Hugging Face's logo

AmberYifan
/
Llama-3.1-8B-Mistral-7B-v0.3-mix

Text Generation
Transformers
Safetensors
llama
Generated from Trainer
trl
dpo
conversational
text-generation-inference
Inference Endpoints
Model card Files Community
AmberYifan commited on
Commit
421b0e4
·
verified ·
1 Parent(s): d5fb442

Training in progress, epoch 2, checkpoint

Browse files
Files changed (19) hide show
  1. last-checkpoint/global_step626/bf16_zero_pp_rank_0_mp_rank_00_optim_states.pt +3 -0
  2. last-checkpoint/global_step626/bf16_zero_pp_rank_1_mp_rank_00_optim_states.pt +3 -0
  3. last-checkpoint/global_step626/bf16_zero_pp_rank_2_mp_rank_00_optim_states.pt +3 -0
  4. last-checkpoint/global_step626/bf16_zero_pp_rank_3_mp_rank_00_optim_states.pt +3 -0
  5. last-checkpoint/global_step626/zero_pp_rank_0_mp_rank_00_model_states.pt +3 -0
  6. last-checkpoint/global_step626/zero_pp_rank_1_mp_rank_00_model_states.pt +3 -0
  7. last-checkpoint/global_step626/zero_pp_rank_2_mp_rank_00_model_states.pt +3 -0
  8. last-checkpoint/global_step626/zero_pp_rank_3_mp_rank_00_model_states.pt +3 -0
  9. last-checkpoint/latest +1 -1
  10. last-checkpoint/model-00001-of-00004.safetensors +1 -1
  11. last-checkpoint/model-00002-of-00004.safetensors +1 -1
  12. last-checkpoint/model-00003-of-00004.safetensors +1 -1
  13. last-checkpoint/model-00004-of-00004.safetensors +1 -1
  14. last-checkpoint/rng_state_0.pth +1 -1
  15. last-checkpoint/rng_state_1.pth +1 -1
  16. last-checkpoint/rng_state_2.pth +1 -1
  17. last-checkpoint/rng_state_3.pth +1 -1
  18. last-checkpoint/scheduler.pt +1 -1
  19. last-checkpoint/trainer_state.json +483 -2
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:cfe9031ffc475de2d5dc048592b2d9e4197457f03313897972bcb52f168eb1df
3
+ size 16060527788
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 sha256:351214b25b8f081935867a9ec7b8e192326d8590b00cfd1771d10d323d1899ab
3
+ size 16060527788
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 sha256:695064794e9af28979681eedce6240ebb1cd7e3376f4e02396b07d428bc02c7e
3
+ size 16060527788
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 sha256:4b5da00f14fbc3eb541a3d56225743d0835fcc37e1e9958d772fd7f5e953ab2f
3
+ size 16060527788
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:abec8762cd151640ecff6bec742b3029077d87c015cc1e6ff2c99913a18e34ac
3
+ size 150693
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:e861607615a2f30cc4a91224dab49c029d4ca20f8e79a03bac84dd4582db1386
3
+ size 150693
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:d3862a1ed2cf9d6e26e9d0d0f03033c2ce320801e6d9f5171504fa1c7896644a
3
+ size 150693
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 sha256:7e6cb6c96bef29946c7aa624f6f1756330136f0b4c04eed1561baaf57b75ea19
3
+ size 150693
last-checkpoint/latest CHANGED
@@ -1 +1 @@
1
- global_step313
 
1
+ global_step626
last-checkpoint/model-00001-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2e43ccfbf09c4ed8af08feda5bcae4e3997daf4d9e193ea0455aea2851ea65f1
3
  size 4976698672
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8e326a1dbd8fc14b10637832fdeb8d0c03940664a8c4f58f99100b04f96138e8
3
  size 4976698672
last-checkpoint/model-00002-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:091f8e1bd9cba29d3dd77cf70431df45558323bcb81475f0cbe3939b90192125
