AmberYifan commited on
Commit
429f159
1 Parent(s): 640db2f

Training in progress, epoch 2, checkpoint

Browse files
last-checkpoint/global_step626/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:ebc04be7384e593372638a16229b531385fa1b6d8e3884ce5841cda644929ae1
3
  size 14496051368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:6170fa8d97fa30b4806ef3343bcb53a84af65a993f324fa5321cf4173e150829
3
  size 14496051368
last-checkpoint/global_step626/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:b0a851bc7dbf381b9415244a3317e1093cc6ce6db168322ac2eb605ee634a467
3
  size 14496051368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:bcd3a2670689e323a10424b337d59878869cdb018d43e79ecf5724ce4675515f
3
  size 14496051368
last-checkpoint/global_step626/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a5d58d2f634e7c064d58547175147411ab23c87779f52b9eadecc3db714cfc59
3
  size 14496051368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d3586a1983f10c04b0a5a285724da3326ba33e283edd212b6385dea38185324d
3
  size 14496051368
last-checkpoint/global_step626/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0c7d6f98f5e6ebd6a6e20367c244f69c74203ca69f2f16f3ab7145a2e910c73f
3
  size 14496051368
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:06d93b2eb30b786046fd464dd5da2ac765044e9f0abae615eeb13b350aa36f67
3
  size 14496051368
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 sha256:61d5356d4a15061ea562977c5a21355be8902685c315c4a848e71d0b3c2850c1
3
  size 4949453792
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:76ca72b83de51129cd7925cbfb1685adb4340322e41bf8b7c4b214cf29510410
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 sha256:437bf184ce59319aaa45a6aab6284a2006b10c6d045d56d64da7f44c15f91fbf
3
  size 4999819336
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e5d25fb197ff0758382a811bf6502cf621a7e6b8a93cd1bbaebe01570f965b7b
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 sha256:86f7cb1d7aecd12d7aec857282a49e1bc06484bc16cd51a24ac3f525df9499b3
3
  size 4546807800
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7c2874469911c4144cd30a5677af2f1136239e505a45a3746e34f027fef833e7
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": 16.065,
504
  "eval_steps_per_second": 0.562,
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": 16.065,
504
  "eval_steps_per_second": 0.562,
505
  "step": 313
506
+ },
507
+ {
508
+ "epoch": 1.0223642172523961,
509
+ "grad_norm": 23.70055727816958,
510
+ "learning_rate": 3.662721893491124e-07,
511
+ "logits/chosen": -3.4375,
512
+ "logits/rejected": -3.375,
513
+ "logps/chosen": -274.0,
514
+ "logps/rejected": -284.0,
515
+ "loss": 0.2261,
516
+ "rewards/accuracies": 0.9125000238418579,
517
+ "rewards/chosen": 0.33984375,
518
+ "rewards/margins": 2.625,
519
+ "rewards/rejected": -2.28125,
520
+ "step": 320
521
+ },
522
+ {
523
