k-r-l commited on
Commit
5593a5a
1 Parent(s): 6ec12e3

Training in progress, step 62, checkpoint

Browse files
last-checkpoint/adapter_config.json CHANGED
@@ -20,8 +20,8 @@
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
23
- "v_proj",
24
- "q_proj"
25
  ],
26
  "task_type": "CAUSAL_LM",
27
  "use_dora": false,
 
20
  "rank_pattern": {},
21
  "revision": null,
22
  "target_modules": [
23
+ "q_proj",
24
+ "v_proj"
25
  ],
26
  "task_type": "CAUSAL_LM",
27
  "use_dora": false,
last-checkpoint/adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:c7da29b3c527f6adfa53cbea2d9f62ad4836f417c6e209d1b791700873e15b72
3
  size 13648432
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:47a08a6e864e2ee1968177f93a78f8afab9ebca87fe279fae45b39b1d4790815
3
  size 13648432
last-checkpoint/optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:9e842bafd437b9c9bd87fe315b19b648d03282bedf2a026f6d1e5f6384398121
3
  size 7309882
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:7711dccfe7177717c6c0d73a485a5f5d766f81d9dbdd88ae79b64c25ed23c69e
3
  size 7309882
last-checkpoint/scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:078374aaaf0ef6477f920c224c32134039e8f06ec3f7925b56035828b5c37d05
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:46e4c05515a68694cd891bd31cda7371a27ef3d19d366250cc2850e7de57780a
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": 0.23439490445859873,
5
  "eval_steps": 500,
6
- "global_step": 23,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -421,6 +421,708 @@
421
  "rewards/margins": 0.010525593534111977,
422
  "rewards/rejected": -0.21030855178833008,
423
  "step": 23
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
424
  }
425
  ],
426
  "logging_steps": 1,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 0.6318471337579618,
5
  "eval_steps": 500,
6
+ "global_step": 62,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
421
  "rewards/margins": 0.010525593534111977,
422
  "rewards/rejected": -0.21030855178833008,
423
  "step": 23
424
+ },
425
+ {
426
+ "epoch": 0.2445859872611465,
427
+ "grad_norm": 2.3382904529571533,
428
+ "learning_rate": 9.183673469387756e-05,
429
+ "log_odds_chosen": 0.043098174035549164,
430
+ "log_odds_ratio": -0.675950825214386,
431
+ "logits/chosen": -0.7050585746765137,
432
+ "logits/rejected": -0.6946679949760437,
433
+ "logps/chosen": -1.8801963329315186,
434
+ "logps/rejected": -1.9172497987747192,
435
+ "loss": 2.0647,
436
+ "nll_loss": 1.9970893859863281,
437
+ "rewards/accuracies": 0.53125,
438
+ "rewards/chosen": -0.18801963329315186,
439
+ "rewards/margins": 0.003705363254994154,
440
+ "rewards/rejected": -0.1917250007390976,
441
+ "step": 24
442
+ },
443
+ {
444
+ "epoch": 0.25477707006369427,
445
+ "grad_norm": 2.2423338890075684,
446
+ "learning_rate": 9.149659863945579e-05,
447
+ "log_odds_chosen": 0.051533281803131104,
448
+ "log_odds_ratio": -0.6742590665817261,
449
+ "logits/chosen": -0.7378531694412231,
450
+ "logits/rejected": -0.7483745217323303,
451
+ "logps/chosen": -1.7976934909820557,
452
+ "logps/rejected": -1.8392703533172607,
453
+ "loss": 1.9856,
454
+ "nll_loss": 1.9182195663452148,
455
+ "rewards/accuracies": 0.5625,
456
+ "rewards/chosen": -0.1797693371772766,
457
+ "rewards/margins": 0.004157697316259146,
458
+ "rewards/rejected": -0.1839270442724228,
459
+ "step": 25
460
+ },
461
+ {
462
+ "epoch": 0.26496815286624203,
463
+ "grad_norm": 2.2076170444488525,
464
+ "learning_rate": 9.115646258503402e-05,
465
+ "log_odds_chosen": 0.14921554923057556,
466