3
  size 4999802720
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd4ce9e4ed9d86334b47f18822649eb3f59b7a5fd11b4110414b589b19b2233e
3
  size 4999802720
last-checkpoint/model-00003-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:0e54fac96c6aa2e0495935eac8ca6ffff6770cdc037c904f9cb5bd493f787da8
3
  size 4915916176
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1b18151b5ab7ae4151110e5c07fedb062428d352acd2bb8b2f6d66b6153fcaa0
3
  size 4915916176
last-checkpoint/model-00004-of-00004.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:a2262b675f2d413afb23fc20133126e1ef385833b70462ddbd7a17b8eb9cd4d0
3
  size 1168138808
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8aef27c1aab1f1cdcc79db0469c18cc56b552142b9f4a6a7c33f91ef67626901
3
  size 1168138808
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 sha256:6b9aad3a748e15719e20120f0a760f22686338e176ee3f1c011b6dab2f3ab19b
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:f7725b44c69725321786f1e58dccd7ea4d3ae5794ea47bd49c0d4a139aec266d
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": 15.771,
504
  "eval_steps_per_second": 0.552,
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": 15.771,
504
  "eval_steps_per_second": 0.552,
505
  "step": 313
506
+ },
507
+ {
508
+ "epoch": 1.0223642172523961,
509
+ "grad_norm": 23.23800693350808,
510
+ "learning_rate": 3.662721893491124e-07,
511
+ "logits/chosen": -1.7265625,
512
+ "logits/rejected": -1.6875,
513
+ "logps/chosen": -238.0,
514
+ "logps/rejected": -266.0,
515
+ "loss": 0.365,
516
+ "rewards/accuracies": 0.9125000238418579,
517
+ "rewards/chosen": 0.012451171875,
518
+ "rewards/margins": 1.2734375,
519
+ "rewards/rejected": -1.265625,
520
+ "step": 320
521
+ },
522
+ {
523
+ "epoch": 1.0543130990415335,
524
+ "grad_norm": 28.085201782352296,
525
+ "learning_rate": 3.603550295857988e-07,
526
+ "logits/chosen": -1.7109375,
527
+ "logits/rejected": -1.734375,
528
+ "logps/chosen": -242.0,
529
+ "logps/rejected": -276.0,
530
+ "loss": 0.3465,
531
+ "rewards/accuracies": 0.862500011920929,
532
+ "rewards/chosen": -0.00872802734375,
533
+ "rewards/margins": 1.46875,
534
+ "rewards/rejected": -1.4765625,
535
+ "step": 330
536
+ },
537
+ {
538
+ "epoch": 1.0862619808306708,
539
+ "grad_norm": 24.247027348577692,
540
+ "learning_rate": 3.544378698224852e-07,
541
+ "logits/chosen": -1.734375,
542
+ "logits/rejected": -1.6953125,
543
+ "logps/chosen": -238.0,
544
+ "logps/rejected": -278.0,
545
+ "loss": 0.2981,
546
+ "rewards/accuracies": 0.862500011920929,
547
+ "rewards/chosen": 0.1435546875,
548
+ "rewards/margins": 1.4453125,
549
+ "rewards/rejected": -1.3046875,
550
+ "step": 340
551
+ },
552
+ {
553
+ "epoch": 1.1182108626198084,
554
+ "grad_norm": 21.346684851723847,
555
+ "learning_rate": 3.485207100591716e-07,
556
+ "logits/chosen": -1.7109375,
557
+ "logits/rejected": -1.6953125,
558
+ "logps/chosen": -239.0,
559
+ "logps/rejected": -274.0,
560
+ "loss": 0.2997,
561
+ "rewards/accuracies": 0.8500000238418579,
562
+ "rewards/chosen": -0.12158203125,
563
+ "rewards/margins": 1.453125,
564
+ "rewards/rejected": -1.5703125,
565
+ "step": 350
566
+ },
567
+ {
568
+ "epoch": 1.1501597444089458,
569
+ "grad_norm": 22.92923404174748,
570