+ "epoch": 1.0543130990415335,
524
+ "grad_norm": 20.6098819793211,
525
+ "learning_rate": 3.603550295857988e-07,
526
+ "logits/chosen": -3.484375,
527
+ "logits/rejected": -3.40625,
528
+ "logps/chosen": -270.0,
529
+ "logps/rejected": -296.0,
530
+ "loss": 0.1038,
531
+ "rewards/accuracies": 0.9750000238418579,
532
+ "rewards/chosen": 0.484375,
533
+ "rewards/margins": 3.328125,
534
+ "rewards/rejected": -2.84375,
535
+ "step": 330
536
+ },
537
+ {
538
+ "epoch": 1.0862619808306708,
539
+ "grad_norm": 12.129974272451467,
540
+ "learning_rate": 3.544378698224852e-07,
541
+ "logits/chosen": -3.46875,
542
+ "logits/rejected": -3.359375,
543
+ "logps/chosen": -278.0,
544
+ "logps/rejected": -300.0,
545
+ "loss": 0.0875,
546
+ "rewards/accuracies": 0.987500011920929,
547
+ "rewards/chosen": 0.4296875,
548
+ "rewards/margins": 3.359375,
549
+ "rewards/rejected": -2.921875,
550
+ "step": 340
551
+ },
552
+ {
553
+ "epoch": 1.1182108626198084,
554
+ "grad_norm": 19.312391737005985,
555
+ "learning_rate": 3.485207100591716e-07,
556
+ "logits/chosen": -3.46875,
557
+ "logits/rejected": -3.3125,
558
+ "logps/chosen": -270.0,
559
+ "logps/rejected": -284.0,
560
+ "loss": 0.1201,
561
+ "rewards/accuracies": 0.9375,
562
+ "rewards/chosen": 0.66015625,
563
+ "rewards/margins": 3.921875,
564
+ "rewards/rejected": -3.265625,
565
+ "step": 350
566
+ },
567
+ {
568
+ "epoch": 1.1501597444089458,
569
+ "grad_norm": 16.056335220522016,
570
+ "learning_rate": 3.42603550295858e-07,
571
+ "logits/chosen": -3.40625,
572
+ "logits/rejected": -3.421875,
573
+ "logps/chosen": -266.0,
574
+ "logps/rejected": -300.0,
575
+ "loss": 0.0965,
576
+ "rewards/accuracies": 0.949999988079071,
577
+ "rewards/chosen": 0.3671875,
578
+ "rewards/margins": 3.296875,
579
+ "rewards/rejected": -2.921875,
580
+ "step": 360
581
+ },
582
+ {
583
+ "epoch": 1.182108626198083,
584
+ "grad_norm": 14.173305651953171,
585
+ "learning_rate": 3.366863905325444e-07,
586
+ "logits/chosen": -3.390625,
587
+ "logits/rejected": -3.421875,
588
+ "logps/chosen": -274.0,
589
+ "logps/rejected": -288.0,
590
+ "loss": 0.1037,
591
+ "rewards/accuracies": 0.949999988079071,
592
+ "rewards/chosen": 0.72265625,
593
+ "rewards/margins": 3.625,
594
+ "rewards/rejected": -2.90625,
595
+ "step": 370
596
+ },
597
+ {
598
+ "epoch": 1.2140575079872205,
599
+ "grad_norm": 28.838482988149984,
600
+ "learning_rate": 3.3076923076923075e-07,
601
+ "logits/chosen": -3.421875,
602
+ "logits/rejected": -3.46875,
603
+ "logps/chosen": -280.0,
604
+ "logps/rejected": -286.0,
605
+ "loss": 0.0885,
606
+ "rewards/accuracies": 0.9750000238418579,
607
+ "rewards/chosen": 0.69140625,
608
+ "rewards/margins": 3.796875,
609
+ "rewards/rejected": -3.109375,
610
+ "step": 380
611
+ },
612
+ {
613
+ "epoch": 1.2460063897763578,
614
+ "grad_norm": 18.796201336624538,
615
+ "learning_rate": 3.2485207100591716e-07,
