narekvslife commited on
Commit
8904848
1 Parent(s): 0e3ef6e

dpo_5wiothfs 5.9

Browse files
Files changed (4) hide show
  1. adapter_model.safetensors +1 -1
  2. optimizer.pt +1 -1
  3. scheduler.pt +1 -1
  4. trainer_state.json +452 -2
adapter_model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:5cd287bba5fde8fc753916b1d432058128613e0bcfc071316ad6378d8a26508e
3
  size 18900240
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9527902bd6aeaf0355fc706a007b7e21ee1a936860b8d1b9bd19824385fc4972
3
  size 18900240
optimizer.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:74d28b77057f2ce067f3506293bc8da387c9a97b206dccc88b87d9cb314b5e32
3
  size 37910458
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e7dfe12af6d9861c0b41c1b9ca0e7b6f45d90828d79882df00f0e054a2f011d0
3
  size 37910458
scheduler.pt CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:e1dcb1c05f8406763f478190e5dde325c77e4a5fb69a17c7b42d79ed2e579e6e
3
  size 1064
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:a0b908a911ffc3dc212618df71c6aa766b5d758bf18eb427c2dcfb767a1b2cba
3
  size 1064
trainer_state.json CHANGED
@@ -1,9 +1,9 @@
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
- "epoch": 1.763710090153931,
5
  "eval_steps": 2000,
6
- "global_step": 5600,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
@@ -8439,6 +8439,456 @@
8439
  "rewards/margins": 0.27365249395370483,
8440
  "rewards/rejected": 0.5168629884719849,
8441
  "step": 5600
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
8442
  }
8443
  ],
8444
  "logging_steps": 10,
 
1
  {
2
  "best_metric": null,
3
  "best_model_checkpoint": null,
4
+ "epoch": 1.85819455926932,
5
  "eval_steps": 2000,
6
+ "global_step": 5900,
7
  "is_hyper_param_search": false,
8
  "is_local_process_zero": true,
9
  "is_world_process_zero": true,
 
8439
  "rewards/margins": 0.27365249395370483,
8440
  "rewards/rejected": 0.5168629884719849,
8441
  "step": 5600
8442
+ },
8443
+ {
8444
+ "epoch": 1.7668595724577774,
8445
+ "grad_norm": 2.46875,
8446
+ "learning_rate": 3.4943208086663183e-06,
8447
+ "logits/chosen": -0.4847659170627594,
8448
+ "logits/rejected": -0.33793026208877563,
8449
+ "logps/chosen": -197.34933471679688,
8450
+ "logps/rejected": -174.9829559326172,
8451
+ "loss": 0.646,
8452
+ "rewards/accuracies": 0.6000000238418579,
8453
+ "rewards/chosen": 0.7014733552932739,
8454
+ "rewards/margins": 0.12839707732200623,
8455
+ "rewards/rejected": 0.5730762481689453,
8456
+ "step": 5610
8457
+ },
8458
+ {
8459
+ "epoch": 1.7700090547616236,
8460
+ "grad_norm": 3.578125,
8461
+ "learning_rate": 3.4894823245512986e-06,
8462
+ "logits/chosen": -0.506749153137207,
8463
+ "logits/rejected": -0.45556968450546265,
8464
+ "logps/chosen": -197.71902465820312,
8465
+ "logps/rejected": -186.50241088867188,
8466
+ "loss": 0.6803,
8467
+ "rewards/accuracies": 0.550000011920929,
8468
+ "rewards/chosen": 0.7117626070976257,
8469
+ "rewards/margins": 0.06737571209669113,
8470
+ "rewards/rejected": 0.644386887550354,
8471
+ "step": 5620
8472
+ },
8473
+ {
8474