+ "learning_rate": 3.42603550295858e-07,
571
+ "logits/chosen": -1.734375,
572
+ "logits/rejected": -1.7109375,
573
+ "logps/chosen": -239.0,
574
+ "logps/rejected": -278.0,
575
+ "loss": 0.2966,
576
+ "rewards/accuracies": 0.875,
577
+ "rewards/chosen": 0.197265625,
578
+ "rewards/margins": 1.6328125,
579
+ "rewards/rejected": -1.4296875,
580
+ "step": 360
581
+ },
582
+ {
583
+ "epoch": 1.182108626198083,
584
+ "grad_norm": 19.074914011714277,
585
+ "learning_rate": 3.366863905325444e-07,
586
+ "logits/chosen": -1.71875,
587
+ "logits/rejected": -1.734375,
588
+ "logps/chosen": -240.0,
589
+ "logps/rejected": -278.0,
590
+ "loss": 0.3032,
591
+ "rewards/accuracies": 0.862500011920929,
592
+ "rewards/chosen": 0.2451171875,
593
+ "rewards/margins": 1.78125,
594
+ "rewards/rejected": -1.5390625,
595
+ "step": 370
596
+ },
597
+ {
598
+ "epoch": 1.2140575079872205,
599
+ "grad_norm": 24.568535159721062,
600
+ "learning_rate": 3.3076923076923075e-07,
601
+ "logits/chosen": -1.734375,
602
+ "logits/rejected": -1.7109375,
603
+ "logps/chosen": -239.0,
604
+ "logps/rejected": -292.0,
605
+ "loss": 0.2915,
606
+ "rewards/accuracies": 0.9125000238418579,
607
+ "rewards/chosen": 0.047607421875,
608
+ "rewards/margins": 1.71875,
609
+ "rewards/rejected": -1.671875,
610
+ "step": 380
611
+ },
612
+ {
613
+ "epoch": 1.2460063897763578,
614
+ "grad_norm": 27.96077981193079,
615
+ "learning_rate": 3.2485207100591716e-07,
616
+ "logits/chosen": -1.71875,
617
+ "logits/rejected": -1.6953125,
618
+ "logps/chosen": -239.0,
619
+ "logps/rejected": -280.0,
620
+ "loss": 0.3166,
621
+ "rewards/accuracies": 0.925000011920929,
622
+ "rewards/chosen": -0.12158203125,
623
+ "rewards/margins": 1.6796875,
624
+ "rewards/rejected": -1.796875,
625
+ "step": 390
626
+ },
627
+ {
628
+ "epoch": 1.2779552715654952,
629
+ "grad_norm": 23.939263929977898,
630
+ "learning_rate": 3.189349112426035e-07,
631
+ "logits/chosen": -1.7421875,
632
+ "logits/rejected": -1.703125,
633
+ "logps/chosen": -247.0,
634
+ "logps/rejected": -278.0,
635
+ "loss": 0.3313,
636
+ "rewards/accuracies": 0.8500000238418579,
637
+ "rewards/chosen": -0.291015625,
638
+ "rewards/margins": 1.5,
639
+ "rewards/rejected": -1.7890625,
640
+ "step": 400
641
+ },
642
+ {
643
+ "epoch": 1.3099041533546325,
644
+ "grad_norm": 23.445460332455205,
645
+ "learning_rate": 3.130177514792899e-07,
646
+ "logits/chosen": -1.765625,
647
+ "logits/rejected": -1.7265625,
648
+ "logps/chosen": -244.0,
649
+ "logps/rejected": -276.0,
650
+ "loss": 0.29,
651
+ "rewards/accuracies": 0.8500000238418579,
652
+ "rewards/chosen": -0.322265625,
653
+ "rewards/margins": 1.59375,
654
+ "rewards/rejected": -1.9140625,
655
+ "step": 410
656
+ },
657
+ {
658
+ "epoch": 1.34185303514377,
659
+ "grad_norm": 15.133152781473944,
660
+ "learning_rate": 3.071005917159763e-07,
661
+ "logits/chosen": -1.71875,
662
+ "logits/rejected": -1.734375,
663
+ "logps/chosen": -250.0,
664
+ "logps/rejected": -278.0,
665
+ "loss": 0.2755,
666
+ "rewards/accuracies": 0.875,
667
+ "rewards/chosen": -0.322265625,
668
+ "rewards/margins": 1.546875,
669
+ "rewards/rejected": -1.8671875,
670
+ "step": 420
671
+ },
672
+ {
673
+ "epoch": 1.3738019169329074,
674
+ "grad_norm": 19.11123594038131,
675