616
+ "logits/chosen": -3.515625,
617
+ "logits/rejected": -3.328125,
618
+ "logps/chosen": -272.0,
619
+ "logps/rejected": -298.0,
620
+ "loss": 0.1175,
621
+ "rewards/accuracies": 0.925000011920929,
622
+ "rewards/chosen": 0.546875,
623
+ "rewards/margins": 3.625,
624
+ "rewards/rejected": -3.078125,
625
+ "step": 390
626
+ },
627
+ {
628
+ "epoch": 1.2779552715654952,
629
+ "grad_norm": 12.284250452584237,
630
+ "learning_rate": 3.189349112426035e-07,
631
+ "logits/chosen": -3.4375,
632
+ "logits/rejected": -3.4375,
633
+ "logps/chosen": -274.0,
634
+ "logps/rejected": -292.0,
635
+ "loss": 0.0702,
636
+ "rewards/accuracies": 0.987500011920929,
637
+ "rewards/chosen": 0.7109375,
638
+ "rewards/margins": 3.859375,
639
+ "rewards/rejected": -3.15625,
640
+ "step": 400
641
+ },
642
+ {
643
+ "epoch": 1.3099041533546325,
644
+ "grad_norm": 13.495027410207776,
645
+ "learning_rate": 3.130177514792899e-07,
646
+ "logits/chosen": -3.484375,
647
+ "logits/rejected": -3.40625,
648
+ "logps/chosen": -272.0,
649
+ "logps/rejected": -294.0,
650
+ "loss": 0.087,
651
+ "rewards/accuracies": 0.949999988079071,
652
+ "rewards/chosen": 0.578125,
653
+ "rewards/margins": 3.640625,
654
+ "rewards/rejected": -3.0625,
655
+ "step": 410
656
+ },
657
+ {
658
+ "epoch": 1.34185303514377,
659
+ "grad_norm": 22.992235645975622,
660
+ "learning_rate": 3.071005917159763e-07,
661
+ "logits/chosen": -3.5,
662
+ "logits/rejected": -3.421875,
663
+ "logps/chosen": -286.0,
664
+ "logps/rejected": -300.0,
665
+ "loss": 0.1101,
666
+ "rewards/accuracies": 0.9375,
667
+ "rewards/chosen": 0.232421875,
668
+ "rewards/margins": 3.765625,
669
+ "rewards/rejected": -3.53125,
670
+ "step": 420
671
+ },
672
+ {
673
+ "epoch": 1.3738019169329074,
674
+ "grad_norm": 21.69400097170527,
675
+ "learning_rate": 3.011834319526627e-07,
676
+ "logits/chosen": -3.40625,
677
+ "logits/rejected": -3.40625,
678
+ "logps/chosen": -272.0,
679
+ "logps/rejected": -296.0,
680
+ "loss": 0.117,
681
+ "rewards/accuracies": 0.9624999761581421,
682
+ "rewards/chosen": 0.47265625,
683
+ "rewards/margins": 3.84375,
684
+ "rewards/rejected": -3.375,
685
+ "step": 430
686
+ },
687
+ {
688
+ "epoch": 1.4057507987220448,
689
+ "grad_norm": 18.285863744766086,
690
+ "learning_rate": 2.952662721893491e-07,
691
+ "logits/chosen": -3.484375,
692
+ "logits/rejected": -3.359375,
693
+ "logps/chosen": -280.0,
694
+ "logps/rejected": -282.0,
695
+ "loss": 0.1335,
696
+ "rewards/accuracies": 0.9624999761581421,
697
+ "rewards/chosen": 0.51171875,
698
+ "rewards/margins": 3.578125,
699
+ "rewards/rejected": -3.0625,
700
+ "step": 440
701
+ },
702
+ {
703
+ "epoch": 1.4376996805111821,
704
+ "grad_norm": 17.34600664501295,
705
+ "learning_rate": 2.893491124260355e-07,
706
+ "logits/chosen": -3.359375,
707
+ "logits/rejected": -3.34375,
708
+ "logps/chosen": -284.0,
