AmberYifan commited on
Commit
2e0be61
1 Parent(s): b649536

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 sha256:3bde31c50544e333701c44fd2fbf4ec0bc102e3aac66bd697be18b8c1a62409f
3
  size 14483468456
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8b4734ca9784c24a7665d975da05cdb36a4eefd14688ee2621840db0fd63d65b
3
  size 14483468456
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:a9139d7ce80b2537ca28ffa07f0746ac41beec9de739e18815a9c5b8b03f5981
3
  size 14483468456
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d5a5e0a79cd8ee52cae75e340780a40e7cd0a1cf4aa50bb97955650d077c0025
3
  size 14483468456
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 sha256:7647f6d0e8cab95a428a423c9c1f56ecd61b87f25a18e24c8a6e354026bbbfaf
3
  size 14483468456
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:feb60faeb08c50f0ab692a107aa414a02d041eb3707d473dc5b29cb63fb11ce5
3
  size 14483468456
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 sha256:42ff90fdf33c0d6769e7767ef6e88f1701b80b3c6e70e602eb3b9cc32d2877d6
3
  size 14483468456
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:49801efea7b881cb10f1b01d5754952e0f3e18533574a1635330a5e9301f2364
3
  size 14483468456
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 sha256:47a89a06839841c7eb87dcbed4e8913f1671af0d8ed467692e12b0a36f7b66d5
3
  size 4943162336
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:43c7154919e968cd92a3d62643f31b03c6b226da33a6e4e99d77a77e3ad6aedd
3
  size 4943162336
last-checkpoint/model-00002-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:fd92879869d7067d8e76adbd5cc290284d880eddaf06396d04c9b4533088766b
3
  size 4999819336
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8aac376842b3031fdb1ef335e0763296c658688d3a6e0c0c216e0129e232e2e2
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 sha256:3a56de2e70c5e45089154d64b749c86385b3aca73991e54c21bce0a5140a6d3a
3
  size 4540516344
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:4e4bfa82d9210c8f5161e3131e3ae52acee7f9c4fea89a80e2dd7554f862dd6b
3
  size 4540516344
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": 16.006,
504
  "eval_steps_per_second": 0.56,
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": 16.006,
504
  "eval_steps_per_second": 0.56,
505
  "step": 313
506
+ },
507
+ {
508
+ "epoch": 1.0223642172523961,
509
+ "grad_norm": 23.67148600061787,
510
+ "learning_rate": 3.662721893491124e-07,
511
+ "logits/chosen": -3.296875,
512
+ "logits/rejected": -3.296875,
513
+ "logps/chosen": -276.0,
514
+ "logps/rejected": -282.0,
515
+ "loss": 0.232,
516
+ "rewards/accuracies": 0.9125000238418579,
517
+ "rewards/chosen": 0.482421875,
518
+ "rewards/margins": 2.34375,
519
+ "rewards/rejected": -1.859375,
520
+ "step": 320
521
+ },
522
+ {
523
+ "epoch": 1.0543130990415335,
524
+ "grad_norm": 16.419046920806387,
525