+ "learning_rate": 3.011834319526627e-07,
676
+ "logits/chosen": -1.765625,
677
+ "logits/rejected": -1.7421875,
678
+ "logps/chosen": -245.0,
679
+ "logps/rejected": -274.0,
680
+ "loss": 0.2506,
681
+ "rewards/accuracies": 0.862500011920929,
682
+ "rewards/chosen": -0.10009765625,
683
+ "rewards/margins": 1.8203125,
684
+ "rewards/rejected": -1.921875,
685
+ "step": 430
686
+ },
687
+ {
688
+ "epoch": 1.4057507987220448,
689
+ "grad_norm": 17.795524020323455,
690
+ "learning_rate": 2.952662721893491e-07,
691
+ "logits/chosen": -1.7578125,
692
+ "logits/rejected": -1.7421875,
693
+ "logps/chosen": -243.0,
694
+ "logps/rejected": -286.0,
695
+ "loss": 0.2583,
696
+ "rewards/accuracies": 0.8999999761581421,
697
+ "rewards/chosen": -0.0123291015625,
698
+ "rewards/margins": 1.8046875,
699
+ "rewards/rejected": -1.8125,
700
+ "step": 440
701
+ },
702
+ {
703
+ "epoch": 1.4376996805111821,
704
+ "grad_norm": 23.74400717983665,
705
+ "learning_rate": 2.893491124260355e-07,
706
+ "logits/chosen": -1.734375,
707
+ "logits/rejected": -1.7109375,
708
+ "logps/chosen": -241.0,
709
+ "logps/rejected": -288.0,
710
+ "loss": 0.2642,
711
+ "rewards/accuracies": 0.9125000238418579,
712
+ "rewards/chosen": -0.03369140625,
713
+ "rewards/margins": 1.9453125,
714
+ "rewards/rejected": -1.984375,
715
+ "step": 450
716
+ },
717
+ {
718
+ "epoch": 1.4696485623003195,
719
+ "grad_norm": 28.078711225527382,
720
+ "learning_rate": 2.834319526627219e-07,
721
+ "logits/chosen": -1.7265625,
722
+ "logits/rejected": -1.734375,
723
+ "logps/chosen": -240.0,
724
+ "logps/rejected": -274.0,
725
+ "loss": 0.2679,
726
+ "rewards/accuracies": 0.862500011920929,
727
+ "rewards/chosen": 0.1669921875,
728
+ "rewards/margins": 1.734375,
729
+ "rewards/rejected": -1.5703125,
730
+ "step": 460
731
+ },
732
+ {
733
+ "epoch": 1.5015974440894568,
734
+ "grad_norm": 22.708510750866107,
735
+ "learning_rate": 2.7751479289940824e-07,
736
+ "logits/chosen": -1.734375,
737
+ "logits/rejected": -1.7265625,
738
+ "logps/chosen": -242.0,
739
+ "logps/rejected": -268.0,
740
+ "loss": 0.2787,
741
+ "rewards/accuracies": 0.887499988079071,
742
+ "rewards/chosen": 0.0673828125,
743
+ "rewards/margins": 1.5703125,
744
+ "rewards/rejected": -1.5,
745
+ "step": 470
746
+ },
747
+ {
748
+ "epoch": 1.5335463258785942,
749
+ "grad_norm": 18.86680359426442,
750
+ "learning_rate": 2.715976331360947e-07,
751
+ "logits/chosen": -1.734375,
752
+ "logits/rejected": -1.7421875,
753
+ "logps/chosen": -241.0,
754
+ "logps/rejected": -288.0,
755
+ "loss": 0.2716,
756
+ "rewards/accuracies": 0.8500000238418579,
757
+ "rewards/chosen": 0.1298828125,
758
+ "rewards/margins": 1.7578125,
759
+ "rewards/rejected": -1.625,
760
+ "step": 480
761
+ },
762
+ {
763
+ "epoch": 1.5654952076677318,
764
+ "grad_norm": 17.156329445818553,
765
+ "learning_rate": 2.6568047337278106e-07,
766
+ "logits/chosen": -1.71875,
767
+ "logits/rejected": -1.7578125,
768
+ "logps/chosen": -237.0,
769
+ "logps/rejected": -276.0,
770
+ "loss": 0.3079,
771
+ "rewards/accuracies": 0.862500011920929,
772
+ "rewards/chosen": 0.0712890625,
773
+ "rewards/margins": 1.75,
774
+ "rewards/rejected": -1.6796875,
775
+ "step": 490
776
+ },
777
+ {
778
+ "epoch": 1.5974440894568689,
779
+ "grad_norm": 19.332118400424566,
780
