{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.999244142101285,
  "eval_steps": 100,
  "global_step": 661,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0015117157974300832,
      "epsilon_dpo/beta": 0.10007250308990479,
      "epsilon_dpo/beta_margin_grad_mean": -0.5023139119148254,
      "epsilon_dpo/beta_margin_grad_std": 0.008728506043553352,
      "epsilon_dpo/beta_margin_mean": -0.009260065853595734,
      "epsilon_dpo/beta_margin_std": 0.034927163273096085,
      "epsilon_dpo/loss_margin_mean": -0.08983081579208374,
      "grad_norm": 20.890338897705078,
      "kl/avg_steps": -0.0625,
      "kl/beta": 0.10000000149011612,
      "kl/n_epsilon_steps": 0.53125,
      "kl/p_epsilon_steps": 0.46875,
      "learning_rate": 0.0,
      "logits/chosen": 0.16419550776481628,
      "logits/rejected": 0.3213843107223511,
      "logps/chosen": -85.75910186767578,
      "logps/ref_chosen": -85.73025512695312,
      "logps/ref_rejected": -88.7523193359375,
      "logps/rejected": -88.69134521484375,
      "loss": 1.3959,
      "rewards/accuracies": 0.46875,
      "rewards/chosen": -0.0029897706117480993,
      "rewards/margins": -0.00926008727401495,
      "rewards/rejected": 0.006270316895097494,
      "step": 1
    },
    {
      "epoch": 0.0030234315948601664,
      "epsilon_dpo/beta": 0.10010381788015366,
      "epsilon_dpo/beta_margin_grad_mean": -0.5008825063705444,
      "epsilon_dpo/beta_margin_grad_std": 0.009928400628268719,
      "epsilon_dpo/beta_margin_mean": -0.0035273784305900335,
      "epsilon_dpo/beta_margin_std": 0.03973941132426262,
      "epsilon_dpo/loss_margin_mean": -0.0322224497795105,
      "grad_norm": 19.60763168334961,
      "kl/avg_steps": -0.03125,
      "kl/beta": 0.1000625416636467,
      "kl/n_epsilon_steps": 0.515625,
      "kl/p_epsilon_steps": 0.484375,
      "learning_rate": 7.462686567164179e-09,
      "logits/chosen": 0.22774043679237366,
      "logits/rejected": 0.2854035198688507,
      "logps/chosen": -79.52250671386719,
      "logps/ref_chosen": -79.5359115600586,
      "logps/ref_rejected": -88.0529556274414,
      "logps/rejected": -88.00732421875,
      "loss": 1.3902,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.00122718489728868,
      "rewards/margins": -0.0035273819230496883,
      "rewards/rejected": 0.0047545661218464375,
      "step": 2
    },
    {
      "epoch": 0.0045351473922902496,
      "epsilon_dpo/beta": 0.10007254779338837,
      "epsilon_dpo/beta_margin_grad_mean": -0.49972668290138245,
      "epsilon_dpo/beta_margin_grad_std": 0.010364462621510029,
      "epsilon_dpo/beta_margin_mean": 0.0010939195053651929,
      "epsilon_dpo/beta_margin_std": 0.04147500917315483,
      "epsilon_dpo/loss_margin_mean": 0.014205396175384521,
      "grad_norm": 22.582014083862305,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.10009381920099258,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 1.4925373134328357e-08,
      "logits/chosen": 0.414395809173584,
      "logits/rejected": 0.5279880166053772,
      "logps/chosen": -88.2738037109375,
      "logps/ref_chosen": -88.24923706054688,
      "logps/ref_rejected": -115.33953857421875,
      "logps/rejected": -115.37831115722656,
      "loss": 1.3856,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.0025792494416236877,
      "rewards/margins": 0.0010940060019493103,
      "rewards/rejected": -0.0036732552107423544,
      "step": 3
    },
    {
      "epoch": 0.006046863189720333,
      "epsilon_dpo/beta": 0.09994746744632721,
      "epsilon_dpo/beta_margin_grad_mean": -0.4983575642108917,
      "epsilon_dpo/beta_margin_grad_std": 0.010976379737257957,
      "epsilon_dpo/beta_margin_mean": 0.006576488725841045,
      "epsilon_dpo/beta_margin_std": 0.04393243417143822,
      "epsilon_dpo/loss_margin_mean": 0.06914472579956055,
      "grad_norm": 22.595983505249023,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.1000625491142273,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 2.2388059701492534e-08,
      "logits/chosen": 0.6811072826385498,
      "logits/rejected": 0.5620886087417603,
      "logps/chosen": -100.42185974121094,
      "logps/ref_chosen": -100.4271240234375,
      "logps/ref_rejected": -107.98687744140625,
      "logps/rejected": -108.0507583618164,
      "loss": 1.3802,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.00037316768430173397,
      "rewards/margins": 0.006576500833034515,
      "rewards/rejected": -0.006203333381563425,
      "step": 4
    },
    {
      "epoch": 0.007558578987150416,
      "epsilon_dpo/beta": 0.09988515079021454,
      "epsilon_dpo/beta_margin_grad_mean": -0.4993220567703247,
      "epsilon_dpo/beta_margin_grad_std": 0.011706396006047726,
      "epsilon_dpo/beta_margin_mean": 0.0027205420192331076,
      "epsilon_dpo/beta_margin_std": 0.046862564980983734,
      "epsilon_dpo/loss_margin_mean": 0.030702859163284302,
      "grad_norm": 21.218976974487305,
      "kl/avg_steps": 0.0625,
      "kl/beta": 0.09993762522935867,
      "kl/n_epsilon_steps": 0.46875,
      "kl/p_epsilon_steps": 0.53125,
      "learning_rate": 2.9850746268656714e-08,
      "logits/chosen": 0.35812437534332275,
      "logits/rejected": 0.536173939704895,
      "logps/chosen": -86.76708984375,
      "logps/ref_chosen": -86.78726959228516,
      "logps/ref_rejected": -105.92151641845703,
      "logps/rejected": -105.93203735351562,
      "loss": 1.3841,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": 0.001866575563326478,
      "rewards/margins": 0.0027204796206206083,
      "rewards/rejected": -0.0008539038244634867,
      "step": 5
    },
    {
      "epoch": 0.009070294784580499,
      "epsilon_dpo/beta": 0.09994761645793915,
      "epsilon_dpo/beta_margin_grad_mean": -0.500268280506134,
      "epsilon_dpo/beta_margin_grad_std": 0.009158617816865444,
      "epsilon_dpo/beta_margin_mean": -0.0010719078127294779,
      "epsilon_dpo/beta_margin_std": 0.036647580564022064,
      "epsilon_dpo/loss_margin_mean": -0.007842868566513062,
      "grad_norm": 21.13234519958496,
      "kl/avg_steps": -0.0625,
      "kl/beta": 0.09987520426511765,
      "kl/n_epsilon_steps": 0.53125,
      "kl/p_epsilon_steps": 0.46875,
      "learning_rate": 3.731343283582089e-08,
      "logits/chosen": 0.33463966846466064,
      "logits/rejected": 0.5480047464370728,
      "logps/chosen": -73.50778198242188,
      "logps/ref_chosen": -73.4627914428711,
      "logps/ref_rejected": -108.36318969726562,
      "logps/rejected": -108.40033721923828,
      "loss": 1.3877,
      "rewards/accuracies": 0.4375,
      "rewards/chosen": -0.004621636122465134,
      "rewards/margins": -0.001071843784302473,
      "rewards/rejected": -0.003549792803823948,
      "step": 6
    },
    {
      "epoch": 0.010582010582010581,
      "epsilon_dpo/beta": 0.10004135221242905,
      "epsilon_dpo/beta_margin_grad_mean": -0.5008215308189392,
      "epsilon_dpo/beta_margin_grad_std": 0.00910845585167408,
      "epsilon_dpo/beta_margin_mean": -0.0032860664650797844,
      "epsilon_dpo/beta_margin_std": 0.03644711524248123,
      "epsilon_dpo/loss_margin_mean": -0.02999284863471985,
      "grad_norm": 20.586753845214844,
      "kl/avg_steps": -0.09375,
      "kl/beta": 0.09993766248226166,
      "kl/n_epsilon_steps": 0.546875,
      "kl/p_epsilon_steps": 0.453125,
      "learning_rate": 4.477611940298507e-08,
      "logits/chosen": 0.38425832986831665,
      "logits/rejected": 0.5074484944343567,
      "logps/chosen": -73.5091552734375,
      "logps/ref_chosen": -73.51263427734375,
      "logps/ref_rejected": -92.73175811767578,
      "logps/rejected": -92.69828796386719,
      "loss": 1.3899,
      "rewards/accuracies": 0.453125,
      "rewards/chosen": 0.00024446635507047176,