+ "learning_rate": 3.603550295857988e-07,
526
+ "logits/chosen": -3.28125,
527
+ "logits/rejected": -3.203125,
528
+ "logps/chosen": -268.0,
529
+ "logps/rejected": -302.0,
530
+ "loss": 0.1279,
531
+ "rewards/accuracies": 0.949999988079071,
532
+ "rewards/chosen": 0.51171875,
533
+ "rewards/margins": 3.375,
534
+ "rewards/rejected": -2.859375,
535
+ "step": 330
536
+ },
537
+ {
538
+ "epoch": 1.0862619808306708,
539
+ "grad_norm": 11.936537156516739,
540
+ "learning_rate": 3.544378698224852e-07,
541
+ "logits/chosen": -3.1875,
542
+ "logits/rejected": -3.203125,
543
+ "logps/chosen": -276.0,
544
+ "logps/rejected": -288.0,
545
+ "loss": 0.1213,
546
+ "rewards/accuracies": 0.9375,
547
+ "rewards/chosen": 0.6015625,
548
+ "rewards/margins": 3.171875,
549
+ "rewards/rejected": -2.578125,
550
+ "step": 340
551
+ },
552
+ {
553
+ "epoch": 1.1182108626198084,
554
+ "grad_norm": 20.043411609324348,
555
+ "learning_rate": 3.485207100591716e-07,
556
+ "logits/chosen": -3.125,
557
+ "logits/rejected": -3.15625,
558
+ "logps/chosen": -276.0,
559
+ "logps/rejected": -300.0,
560
+ "loss": 0.1216,
561
+ "rewards/accuracies": 0.9750000238418579,
562
+ "rewards/chosen": 0.484375,
563
+ "rewards/margins": 3.40625,
564
+ "rewards/rejected": -2.921875,
565
+ "step": 350
566
+ },
567
+ {
568
+ "epoch": 1.1501597444089458,
569
+ "grad_norm": 36.206031638132856,
570
+ "learning_rate": 3.42603550295858e-07,
571
+ "logits/chosen": -3.171875,
572
+ "logits/rejected": -3.265625,
573
+ "logps/chosen": -266.0,
574
+ "logps/rejected": -298.0,
575
+ "loss": 0.1088,
576
+ "rewards/accuracies": 0.9750000238418579,
577
+ "rewards/chosen": 0.4375,
578
+ "rewards/margins": 3.390625,
579
+ "rewards/rejected": -2.953125,
580
+ "step": 360
581
+ },
582
+ {
583
+ "epoch": 1.182108626198083,
584
+ "grad_norm": 30.276531485727194,
585
+ "learning_rate": 3.366863905325444e-07,
586
+ "logits/chosen": -3.28125,
587
+ "logits/rejected": -3.234375,
588
+ "logps/chosen": -270.0,
589
+ "logps/rejected": -300.0,
590
+ "loss": 0.1304,
591
+ "rewards/accuracies": 0.925000011920929,
592
+ "rewards/chosen": 0.51953125,
593
+ "rewards/margins": 3.234375,
594
+ "rewards/rejected": -2.71875,
595
+ "step": 370
596
+ },
597
+ {
598
+ "epoch": 1.2140575079872205,
599
+ "grad_norm": 36.79686407358778,
600
+ "learning_rate": 3.3076923076923075e-07,
601
+ "logits/chosen": -3.171875,
602
+ "logits/rejected": -3.234375,
603
+ "logps/chosen": -274.0,
604
+ "logps/rejected": -288.0,
605
+ "loss": 0.1287,
606
+ "rewards/accuracies": 0.9375,
607
+ "rewards/chosen": 0.466796875,
608
+ "rewards/margins": 3.15625,
609
+ "rewards/rejected": -2.6875,
610
+ "step": 380
611
+ },
612
+ {
613
+ "epoch": 1.2460063897763578,
614
+ "grad_norm": 45.670162060448156,
615
+ "learning_rate": 3.2485207100591716e-07,
616
+ "logits/chosen": -3.09375,
617
+ "logits/rejected": -3.265625,