+ "learning_rate": 2.5976331360946746e-07,
781
+ "logits/chosen": -1.7265625,
782
+ "logits/rejected": -1.6953125,
783
+ "logps/chosen": -243.0,
784
+ "logps/rejected": -280.0,
785
+ "loss": 0.2851,
786
+ "rewards/accuracies": 0.875,
787
+ "rewards/chosen": 0.080078125,
788
+ "rewards/margins": 1.7734375,
789
+ "rewards/rejected": -1.6953125,
790
+ "step": 500
791
+ },
792
+ {
793
+ "epoch": 1.6293929712460065,
794
+ "grad_norm": 19.62634703131097,
795
+ "learning_rate": 2.538461538461538e-07,
796
+ "logits/chosen": -1.7421875,
797
+ "logits/rejected": -1.734375,
798
+ "logps/chosen": -239.0,
799
+ "logps/rejected": -272.0,
800
+ "loss": 0.2776,
801
+ "rewards/accuracies": 0.887499988079071,
802
+ "rewards/chosen": 0.030029296875,
803
+ "rewards/margins": 1.5546875,
804
+ "rewards/rejected": -1.5234375,
805
+ "step": 510
806
+ },
807
+ {
808
+ "epoch": 1.6613418530351438,
809
+ "grad_norm": 30.286972397278998,
810
+ "learning_rate": 2.479289940828402e-07,
811
+ "logits/chosen": -1.7109375,
812
+ "logits/rejected": -1.6875,
813
+ "logps/chosen": -239.0,
814
+ "logps/rejected": -272.0,
815
+ "loss": 0.2678,
816
+ "rewards/accuracies": 0.875,
817
+ "rewards/chosen": 0.14453125,
818
+ "rewards/margins": 1.953125,
819
+ "rewards/rejected": -1.8125,
820
+ "step": 520
821
+ },
822
+ {
823
+ "epoch": 1.6932907348242812,
824
+ "grad_norm": 26.68217050843353,
825
+ "learning_rate": 2.420118343195266e-07,
826
+ "logits/chosen": -1.75,
827
+ "logits/rejected": -1.7734375,
828
+ "logps/chosen": -234.0,
829
+ "logps/rejected": -272.0,
830
+ "loss": 0.2765,
831
+ "rewards/accuracies": 0.887499988079071,
832
+ "rewards/chosen": 0.404296875,
833
+ "rewards/margins": 1.9140625,
834
+ "rewards/rejected": -1.5078125,
835
+ "step": 530
836
+ },
837
+ {
838
+ "epoch": 1.7252396166134185,
839
+ "grad_norm": 22.03681852867734,
840
+ "learning_rate": 2.36094674556213e-07,
841
+ "logits/chosen": -1.734375,
842
+ "logits/rejected": -1.7421875,
843
+ "logps/chosen": -243.0,
844
+ "logps/rejected": -276.0,
845
+ "loss": 0.2823,
846
+ "rewards/accuracies": 0.8374999761581421,
847
+ "rewards/chosen": -0.09716796875,
848
+ "rewards/margins": 1.7421875,
849
+ "rewards/rejected": -1.8359375,
850
+ "step": 540
851
+ },
852
+ {
853
+ "epoch": 1.7571884984025559,
854
+ "grad_norm": 19.88206815145969,
855
+ "learning_rate": 2.301775147928994e-07,
856
+ "logits/chosen": -1.71875,
857
+ "logits/rejected": -1.7265625,
858
+ "logps/chosen": -237.0,
859
+ "logps/rejected": -274.0,
860
+ "loss": 0.258,
861
+ "rewards/accuracies": 0.8999999761581421,
862
+ "rewards/chosen": -0.0025634765625,
863
+ "rewards/margins": 1.796875,
864
+ "rewards/rejected": -1.796875,
865
+ "step": 550
866
+ },
867
+ {
868
+ "epoch": 1.7891373801916934,
869
+ "grad_norm": 24.819444792106694,
870
+ "learning_rate": 2.242603550295858e-07,
871
+ "logits/chosen": -1.7578125,
872
+ "logits/rejected": -1.7109375,
873
+ "logps/chosen": -242.0,
874
+ "logps/rejected": -282.0,
875
+ "loss": 0.292,
876
+ "rewards/accuracies": 0.862500011920929,
877
+ "rewards/chosen": 0.2470703125,
878
+ "rewards/margins": 1.9765625,
879
+ "rewards/rejected": -1.7265625,
880
+ "step": 560
881
+ },
882
+ {
883
+ "epoch": 1.8210862619808306,
884
+ "grad_norm": 25.1506153366333,
885