+ "epoch": 1.7731585370654699,
8475
+ "grad_norm": 2.578125,
8476
+ "learning_rate": 3.484639441627448e-06,
8477
+ "logits/chosen": -0.5070594549179077,
8478
+ "logits/rejected": -0.3329693078994751,
8479
+ "logps/chosen": -220.60986328125,
8480
+ "logps/rejected": -183.98416137695312,
8481
+ "loss": 0.6042,
8482
+ "rewards/accuracies": 0.737500011920929,
8483
+ "rewards/chosen": 0.7875211834907532,
8484
+ "rewards/margins": 0.2286391705274582,
8485
+ "rewards/rejected": 0.5588821172714233,
8486
+ "step": 5630
8487
+ },
8488
+ {
8489
+ "epoch": 1.7763080193693161,
8490
+ "grad_norm": 2.546875,
8491
+ "learning_rate": 3.4797921814241196e-06,
8492
+ "logits/chosen": -0.48938584327697754,
8493
+ "logits/rejected": -0.37643399834632874,
8494
+ "logps/chosen": -194.7692413330078,
8495
+ "logps/rejected": -171.0836944580078,
8496
+ "loss": 0.6345,
8497
+ "rewards/accuracies": 0.637499988079071,
8498
+ "rewards/chosen": 0.7208179235458374,
8499
+ "rewards/margins": 0.17952939867973328,
8500
+ "rewards/rejected": 0.5412884950637817,
8501
+ "step": 5640
8502
+ },
8503
+ {
8504
+ "epoch": 1.7794575016731624,
8505
+ "grad_norm": 2.71875,
8506
+ "learning_rate": 3.4749405654901297e-06,
8507
+ "logits/chosen": -0.5021311044692993,
8508
+ "logits/rejected": -0.3592470586299896,
8509
+ "logps/chosen": -203.04798889160156,
8510
+ "logps/rejected": -170.28916931152344,
8511
+ "loss": 0.6468,
8512
+ "rewards/accuracies": 0.5874999761581421,
8513
+ "rewards/chosen": 0.7304830551147461,
8514
+ "rewards/margins": 0.14200101792812347,
8515
+ "rewards/rejected": 0.5884820222854614,
8516
+ "step": 5650
8517
+ },
8518
+ {
8519
+ "epoch": 1.7826069839770087,
8520
+ "grad_norm": 1.8125,
8521
+ "learning_rate": 3.470084615393655e-06,
8522
+ "logits/chosen": -0.5099314451217651,
8523
+ "logits/rejected": -0.36777496337890625,
8524
+ "logps/chosen": -188.96286010742188,
8525
+ "logps/rejected": -158.13487243652344,
8526
+ "loss": 0.5854,
8527
+ "rewards/accuracies": 0.8374999761581421,
8528
+ "rewards/chosen": 0.7638787031173706,
8529
+ "rewards/margins": 0.25533777475357056,
8530
+ "rewards/rejected": 0.5085408687591553,
8531
+ "step": 5660
8532
+ },
8533
+ {
8534
+ "epoch": 1.785756466280855,
8535
+ "grad_norm": 2.71875,
8536
+ "learning_rate": 3.4652243527221423e-06,
8537
+ "logits/chosen": -0.4756031632423401,
8538
+ "logits/rejected": -0.44920986890792847,
8539
+ "logps/chosen": -185.1388397216797,
8540
+ "logps/rejected": -172.55137634277344,
8541
+ "loss": 0.6583,
8542
+ "rewards/accuracies": 0.5375000238418579,
8543
+ "rewards/chosen": 0.7094627618789673,
8544
+ "rewards/margins": 0.13025884330272675,
8545
+ "rewards/rejected": 0.5792039036750793,
8546
+ "step": 5670
8547
+ },
8548
+ {
8549
+ "epoch": 1.7889059485847014,
8550
+ "grad_norm": 3.171875,
8551
+ "learning_rate": 3.460359799082209e-06,
8552
+ "logits/chosen": -0.47689515352249146,
8553
+ "logits/rejected": -0.34241801500320435,
8554
+ "logps/chosen": -204.8109588623047,
8555
+ "logps/rejected": -166.13514709472656,
8556
+ "loss": 0.615,
8557
