -156.0,
949
+ "logps/rejected": -492.0,
950
+ "loss": 0.0,
951
+ "rewards/accuracies": 1.0,
952
+ "rewards/chosen": -0.470703125,
953
+ "rewards/margins": 33.5,
954
+ "rewards/rejected": -33.75,
955
+ "step": 610
956
+ },
957
+ {
958
+ "epoch": 1.9808306709265175,
959
+ "grad_norm": 1.8988577382856775e-06,
960
+ "learning_rate": 1.8875739644970412e-07,
961
+ "logits/chosen": -3.46875,
962
+ "logits/rejected": -3.3125,
963
+ "logps/chosen": -146.0,
964
+ "logps/rejected": -516.0,
965
+ "loss": 0.0,
966
+ "rewards/accuracies": 1.0,
967
+ "rewards/chosen": -0.326171875,
968
+ "rewards/margins": 34.5,
969
+ "rewards/rejected": -34.75,
970
+ "step": 620
971
+ },
972
+ {
973
+ "epoch": 2.0,
974
+ "eval_logits/chosen": -3.53125,
975
+ "eval_logits/rejected": -3.296875,
976
+ "eval_logps/chosen": -150.0,
977
+ "eval_logps/rejected": -498.0,
978
+ "eval_loss": 3.6485798204921593e-07,
979
+ "eval_rewards/accuracies": 1.0,
980
+ "eval_rewards/chosen": -0.265625,
981
+ "eval_rewards/margins": 35.25,
982
+ "eval_rewards/rejected": -35.5,
983
+ "eval_runtime": 15.435,
984
+ "eval_samples_per_second": 12.958,
985
+ "eval_steps_per_second": 0.454,
986
+ "step": 626
987
  }
988
  ],
989
  "logging_steps": 10,