+ "log_odds_ratio": -0.6260839700698853,
467
+ "logits/chosen": -0.7678406834602356,
468
+ "logits/rejected": -0.7855570316314697,
469
+ "logps/chosen": -1.690258264541626,
470
+ "logps/rejected": -1.813413143157959,
471
+ "loss": 1.8961,
472
+ "nll_loss": 1.833521842956543,
473
+ "rewards/accuracies": 0.78125,
474
+ "rewards/chosen": -0.16902580857276917,
475
+ "rewards/margins": 0.012315504252910614,
476
+ "rewards/rejected": -0.18134130537509918,
477
+ "step": 26
478
+ },
479
+ {
480
+ "epoch": 0.2751592356687898,
481
+ "grad_norm": 2.486793279647827,
482
+ "learning_rate": 9.081632653061225e-05,
483
+ "log_odds_chosen": 0.15732049942016602,
484
+ "log_odds_ratio": -0.6271807551383972,
485
+ "logits/chosen": -0.7320041060447693,
486
+ "logits/rejected": -0.7608926296234131,
487
+ "logps/chosen": -1.6371859312057495,
488
+ "logps/rejected": -1.7642821073532104,
489
+ "loss": 1.8287,
490
+ "nll_loss": 1.7660211324691772,
491
+ "rewards/accuracies": 0.6875,
492
+ "rewards/chosen": -0.163718581199646,
493
+ "rewards/margins": 0.01270962879061699,
494
+ "rewards/rejected": -0.17642822861671448,
495
+ "step": 27
496
+ },
497
+ {
498
+ "epoch": 0.28535031847133757,
499
+ "grad_norm": 2.56262469291687,
500
+ "learning_rate": 9.047619047619048e-05,
501
+ "log_odds_chosen": 0.16445228457450867,
502
+ "log_odds_ratio": -0.6198588609695435,
503
+ "logits/chosen": -0.7562386989593506,
504
+ "logits/rejected": -0.7781034708023071,
505
+ "logps/chosen": -1.5056514739990234,
506
+ "logps/rejected": -1.634886622428894,
507
+ "loss": 1.7373,
508
+ "nll_loss": 1.6753149032592773,
509
+ "rewards/accuracies": 0.6875,
510
+ "rewards/chosen": -0.15056513249874115,
511
+ "rewards/margins": 0.012923511676490307,
512
+ "rewards/rejected": -0.16348864138126373,
513
+ "step": 28
514
+ },
515
+ {
516
+ "epoch": 0.29554140127388534,
517
+ "grad_norm": 2.2155702114105225,
518
+ "learning_rate": 9.01360544217687e-05,
519
+ "log_odds_chosen": 0.07792021334171295,
520
+ "log_odds_ratio": -0.6598246693611145,
521
+ "logits/chosen": -0.7097214460372925,
522
+ "logits/rejected": -0.7517991065979004,
523
+ "logps/chosen": -1.4955050945281982,
524
+ "logps/rejected": -1.5548988580703735,
525
+ "loss": 1.7355,
526
+ "nll_loss": 1.669482707977295,
527
+ "rewards/accuracies": 0.71875,
528
+ "rewards/chosen": -0.14955051243305206,
529
+ "rewards/margins": 0.005939368158578873,
530
+ "rewards/rejected": -0.15548989176750183,
531
+ "step": 29
532
+ },
533
+ {
534
+ "epoch": 0.3057324840764331,
535
+ "grad_norm": 2.1764378547668457,
536
+ "learning_rate": 8.979591836734695e-05,
537
+ "log_odds_chosen": 0.03990812599658966,
538
+ "log_odds_ratio": -0.6799441576004028,
539
+ "logits/chosen": -0.7283363938331604,
540
+ "logits/rejected": -0.7530458569526672,
541
+ "logps/chosen": -1.4264872074127197,
542
+ "logps/rejected": -1.4550918340682983,
543
+ "loss": 1.6459,
544
+ "nll_loss": 1.5778682231903076,
545
+ "rewards/accuracies": 0.625,
546
+ "rewards/chosen": -0.14264871180057526,
547
+ "rewards/margins": 0.0028604594990611076,
548
+ "rewards/rejected": -0.14550916850566864,
549
+ "step": 30
550
+ },
551
+ {
552
+ "epoch": 0.3159235668789809,
553
+ "grad_norm": 2.089293956756592,
554
+ "learning_rate": 8.945578231292518e-05,
555
+ "log_odds_chosen": 0.2132435142993927,
556
+ "log_odds_ratio": -0.5990588068962097,
557
+ "logits/chosen": -0.7453004121780396,
558
+ "logits/rejected": -0.7864771485328674,
559
+ "logps/chosen": -1.330241322517395,
560
+ "logps/rejected": -1.4900933504104614,
561
+ "loss": 1.5747,