+ "learning_rate": 2.1834319526627217e-07,
886
+ "logits/chosen": -1.7265625,
887
+ "logits/rejected": -1.71875,
888
+ "logps/chosen": -235.0,
889
+ "logps/rejected": -280.0,
890
+ "loss": 0.3011,
891
+ "rewards/accuracies": 0.949999988079071,
892
+ "rewards/chosen": 0.24609375,
893
+ "rewards/margins": 1.921875,
894
+ "rewards/rejected": -1.6796875,
895
+ "step": 570
896
+ },
897
+ {
898
+ "epoch": 1.8530351437699681,
899
+ "grad_norm": 25.408870033012057,
900
+ "learning_rate": 2.1242603550295858e-07,
901
+ "logits/chosen": -1.734375,
902
+ "logits/rejected": -1.7421875,
903
+ "logps/chosen": -243.0,
904
+ "logps/rejected": -284.0,
905
+ "loss": 0.2759,
906
+ "rewards/accuracies": 0.875,
907
+ "rewards/chosen": 0.060302734375,
908
+ "rewards/margins": 1.953125,
909
+ "rewards/rejected": -1.8984375,
910
+ "step": 580
911
+ },
912
+ {
913
+ "epoch": 1.8849840255591053,
914
+ "grad_norm": 30.243845189152044,
915
+ "learning_rate": 2.0650887573964496e-07,
916
+ "logits/chosen": -1.78125,
917
+ "logits/rejected": -1.75,
918
+ "logps/chosen": -248.0,
919
+ "logps/rejected": -288.0,
920
+ "loss": 0.3033,
921
+ "rewards/accuracies": 0.875,
922
+ "rewards/chosen": -0.072265625,
923
+ "rewards/margins": 1.8203125,
924
+ "rewards/rejected": -1.890625,
925
+ "step": 590
926
+ },
927
+ {
928
+ "epoch": 1.9169329073482428,
929
+ "grad_norm": 25.522451943807717,
930
+ "learning_rate": 2.0059171597633133e-07,
931
+ "logits/chosen": -1.734375,
932
+ "logits/rejected": -1.765625,
933
+ "logps/chosen": -236.0,
934
+ "logps/rejected": -282.0,
935
+ "loss": 0.2878,
936
+ "rewards/accuracies": 0.8999999761581421,
937
+ "rewards/chosen": 0.2021484375,
938
+ "rewards/margins": 1.9296875,
939
+ "rewards/rejected": -1.7265625,
940
+ "step": 600
941
+ },
942
+ {
943
+ "epoch": 1.9488817891373802,
944
+ "grad_norm": 25.361411149259393,
945
+ "learning_rate": 1.9467455621301774e-07,
946
+ "logits/chosen": -1.734375,
947
+ "logits/rejected": -1.734375,
948
+ "logps/chosen": -241.0,
949
+ "logps/rejected": -282.0,
950
+ "loss": 0.2881,
951
+ "rewards/accuracies": 0.887499988079071,
952
+ "rewards/chosen": 0.11767578125,
953
+ "rewards/margins": 1.8046875,
954
+ "rewards/rejected": -1.6953125,
955
+ "step": 610
956
+ },
957
+ {
958
+ "epoch": 1.9808306709265175,
959
+ "grad_norm": 18.620086897639467,
960
+ "learning_rate": 1.8875739644970412e-07,
961
+ "logits/chosen": -1.703125,
962
+ "logits/rejected": -1.703125,
963
+ "logps/chosen": -243.0,
964
+ "logps/rejected": -278.0,
965
+ "loss": 0.275,
966
+ "rewards/accuracies": 0.8999999761581421,
967
+ "rewards/chosen": -0.072265625,
968
+ "rewards/margins": 1.5859375,
969
+ "rewards/rejected": -1.6640625,
970
+ "step": 620
971
+ },
972
+ {
973
+ "epoch": 2.0,
974
+ "eval_logits/chosen": -1.703125,
975
+ "eval_logits/rejected": -1.7421875,
976
+ "eval_logps/chosen": -240.0,
977
+ "eval_logps/rejected": -280.0,
978
+ "eval_loss": 0.5277343988418579,
979
+ "eval_rewards/accuracies": 0.7678571343421936,
980
+ "eval_rewards/chosen": -0.37109375,
981
+ "eval_rewards/margins": 0.859375,
982
+ "eval_rewards/rejected": -1.234375,
983
+ "eval_runtime": 12.6,
984
+ "eval_samples_per_second": 15.873,
985
+ "eval_steps_per_second": 0.556,
986
+ "step": 626
987
  }
988
  ],
989
  "logging_steps": 10,