618
+ "logps/chosen": -272.0,
619
+ "logps/rejected": -304.0,
620
+ "loss": 0.1189,
621
+ "rewards/accuracies": 0.9750000238418579,
622
+ "rewards/chosen": 0.87109375,
623
+ "rewards/margins": 4.125,
624
+ "rewards/rejected": -3.25,
625
+ "step": 390
626
+ },
627
+ {
628
+ "epoch": 1.2779552715654952,
629
+ "grad_norm": 44.93419834702028,
630
+ "learning_rate": 3.189349112426035e-07,
631
+ "logits/chosen": -3.25,
632
+ "logits/rejected": -3.28125,
633
+ "logps/chosen": -274.0,
634
+ "logps/rejected": -298.0,
635
+ "loss": 0.139,
636
+ "rewards/accuracies": 0.925000011920929,
637
+ "rewards/chosen": 0.63671875,
638
+ "rewards/margins": 3.140625,
639
+ "rewards/rejected": -2.5,
640
+ "step": 400
641
+ },
642
+ {
643
+ "epoch": 1.3099041533546325,
644
+ "grad_norm": 34.60440373360165,
645
+ "learning_rate": 3.130177514792899e-07,
646
+ "logits/chosen": -3.1875,
647
+ "logits/rejected": -3.21875,
648
+ "logps/chosen": -266.0,
649
+ "logps/rejected": -300.0,
650
+ "loss": 0.1653,
651
+ "rewards/accuracies": 0.949999988079071,
652
+ "rewards/chosen": 0.55078125,
653
+ "rewards/margins": 3.375,
654
+ "rewards/rejected": -2.828125,
655
+ "step": 410
656
+ },
657
+ {
658
+ "epoch": 1.34185303514377,
659
+ "grad_norm": 23.20153131751831,
660
+ "learning_rate": 3.071005917159763e-07,
661
+ "logits/chosen": -3.203125,
662
+ "logits/rejected": -3.171875,
663
+ "logps/chosen": -268.0,
664
+ "logps/rejected": -306.0,
665
+ "loss": 0.1342,
666
+ "rewards/accuracies": 0.925000011920929,
667
+ "rewards/chosen": 0.578125,
668
+ "rewards/margins": 3.359375,
669
+ "rewards/rejected": -2.796875,
670
+ "step": 420
671
+ },
672
+ {
673
+ "epoch": 1.3738019169329074,
674
+ "grad_norm": 27.675463987741193,
675
+ "learning_rate": 3.011834319526627e-07,
676
+ "logits/chosen": -3.171875,
677
+ "logits/rejected": -3.15625,
678
+ "logps/chosen": -266.0,
679
+ "logps/rejected": -304.0,
680
+ "loss": 0.1054,
681
+ "rewards/accuracies": 0.9624999761581421,
682
+ "rewards/chosen": 0.6328125,
683
+ "rewards/margins": 3.71875,
684
+ "rewards/rejected": -3.09375,
685
+ "step": 430
686
+ },
687
+ {
688
+ "epoch": 1.4057507987220448,
689
+ "grad_norm": 40.196786857901714,
690
+ "learning_rate": 2.952662721893491e-07,
691
+ "logits/chosen": -3.15625,
692
+ "logits/rejected": -3.203125,
693
+ "logps/chosen": -270.0,
694
+ "logps/rejected": -292.0,
695
+ "loss": 0.1471,
696
+ "rewards/accuracies": 0.8999999761581421,
697
+ "rewards/chosen": 0.71875,
698
+ "rewards/margins": 3.21875,
699
+ "rewards/rejected": -2.5,
700
+ "step": 440
701
+ },
702
+ {
703
+ "epoch": 1.4376996805111821,
704
+ "grad_norm": 15.707432621538732,
705
+ "learning_rate": 2.893491124260355e-07,
706
+ "logits/chosen": -3.203125,
707
+ "logits/rejected": -3.203125,
708
+ "logps/chosen": -268.0,
709
+ "logps/rejected": -296.0,
710
+ "loss": 0.093,