562
+ "nll_loss": 1.5148346424102783,
563
+ "rewards/accuracies": 0.75,
564
+ "rewards/chosen": -0.1330241560935974,
565
+ "rewards/margins": 0.01598520018160343,
566
+ "rewards/rejected": -0.1490093469619751,
567
+ "step": 31
568
+ },
569
+ {
570
+ "epoch": 0.32611464968152865,
571
+ "grad_norm": 1.9524860382080078,
572
+ "learning_rate": 8.911564625850341e-05,
573
+ "log_odds_chosen": 0.07104514539241791,
574
+ "log_odds_ratio": -0.6634508371353149,
575
+ "logits/chosen": -0.7562890648841858,
576
+ "logits/rejected": -0.762592077255249,
577
+ "logps/chosen": -1.3261951208114624,
578
+ "logps/rejected": -1.3763710260391235,
579
+ "loss": 1.5646,
580
+ "nll_loss": 1.4982781410217285,
581
+ "rewards/accuracies": 0.625,
582
+ "rewards/chosen": -0.13261951506137848,
583
+ "rewards/margins": 0.005017599556595087,
584
+ "rewards/rejected": -0.13763710856437683,
585
+ "step": 32
586
+ },
587
+ {
588
+ "epoch": 0.3363057324840764,
589
+ "grad_norm": 1.859688401222229,
590
+ "learning_rate": 8.877551020408164e-05,
591
+ "log_odds_chosen": 0.2007061243057251,
592
+ "log_odds_ratio": -0.6077967882156372,
593
+ "logits/chosen": -0.7153605222702026,
594
+ "logits/rejected": -0.7660464644432068,
595
+ "logps/chosen": -1.1958274841308594,
596
+ "logps/rejected": -1.3376389741897583,
597
+ "loss": 1.473,
598
+ "nll_loss": 1.4121888875961304,
599
+ "rewards/accuracies": 0.71875,
600
+ "rewards/chosen": -0.11958275735378265,
601
+ "rewards/margins": 0.01418115571141243,
602
+ "rewards/rejected": -0.13376390933990479,
603
+ "step": 33
604
+ },
605
+ {
606
+ "epoch": 0.3464968152866242,
607
+ "grad_norm": 1.627384901046753,
608
+ "learning_rate": 8.843537414965987e-05,
609
+ "log_odds_chosen": 0.20318442583084106,
610
+ "log_odds_ratio": -0.60548335313797,
611
+ "logits/chosen": -0.6687170267105103,
612
+ "logits/rejected": -0.6979736089706421,
613
+ "logps/chosen": -1.1998878717422485,
614
+ "logps/rejected": -1.3429744243621826,
615
+ "loss": 1.4198,
616
+ "nll_loss": 1.3592194318771362,
617
+ "rewards/accuracies": 0.75,
618
+ "rewards/chosen": -0.11998879909515381,
619
+ "rewards/margins": 0.014308637008070946,
620
+ "rewards/rejected": -0.1342974454164505,
621
+ "step": 34
622
+ },
623
+ {
624
+ "epoch": 0.35668789808917195,
625
+ "grad_norm": 1.6395893096923828,
626
+ "learning_rate": 8.80952380952381e-05,
627
+ "log_odds_chosen": 0.1479332149028778,
628
+ "log_odds_ratio": -0.6287116408348083,
629
+ "logits/chosen": -0.6793037056922913,
630
+ "logits/rejected": -0.7141229510307312,
631
+ "logps/chosen": -1.1636970043182373,
632
+ "logps/rejected": -1.2673585414886475,
633
+ "loss": 1.3737,
634
+ "nll_loss": 1.310793161392212,
635
+ "rewards/accuracies": 0.6875,
636
+ "rewards/chosen": -0.11636970937252045,
637
+ "rewards/margins": 0.01036614365875721,
638
+ "rewards/rejected": -0.1267358511686325,
639
+ "step": 35
640
+ },
641
+ {
642
+ "epoch": 0.3668789808917197,
643
+ "grad_norm": 1.3266431093215942,
644
+ "learning_rate": 8.775510204081632e-05,
645
+ "log_odds_chosen": 0.16748954355716705,
646
+ "log_odds_ratio": -0.6182947754859924,
647
+ "logits/chosen": -0.6686299443244934,
648
+ "logits/rejected": -0.6554075479507446,
649
+ "logps/chosen": -1.0804071426391602,
650
+ "logps/rejected": -1.1927094459533691,
651
+ "loss": 1.328,
652
+ "nll_loss": 1.2661728858947754,
653
+ "rewards/accuracies": 0.75,
654
+ "rewards/chosen": -0.1080407053232193,
655
+ "rewards/margins": 0.011230221018195152,
656
+ "rewards/rejected": -0.1192709356546402,
657
+ "step": 36
658