      "rewards/margins": -0.003286009654402733,
      "rewards/rejected": 0.0035304762423038483,
      "step": 7
    },
    {
      "epoch": 0.012093726379440665,
      "epsilon_dpo/beta": 0.10005691647529602,
      "epsilon_dpo/beta_margin_grad_mean": -0.5013768076896667,
      "epsilon_dpo/beta_margin_grad_std": 0.010522288270294666,
      "epsilon_dpo/beta_margin_mean": -0.005510755814611912,
      "epsilon_dpo/beta_margin_std": 0.04210984334349632,
      "epsilon_dpo/loss_margin_mean": -0.051828861236572266,
      "grad_norm": 21.454986572265625,
      "kl/avg_steps": -0.015625,
      "kl/beta": 0.10003144294023514,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.484375,
      "learning_rate": 5.223880597014925e-08,
      "logits/chosen": 0.44094252586364746,
      "logits/rejected": 0.6611981391906738,
      "logps/chosen": -86.21814727783203,
      "logps/ref_chosen": -86.1173095703125,
      "logps/ref_rejected": -97.60299682617188,
      "logps/rejected": -97.65200805664062,
      "loss": 1.3923,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": -0.010229799896478653,
      "rewards/margins": -0.005510762333869934,
      "rewards/rejected": -0.004719037562608719,
      "step": 8
    },
    {
      "epoch": 0.013605442176870748,
      "epsilon_dpo/beta": 0.10011961311101913,
      "epsilon_dpo/beta_margin_grad_mean": -0.5023381114006042,
      "epsilon_dpo/beta_margin_grad_std": 0.010044464841485023,
      "epsilon_dpo/beta_margin_mean": -0.009359496645629406,
      "epsilon_dpo/beta_margin_std": 0.04019934684038162,
      "epsilon_dpo/loss_margin_mean": -0.0903191864490509,
      "grad_norm": 22.43557357788086,
      "kl/avg_steps": -0.0625,
      "kl/beta": 0.10004707425832748,
      "kl/n_epsilon_steps": 0.53125,
      "kl/p_epsilon_steps": 0.46875,
      "learning_rate": 5.970149253731343e-08,
      "logits/chosen": 0.33967816829681396,
      "logits/rejected": 0.5344347953796387,
      "logps/chosen": -76.60382843017578,
      "logps/ref_chosen": -76.54713439941406,
      "logps/ref_rejected": -113.62154388427734,
      "logps/rejected": -113.5879135131836,
      "loss": 1.3961,
      "rewards/accuracies": 0.453125,
      "rewards/chosen": -0.005803982727229595,
      "rewards/margins": -0.009359488263726234,
      "rewards/rejected": 0.0035555048380047083,
      "step": 9
    },
    {
      "epoch": 0.015117157974300832,
      "epsilon_dpo/beta": 0.09994741529226303,
      "epsilon_dpo/beta_margin_grad_mean": -0.498788446187973,
      "epsilon_dpo/beta_margin_grad_std": 0.013752726837992668,
      "epsilon_dpo/beta_margin_mean": 0.0048447358421981335,
      "epsilon_dpo/beta_margin_std": 0.05508105456829071,
      "epsilon_dpo/loss_margin_mean": 0.05251148343086243,
      "grad_norm": 21.88376235961914,
      "kl/avg_steps": 0.171875,
      "kl/beta": 0.10010964423418045,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 6.71641791044776e-08,
      "logits/chosen": 0.3774455189704895,
      "logits/rejected": 0.5606910586357117,
      "logps/chosen": -86.106201171875,
      "logps/ref_chosen": -86.14695739746094,
      "logps/ref_rejected": -112.69779968261719,
      "logps/rejected": -112.70955657958984,
      "loss": 1.3822,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.003937458153814077,
      "rewards/margins": 0.004844787064939737,
      "rewards/rejected": -0.0009073292021639645,
      "step": 10
    },
    {
      "epoch": 0.016628873771730914,
      "epsilon_dpo/beta": 0.09985417127609253,
      "epsilon_dpo/beta_margin_grad_mean": -0.5002174377441406,
      "epsilon_dpo/beta_margin_grad_std": 0.011608804576098919,
      "epsilon_dpo/beta_margin_mean": -0.0008744366350583732,
      "epsilon_dpo/beta_margin_std": 0.04646685719490051,
      "epsilon_dpo/loss_margin_mean": -0.00528264045715332,
      "grad_norm": 20.917577743530273,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.09993787854909897,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 7.462686567164178e-08,
      "logits/chosen": 0.46160414814949036,
      "logits/rejected": 0.4182695746421814,
      "logps/chosen": -93.89750671386719,
      "logps/ref_chosen": -93.89774322509766,
      "logps/ref_rejected": -91.40093994140625,
      "logps/rejected": -91.39541625976562,
      "loss": 1.3877,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.0001967300195246935,
      "rewards/margins": -0.0008744364604353905,
      "rewards/rejected": 0.0006777062080800533,
      "step": 11
    },
    {
      "epoch": 0.018140589569160998,
      "epsilon_dpo/beta": 0.09980729967355728,
      "epsilon_dpo/beta_margin_grad_mean": -0.4989064633846283,
      "epsilon_dpo/beta_margin_grad_std": 0.011306311003863811,
      "epsilon_dpo/beta_margin_mean": 0.004372739233076572,
      "epsilon_dpo/beta_margin_std": 0.045261066406965256,
      "epsilon_dpo/loss_margin_mean": 0.047140806913375854,
      "grad_norm": 19.881134033203125,
      "kl/avg_steps": 0.046875,
      "kl/beta": 0.09984427690505981,
      "kl/n_epsilon_steps": 0.46875,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 8.208955223880596e-08,
      "logits/chosen": 0.39477378129959106,
      "logits/rejected": 0.4941557049751282,
      "logps/chosen": -86.1840591430664,
      "logps/ref_chosen": -86.17680358886719,
      "logps/ref_rejected": -97.00653076171875,
      "logps/rejected": -97.0609359741211,
      "loss": 1.3824,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.0008587964694015682,
      "rewards/margins": 0.004372742492705584,
      "rewards/rejected": -0.005231539253145456,
      "step": 12
    },
    {
      "epoch": 0.019652305366591082,
      "epsilon_dpo/beta": 0.09980747103691101,
      "epsilon_dpo/beta_margin_grad_mean": -0.5011844635009766,
      "epsilon_dpo/beta_margin_grad_std": 0.010257571935653687,
      "epsilon_dpo/beta_margin_mean": -0.004736810456961393,
      "epsilon_dpo/beta_margin_std": 0.041052356362342834,
      "epsilon_dpo/loss_margin_mean": -0.0443558394908905,
      "grad_norm": 20.99261474609375,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.09979749470949173,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 8.955223880597014e-08,
      "logits/chosen": 0.04624774307012558,
      "logits/rejected": 0.36173152923583984,
      "logps/chosen": -67.12196350097656,
      "logps/ref_chosen": -67.10835266113281,
      "logps/ref_rejected": -114.23330688476562,
      "logps/rejected": -114.20256042480469,
      "loss": 1.3915,
      "rewards/accuracies": 0.453125,
      "rewards/chosen": -0.0015045893378555775,
      "rewards/margins": -0.00473686121404171,
      "rewards/rejected": 0.003232272807508707,
      "step": 13
    },
    {
      "epoch": 0.021164021164021163,
      "epsilon_dpo/beta": 0.09983865916728973,
      "epsilon_dpo/beta_margin_grad_mean": -0.5012878775596619,
      "epsilon_dpo/beta_margin_grad_std": 0.010014628060162067,
      "epsilon_dpo/beta_margin_mean": -0.005156196188181639,
      "epsilon_dpo/beta_margin_std": 0.040078554302453995,
      "epsilon_dpo/loss_margin_mean": -0.0487152636051178,
      "grad_norm": 22.20970344543457,
      "kl/avg_steps": -0.03125,
      "kl/beta": 0.09979749470949173,
      "kl/n_epsilon_steps": 0.515625,
      "kl/p_epsilon_steps": 0.484375,
      "learning_rate": 9.701492537313432e-08,
      "logits/chosen": 0.3906971514225006,
      "logits/rejected": 0.523829460144043,
      "logps/chosen": -87.2183837890625,
      "logps/ref_chosen": -87.24519348144531,
      "logps/ref_rejected": -96.43710327148438,
      "logps/rejected": -96.36158752441406,
      "loss": 1.3919,
      "rewards/accuracies": 0.46875,
      "rewards/chosen": 0.0025335517711937428,
      "rewards/margins": -0.005156217608600855,
      "rewards/rejected": 0.007689769379794598,