709
+ "logps/rejected": -292.0,
710
+ "loss": 0.083,
711
+ "rewards/accuracies": 1.0,
712
+ "rewards/chosen": 0.404296875,
713
+ "rewards/margins": 4.1875,
714
+ "rewards/rejected": -3.78125,
715
+ "step": 450
716
+ },
717
+ {
718
+ "epoch": 1.4696485623003195,
719
+ "grad_norm": 36.80402921857842,
720
+ "learning_rate": 2.834319526627219e-07,
721
+ "logits/chosen": -3.4375,
722
+ "logits/rejected": -3.3125,
723
+ "logps/chosen": -268.0,
724
+ "logps/rejected": -302.0,
725
+ "loss": 0.116,
726
+ "rewards/accuracies": 0.9624999761581421,
727
+ "rewards/chosen": 0.287109375,
728
+ "rewards/margins": 3.671875,
729
+ "rewards/rejected": -3.390625,
730
+ "step": 460
731
+ },
732
+ {
733
+ "epoch": 1.5015974440894568,
734
+ "grad_norm": 15.918868054649243,
735
+ "learning_rate": 2.7751479289940824e-07,
736
+ "logits/chosen": -3.375,
737
+ "logits/rejected": -3.375,
738
+ "logps/chosen": -272.0,
739
+ "logps/rejected": -292.0,
740
+ "loss": 0.0565,
741
+ "rewards/accuracies": 1.0,
742
+ "rewards/chosen": 0.75390625,
743
+ "rewards/margins": 4.15625,
744
+ "rewards/rejected": -3.40625,
745
+ "step": 470
746
+ },
747
+ {
748
+ "epoch": 1.5335463258785942,
749
+ "grad_norm": 22.888062361116138,
750
+ "learning_rate": 2.715976331360947e-07,
751
+ "logits/chosen": -3.390625,
752
+ "logits/rejected": -3.3125,
753
+ "logps/chosen": -278.0,
754
+ "logps/rejected": -292.0,
755
+ "loss": 0.1266,
756
+ "rewards/accuracies": 0.925000011920929,
757
+ "rewards/chosen": 0.640625,
758
+ "rewards/margins": 3.875,
759
+ "rewards/rejected": -3.234375,
760
+ "step": 480
761
+ },
762
+ {
763
+ "epoch": 1.5654952076677318,
764
+ "grad_norm": 30.24738555666603,
765
+ "learning_rate": 2.6568047337278106e-07,
766
+ "logits/chosen": -3.34375,
767
+ "logits/rejected": -3.3125,
768
+ "logps/chosen": -272.0,
769
+ "logps/rejected": -292.0,
770
+ "loss": 0.1042,
771
+ "rewards/accuracies": 0.949999988079071,
772
+ "rewards/chosen": 0.78125,
773
+ "rewards/margins": 4.1875,
774
+ "rewards/rejected": -3.40625,
775
+ "step": 490
776
+ },
777
+ {
778
+ "epoch": 1.5974440894568689,
779
+ "grad_norm": 23.550496082239732,
780
+ "learning_rate": 2.5976331360946746e-07,
781
+ "logits/chosen": -3.453125,
782
+ "logits/rejected": -3.421875,
783
+ "logps/chosen": -278.0,
784
+ "logps/rejected": -284.0,
785
+ "loss": 0.1374,
786
+ "rewards/accuracies": 0.949999988079071,
787
+ "rewards/chosen": 0.671875,
788
+ "rewards/margins": 3.6875,
789
+ "rewards/rejected": -3.015625,
790
+ "step": 500
791
+ },
792
+ {
793
+ "epoch": 1.6293929712460065,
794
+ "grad_norm": 25.63710072069056,
795
+ "learning_rate": 2.538461538461538e-07,
796
+ "logits/chosen": -3.328125,
797
+ "logits/rejected": -3.34375,
798
+ "logps/chosen": -268.0,
799
+ "logps/rejected": -292.0,
800
+ "loss": 0.1023,
801
+ "rewards/accuracies": 0.949999988079071,
802