+ },
659
+ {
660
+ "epoch": 0.37707006369426754,
661
+ "grad_norm": 1.6111111640930176,
662
+ "learning_rate": 8.741496598639455e-05,
663
+ "log_odds_chosen": 0.09395941346883774,
664
+ "log_odds_ratio": -0.6549113988876343,
665
+ "logits/chosen": -0.7055931091308594,
666
+ "logits/rejected": -0.6843895316123962,
667
+ "logps/chosen": -1.123265027999878,
668
+ "logps/rejected": -1.1842732429504395,
669
+ "loss": 1.3436,
670
+ "nll_loss": 1.2781226634979248,
671
+ "rewards/accuracies": 0.59375,
672
+ "rewards/chosen": -0.1123264953494072,
673
+ "rewards/margins": 0.006100829225033522,
674
+ "rewards/rejected": -0.1184273213148117,
675
+ "step": 37
676
+ },
677
+ {
678
+ "epoch": 0.3872611464968153,
679
+ "grad_norm": 1.3853479623794556,
680
+ "learning_rate": 8.70748299319728e-05,
681
+ "log_odds_chosen": 0.22841353714466095,
682
+ "log_odds_ratio": -0.5929134488105774,
683
+ "logits/chosen": -0.6214150786399841,
684
+ "logits/rejected": -0.6320711374282837,
685
+ "logps/chosen": -0.9967460632324219,
686
+ "logps/rejected": -1.1483112573623657,
687
+ "loss": 1.239,
688
+ "nll_loss": 1.179752230644226,
689
+ "rewards/accuracies": 0.84375,
690
+ "rewards/chosen": -0.09967461228370667,
691
+ "rewards/margins": 0.01515651773661375,
692
+ "rewards/rejected": -0.11483113467693329,
693
+ "step": 38
694
+ },
695
+ {
696
+ "epoch": 0.3974522292993631,
697
+ "grad_norm": 1.4109987020492554,
698
+ "learning_rate": 8.673469387755102e-05,
699
+ "log_odds_chosen": 0.15123127400875092,
700
+ "log_odds_ratio": -0.6313170194625854,
701
+ "logits/chosen": -0.5636093020439148,
702
+ "logits/rejected": -0.5879440903663635,
703
+ "logps/chosen": -0.9781452417373657,
704
+ "logps/rejected": -1.0714449882507324,
705
+ "loss": 1.2294,
706
+ "nll_loss": 1.1662321090698242,
707
+ "rewards/accuracies": 0.65625,
708
+ "rewards/chosen": -0.09781452268362045,
709
+ "rewards/margins": 0.00932997465133667,
710
+ "rewards/rejected": -0.10714450478553772,
711
+ "step": 39
712
+ },
713
+ {
714
+ "epoch": 0.40764331210191085,
715
+ "grad_norm": 1.505708932876587,
716
+ "learning_rate": 8.639455782312925e-05,
717
+ "log_odds_chosen": 0.15932558476924896,
718
+ "log_odds_ratio": -0.625272274017334,
719
+ "logits/chosen": -0.5791752934455872,
720
+ "logits/rejected": -0.6104503870010376,
721
+ "logps/chosen": -0.9371697902679443,
722
+ "logps/rejected": -1.0375370979309082,
723
+ "loss": 1.1666,
724
+ "nll_loss": 1.104070782661438,
725
+ "rewards/accuracies": 0.71875,
726
+ "rewards/chosen": -0.09371697902679443,
727
+ "rewards/margins": 0.010036729276180267,
728
+ "rewards/rejected": -0.1037537157535553,
729
+ "step": 40
730
+ },
731
+ {
732
+ "epoch": 0.4178343949044586,
733
+ "grad_norm": 1.3721827268600464,
734
+ "learning_rate": 8.605442176870748e-05,
735
+ "log_odds_chosen": 0.15735837817192078,
736
+ "log_odds_ratio": -0.6272850036621094,
737
+ "logits/chosen": -0.5897191762924194,
738
+ "logits/rejected": -0.5843300819396973,
739
+ "logps/chosen": -0.9254653453826904,
740
+ "logps/rejected": -1.0147194862365723,
741
+ "loss": 1.1898,
742
+ "nll_loss": 1.127084732055664,
743
+ "rewards/accuracies": 0.75,
744
+ "rewards/chosen": -0.09254653751850128,
745
+ "rewards/margins": 0.008925420232117176,
746
+ "rewards/rejected": -0.10147196054458618,
747
+ "step": 41
748
+ },
749
+ {
750
+ "epoch": 0.4280254777070064,
751
+ "grad_norm": 1.4123814105987549,
752
+ "learning_rate": 8.571428571428571e-05,
753
+ "log_odds_chosen": 0.1627001166343689,
754
+ "log_odds_ratio": -0.6306536197662354,
755