+ "rewards/accuracies": 0.7250000238418579,
8558
+ "rewards/chosen": 0.7595565915107727,
8559
+ "rewards/margins": 0.21238622069358826,
8560
+ "rewards/rejected": 0.5471702814102173,
8561
+ "step": 5680
8562
+ },
8563
+ {
8564
+ "epoch": 1.7920554308885477,
8565
+ "grad_norm": 3.765625,
8566
+ "learning_rate": 3.4554909760995485e-06,
8567
+ "logits/chosen": -0.5418170094490051,
8568
+ "logits/rejected": -0.41362690925598145,
8569
+ "logps/chosen": -187.98043823242188,
8570
+ "logps/rejected": -167.5854034423828,
8571
+ "loss": 0.6338,
8572
+ "rewards/accuracies": 0.699999988079071,
8573
+ "rewards/chosen": 0.7204712629318237,
8574
+ "rewards/margins": 0.1737706959247589,
8575
+ "rewards/rejected": 0.5467005968093872,
8576
+ "step": 5690
8577
+ },
8578
+ {
8579
+ "epoch": 1.795204913192394,
8580
+ "grad_norm": 3.3125,
8581
+ "learning_rate": 3.450617905418834e-06,
8582
+ "logits/chosen": -0.442087322473526,
8583
+ "logits/rejected": -0.3480719029903412,
8584
+ "logps/chosen": -205.0787353515625,
8585
+ "logps/rejected": -176.585693359375,
8586
+ "loss": 0.6078,
8587
+ "rewards/accuracies": 0.699999988079071,
8588
+ "rewards/chosen": 0.7968889474868774,
8589
+ "rewards/margins": 0.2236328423023224,
8590
+ "rewards/rejected": 0.5732561349868774,
8591
+ "step": 5700
8592
+ },
8593
+ {
8594
+ "epoch": 1.7983543954962404,
8595
+ "grad_norm": 3.125,
8596
+ "learning_rate": 3.4457406087036233e-06,
8597
+ "logits/chosen": -0.4669428765773773,
8598
+ "logits/rejected": -0.379183828830719,
8599
+ "logps/chosen": -183.84532165527344,
8600
+ "logps/rejected": -169.44937133789062,
8601
+ "loss": 0.6755,
8602
+ "rewards/accuracies": 0.637499988079071,
8603
+ "rewards/chosen": 0.6309347748756409,
8604
+ "rewards/margins": 0.07157482206821442,
8605
+ "rewards/rejected": 0.5593599081039429,
8606
+ "step": 5710
8607
+ },
8608
+ {
8609
+ "epoch": 1.8015038778000867,
8610
+ "grad_norm": 2.984375,
8611
+ "learning_rate": 3.4408591076362585e-06,
8612
+ "logits/chosen": -0.5323187112808228,
8613
+ "logits/rejected": -0.45780545473098755,
8614
+ "logps/chosen": -205.9134521484375,
8615
+ "logps/rejected": -180.65916442871094,
8616
+ "loss": 0.6566,
8617
+ "rewards/accuracies": 0.612500011920929,
8618
+ "rewards/chosen": 0.7317408323287964,
8619
+ "rewards/margins": 0.11702696233987808,
8620
+ "rewards/rejected": 0.6147138476371765,
8621
+ "step": 5720
8622
+ },
8623
+ {
8624
+ "epoch": 1.804653360103933,
8625
+ "grad_norm": 2.859375,
8626
+ "learning_rate": 3.435973423917774e-06,
8627
+ "logits/chosen": -0.48551005125045776,
8628
+ "logits/rejected": -0.40477806329727173,
8629
+ "logps/chosen": -195.50228881835938,
8630
+ "logps/rejected": -173.91912841796875,
8631
+ "loss": 0.6842,
8632
+ "rewards/accuracies": 0.5375000238418579,
8633
+ "rewards/chosen": 0.7036144137382507,
8634
+ "rewards/margins": 0.06239970773458481,
8635
+ "rewards/rejected": 0.6412147283554077,
8636
+ "step": 5730
8637
+ },
8638
+ {
8639
+ "epoch": 1.8078028424077792,
8640
+ "grad_norm": 2.40625,
8641
+ "learning_rate": 3.4310835792677995e-06,
8642