711
+ "rewards/accuracies": 0.9750000238418579,
712
+ "rewards/chosen": 0.8828125,
713
+ "rewards/margins": 3.59375,
714
+ "rewards/rejected": -2.703125,
715
+ "step": 450
716
+ },
717
+ {
718
+ "epoch": 1.4696485623003195,
719
+ "grad_norm": 28.76561180608655,
720
+ "learning_rate": 2.834319526627219e-07,
721
+ "logits/chosen": -3.1875,
722
+ "logits/rejected": -3.140625,
723
+ "logps/chosen": -264.0,
724
+ "logps/rejected": -300.0,
725
+ "loss": 0.1042,
726
+ "rewards/accuracies": 0.9750000238418579,
727
+ "rewards/chosen": 0.6328125,
728
+ "rewards/margins": 3.671875,
729
+ "rewards/rejected": -3.046875,
730
+ "step": 460
731
+ },
732
+ {
733
+ "epoch": 1.5015974440894568,
734
+ "grad_norm": 37.202582805520755,
735
+ "learning_rate": 2.7751479289940824e-07,
736
+ "logits/chosen": -3.15625,
737
+ "logits/rejected": -3.171875,
738
+ "logps/chosen": -268.0,
739
+ "logps/rejected": -302.0,
740
+ "loss": 0.0995,
741
+ "rewards/accuracies": 0.9624999761581421,
742
+ "rewards/chosen": 0.6796875,
743
+ "rewards/margins": 3.90625,
744
+ "rewards/rejected": -3.21875,
745
+ "step": 470
746
+ },
747
+ {
748
+ "epoch": 1.5335463258785942,
749
+ "grad_norm": 36.12380615792283,
750
+ "learning_rate": 2.715976331360947e-07,
751
+ "logits/chosen": -3.21875,
752
+ "logits/rejected": -3.078125,
753
+ "logps/chosen": -262.0,
754
+ "logps/rejected": -312.0,
755
+ "loss": 0.168,
756
+ "rewards/accuracies": 0.9125000238418579,
757
+ "rewards/chosen": 0.59765625,
758
+ "rewards/margins": 3.4375,
759
+ "rewards/rejected": -2.828125,
760
+ "step": 480
761
+ },
762
+ {
763
+ "epoch": 1.5654952076677318,
764
+ "grad_norm": 31.441359412649966,
765
+ "learning_rate": 2.6568047337278106e-07,
766
+ "logits/chosen": -3.203125,
767
+ "logits/rejected": -3.171875,
768
+ "logps/chosen": -266.0,
769
+ "logps/rejected": -296.0,
770
+ "loss": 0.1083,
771
+ "rewards/accuracies": 0.949999988079071,
772
+ "rewards/chosen": 0.494140625,
773
+ "rewards/margins": 3.625,
774
+ "rewards/rejected": -3.125,
775
+ "step": 490
776
+ },
777
+ {
778
+ "epoch": 1.5974440894568689,
779
+ "grad_norm": 56.69663253139224,
780
+ "learning_rate": 2.5976331360946746e-07,
781
+ "logits/chosen": -3.125,
782
+ "logits/rejected": -3.125,
783
+ "logps/chosen": -264.0,
784
+ "logps/rejected": -298.0,
785
+ "loss": 0.1199,
786
+ "rewards/accuracies": 0.9750000238418579,
787
+ "rewards/chosen": 0.80078125,
788
+ "rewards/margins": 3.859375,
789
+ "rewards/rejected": -3.0625,
790
+ "step": 500
791
+ },
792
+ {
793
+ "epoch": 1.6293929712460065,
794
+ "grad_norm": 42.26084157818348,
795
+ "learning_rate": 2.538461538461538e-07,
796
+ "logits/chosen": -3.140625,
797
+ "logits/rejected": -3.21875,
798
+ "logps/chosen": -268.0,
799
+ "logps/rejected": -284.0,
800
+ "loss": 0.1411,
801
+ "rewards/accuracies": 0.9125000238418579,
802
+ "rewards/chosen": 0.7421875,