+ "logits/chosen": -0.5717113614082336,
756
+ "logits/rejected": -0.5668535232543945,
757
+ "logps/chosen": -0.877647876739502,
758
+ "logps/rejected": -0.9752596020698547,
759
+ "loss": 1.1148,
760
+ "nll_loss": 1.0516966581344604,
761
+ "rewards/accuracies": 0.71875,
762
+ "rewards/chosen": -0.08776478469371796,
763
+ "rewards/margins": 0.00976115744560957,
764
+ "rewards/rejected": -0.0975259393453598,
765
+ "step": 42
766
+ },
767
+ {
768
+ "epoch": 0.43821656050955415,
769
+ "grad_norm": 1.4072459936141968,
770
+ "learning_rate": 8.537414965986394e-05,
771
+ "log_odds_chosen": 0.26022908091545105,
772
+ "log_odds_ratio": -0.5852185487747192,
773
+ "logits/chosen": -0.5471714735031128,
774
+ "logits/rejected": -0.5991571545600891,
775
+ "logps/chosen": -0.8360999822616577,
776
+ "logps/rejected": -0.9906545281410217,
777
+ "loss": 1.0647,
778
+ "nll_loss": 1.0061821937561035,
779
+ "rewards/accuracies": 0.71875,
780
+ "rewards/chosen": -0.08360999822616577,
781
+ "rewards/margins": 0.015455447137355804,
782
+ "rewards/rejected": -0.09906545281410217,
783
+ "step": 43
784
+ },
785
+ {
786
+ "epoch": 0.4484076433121019,
787
+ "grad_norm": 1.263767957687378,
788
+ "learning_rate": 8.503401360544217e-05,
789
+ "log_odds_chosen": 0.15025389194488525,
790
+ "log_odds_ratio": -0.6313925981521606,
791
+ "logits/chosen": -0.5081976652145386,
792
+ "logits/rejected": -0.5151488184928894,
793
+ "logps/chosen": -0.860946774482727,
794
+ "logps/rejected": -0.9522109031677246,
795
+ "loss": 1.0628,
796
+ "nll_loss": 0.9996945858001709,
797
+ "rewards/accuracies": 0.6875,
798
+ "rewards/chosen": -0.0860946774482727,
799
+ "rewards/margins": 0.00912640430033207,
800
+ "rewards/rejected": -0.09522109478712082,
801
+ "step": 44
802
+ },
803
+ {
804
+ "epoch": 0.4585987261146497,
805
+ "grad_norm": 1.1564173698425293,
806
+ "learning_rate": 8.469387755102041e-05,
807
+ "log_odds_chosen": 0.25856250524520874,
808
+ "log_odds_ratio": -0.5882014632225037,
809
+ "logits/chosen": -0.47218459844589233,
810
+ "logits/rejected": -0.47766923904418945,
811
+ "logps/chosen": -0.786893904209137,
812
+ "logps/rejected": -0.9319732785224915,
813
+ "loss": 1.0045,
814
+ "nll_loss": 0.9457147121429443,
815
+ "rewards/accuracies": 0.78125,
816
+ "rewards/chosen": -0.07868938893079758,
817
+ "rewards/margins": 0.01450793445110321,
818
+ "rewards/rejected": -0.09319733083248138,
819
+ "step": 45
820
+ },
821
+ {
822
+ "epoch": 0.46878980891719746,
823
+ "grad_norm": 1.1476246118545532,
824
+ "learning_rate": 8.435374149659864e-05,
825
+ "log_odds_chosen": 0.1751195192337036,
826
+ "log_odds_ratio": -0.6188879013061523,
827
+ "logits/chosen": -0.4528760015964508,
828
+ "logits/rejected": -0.47072282433509827,
829
+ "logps/chosen": -0.7283643484115601,
830
+ "logps/rejected": -0.8211995363235474,
831
+ "loss": 0.9499,
832
+ "nll_loss": 0.8879653811454773,
833
+ "rewards/accuracies": 0.78125,
834
+ "rewards/chosen": -0.07283642888069153,
835
+ "rewards/margins": 0.009283517487347126,
836
+ "rewards/rejected": -0.08211994916200638,
837
+ "step": 46
838
+ },
839
+ {
840
+ "epoch": 0.4789808917197452,
841
+ "grad_norm": 1.082606554031372,
842
+ "learning_rate": 8.401360544217689e-05,
843
+ "log_odds_chosen": 0.239468514919281,
844
+ "log_odds_ratio": -0.5954229831695557,
845
+ "logits/chosen": -0.4875160753726959,
846
+ "logits/rejected": -0.48623812198638916,
847
+ "logps/chosen": -0.7523530125617981,
848
+ "logps/rejected": -0.877802848815918,
849
+ "loss": 0.954,
850
+ "nll_loss": 0.8944483399391174,