+ "logits/chosen": -0.4431411623954773,
8643
+ "logits/rejected": -0.3337770104408264,
8644
+ "logps/chosen": -198.4442138671875,
8645
+ "logps/rejected": -162.93258666992188,
8646
+ "loss": 0.6348,
8647
+ "rewards/accuracies": 0.637499988079071,
8648
+ "rewards/chosen": 0.6712988018989563,
8649
+ "rewards/margins": 0.1662341058254242,
8650
+ "rewards/rejected": 0.5050647854804993,
8651
+ "step": 5740
8652
+ },
8653
+ {
8654
+ "epoch": 1.8109523247116255,
8655
+ "grad_norm": 3.015625,
8656
+ "learning_rate": 3.4261895954244613e-06,
8657
+ "logits/chosen": -0.4226387143135071,
8658
+ "logits/rejected": -0.3787776827812195,
8659
+ "logps/chosen": -173.4969024658203,
8660
+ "logps/rejected": -161.3011932373047,
8661
+ "loss": 0.6435,
8662
+ "rewards/accuracies": 0.6499999761581421,
8663
+ "rewards/chosen": 0.6374837160110474,
8664
+ "rewards/margins": 0.13156263530254364,
8665
+ "rewards/rejected": 0.5059210658073425,
8666
+ "step": 5750
8667
+ },
8668
+ {
8669
+ "epoch": 1.8141018070154717,
8670
+ "grad_norm": 3.09375,
8671
+ "learning_rate": 3.4212914941442866e-06,
8672
+ "logits/chosen": -0.48183003067970276,
8673
+ "logits/rejected": -0.3869970142841339,
8674
+ "logps/chosen": -199.9102020263672,
8675
+ "logps/rejected": -183.46273803710938,
8676
+ "loss": 0.6739,
8677
+ "rewards/accuracies": 0.512499988079071,
8678
+ "rewards/chosen": 0.7144922614097595,
8679
+ "rewards/margins": 0.07754186540842056,
8680
+ "rewards/rejected": 0.6369503736495972,
8681
+ "step": 5760
8682
+ },
8683
+ {
8684
+ "epoch": 1.817251289319318,
8685
+ "grad_norm": 2.796875,
8686
+ "learning_rate": 3.416389297202107e-06,
8687
+ "logits/chosen": -0.435200035572052,
8688
+ "logits/rejected": -0.273305743932724,
8689
+ "logps/chosen": -200.13018798828125,
8690
+ "logps/rejected": -172.42526245117188,
8691
+ "loss": 0.6273,
8692
+ "rewards/accuracies": 0.6499999761581421,
8693
+ "rewards/chosen": 0.7225381135940552,
8694
+ "rewards/margins": 0.19170936942100525,
8695
+ "rewards/rejected": 0.5308286547660828,
8696
+ "step": 5770
8697
+ },
8698
+ {
8699
+ "epoch": 1.8204007716231645,
8700
+ "grad_norm": 3.203125,
8701
+ "learning_rate": 3.4114830263909615e-06,
8702
+ "logits/chosen": -0.488565593957901,
8703
+ "logits/rejected": -0.3196925222873688,
8704
+ "logps/chosen": -203.71237182617188,
8705
+ "logps/rejected": -175.8201141357422,
8706
+ "loss": 0.6425,
8707
+ "rewards/accuracies": 0.6625000238418579,
8708
+ "rewards/chosen": 0.6835566759109497,
8709
+ "rewards/margins": 0.13755542039871216,
8710
+ "rewards/rejected": 0.5460013151168823,
8711
+ "step": 5780
8712
+ },
8713
+ {
8714
+ "epoch": 1.8235502539270108,
8715
+ "grad_norm": 2.84375,
8716
+ "learning_rate": 3.4065727035220013e-06,
8717
+ "logits/chosen": -0.48802971839904785,
8718
+ "logits/rejected": -0.401599645614624,
8719
+ "logps/chosen": -203.4430694580078,
8720
+ "logps/rejected": -178.24978637695312,
8721
+ "loss": 0.6509,
8722
+ "rewards/accuracies": 0.574999988079071,
8723
+ "rewards/chosen": 0.7013251185417175,
8724
+ "rewards/margins": 0.12661480903625488,