+ "rewards/chosen": 0.5546875,
803
+ "rewards/margins": 3.640625,
804
+ "rewards/rejected": -3.09375,
805
+ "step": 510
806
+ },
807
+ {
808
+ "epoch": 1.6613418530351438,
809
+ "grad_norm": 24.141726605352023,
810
+ "learning_rate": 2.479289940828402e-07,
811
+ "logits/chosen": -3.28125,
812
+ "logits/rejected": -3.296875,
813
+ "logps/chosen": -270.0,
814
+ "logps/rejected": -286.0,
815
+ "loss": 0.082,
816
+ "rewards/accuracies": 0.987500011920929,
817
+ "rewards/chosen": 0.890625,
818
+ "rewards/margins": 4.375,
819
+ "rewards/rejected": -3.5,
820
+ "step": 520
821
+ },
822
+ {
823
+ "epoch": 1.6932907348242812,
824
+ "grad_norm": 34.35651000503782,
825
+ "learning_rate": 2.420118343195266e-07,
826
+ "logits/chosen": -3.421875,
827
+ "logits/rejected": -3.265625,
828
+ "logps/chosen": -268.0,
829
+ "logps/rejected": -294.0,
830
+ "loss": 0.1029,
831
+ "rewards/accuracies": 0.9750000238418579,
832
+ "rewards/chosen": 0.87109375,
833
+ "rewards/margins": 4.15625,
834
+ "rewards/rejected": -3.28125,
835
+ "step": 530
836
+ },
837
+ {
838
+ "epoch": 1.7252396166134185,
839
+ "grad_norm": 44.21100968000028,
840
+ "learning_rate": 2.36094674556213e-07,
841
+ "logits/chosen": -3.375,
842
+ "logits/rejected": -3.34375,
843
+ "logps/chosen": -262.0,
844
+ "logps/rejected": -292.0,
845
+ "loss": 0.1529,
846
+ "rewards/accuracies": 0.949999988079071,
847
+ "rewards/chosen": 1.0,
848
+ "rewards/margins": 4.25,
849
+ "rewards/rejected": -3.25,
850
+ "step": 540
851
+ },
852
+ {
853
+ "epoch": 1.7571884984025559,
854
+ "grad_norm": 47.981400895385406,
855
+ "learning_rate": 2.301775147928994e-07,
856
+ "logits/chosen": -3.375,
857
+ "logits/rejected": -3.3125,
858
+ "logps/chosen": -270.0,
859
+ "logps/rejected": -292.0,
860
+ "loss": 0.1231,
861
+ "rewards/accuracies": 0.9750000238418579,
862
+ "rewards/chosen": 0.83984375,
863
+ "rewards/margins": 3.6875,
864
+ "rewards/rejected": -2.859375,
865
+ "step": 550
866
+ },
867
+ {
868
+ "epoch": 1.7891373801916934,
869
+ "grad_norm": 38.43543934960892,
870
+ "learning_rate": 2.242603550295858e-07,
871
+ "logits/chosen": -3.4375,
872
+ "logits/rejected": -3.34375,
873
+ "logps/chosen": -258.0,
874
+ "logps/rejected": -294.0,
875
+ "loss": 0.135,
876
+ "rewards/accuracies": 0.9624999761581421,
877
+ "rewards/chosen": 0.7578125,
878
+ "rewards/margins": 3.65625,
879
+ "rewards/rejected": -2.890625,
880
+ "step": 560
881
+ },
882
+ {
883
+ "epoch": 1.8210862619808306,
884
+ "grad_norm": 11.780750523956426,
885
+ "learning_rate": 2.1834319526627217e-07,
886
+ "logits/chosen": -3.375,
887
+ "logits/rejected": -3.34375,
888
+ "logps/chosen": -276.0,
889
+ "logps/rejected": -288.0,
890
+ "loss": 0.1099,
891
+ "rewards/accuracies": 0.9624999761581421,
892
+ "rewards/chosen": 0.5625,
893
+ "rewards/margins": 3.859375,
894
+ "rewards/rejected": -3.296875,
895
+ "step": 570
896
+ },