851
+ "rewards/accuracies": 0.78125,
852
+ "rewards/chosen": -0.07523529976606369,
853
+ "rewards/margins": 0.012544991448521614,
854
+ "rewards/rejected": -0.08778029680252075,
855
+ "step": 47
856
+ },
857
+ {
858
+ "epoch": 0.489171974522293,
859
+ "grad_norm": 1.1768563985824585,
860
+ "learning_rate": 8.367346938775511e-05,
861
+ "log_odds_chosen": 0.11137399077415466,
862
+ "log_odds_ratio": -0.6481903791427612,
863
+ "logits/chosen": -0.47852304577827454,
864
+ "logits/rejected": -0.49898433685302734,
865
+ "logps/chosen": -0.7479023933410645,
866
+ "logps/rejected": -0.8043348789215088,
867
+ "loss": 0.9558,
868
+ "nll_loss": 0.8910031914710999,
869
+ "rewards/accuracies": 0.59375,
870
+ "rewards/chosen": -0.07479023933410645,
871
+ "rewards/margins": 0.005643244832754135,
872
+ "rewards/rejected": -0.08043348044157028,
873
+ "step": 48
874
+ },
875
+ {
876
+ "epoch": 0.49936305732484076,
877
+ "grad_norm": 1.0716862678527832,
878
+ "learning_rate": 8.333333333333334e-05,
879
+ "log_odds_chosen": 0.20709852874279022,
880
+ "log_odds_ratio": -0.6083564758300781,
881
+ "logits/chosen": -0.4109807312488556,
882
+ "logits/rejected": -0.44194647669792175,
883
+ "logps/chosen": -0.7731195688247681,
884
+ "logps/rejected": -0.8942749500274658,
885
+ "loss": 0.9623,
886
+ "nll_loss": 0.9014848470687866,
887
+ "rewards/accuracies": 0.6875,
888
+ "rewards/chosen": -0.07731196284294128,
889
+ "rewards/margins": 0.012115536257624626,
890
+ "rewards/rejected": -0.08942750096321106,
891
+ "step": 49
892
+ },
893
+ {
894
+ "epoch": 0.5095541401273885,
895
+ "grad_norm": 0.9973980188369751,
896
+ "learning_rate": 8.299319727891157e-05,
897
+ "log_odds_chosen": 0.18566134572029114,
898
+ "log_odds_ratio": -0.61891108751297,
899
+ "logits/chosen": -0.4191117286682129,
900
+ "logits/rejected": -0.41958895325660706,
901
+ "logps/chosen": -0.702156662940979,
902
+ "logps/rejected": -0.7984734177589417,
903
+ "loss": 0.8786,
904
+ "nll_loss": 0.8166758418083191,
905
+ "rewards/accuracies": 0.71875,
906
+ "rewards/chosen": -0.07021567225456238,
907
+ "rewards/margins": 0.00963167380541563,
908
+ "rewards/rejected": -0.07984734326601028,
909
+ "step": 50
910
+ },
911
+ {
912
+ "epoch": 0.5197452229299363,
913
+ "grad_norm": 0.8124039769172668,
914
+ "learning_rate": 8.26530612244898e-05,
915
+ "log_odds_chosen": 0.18215714395046234,
916
+ "log_odds_ratio": -0.6213276386260986,
917
+ "logits/chosen": -0.4588465690612793,
918
+ "logits/rejected": -0.46019288897514343,
919
+ "logps/chosen": -0.7112851142883301,
920
+ "logps/rejected": -0.7996297478675842,
921
+ "loss": 0.8886,
922
+ "nll_loss": 0.8265119194984436,
923
+ "rewards/accuracies": 0.6875,
924
+ "rewards/chosen": -0.07112851738929749,
925
+ "rewards/margins": 0.008834462612867355,
926
+ "rewards/rejected": -0.07996297627687454,
927
+ "step": 51
928
+ },
929
+ {
930
+ "epoch": 0.5299363057324841,
931
+ "grad_norm": 0.777076780796051,
932
+ "learning_rate": 8.231292517006803e-05,
933
+ "log_odds_chosen": 0.2624254524707794,
934
+ "log_odds_ratio": -0.5843685269355774,
935
+ "logits/chosen": -0.3662329912185669,
936
+ "logits/rejected": -0.34878844022750854,
937
+ "logps/chosen": -0.6983358860015869,
938
+ "logps/rejected": -0.838935911655426,
939
+ "loss": 0.8633,
940
+ "nll_loss": 0.8049026131629944,
941
+ "rewards/accuracies": 0.75,
942
+ "rewards/chosen": -0.06983359158039093,
943
+ "rewards/margins": 0.014060002751648426,
944
+ "rewards/rejected": -0.08389358967542648,
945
+ "step": 52
946
+ },
947
+ {
948