      "step": 14
    },
    {
      "epoch": 0.022675736961451247,
      "epsilon_dpo/beta": 0.09971387684345245,
      "epsilon_dpo/beta_margin_grad_mean": -0.4991253614425659,
      "epsilon_dpo/beta_margin_grad_std": 0.012193976901471615,
      "epsilon_dpo/beta_margin_mean": 0.0035042737144976854,
      "epsilon_dpo/beta_margin_std": 0.0488242469727993,
      "epsilon_dpo/loss_margin_mean": 0.038533955812454224,
      "grad_norm": 22.38343048095703,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.09982869029045105,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 1.044776119402985e-07,
      "logits/chosen": 0.41666990518569946,
      "logits/rejected": 0.5997655391693115,
      "logps/chosen": -86.97470092773438,
      "logps/ref_chosen": -87.02511596679688,
      "logps/ref_rejected": -115.12210845947266,
      "logps/rejected": -115.1102294921875,
      "loss": 1.3834,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": 0.004882176406681538,
      "rewards/margins": 0.003504345426335931,
      "rewards/rejected": 0.0013778312131762505,
      "step": 15
    },
    {
      "epoch": 0.02418745275888133,
      "epsilon_dpo/beta": 0.09968286752700806,
      "epsilon_dpo/beta_margin_grad_mean": -0.4998158812522888,
      "epsilon_dpo/beta_margin_grad_std": 0.007838121615350246,
      "epsilon_dpo/beta_margin_mean": 0.0007361461757682264,
      "epsilon_dpo/beta_margin_std": 0.03135906532406807,
      "epsilon_dpo/loss_margin_mean": 0.009959220886230469,
      "grad_norm": 19.675779342651367,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.09970405697822571,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 1.1194029850746268e-07,
      "logits/chosen": 0.36139073967933655,
      "logits/rejected": 0.5083924531936646,
      "logps/chosen": -65.78361511230469,
      "logps/ref_chosen": -65.84059143066406,
      "logps/ref_rejected": -95.32190704345703,
      "logps/rejected": -95.27488708496094,
      "loss": 1.3858,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": 0.005595714319497347,
      "rewards/margins": 0.0007362178876064718,
      "rewards/rejected": 0.004859496373683214,
      "step": 16
    },
    {
      "epoch": 0.025699168556311415,
      "epsilon_dpo/beta": 0.09983862936496735,
      "epsilon_dpo/beta_margin_grad_mean": -0.5021052956581116,
      "epsilon_dpo/beta_margin_grad_std": 0.0091917235404253,
      "epsilon_dpo/beta_margin_mean": -0.008424973115324974,
      "epsilon_dpo/beta_margin_std": 0.03678138181567192,
      "epsilon_dpo/loss_margin_mean": -0.08158034086227417,
      "grad_norm": 21.948225021362305,
      "kl/avg_steps": -0.15625,
      "kl/beta": 0.09967291355133057,
      "kl/n_epsilon_steps": 0.578125,
      "kl/p_epsilon_steps": 0.421875,
      "learning_rate": 1.1940298507462686e-07,
      "logits/chosen": 0.265707790851593,
      "logits/rejected": 0.5080006122589111,
      "logps/chosen": -91.50019836425781,
      "logps/ref_chosen": -91.41425323486328,
      "logps/ref_rejected": -109.13850402832031,
      "logps/rejected": -109.14286804199219,
      "loss": 1.3951,
      "rewards/accuracies": 0.421875,
      "rewards/chosen": -0.008698860183358192,
      "rewards/margins": -0.008425043895840645,
      "rewards/rejected": -0.00027381590916775167,
      "step": 17
    },
    {
      "epoch": 0.027210884353741496,
      "epsilon_dpo/beta": 0.09996367245912552,
      "epsilon_dpo/beta_margin_grad_mean": -0.5014711022377014,
      "epsilon_dpo/beta_margin_grad_std": 0.010209612548351288,
      "epsilon_dpo/beta_margin_mean": -0.005889255087822676,
      "epsilon_dpo/beta_margin_std": 0.0408661924302578,
      "epsilon_dpo/loss_margin_mean": -0.055916786193847656,
      "grad_norm": 20.114952087402344,
      "kl/avg_steps": -0.125,
      "kl/beta": 0.09982889145612717,
      "kl/n_epsilon_steps": 0.5625,
      "kl/p_epsilon_steps": 0.4375,
      "learning_rate": 1.2686567164179106e-07,
      "logits/chosen": 0.295833557844162,
      "logits/rejected": 0.4196588397026062,
      "logps/chosen": -67.54663848876953,
      "logps/ref_chosen": -67.49713134765625,
      "logps/ref_rejected": -91.58694458007812,
      "logps/rejected": -91.58052825927734,
      "loss": 1.3926,
      "rewards/accuracies": 0.46875,
      "rewards/chosen": -0.005046804901212454,
      "rewards/margins": -0.005889310501515865,
      "rewards/rejected": 0.0008425057167187333,
      "step": 18
    },
    {
      "epoch": 0.02872260015117158,
      "epsilon_dpo/beta": 0.09993259608745575,
      "epsilon_dpo/beta_margin_grad_mean": -0.5004714727401733,
      "epsilon_dpo/beta_margin_grad_std": 0.01001923531293869,
      "epsilon_dpo/beta_margin_mean": -0.001890135812573135,
      "epsilon_dpo/beta_margin_std": 0.040112897753715515,
      "epsilon_dpo/loss_margin_mean": -0.01608338952064514,
      "grad_norm": 21.74436378479004,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.09995383769273758,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 1.343283582089552e-07,
      "logits/chosen": 0.534844696521759,
      "logits/rejected": 0.6198501586914062,
      "logps/chosen": -92.06356811523438,
      "logps/ref_chosen": -92.03971862792969,
      "logps/ref_rejected": -106.87828063964844,
      "logps/rejected": -106.88603210449219,
      "loss": 1.3886,
      "rewards/accuracies": 0.5,
      "rewards/chosen": -0.002494135871529579,
      "rewards/margins": -0.001890112180262804,
      "rewards/rejected": -0.0006040236912667751,
      "step": 19
    },
    {
      "epoch": 0.030234315948601664,
      "epsilon_dpo/beta": 0.09987014532089233,
      "epsilon_dpo/beta_margin_grad_mean": -0.49894291162490845,
      "epsilon_dpo/beta_margin_grad_std": 0.008131814189255238,
      "epsilon_dpo/beta_margin_mean": 0.004229863174259663,
      "epsilon_dpo/beta_margin_std": 0.032535601407289505,
      "epsilon_dpo/loss_margin_mean": 0.04488375782966614,
      "grad_norm": 21.32170295715332,
      "kl/avg_steps": 0.0625,
      "kl/beta": 0.09992261230945587,
      "kl/n_epsilon_steps": 0.46875,
      "kl/p_epsilon_steps": 0.53125,
      "learning_rate": 1.4179104477611938e-07,
      "logits/chosen": 0.298221617937088,
      "logits/rejected": 0.36488860845565796,
      "logps/chosen": -79.46307373046875,
      "logps/ref_chosen": -79.5280532836914,
      "logps/ref_rejected": -96.36955261230469,
      "logps/rejected": -96.34945678710938,
      "loss": 1.3823,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": 0.006376877427101135,
      "rewards/margins": 0.00422988086938858,
      "rewards/rejected": 0.002146995859220624,
      "step": 20
    },
    {
      "epoch": 0.031746031746031744,
      "epsilon_dpo/beta": 0.09987018257379532,
      "epsilon_dpo/beta_margin_grad_mean": -0.5001654624938965,
      "epsilon_dpo/beta_margin_grad_std": 0.010333231650292873,
      "epsilon_dpo/beta_margin_mean": -0.0006611068965867162,
      "epsilon_dpo/beta_margin_std": 0.04135696962475777,
      "epsilon_dpo/loss_margin_mean": -0.003520876169204712,
      "grad_norm": 20.74885368347168,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.09986019879579544,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 1.4925373134328355e-07,
      "logits/chosen": 0.32211834192276,
      "logits/rejected": 0.4015572667121887,
      "logps/chosen": -87.98433685302734,
      "logps/ref_chosen": -87.99819946289062,
      "logps/ref_rejected": -103.26502990722656,
      "logps/rejected": -103.24763488769531,
      "loss": 1.3874,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.0012191254645586014,
      "rewards/margins": -0.0006612140568904579,
      "rewards/rejected": 0.0018803395796567202,
      "step": 21
    },
    {
      "epoch": 0.03325774754346183,
      "epsilon_dpo/beta": 0.0998389720916748,