8725
+ "rewards/rejected": 0.5747103095054626,
8726
+ "step": 5790
8727
+ },
8728
+ {
8729
+ "epoch": 1.826699736230857,
8730
+ "grad_norm": 3.15625,
8731
+ "learning_rate": 3.4016583504243892e-06,
8732
+ "logits/chosen": -0.39509814977645874,
8733
+ "logits/rejected": -0.3049541115760803,
8734
+ "logps/chosen": -193.34628295898438,
8735
+ "logps/rejected": -168.88990783691406,
8736
+ "loss": 0.6467,
8737
+ "rewards/accuracies": 0.625,
8738
+ "rewards/chosen": 0.6939215660095215,
8739
+ "rewards/margins": 0.13865116238594055,
8740
+ "rewards/rejected": 0.5552703738212585,
8741
+ "step": 5800
8742
+ },
8743
+ {
8744
+ "epoch": 1.8298492185347035,
8745
+ "grad_norm": 2.609375,
8746
+ "learning_rate": 3.3967399889452056e-06,
8747
+ "logits/chosen": -0.5302572250366211,
8748
+ "logits/rejected": -0.42114171385765076,
8749
+ "logps/chosen": -187.310791015625,
8750
+ "logps/rejected": -158.18551635742188,
8751
+ "loss": 0.62,
8752
+ "rewards/accuracies": 0.699999988079071,
8753
+ "rewards/chosen": 0.6706022024154663,
8754
+ "rewards/margins": 0.18203167617321014,
8755
+ "rewards/rejected": 0.48857051134109497,
8756
+ "step": 5810
8757
+ },
8758
+ {
8759
+ "epoch": 1.8329987008385498,
8760
+ "grad_norm": 2.359375,
8761
+ "learning_rate": 3.3918176409493498e-06,
8762
+ "logits/chosen": -0.4302283227443695,
8763
+ "logits/rejected": -0.3126838207244873,
8764
+ "logps/chosen": -207.9413604736328,
8765
+ "logps/rejected": -186.14862060546875,
8766
+ "loss": 0.6106,
8767
+ "rewards/accuracies": 0.737500011920929,
8768
+ "rewards/chosen": 0.8021620512008667,
8769
+ "rewards/margins": 0.22595825791358948,
8770
+ "rewards/rejected": 0.5762038826942444,
8771
+ "step": 5820
8772
+ },
8773
+ {
8774
+ "epoch": 1.836148183142396,
8775
+ "grad_norm": 3.15625,
8776
+ "learning_rate": 3.3868913283194445e-06,
8777
+ "logits/chosen": -0.4245404303073883,
8778
+ "logits/rejected": -0.3099447190761566,
8779
+ "logps/chosen": -215.6573486328125,
8780
+ "logps/rejected": -180.88473510742188,
8781
+ "loss": 0.6243,
8782
+ "rewards/accuracies": 0.6499999761581421,
8783
+ "rewards/chosen": 0.8092790842056274,
8784
+ "rewards/margins": 0.21156442165374756,
8785
+ "rewards/rejected": 0.5977145433425903,
8786
+ "step": 5830
8787
+ },
8788
+ {
8789
+ "epoch": 1.8392976654462423,
8790
+ "grad_norm": 2.203125,
8791
+ "learning_rate": 3.381961072955737e-06,
8792
+ "logits/chosen": -0.4956479072570801,
8793
+ "logits/rejected": -0.4022194743156433,
8794
+ "logps/chosen": -181.72386169433594,
8795
+ "logps/rejected": -157.3038330078125,
8796
+ "loss": 0.6444,
8797
+ "rewards/accuracies": 0.699999988079071,
8798
+ "rewards/chosen": 0.6013648509979248,
8799
+ "rewards/margins": 0.13577811419963837,
8800
+ "rewards/rejected": 0.46558675169944763,
8801
+ "step": 5840
8802
+ },
8803
+ {
8804
+ "epoch": 1.8424471477500886,
8805
+ "grad_norm": 2.828125,
8806
+ "learning_rate": 3.3770268967760026e-06,
8807
+ "logits/chosen": -0.4699929356575012,
8808
+ "logits/rejected": -0.38960105180740356,
8809
+ "logps/chosen": -190.84512329101562,