+ "epoch": 0.5401273885350318,
949
+ "grad_norm": 0.5475600957870483,
950
+ "learning_rate": 8.197278911564626e-05,
951
+ "log_odds_chosen": 0.23267458379268646,
952
+ "log_odds_ratio": -0.6021015644073486,
953
+ "logits/chosen": -0.3862481415271759,
954
+ "logits/rejected": -0.38804885745048523,
955
+ "logps/chosen": -0.7031423449516296,
956
+ "logps/rejected": -0.8290390968322754,
957
+ "loss": 0.8748,
958
+ "nll_loss": 0.8145608901977539,
959
+ "rewards/accuracies": 0.71875,
960
+ "rewards/chosen": -0.07031424343585968,
961
+ "rewards/margins": 0.012589672580361366,
962
+ "rewards/rejected": -0.0829039141535759,
963
+ "step": 53
964
+ },
965
+ {
966
+ "epoch": 0.5503184713375796,
967
+ "grad_norm": 0.5666794776916504,
968
+ "learning_rate": 8.163265306122449e-05,
969
+ "log_odds_chosen": 0.25999322533607483,
970
+ "log_odds_ratio": -0.5887346863746643,
971
+ "logits/chosen": -0.347758412361145,
972
+ "logits/rejected": -0.3341512084007263,
973
+ "logps/chosen": -0.6920559406280518,
974
+ "logps/rejected": -0.8253049254417419,
975
+ "loss": 0.8235,
976
+ "nll_loss": 0.7646136283874512,
977
+ "rewards/accuracies": 0.78125,
978
+ "rewards/chosen": -0.06920559704303741,
979
+ "rewards/margins": 0.013324896804988384,
980
+ "rewards/rejected": -0.08253049850463867,
981
+ "step": 54
982
+ },
983
+ {
984
+ "epoch": 0.5605095541401274,
985
+ "grad_norm": 0.5203105211257935,
986
+ "learning_rate": 8.129251700680273e-05,
987
+ "log_odds_chosen": 0.18836164474487305,
988
+ "log_odds_ratio": -0.6200145483016968,
989
+ "logits/chosen": -0.3182459771633148,
990
+ "logits/rejected": -0.3383919596672058,
991
+ "logps/chosen": -0.6551133394241333,
992
+ "logps/rejected": -0.7530025839805603,
993
+ "loss": 0.8446,
994
+ "nll_loss": 0.7825708389282227,
995
+ "rewards/accuracies": 0.65625,
996
+ "rewards/chosen": -0.06551133096218109,
997
+ "rewards/margins": 0.009788922034204006,
998
+ "rewards/rejected": -0.07530025392770767,
999
+ "step": 55
1000
+ },
1001
+ {
1002
+ "epoch": 0.5707006369426751,
1003
+ "grad_norm": 0.5612345933914185,
1004
+ "learning_rate": 8.095238095238096e-05,
1005
+ "log_odds_chosen": 0.1355578601360321,
1006
+ "log_odds_ratio": -0.6451556086540222,
1007
+ "logits/chosen": -0.29888439178466797,
1008
+ "logits/rejected": -0.2906675338745117,
1009
+ "logps/chosen": -0.6383944153785706,
1010
+ "logps/rejected": -0.7033387422561646,
1011
+ "loss": 0.8437,
1012
+ "nll_loss": 0.7791876196861267,
1013
+ "rewards/accuracies": 0.59375,
1014
+ "rewards/chosen": -0.06383944302797318,
1015
+ "rewards/margins": 0.006494428031146526,
1016
+ "rewards/rejected": -0.07033387571573257,
1017
+ "step": 56
1018
+ },
1019
+ {
1020
+ "epoch": 0.5808917197452229,
1021
+ "grad_norm": 0.8524444103240967,
1022
+ "learning_rate": 8.061224489795919e-05,
1023
+ "log_odds_chosen": 0.349676251411438,
1024
+ "log_odds_ratio": -0.5498685836791992,
1025
+ "logits/chosen": -0.2660248577594757,
1026
+ "logits/rejected": -0.29482465982437134,
1027
+ "logps/chosen": -0.5994170904159546,
1028
+ "logps/rejected": -0.770711362361908,
1029
+ "loss": 0.7653,
1030
+ "nll_loss": 0.7103118300437927,
1031
+ "rewards/accuracies": 0.84375,
1032
+ "rewards/chosen": -0.05994170904159546,
1033
+ "rewards/margins": 0.017129434272646904,
1034
+ "rewards/rejected": -0.07707113772630692,
1035
+ "step": 57
1036
+ },
1037
+ {
1038
+ "epoch": 0.5910828025477707,
1039
+ "grad_norm": 0.5966730713844299,
1040
+ "learning_rate": 8.027210884353742e-05,
1041
+ "log_odds_chosen": 0.4513949453830719,
1042