      "epsilon_dpo/beta_margin_grad_mean": -0.49859529733657837,
      "epsilon_dpo/beta_margin_grad_std": 0.00828811526298523,
      "epsilon_dpo/beta_margin_mean": 0.005621983669698238,
      "epsilon_dpo/beta_margin_std": 0.033166419714689255,
      "epsilon_dpo/loss_margin_mean": 0.05865293741226196,
      "grad_norm": 20.97919464111328,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.09986019879579544,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 1.5671641791044775e-07,
      "logits/chosen": 0.23899997770786285,
      "logits/rejected": 0.5824065804481506,
      "logps/chosen": -81.6365966796875,
      "logps/ref_chosen": -81.64871978759766,
      "logps/ref_rejected": -100.75686645507812,
      "logps/rejected": -100.80340576171875,
      "loss": 1.381,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": 0.0011308013927191496,
      "rewards/margins": 0.0056219929829239845,
      "rewards/rejected": -0.004491191357374191,
      "step": 22
    },
    {
      "epoch": 0.03476946334089191,
      "epsilon_dpo/beta": 0.099838986992836,
      "epsilon_dpo/beta_margin_grad_mean": -0.4990421533584595,
      "epsilon_dpo/beta_margin_grad_std": 0.010845753364264965,
      "epsilon_dpo/beta_margin_mean": 0.003838536562398076,
      "epsilon_dpo/beta_margin_std": 0.043410323560237885,
      "epsilon_dpo/loss_margin_mean": 0.0417940616607666,
      "grad_norm": 21.895824432373047,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.09982900321483612,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 1.6417910447761193e-07,
      "logits/chosen": 0.3985062539577484,
      "logits/rejected": 0.523756742477417,
      "logps/chosen": -92.14674377441406,
      "logps/ref_chosen": -92.16307830810547,
      "logps/ref_rejected": -113.73563385009766,
      "logps/rejected": -113.76109313964844,
      "loss": 1.3829,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.0014242115430533886,
      "rewards/margins": 0.0038385987281799316,
      "rewards/rejected": -0.0024143874179571867,
      "step": 23
    },
    {
      "epoch": 0.036281179138321996,
      "epsilon_dpo/beta": 0.099838986992836,
      "epsilon_dpo/beta_margin_grad_mean": -0.49900925159454346,
      "epsilon_dpo/beta_margin_grad_std": 0.01059596799314022,
      "epsilon_dpo/beta_margin_mean": 0.0039714789018034935,
      "epsilon_dpo/beta_margin_std": 0.04242187365889549,
      "epsilon_dpo/loss_margin_mean": 0.04267817735671997,
      "grad_norm": 20.0445556640625,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.09982900321483612,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 1.716417910447761e-07,
      "logits/chosen": 0.23830129206180573,
      "logits/rejected": 0.3382936120033264,
      "logps/chosen": -74.7188491821289,
      "logps/ref_chosen": -74.73088073730469,
      "logps/ref_rejected": -90.28663635253906,
      "logps/rejected": -90.31729125976562,
      "loss": 1.3828,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.0010350291850045323,
      "rewards/margins": 0.003971490543335676,
      "rewards/rejected": -0.0029364614747464657,
      "step": 24
    },
    {
      "epoch": 0.03779289493575208,
      "epsilon_dpo/beta": 0.09974538534879684,
      "epsilon_dpo/beta_margin_grad_mean": -0.5002205967903137,
      "epsilon_dpo/beta_margin_grad_std": 0.012343145906925201,
      "epsilon_dpo/beta_margin_mean": -0.0008880018722265959,
      "epsilon_dpo/beta_margin_std": 0.04941296949982643,
      "epsilon_dpo/loss_margin_mean": -0.005264639854431152,
      "grad_norm": 21.26227569580078,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.09982900321483612,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 1.7910447761194027e-07,
      "logits/chosen": 0.38851284980773926,
      "logits/rejected": 0.5324080586433411,
      "logps/chosen": -81.98751068115234,
      "logps/ref_chosen": -81.99627685546875,
      "logps/ref_rejected": -114.30165100097656,
      "logps/rejected": -114.2876205444336,
      "loss": 1.3878,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.0007206163718365133,
      "rewards/margins": -0.0008880080422386527,
      "rewards/rejected": 0.001608624355867505,
      "step": 25
    },
    {
      "epoch": 0.039304610733182165,
      "epsilon_dpo/beta": 0.09977664053440094,
      "epsilon_dpo/beta_margin_grad_mean": -0.500114381313324,
      "epsilon_dpo/beta_margin_grad_std": 0.009109625592827797,
      "epsilon_dpo/beta_margin_mean": -0.0004547676944639534,
      "epsilon_dpo/beta_margin_std": 0.03645933419466019,
      "epsilon_dpo/loss_margin_mean": -0.0018818974494934082,
      "grad_norm": 21.053829193115234,
      "kl/avg_steps": -0.03125,
      "kl/beta": 0.09973549842834473,
      "kl/n_epsilon_steps": 0.515625,
      "kl/p_epsilon_steps": 0.484375,
      "learning_rate": 1.8656716417910447e-07,
      "logits/chosen": 0.4273748993873596,
      "logits/rejected": 0.6181799173355103,
      "logps/chosen": -80.81678771972656,
      "logps/ref_chosen": -80.84437561035156,
      "logps/ref_rejected": -99.3437271118164,
      "logps/rejected": -99.31425476074219,
      "loss": 1.3871,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.0025958300102502108,
      "rewards/margins": -0.0004547259886749089,
      "rewards/rejected": 0.0030505559407174587,
      "step": 26
    },
    {
      "epoch": 0.04081632653061224,
      "epsilon_dpo/beta": 0.09987019002437592,
      "epsilon_dpo/beta_margin_grad_mean": -0.5037402510643005,
      "epsilon_dpo/beta_margin_grad_std": 0.01256086491048336,
      "epsilon_dpo/beta_margin_mean": -0.014985705725848675,
      "epsilon_dpo/beta_margin_std": 0.050342340022325516,
      "epsilon_dpo/loss_margin_mean": -0.14665690064430237,
      "grad_norm": 21.79582977294922,
      "kl/avg_steps": -0.09375,
      "kl/beta": 0.09976667910814285,
      "kl/n_epsilon_steps": 0.546875,
      "kl/p_epsilon_steps": 0.453125,
      "learning_rate": 1.9402985074626865e-07,
      "logits/chosen": 0.2670428454875946,
      "logits/rejected": 0.4372353255748749,
      "logps/chosen": -85.55349731445312,
      "logps/ref_chosen": -85.45787811279297,
      "logps/ref_rejected": -100.81729888916016,
      "logps/rejected": -100.76626586914062,
      "loss": 1.402,
      "rewards/accuracies": 0.453125,
      "rewards/chosen": -0.00972618255764246,
      "rewards/margins": -0.014985740184783936,
      "rewards/rejected": 0.005259557627141476,
      "step": 27
    },
    {
      "epoch": 0.042328042328042326,
      "epsilon_dpo/beta": 0.09996391087770462,
      "epsilon_dpo/beta_margin_grad_mean": -0.5001952648162842,
      "epsilon_dpo/beta_margin_grad_std": 0.011885374784469604,
      "epsilon_dpo/beta_margin_mean": -0.0007757341954857111,
      "epsilon_dpo/beta_margin_std": 0.04760770499706268,
      "epsilon_dpo/loss_margin_mean": -0.004470318555831909,
      "grad_norm": 20.6678524017334,
      "kl/avg_steps": -0.09375,
      "kl/beta": 0.0998602956533432,
      "kl/n_epsilon_steps": 0.546875,
      "kl/p_epsilon_steps": 0.453125,
      "learning_rate": 2.0149253731343282e-07,
      "logits/chosen": 0.343519926071167,
      "logits/rejected": 0.44987940788269043,
      "logps/chosen": -82.88597106933594,
      "logps/ref_chosen": -82.87960815429688,
      "logps/ref_rejected": -95.13070678710938,
      "logps/rejected": -95.13259887695312,
      "loss": 1.3876,
      "rewards/accuracies": 0.4375,
      "rewards/chosen": -0.0007612911285832524,
      "rewards/margins": -0.0007756366394460201,
      "rewards/rejected": 1.434539444744587e-05,
      "step": 28
    },
    {
      "epoch": 0.04383975812547241,
      "epsilon_dpo/beta": 0.09996399283409119,
      "epsilon_dpo/beta_margin_grad_mean": -0.49949145317077637,
      "epsilon_dpo/beta_margin_grad_std": 0.010683816857635975,