897
+ {
898
+ "epoch": 1.8530351437699681,
899
+ "grad_norm": 16.56566595786949,
900
+ "learning_rate": 2.1242603550295858e-07,
901
+ "logits/chosen": -3.3125,
902
+ "logits/rejected": -3.28125,
903
+ "logps/chosen": -274.0,
904
+ "logps/rejected": -294.0,
905
+ "loss": 0.1071,
906
+ "rewards/accuracies": 0.949999988079071,
907
+ "rewards/chosen": 0.54296875,
908
+ "rewards/margins": 3.859375,
909
+ "rewards/rejected": -3.3125,
910
+ "step": 580
911
+ },
912
+ {
913
+ "epoch": 1.8849840255591053,
914
+ "grad_norm": 18.132322544979896,
915
+ "learning_rate": 2.0650887573964496e-07,
916
+ "logits/chosen": -3.3125,
917
+ "logits/rejected": -3.21875,
918
+ "logps/chosen": -270.0,
919
+ "logps/rejected": -304.0,
920
+ "loss": 0.0811,
921
+ "rewards/accuracies": 0.9750000238418579,
922
+ "rewards/chosen": 1.0078125,
923
+ "rewards/margins": 4.59375,
924
+ "rewards/rejected": -3.5625,
925
+ "step": 590
926
+ },
927
+ {
928
+ "epoch": 1.9169329073482428,
929
+ "grad_norm": 32.52764727647769,
930
+ "learning_rate": 2.0059171597633133e-07,
931
+ "logits/chosen": -3.359375,
932
+ "logits/rejected": -3.25,
933
+ "logps/chosen": -274.0,
934
+ "logps/rejected": -292.0,
935
+ "loss": 0.114,
936
+ "rewards/accuracies": 0.9624999761581421,
937
+ "rewards/chosen": 0.6953125,
938
+ "rewards/margins": 4.15625,
939
+ "rewards/rejected": -3.46875,
940
+ "step": 600
941
+ },
942
+ {
943
+ "epoch": 1.9488817891373802,
944
+ "grad_norm": 22.068866501441573,
945
+ "learning_rate": 1.9467455621301774e-07,
946
+ "logits/chosen": -3.3125,
947
+ "logits/rejected": -3.265625,
948
+ "logps/chosen": -276.0,
949
+ "logps/rejected": -296.0,
950
+ "loss": 0.0821,
951
+ "rewards/accuracies": 0.987500011920929,
952
+ "rewards/chosen": 0.88671875,
953
+ "rewards/margins": 4.4375,
954
+ "rewards/rejected": -3.546875,
955
+ "step": 610
956
+ },
957
+ {
958
+ "epoch": 1.9808306709265175,
959
+ "grad_norm": 41.19041486210437,
960
+ "learning_rate": 1.8875739644970412e-07,
961
+ "logits/chosen": -3.328125,
962
+ "logits/rejected": -3.203125,
963
+ "logps/chosen": -270.0,
964
+ "logps/rejected": -296.0,
965
+ "loss": 0.1093,
966
+ "rewards/accuracies": 0.9375,
967
+ "rewards/chosen": 0.59765625,
968
+ "rewards/margins": 4.28125,
969
+ "rewards/rejected": -3.671875,
970
+ "step": 620
971
+ },
972
+ {
973
+ "epoch": 2.0,
974
+ "eval_logits/chosen": -3.359375,
975
+ "eval_logits/rejected": -3.234375,
976
+ "eval_logps/chosen": -282.0,
977
+ "eval_logps/rejected": -288.0,
978
+ "eval_loss": 0.48261719942092896,
979
+ "eval_rewards/accuracies": 0.7142857313156128,
980
+ "eval_rewards/chosen": -0.46484375,
981
+ "eval_rewards/margins": 1.875,
982
+ "eval_rewards/rejected": -2.34375,
983
+ "eval_runtime": 12.43,
984
+ "eval_samples_per_second": 16.09,
985
+ "eval_steps_per_second": 0.563,
986
+ "step": 626
987
  }
988
  ],
989
  "logging_steps": 10,