8810
+ "logps/rejected": -165.31561279296875,
8811
+ "loss": 0.6521,
8812
+ "rewards/accuracies": 0.6000000238418579,
8813
+ "rewards/chosen": 0.7368890047073364,
8814
+ "rewards/margins": 0.1246052160859108,
8815
+ "rewards/rejected": 0.6122837662696838,
8816
+ "step": 5850
8817
+ },
8818
+ {
8819
+ "epoch": 1.8455966300539348,
8820
+ "grad_norm": 3.640625,
8821
+ "learning_rate": 3.372088821715446e-06,
8822
+ "logits/chosen": -0.5164574384689331,
8823
+ "logits/rejected": -0.40460482239723206,
8824
+ "logps/chosen": -215.09130859375,
8825
+ "logps/rejected": -181.18551635742188,
8826
+ "loss": 0.6583,
8827
+ "rewards/accuracies": 0.6000000238418579,
8828
+ "rewards/chosen": 0.7538167238235474,
8829
+ "rewards/margins": 0.11776645481586456,
8830
+ "rewards/rejected": 0.636050283908844,
8831
+ "step": 5860
8832
+ },
8833
+ {
8834
+ "epoch": 1.848746112357781,
8835
+ "grad_norm": 2.65625,
8836
+ "learning_rate": 3.3671468697266048e-06,
8837
+ "logits/chosen": -0.486356645822525,
8838
+ "logits/rejected": -0.45697417855262756,
8839
+ "logps/chosen": -189.52955627441406,
8840
+ "logps/rejected": -172.86190795898438,
8841
+ "loss": 0.6822,
8842
+ "rewards/accuracies": 0.5375000238418579,
8843
+ "rewards/chosen": 0.6347873210906982,
8844
+ "rewards/margins": 0.054320335388183594,
8845
+ "rewards/rejected": 0.5804670453071594,
8846
+ "step": 5870
8847
+ },
8848
+ {
8849
+ "epoch": 1.8518955946616273,
8850
+ "grad_norm": 3.375,
8851
+ "learning_rate": 3.3622010627792513e-06,
8852
+ "logits/chosen": -0.5492820143699646,
8853
+ "logits/rejected": -0.38086193799972534,
8854
+ "logps/chosen": -194.9511260986328,
8855
+ "logps/rejected": -161.57528686523438,
8856
+ "loss": 0.6699,
8857
+ "rewards/accuracies": 0.550000011920929,
8858
+ "rewards/chosen": 0.6926398873329163,
8859
+ "rewards/margins": 0.09596933424472809,
8860
+ "rewards/rejected": 0.5966705083847046,
8861
+ "step": 5880
8862
+ },
8863
+ {
8864
+ "epoch": 1.8550450769654738,
8865
+ "grad_norm": 2.84375,
8866
+ "learning_rate": 3.3572514228602977e-06,
8867
+ "logits/chosen": -0.4424726366996765,
8868
+ "logits/rejected": -0.35579612851142883,
8869
+ "logps/chosen": -196.1681671142578,
8870
+ "logps/rejected": -165.40811157226562,
8871
+ "loss": 0.6129,
8872
+ "rewards/accuracies": 0.6875,
8873
+ "rewards/chosen": 0.7343538999557495,
8874
+ "rewards/margins": 0.21051523089408875,
8875
+ "rewards/rejected": 0.5238386392593384,
8876
+ "step": 5890
8877
+ },
8878
+ {
8879
+ "epoch": 1.85819455926932,
8880
+ "grad_norm": 2.96875,
8881
+ "learning_rate": 3.3522979719736923e-06,
8882
+ "logits/chosen": -0.4300655722618103,
8883
+ "logits/rejected": -0.23585304617881775,
8884
+ "logps/chosen": -209.92355346679688,
8885
+ "logps/rejected": -173.3553924560547,
8886
+ "loss": 0.639,
8887
+ "rewards/accuracies": 0.5874999761581421,
8888
+ "rewards/chosen": 0.7228736877441406,
8889
+ "rewards/margins": 0.16360947489738464,
8890
+ "rewards/rejected": 0.5592643022537231,
8891
+ "step": 5900
8892
  }
8893
  ],
8894
  "logging_steps": 10,