      "epsilon_dpo/beta_margin_mean": 0.002037009224295616,
      "epsilon_dpo/beta_margin_std": 0.042756691575050354,
      "epsilon_dpo/loss_margin_mean": 0.023626625537872314,
      "grad_norm": 23.955188751220703,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.09995400160551071,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 2.08955223880597e-07,
      "logits/chosen": 0.44967925548553467,
      "logits/rejected": 0.4150964021682739,
      "logps/chosen": -96.74322509765625,
      "logps/ref_chosen": -96.73393249511719,
      "logps/ref_rejected": -106.6417007446289,
      "logps/rejected": -106.67463684082031,
      "loss": 1.3847,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": -0.0010466062230989337,
      "rewards/margins": 0.0020370446145534515,
      "rewards/rejected": -0.003083650954067707,
      "step": 29
    },
    {
      "epoch": 0.045351473922902494,
      "epsilon_dpo/beta": 0.0999482199549675,
      "epsilon_dpo/beta_margin_grad_mean": -0.5016565918922424,
      "epsilon_dpo/beta_margin_grad_std": 0.01134086586534977,
      "epsilon_dpo/beta_margin_mean": -0.006640481762588024,
      "epsilon_dpo/beta_margin_std": 0.045420292764902115,
      "epsilon_dpo/loss_margin_mean": -0.06311115622520447,
      "grad_norm": 22.373016357421875,
      "kl/avg_steps": 0.015625,
      "kl/beta": 0.09995400160551071,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 2.1641791044776117e-07,
      "logits/chosen": 0.2677932381629944,
      "logits/rejected": 0.44384220242500305,
      "logps/chosen": -73.20950317382812,
      "logps/ref_chosen": -73.2177734375,
      "logps/ref_rejected": -111.69508361816406,
      "logps/rejected": -111.62370300292969,
      "loss": 1.3935,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.0007222597487270832,
      "rewards/margins": -0.006640553940087557,
      "rewards/rejected": 0.00736281368881464,
      "step": 30
    },
    {
      "epoch": 0.04686318972033258,
      "epsilon_dpo/beta": 0.09985467791557312,
      "epsilon_dpo/beta_margin_grad_mean": -0.4987980127334595,
      "epsilon_dpo/beta_margin_grad_std": 0.01364812720566988,
      "epsilon_dpo/beta_margin_mean": 0.0048094564117491245,
      "epsilon_dpo/beta_margin_std": 0.05463935062289238,
      "epsilon_dpo/loss_margin_mean": 0.052467405796051025,
      "grad_norm": 24.33083724975586,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.09993838518857956,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 2.2388059701492537e-07,
      "logits/chosen": 0.2801819145679474,
      "logits/rejected": 0.494117796421051,
      "logps/chosen": -91.95301818847656,
      "logps/ref_chosen": -91.93222045898438,
      "logps/ref_rejected": -126.33772277832031,
      "logps/rejected": -126.41098022460938,
      "loss": 1.3822,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.0022574204485863447,
      "rewards/margins": 0.004809522069990635,
      "rewards/rejected": -0.007066942285746336,
      "step": 31
    },
    {
      "epoch": 0.04837490551776266,
      "epsilon_dpo/beta": 0.09991717338562012,
      "epsilon_dpo/beta_margin_grad_mean": -0.5006169080734253,
      "epsilon_dpo/beta_margin_grad_std": 0.009541872888803482,
      "epsilon_dpo/beta_margin_mean": -0.0024695084430277348,
      "epsilon_dpo/beta_margin_std": 0.038185965269804,
      "epsilon_dpo/loss_margin_mean": -0.021846860647201538,
      "grad_norm": 21.380720138549805,
      "kl/avg_steps": -0.0625,
      "kl/beta": 0.0998447835445404,
      "kl/n_epsilon_steps": 0.53125,
      "kl/p_epsilon_steps": 0.46875,
      "learning_rate": 2.3134328358208954e-07,
      "logits/chosen": 0.04479830712080002,
      "logits/rejected": 0.5272079110145569,
      "logps/chosen": -94.07949829101562,
      "logps/ref_chosen": -94.01815795898438,
      "logps/ref_rejected": -106.67022705078125,
      "logps/rejected": -106.709716796875,
      "loss": 1.3891,
      "rewards/accuracies": 0.421875,
      "rewards/chosen": -0.006249488331377506,
      "rewards/margins": -0.0024695044849067926,
      "rewards/rejected": -0.0037799840793013573,
      "step": 32
    },
    {
      "epoch": 0.049886621315192746,
      "epsilon_dpo/beta": 0.09966742247343063,
      "epsilon_dpo/beta_margin_grad_mean": -0.4978310465812683,
      "epsilon_dpo/beta_margin_grad_std": 0.011500690132379532,
      "epsilon_dpo/beta_margin_mean": 0.008678106591105461,
      "epsilon_dpo/beta_margin_std": 0.04604245722293854,
      "epsilon_dpo/loss_margin_mean": 0.09031841158866882,
      "grad_norm": 22.49273681640625,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.09990722686052322,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.388059701492537e-07,
      "logits/chosen": 0.21207210421562195,
      "logits/rejected": 0.4895442724227905,
      "logps/chosen": -89.40026092529297,
      "logps/ref_chosen": -89.4225082397461,
      "logps/ref_rejected": -105.73570251464844,
      "logps/rejected": -105.80377197265625,
      "loss": 1.3782,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": 0.002066537970677018,
      "rewards/margins": 0.008678080514073372,
      "rewards/rejected": -0.006611541844904423,
      "step": 33
    },
    {
      "epoch": 0.05139833711262283,
      "epsilon_dpo/beta": 0.0996057540178299,
      "epsilon_dpo/beta_margin_grad_mean": -0.4997608959674835,
      "epsilon_dpo/beta_margin_grad_std": 0.009196538478136063,
      "epsilon_dpo/beta_margin_mean": 0.0009559081518091261,
      "epsilon_dpo/beta_margin_std": 0.03679739311337471,
      "epsilon_dpo/loss_margin_mean": 0.012517750263214111,
      "grad_norm": 20.236793518066406,
      "kl/avg_steps": 0.0625,
      "kl/beta": 0.09965807944536209,
      "kl/n_epsilon_steps": 0.46875,
      "kl/p_epsilon_steps": 0.53125,
      "learning_rate": 2.4626865671641786e-07,
      "logits/chosen": 0.2664468288421631,
      "logits/rejected": 0.48599302768707275,
      "logps/chosen": -75.28192138671875,
      "logps/ref_chosen": -75.25447845458984,
      "logps/ref_rejected": -90.25471496582031,
      "logps/rejected": -90.294677734375,
      "loss": 1.3857,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.002886722795665264,
      "rewards/margins": 0.0009559143800288439,
      "rewards/rejected": -0.0038426369428634644,
      "step": 34
    },
    {
      "epoch": 0.05291005291005291,
      "epsilon_dpo/beta": 0.09963691234588623,
      "epsilon_dpo/beta_margin_grad_mean": -0.500873327255249,
      "epsilon_dpo/beta_margin_grad_std": 0.009642170742154121,
      "epsilon_dpo/beta_margin_mean": -0.003494812408462167,
      "epsilon_dpo/beta_margin_std": 0.03858347609639168,
      "epsilon_dpo/loss_margin_mean": -0.03209325671195984,
      "grad_norm": 20.750717163085938,
      "kl/avg_steps": -0.03125,
      "kl/beta": 0.09959582984447479,
      "kl/n_epsilon_steps": 0.515625,
      "kl/p_epsilon_steps": 0.484375,
      "learning_rate": 2.537313432835821e-07,
      "logits/chosen": 0.2340129315853119,
      "logits/rejected": 0.3774367570877075,
      "logps/chosen": -84.54783630371094,
      "logps/ref_chosen": -84.49177551269531,
      "logps/ref_rejected": -96.995849609375,
      "logps/rejected": -97.01981353759766,
      "loss": 1.3902,
      "rewards/accuracies": 0.5,
      "rewards/chosen": -0.005702049471437931,
      "rewards/margins": -0.003494792152196169,
      "rewards/rejected": -0.002207257319241762,
      "step": 35
    },
    {
      "epoch": 0.05442176870748299,
      "epsilon_dpo/beta": 0.09960579127073288,
      "epsilon_dpo/beta_margin_grad_mean": -0.4998779594898224,
      "epsilon_dpo/beta_margin_grad_std": 0.012096166610717773,
      "epsilon_dpo/beta_margin_mean": 0.000493088096845895,
      "epsilon_dpo/beta_margin_std": 0.04841361939907074,