+ "log_odds_ratio": -0.5082166194915771,
1043
+ "logits/chosen": -0.35525086522102356,
1044
+ "logits/rejected": -0.39397263526916504,
1045
+ "logps/chosen": -0.6463137865066528,
1046
+ "logps/rejected": -0.8910082578659058,
1047
+ "loss": 0.7776,
1048
+ "nll_loss": 0.7267977595329285,
1049
+ "rewards/accuracies": 0.96875,
1050
+ "rewards/chosen": -0.0646313801407814,
1051
+ "rewards/margins": 0.02446945384144783,
1052
+ "rewards/rejected": -0.08910083770751953,
1053
+ "step": 58
1054
+ },
1055
+ {
1056
+ "epoch": 0.6012738853503184,
1057
+ "grad_norm": 0.6914296746253967,
1058
+ "learning_rate": 7.993197278911565e-05,
1059
+ "log_odds_chosen": 0.2607879042625427,
1060
+ "log_odds_ratio": -0.5879091024398804,
1061
+ "logits/chosen": -0.26178666949272156,
1062
+ "logits/rejected": -0.2836032807826996,
1063
+ "logps/chosen": -0.60906583070755,
1064
+ "logps/rejected": -0.7373634576797485,
1065
+ "loss": 0.7941,
1066
+ "nll_loss": 0.7352855801582336,
1067
+ "rewards/accuracies": 0.75,
1068
+ "rewards/chosen": -0.060906581580638885,
1069
+ "rewards/margins": 0.012829765677452087,
1070
+ "rewards/rejected": -0.07373635470867157,
1071
+ "step": 59
1072
+ },
1073
+ {
1074
+ "epoch": 0.6114649681528662,
1075
+ "grad_norm": 0.49855294823646545,
1076
+ "learning_rate": 7.959183673469388e-05,
1077
+ "log_odds_chosen": 0.17502906918525696,
1078
+ "log_odds_ratio": -0.6250565648078918,
1079
+ "logits/chosen": -0.27342450618743896,
1080
+ "logits/rejected": -0.2793465256690979,
1081
+ "logps/chosen": -0.643248975276947,
1082
+ "logps/rejected": -0.7312005162239075,
1083
+ "loss": 0.8101,
1084
+ "nll_loss": 0.7475584149360657,
1085
+ "rewards/accuracies": 0.625,
1086
+ "rewards/chosen": -0.06432489305734634,
1087
+ "rewards/margins": 0.008795149624347687,
1088
+ "rewards/rejected": -0.07312005013227463,
1089
+ "step": 60
1090
+ },
1091
+ {
1092
+ "epoch": 0.621656050955414,
1093
+ "grad_norm": 0.623975396156311,
1094
+ "learning_rate": 7.925170068027211e-05,
1095
+ "log_odds_chosen": 0.20475000143051147,
1096
+ "log_odds_ratio": -0.6119533181190491,
1097
+ "logits/chosen": -0.2819938361644745,
1098
+ "logits/rejected": -0.2693704068660736,
1099
+ "logps/chosen": -0.6357014179229736,
1100
+ "logps/rejected": -0.7285926342010498,
1101
+ "loss": 0.7994,
1102
+ "nll_loss": 0.7381711602210999,
1103
+ "rewards/accuracies": 0.625,
1104
+ "rewards/chosen": -0.06357015669345856,
1105
+ "rewards/margins": 0.009289107285439968,
1106
+ "rewards/rejected": -0.0728592500090599,
1107
+ "step": 61
1108
+ },
1109
+ {
1110
+ "epoch": 0.6318471337579618,
1111
+ "grad_norm": 0.63492351770401,
1112
+ "learning_rate": 7.891156462585034e-05,
1113
+ "log_odds_chosen": 0.22653746604919434,
1114
+ "log_odds_ratio": -0.5998401045799255,
1115
+ "logits/chosen": -0.2391633242368698,
1116
+ "logits/rejected": -0.22683343291282654,
1117
+ "logps/chosen": -0.6707414388656616,
1118
+ "logps/rejected": -0.7814602255821228,
1119
+ "loss": 0.8376,
1120
+ "nll_loss": 0.7776267528533936,
1121
+ "rewards/accuracies": 0.78125,
1122
+ "rewards/chosen": -0.06707413494586945,
1123
+ "rewards/margins": 0.011071886867284775,
1124
+ "rewards/rejected": -0.07814602553844452,
1125
+ "step": 62
1126
  }
1127
  ],
1128
  "logging_steps": 1,
last-checkpoint/training_args.bin CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:8ee18ea0c5cd8ab4dc26bcfd2b5da8b145de03666713371dc25009f17b330bef
3
  size 5560
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b6e9bd93f66e1ed0f17812a85321161533aa48011f619aebcabb9399556f35a5
3
  size 5560