803
+ "rewards/margins": 3.265625,
804
+ "rewards/rejected": -2.515625,
805
+ "step": 510
806
+ },
807
+ {
808
+ "epoch": 1.6613418530351438,
809
+ "grad_norm": 24.290936395597605,
810
+ "learning_rate": 2.479289940828402e-07,
811
+ "logits/chosen": -3.15625,
812
+ "logits/rejected": -3.125,
813
+ "logps/chosen": -264.0,
814
+ "logps/rejected": -294.0,
815
+ "loss": 0.0927,
816
+ "rewards/accuracies": 0.949999988079071,
817
+ "rewards/chosen": 0.74609375,
818
+ "rewards/margins": 3.609375,
819
+ "rewards/rejected": -2.859375,
820
+ "step": 520
821
+ },
822
+ {
823
+ "epoch": 1.6932907348242812,
824
+ "grad_norm": 30.097818527606982,
825
+ "learning_rate": 2.420118343195266e-07,
826
+ "logits/chosen": -3.125,
827
+ "logits/rejected": -3.125,
828
+ "logps/chosen": -268.0,
829
+ "logps/rejected": -294.0,
830
+ "loss": 0.132,
831
+ "rewards/accuracies": 0.9750000238418579,
832
+ "rewards/chosen": 0.9140625,
833
+ "rewards/margins": 3.96875,
834
+ "rewards/rejected": -3.0625,
835
+ "step": 530
836
+ },
837
+ {
838
+ "epoch": 1.7252396166134185,
839
+ "grad_norm": 25.610518794255373,
840
+ "learning_rate": 2.36094674556213e-07,
841
+ "logits/chosen": -3.09375,
842
+ "logits/rejected": -3.140625,
843
+ "logps/chosen": -270.0,
844
+ "logps/rejected": -298.0,
845
+ "loss": 0.135,
846
+ "rewards/accuracies": 0.9624999761581421,
847
+ "rewards/chosen": 0.5234375,
848
+ "rewards/margins": 3.796875,
849
+ "rewards/rejected": -3.28125,
850
+ "step": 540
851
+ },
852
+ {
853
+ "epoch": 1.7571884984025559,
854
+ "grad_norm": 32.86888420257977,
855
+ "learning_rate": 2.301775147928994e-07,
856
+ "logits/chosen": -3.1875,
857
+ "logits/rejected": -3.21875,
858
+ "logps/chosen": -268.0,
859
+ "logps/rejected": -294.0,
860
+ "loss": 0.1477,
861
+ "rewards/accuracies": 0.925000011920929,
862
+ "rewards/chosen": 0.921875,
863
+ "rewards/margins": 3.546875,
864
+ "rewards/rejected": -2.625,
865
+ "step": 550
866
+ },
867
+ {
868
+ "epoch": 1.7891373801916934,
869
+ "grad_norm": 38.670593041590834,
870
+ "learning_rate": 2.242603550295858e-07,
871
+ "logits/chosen": -3.09375,
872
+ "logits/rejected": -3.078125,
873
+ "logps/chosen": -272.0,
874
+ "logps/rejected": -300.0,
875
+ "loss": 0.0912,
876
+ "rewards/accuracies": 0.949999988079071,
877
+ "rewards/chosen": 0.66796875,
878
+ "rewards/margins": 3.734375,
879
+ "rewards/rejected": -3.0625,
880
+ "step": 560
881
+ },
882
+ {
883
+ "epoch": 1.8210862619808306,
884
+ "grad_norm": 30.06581434037245,
885
+ "learning_rate": 2.1834319526627217e-07,
886
+ "logits/chosen": -3.125,
887
+ "logits/rejected": -3.109375,
888
+ "logps/chosen": -266.0,
889
+ "logps/rejected": -312.0,
890
+ "loss": 0.1155,
891
+ "rewards/accuracies": 1.0,
892
+ "rewards/chosen": 0.81640625,
893
+ "rewards/margins": 4.03125,
894
+ "rewards/rejected": -3.21875,
895
+ "step": 570