      "epsilon_dpo/loss_margin_mean": 0.008720353245735168,
      "grad_norm": 21.406538009643555,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.09962696582078934,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 2.611940298507462e-07,
      "logits/chosen": 0.23152907192707062,
      "logits/rejected": 0.601810097694397,
      "logps/chosen": -72.32382202148438,
      "logps/ref_chosen": -72.29872131347656,
      "logps/ref_rejected": -106.939453125,
      "logps/rejected": -106.9732894897461,
      "loss": 1.3864,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": -0.0026765610091388226,
      "rewards/margins": 0.000493088096845895,
      "rewards/rejected": -0.0031696492806077003,
      "step": 36
    },
    {
      "epoch": 0.055933484504913075,
      "epsilon_dpo/beta": 0.09951242059469223,
      "epsilon_dpo/beta_margin_grad_mean": -0.49880462884902954,
      "epsilon_dpo/beta_margin_grad_std": 0.011057516559958458,
      "epsilon_dpo/beta_margin_mean": 0.004784699063748121,
      "epsilon_dpo/beta_margin_std": 0.0442543663084507,
      "epsilon_dpo/loss_margin_mean": 0.051472991704940796,
      "grad_norm": 26.469751358032227,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.09959584474563599,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 2.686567164179104e-07,
      "logits/chosen": 0.2194148302078247,
      "logits/rejected": 0.3654429614543915,
      "logps/chosen": -81.1114501953125,
      "logps/ref_chosen": -81.10102844238281,
      "logps/ref_rejected": -144.65945434570312,
      "logps/rejected": -144.72134399414062,
      "loss": 1.382,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.0012095842976123095,
      "rewards/margins": 0.004784717224538326,
      "rewards/rejected": -0.005994301289319992,
      "step": 37
    },
    {
      "epoch": 0.05744520030234316,
      "epsilon_dpo/beta": 0.09941920638084412,
      "epsilon_dpo/beta_margin_grad_mean": -0.5004234313964844,
      "epsilon_dpo/beta_margin_grad_std": 0.010645652189850807,
      "epsilon_dpo/beta_margin_mean": -0.0017045249696820974,
      "epsilon_dpo/beta_margin_std": 0.04264267161488533,
      "epsilon_dpo/loss_margin_mean": -0.014012634754180908,
      "grad_norm": 21.179275512695312,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.0995025560259819,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 2.761194029850746e-07,
      "logits/chosen": 0.7173947095870972,
      "logits/rejected": 0.7745821475982666,
      "logps/chosen": -90.1217041015625,
      "logps/ref_chosen": -90.05032348632812,
      "logps/ref_rejected": -96.94110107421875,
      "logps/rejected": -96.99847412109375,
      "loss": 1.3885,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.007230603136122227,
      "rewards/margins": -0.0017045974964275956,
      "rewards/rejected": -0.005526005756109953,
      "step": 38
    },
    {
      "epoch": 0.05895691609977324,
      "epsilon_dpo/beta": 0.09935716539621353,
      "epsilon_dpo/beta_margin_grad_mean": -0.4987095892429352,
      "epsilon_dpo/beta_margin_grad_std": 0.009196286089718342,
      "epsilon_dpo/beta_margin_mean": 0.005164078436791897,
      "epsilon_dpo/beta_margin_std": 0.03680035471916199,
      "epsilon_dpo/loss_margin_mean": 0.0548284649848938,
      "grad_norm": 21.44578742980957,
      "kl/avg_steps": 0.0625,
      "kl/beta": 0.09940936416387558,
      "kl/n_epsilon_steps": 0.46875,
      "kl/p_epsilon_steps": 0.53125,
      "learning_rate": 2.8358208955223876e-07,
      "logits/chosen": 0.5785679221153259,
      "logits/rejected": 0.6269518136978149,
      "logps/chosen": -81.97523498535156,
      "logps/ref_chosen": -81.93487548828125,
      "logps/ref_rejected": -103.41276550292969,
      "logps/rejected": -103.50794982910156,
      "loss": 1.3815,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.004156146664172411,
      "rewards/margins": 0.005164049565792084,
      "rewards/rejected": -0.009320196695625782,
      "step": 39
    },
    {
      "epoch": 0.06046863189720333,
      "epsilon_dpo/beta": 0.09920196235179901,
      "epsilon_dpo/beta_margin_grad_mean": -0.49731066823005676,
      "epsilon_dpo/beta_margin_grad_std": 0.011403567157685757,
      "epsilon_dpo/beta_margin_mean": 0.010767289437353611,
      "epsilon_dpo/beta_margin_std": 0.04566150903701782,
      "epsilon_dpo/loss_margin_mean": 0.11194059252738953,
      "grad_norm": 20.931968688964844,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.09934727102518082,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 2.9104477611940296e-07,
      "logits/chosen": 0.4167971611022949,
      "logits/rejected": 0.3265213370323181,
      "logps/chosen": -83.24398803710938,
      "logps/ref_chosen": -83.23312377929688,
      "logps/ref_rejected": -104.69706726074219,
      "logps/rejected": -104.81986999511719,
      "loss": 1.3761,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.0012151999399065971,
      "rewards/margins": 0.010767241939902306,
      "rewards/rejected": -0.011982442811131477,
      "step": 40
    },
    {
      "epoch": 0.06198034769463341,
      "epsilon_dpo/beta": 0.09929520636796951,
      "epsilon_dpo/beta_margin_grad_mean": -0.5001014471054077,
      "epsilon_dpo/beta_margin_grad_std": 0.00980335846543312,
      "epsilon_dpo/beta_margin_mean": -0.00040489688399247825,
      "epsilon_dpo/beta_margin_std": 0.03922676667571068,
      "epsilon_dpo/loss_margin_mean": -0.000916600227355957,
      "grad_norm": 21.39777183532715,
      "kl/avg_steps": -0.09375,
      "kl/beta": 0.09919228404760361,
      "kl/n_epsilon_steps": 0.546875,
      "kl/p_epsilon_steps": 0.453125,
      "learning_rate": 2.985074626865671e-07,
      "logits/chosen": 0.39663296937942505,
      "logits/rejected": 0.3223673105239868,
      "logps/chosen": -92.5504150390625,
      "logps/ref_chosen": -92.5335693359375,
      "logps/ref_rejected": -107.55760192871094,
      "logps/rejected": -107.57351684570312,
      "loss": 1.3871,
      "rewards/accuracies": 0.46875,
      "rewards/chosen": -0.0017897419165819883,
      "rewards/margins": -0.0004048999398946762,
      "rewards/rejected": -0.0013848419766873121,
      "step": 41
    },
    {
      "epoch": 0.06349206349206349,
      "epsilon_dpo/beta": 0.09917117655277252,
      "epsilon_dpo/beta_margin_grad_mean": -0.5006127953529358,
      "epsilon_dpo/beta_margin_grad_std": 0.011890949681401253,
      "epsilon_dpo/beta_margin_mean": -0.0024510445073246956,
      "epsilon_dpo/beta_margin_std": 0.047598425298929214,
      "epsilon_dpo/loss_margin_mean": -0.021076232194900513,
      "grad_norm": 21.89639663696289,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.09928536415100098,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 3.059701492537313e-07,
      "logits/chosen": 0.35543912649154663,
      "logits/rejected": 0.3687231242656708,
      "logps/chosen": -84.61444091796875,
      "logps/ref_chosen": -84.56074523925781,
      "logps/ref_rejected": -98.42295837402344,
      "logps/rejected": -98.45557403564453,
      "loss": 1.3893,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.005525700282305479,
      "rewards/margins": -0.002451097359880805,
      "rewards/rejected": -0.0030746031552553177,
      "step": 42
    },
    {
      "epoch": 0.06500377928949358,
      "epsilon_dpo/beta": 0.09910934418439865,
      "epsilon_dpo/beta_margin_grad_mean": -0.4985501766204834,
      "epsilon_dpo/beta_margin_grad_std": 0.011431388556957245,
      "epsilon_dpo/beta_margin_mean": 0.0058038788847625256,
      "epsilon_dpo/beta_margin_std": 0.04577384516596794,
      "epsilon_dpo/loss_margin_mean": 0.06193992495536804,
      "grad_norm": 21.574052810668945,
      "kl/avg_steps": 0.0625,