896
+ },
897
+ {
898
+ "epoch": 1.8530351437699681,
899
+ "grad_norm": 37.09573180706117,
900
+ "learning_rate": 2.1242603550295858e-07,
901
+ "logits/chosen": -3.0625,
902
+ "logits/rejected": -3.0625,
903
+ "logps/chosen": -272.0,
904
+ "logps/rejected": -306.0,
905
+ "loss": 0.1263,
906
+ "rewards/accuracies": 0.949999988079071,
907
+ "rewards/chosen": 0.71484375,
908
+ "rewards/margins": 3.84375,
909
+ "rewards/rejected": -3.125,
910
+ "step": 580
911
+ },
912
+ {
913
+ "epoch": 1.8849840255591053,
914
+ "grad_norm": 38.562399882898944,
915
+ "learning_rate": 2.0650887573964496e-07,
916
+ "logits/chosen": -3.125,
917
+ "logits/rejected": -3.078125,
918
+ "logps/chosen": -258.0,
919
+ "logps/rejected": -302.0,
920
+ "loss": 0.1334,
921
+ "rewards/accuracies": 0.9375,
922
+ "rewards/chosen": 0.9921875,
923
+ "rewards/margins": 3.828125,
924
+ "rewards/rejected": -2.84375,
925
+ "step": 590
926
+ },
927
+ {
928
+ "epoch": 1.9169329073482428,
929
+ "grad_norm": 44.104454957376895,
930
+ "learning_rate": 2.0059171597633133e-07,
931
+ "logits/chosen": -3.140625,
932
+ "logits/rejected": -3.1875,
933
+ "logps/chosen": -270.0,
934
+ "logps/rejected": -298.0,
935
+ "loss": 0.1351,
936
+ "rewards/accuracies": 0.9750000238418579,
937
+ "rewards/chosen": 0.74609375,
938
+ "rewards/margins": 3.765625,
939
+ "rewards/rejected": -3.015625,
940
+ "step": 600
941
+ },
942
+ {
943
+ "epoch": 1.9488817891373802,
944
+ "grad_norm": 36.65323926007583,
945
+ "learning_rate": 1.9467455621301774e-07,
946
+ "logits/chosen": -3.15625,
947
+ "logits/rejected": -3.21875,
948
+ "logps/chosen": -266.0,
949
+ "logps/rejected": -296.0,
950
+ "loss": 0.1479,
951
+ "rewards/accuracies": 0.9375,
952
+ "rewards/chosen": 0.8046875,
953
+ "rewards/margins": 3.578125,
954
+ "rewards/rejected": -2.78125,
955
+ "step": 610
956
+ },
957
+ {
958
+ "epoch": 1.9808306709265175,
959
+ "grad_norm": 15.898706379798815,
960
+ "learning_rate": 1.8875739644970412e-07,
961
+ "logits/chosen": -3.09375,
962
+ "logits/rejected": -3.140625,
963
+ "logps/chosen": -272.0,
964
+ "logps/rejected": -302.0,
965
+ "loss": 0.11,
966
+ "rewards/accuracies": 0.9375,
967
+ "rewards/chosen": 0.3515625,
968
+ "rewards/margins": 3.75,
969
+ "rewards/rejected": -3.390625,
970
+ "step": 620
971
+ },
972
+ {
973
+ "epoch": 2.0,
974
+ "eval_logits/chosen": -3.125,
975
+ "eval_logits/rejected": -3.15625,
976
+ "eval_logps/chosen": -278.0,
977
+ "eval_logps/rejected": -282.0,
978
+ "eval_loss": 0.6153125166893005,
979
+ "eval_rewards/accuracies": 0.6428571343421936,
980
+ "eval_rewards/chosen": -0.5234375,
981
+ "eval_rewards/margins": 1.0078125,
982
+ "eval_rewards/rejected": -1.53125,
983
+ "eval_runtime": 12.4752,
984
+ "eval_samples_per_second": 16.032,
985
+ "eval_steps_per_second": 0.561,
986
+ "step": 626
987
  }
988
  ],
989
  "logging_steps": 10,