      "kl/beta": 0.09916140884160995,
      "kl/n_epsilon_steps": 0.46875,
      "kl/p_epsilon_steps": 0.53125,
      "learning_rate": 3.134328358208955e-07,
      "logits/chosen": 0.5684385299682617,
      "logits/rejected": 0.4524195194244385,
      "logps/chosen": -97.50875854492188,
      "logps/ref_chosen": -97.4771728515625,
      "logps/ref_rejected": -100.22068786621094,
      "logps/rejected": -100.314208984375,
      "loss": 1.381,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.0032527425792068243,
      "rewards/margins": 0.00580390403047204,
      "rewards/rejected": -0.00905664637684822,
      "step": 43
    },
    {
      "epoch": 0.06651549508692366,
      "epsilon_dpo/beta": 0.09895452111959457,
      "epsilon_dpo/beta_margin_grad_mean": -0.4991898238658905,
      "epsilon_dpo/beta_margin_grad_std": 0.013249047100543976,
      "epsilon_dpo/beta_margin_mean": 0.0032492447644472122,
      "epsilon_dpo/beta_margin_std": 0.05304902419447899,
      "epsilon_dpo/loss_margin_mean": 0.03694462776184082,
      "grad_norm": 21.381933212280273,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.09909947216510773,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 3.2089552238805965e-07,
      "logits/chosen": 0.23688527941703796,
      "logits/rejected": 0.36477425694465637,
      "logps/chosen": -94.30439758300781,
      "logps/ref_chosen": -94.24890899658203,
      "logps/ref_rejected": -107.57559967041016,
      "logps/rejected": -107.66804504394531,
      "loss": 1.3838,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.005697104148566723,
      "rewards/margins": 0.0032492445316165686,
      "rewards/rejected": -0.00894634798169136,
      "step": 44
    },
    {
      "epoch": 0.06802721088435375,
      "epsilon_dpo/beta": 0.09866083413362503,
      "epsilon_dpo/beta_margin_grad_mean": -0.495146244764328,
      "epsilon_dpo/beta_margin_grad_std": 0.011300712823867798,
      "epsilon_dpo/beta_margin_mean": 0.01942691206932068,
      "epsilon_dpo/beta_margin_std": 0.04522864893078804,
      "epsilon_dpo/loss_margin_mean": 0.20017725229263306,
      "grad_norm": 22.540977478027344,
      "kl/avg_steps": 0.296875,
      "kl/beta": 0.09894487261772156,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 3.2835820895522385e-07,
      "logits/chosen": 0.27466094493865967,
      "logits/rejected": 0.6556792259216309,
      "logps/chosen": -84.09241485595703,
      "logps/ref_chosen": -84.11470794677734,
      "logps/ref_rejected": -120.240234375,
      "logps/rejected": -120.41810607910156,
      "loss": 1.3675,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.0020693570841103792,
      "rewards/margins": 0.019426988437771797,
      "rewards/rejected": -0.017357632517814636,
      "step": 45
    },
    {
      "epoch": 0.06953892668178382,
      "epsilon_dpo/beta": 0.09853853285312653,
      "epsilon_dpo/beta_margin_grad_mean": -0.4979417026042938,
      "epsilon_dpo/beta_margin_grad_std": 0.010288399644196033,
      "epsilon_dpo/beta_margin_mean": 0.008237614296376705,
      "epsilon_dpo/beta_margin_std": 0.041169725358486176,
      "epsilon_dpo/loss_margin_mean": 0.08688095211982727,
      "grad_norm": 21.157970428466797,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.09865199774503708,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 3.3582089552238805e-07,
      "logits/chosen": 0.3935273587703705,
      "logits/rejected": 0.28722450137138367,
      "logps/chosen": -84.30718231201172,
      "logps/ref_chosen": -84.37173461914062,
      "logps/ref_rejected": -106.11967468261719,
      "logps/rejected": -106.14201354980469,
      "loss": 1.3785,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": 0.006227842066437006,
      "rewards/margins": 0.008237641304731369,
      "rewards/rejected": -0.0020097994711250067,
      "step": 46
    },
    {
      "epoch": 0.0710506424792139,
      "epsilon_dpo/beta": 0.09841551631689072,
      "epsilon_dpo/beta_margin_grad_mean": -0.49714115262031555,
      "epsilon_dpo/beta_margin_grad_std": 0.011357166804373264,
      "epsilon_dpo/beta_margin_mean": 0.011449804529547691,
      "epsilon_dpo/beta_margin_std": 0.04547470435500145,
      "epsilon_dpo/loss_margin_mean": 0.11970293521881104,
      "grad_norm": 20.536399841308594,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.09852883964776993,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 3.432835820895522e-07,
      "logits/chosen": 0.38530057668685913,
      "logits/rejected": 0.711715579032898,
      "logps/chosen": -74.67720031738281,
      "logps/ref_chosen": -74.71006774902344,
      "logps/ref_rejected": -97.41304779052734,
      "logps/rejected": -97.4998779296875,
      "loss": 1.3754,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": 0.003110171528533101,
      "rewards/margins": 0.011449852026998997,
      "rewards/rejected": -0.008339679799973965,
      "step": 47
    },
    {
      "epoch": 0.07256235827664399,
      "epsilon_dpo/beta": 0.09835416078567505,
      "epsilon_dpo/beta_margin_grad_mean": -0.4995853304862976,
      "epsilon_dpo/beta_margin_grad_std": 0.010593732818961143,
      "epsilon_dpo/beta_margin_mean": 0.0016609171871095896,
      "epsilon_dpo/beta_margin_std": 0.04239671304821968,
      "epsilon_dpo/loss_margin_mean": 0.020181596279144287,
      "grad_norm": 19.753515243530273,
      "kl/avg_steps": 0.0625,
      "kl/beta": 0.09840583056211472,
      "kl/n_epsilon_steps": 0.46875,
      "kl/p_epsilon_steps": 0.53125,
      "learning_rate": 3.507462686567164e-07,
      "logits/chosen": 0.2559836506843567,
      "logits/rejected": 0.528465986251831,
      "logps/chosen": -71.79025268554688,
      "logps/ref_chosen": -71.76451873779297,
      "logps/ref_rejected": -85.74510192871094,
      "logps/rejected": -85.79102325439453,
      "loss": 1.3851,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.0027173953130841255,
      "rewards/margins": 0.001660920213907957,
      "rewards/rejected": -0.004378315526992083,
      "step": 48
    },
    {
      "epoch": 0.07407407407407407,
      "epsilon_dpo/beta": 0.0982312485575676,
      "epsilon_dpo/beta_margin_grad_mean": -0.4984458088874817,
      "epsilon_dpo/beta_margin_grad_std": 0.009596822783350945,
      "epsilon_dpo/beta_margin_mean": 0.0062199123203754425,
      "epsilon_dpo/beta_margin_std": 0.03840209171175957,
      "epsilon_dpo/loss_margin_mean": 0.06633976101875305,
      "grad_norm": 18.510061264038086,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.0983443632721901,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 3.5820895522388055e-07,
      "logits/chosen": 0.526823878288269,
      "logits/rejected": 0.5619438886642456,
      "logps/chosen": -79.90139770507812,
      "logps/ref_chosen": -79.86338806152344,
      "logps/ref_rejected": -86.35136413574219,
      "logps/rejected": -86.45571899414062,
      "loss": 1.3805,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.003875012043863535,
      "rewards/margins": 0.006219968199729919,
      "rewards/rejected": -0.010094979777932167,
      "step": 49
    },
    {
      "epoch": 0.07558578987150416,
      "epsilon_dpo/beta": 0.09801653027534485,
      "epsilon_dpo/beta_margin_grad_mean": -0.49809837341308594,
      "epsilon_dpo/beta_margin_grad_std": 0.009929750114679337,
      "epsilon_dpo/beta_margin_mean": 0.007609872613102198,
      "epsilon_dpo/beta_margin_std": 0.03973739221692085,
      "epsilon_dpo/loss_margin_mean": 0.08075441420078278,
      "grad_norm": 20.460899353027344,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.09822158515453339,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 3.6567164179104475e-07,
      "logits/chosen": 0.2808557152748108,
      "logits/rejected": 0.47312384843826294,
      "logps/chosen": -76.44319152832031,
      "logps/ref_chosen": -76.4841537475586,
      "logps/ref_rejected": -93.49752044677734,
      "logps/rejected": -93.53730773925781,
      "loss": 1.3791,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": 0.003850931068882346,
      "rewards/margins": 0.0076098134741187096,
      "rewards/rejected": -0.003758882638067007,
      "step": 50
    },
    {
      "epoch": 0.07709750566893424,
      "epsilon_dpo/beta": 0.09789447486400604,
      "epsilon_dpo/beta_margin_grad_mean": -0.4988565742969513,
      "epsilon_dpo/beta_margin_grad_std": 0.012610708363354206,
      "epsilon_dpo/beta_margin_mean": 0.004573495592921972,
      "epsilon_dpo/beta_margin_std": 0.050480011850595474,
      "epsilon_dpo/loss_margin_mean": 0.05065241456031799,
      "grad_norm": 21.301637649536133,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.0980071946978569,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 3.7313432835820895e-07,
      "logits/chosen": 0.21216990053653717,
      "logits/rejected": 0.41427797079086304,
      "logps/chosen": -86.60366821289062,
      "logps/ref_chosen": -86.52101135253906,
      "logps/ref_rejected": -109.46188354492188,
      "logps/rejected": -109.59519958496094,
      "loss": 1.3824,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.008307162672281265,
      "rewards/margins": 0.004573536571115255,
      "rewards/rejected": -0.012880699709057808,
      "step": 51
    },
    {
      "epoch": 0.07860922146636433,
      "epsilon_dpo/beta": 0.09764989465475082,
      "epsilon_dpo/beta_margin_grad_mean": -0.49518582224845886,
      "epsilon_dpo/beta_margin_grad_std": 0.013356276787817478,
      "epsilon_dpo/beta_margin_mean": 0.0192727018147707,
      "epsilon_dpo/beta_margin_std": 0.05347498133778572,
      "epsilon_dpo/loss_margin_mean": 0.2012864351272583,
      "grad_norm": 22.263221740722656,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.09788484126329422,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 3.805970149253731e-07,
      "logits/chosen": 0.2791569232940674,
      "logits/rejected": 0.3946777582168579,
      "logps/chosen": -95.08229064941406,
      "logps/ref_chosen": -95.06444549560547,
      "logps/ref_rejected": -123.63211059570312,
      "logps/rejected": -123.85124206542969,
      "loss": 1.3678,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.0018905354663729668,
      "rewards/margins": 0.019272757694125175,
      "rewards/rejected": -0.021163294091820717,
      "step": 52
    },
    {
      "epoch": 0.0801209372637944,
      "epsilon_dpo/beta": 0.09765049815177917,
      "epsilon_dpo/beta_margin_grad_mean": -0.499002605676651,
      "epsilon_dpo/beta_margin_grad_std": 0.008922034874558449,
      "epsilon_dpo/beta_margin_mean": 0.003993574995547533,
      "epsilon_dpo/beta_margin_std": 0.03570278361439705,
      "epsilon_dpo/loss_margin_mean": 0.0436423122882843,
      "grad_norm": 19.184598922729492,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.0976407378911972,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 3.880597014925373e-07,
      "logits/chosen": 0.19146090745925903,
      "logits/rejected": 0.24986706674098969,
      "logps/chosen": -79.03793334960938,
      "logps/ref_chosen": -79.04600524902344,
      "logps/ref_rejected": -83.90208435058594,
      "logps/rejected": -83.93766784667969,
      "loss": 1.3826,
      "rewards/accuracies": 0.46875,
      "rewards/chosen": 0.0006839814595878124,
      "rewards/margins": 0.003993605263531208,
      "rewards/rejected": -0.0033096238039433956,
      "step": 53
    },
    {
      "epoch": 0.08163265306122448,
      "epsilon_dpo/beta": 0.09755895286798477,
      "epsilon_dpo/beta_margin_grad_mean": -0.4974757432937622,
      "epsilon_dpo/beta_margin_grad_std": 0.009732147678732872,
      "epsilon_dpo/beta_margin_mean": 0.010100303217768669,
      "epsilon_dpo/beta_margin_std": 0.038946740329265594,
      "epsilon_dpo/loss_margin_mean": 0.10647621750831604,
      "grad_norm": 18.53166961669922,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.0976407378911972,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 3.9552238805970144e-07,
      "logits/chosen": 0.24084338545799255,
      "logits/rejected": 0.5024117827415466,
      "logps/chosen": -71.29666137695312,
      "logps/ref_chosen": -71.31385803222656,
      "logps/ref_rejected": -82.59730529785156,
      "logps/rejected": -82.68658447265625,
      "loss": 1.3766,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": 0.0015400029951706529,
      "rewards/margins": 0.010100292041897774,
      "rewards/rejected": -0.008560288697481155,
      "step": 54
    },
    {
      "epoch": 0.08314436885865457,
      "epsilon_dpo/beta": 0.09726925939321518,
      "epsilon_dpo/beta_margin_grad_mean": -0.49761292338371277,
      "epsilon_dpo/beta_margin_grad_std": 0.01231041457504034,
      "epsilon_dpo/beta_margin_mean": 0.009554947726428509,
      "epsilon_dpo/beta_margin_std": 0.04927734658122063,
      "epsilon_dpo/loss_margin_mean": 0.10191529989242554,
      "grad_norm": 21.672401428222656,
      "kl/avg_steps": 0.296875,
      "kl/beta": 0.09754928946495056,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.0298507462686564e-07,
      "logits/chosen": 0.5175954103469849,
      "logits/rejected": 0.7563967704772949,
      "logps/chosen": -72.74899291992188,
      "logps/ref_chosen": -72.7046127319336,
      "logps/ref_rejected": -117.64811706542969,
      "logps/rejected": -117.7944107055664,
      "loss": 1.3774,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.004462738521397114,
      "rewards/margins": 0.009554837830364704,
      "rewards/rejected": -0.014017576351761818,
      "step": 55
    },
    {
      "epoch": 0.08465608465608465,
      "epsilon_dpo/beta": 0.09714867919683456,
      "epsilon_dpo/beta_margin_grad_mean": -0.4990710914134979,
      "epsilon_dpo/beta_margin_grad_std": 0.011907228268682957,
      "epsilon_dpo/beta_margin_mean": 0.0037128631956875324,
      "epsilon_dpo/beta_margin_std": 0.0476621612906456,
      "epsilon_dpo/loss_margin_mean": 0.04199719429016113,
      "grad_norm": 19.10873794555664,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.09726054221391678,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 4.1044776119402984e-07,
      "logits/chosen": 0.301601767539978,
      "logits/rejected": 0.4184657335281372,
      "logps/chosen": -80.00164794921875,
      "logps/ref_chosen": -79.95884704589844,
      "logps/ref_rejected": -86.45406341552734,
      "logps/rejected": -86.53886413574219,
      "loss": 1.3832,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.0043641347438097,
      "rewards/margins": 0.0037128704134374857,
      "rewards/rejected": -0.008077004924416542,
      "step": 56
    },
    {
      "epoch": 0.08616780045351474,
      "epsilon_dpo/beta": 0.09693632274866104,
      "epsilon_dpo/beta_margin_grad_mean": -0.4949696958065033,
      "epsilon_dpo/beta_margin_grad_std": 0.014986831694841385,
      "epsilon_dpo/beta_margin_mean": 0.02014934830367565,
      "epsilon_dpo/beta_margin_std": 0.06002877652645111,
      "epsilon_dpo/loss_margin_mean": 0.2123461365699768,
      "grad_norm": 23.030841827392578,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.09713912010192871,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.17910447761194e-07,
      "logits/chosen": 0.3571116328239441,
      "logits/rejected": 0.2162424772977829,
      "logps/chosen": -80.38567352294922,
      "logps/ref_chosen": -80.30390930175781,
      "logps/ref_rejected": -119.16410827636719,
      "logps/rejected": -119.45822143554688,
      "loss": 1.3671,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.008071388117969036,
      "rewards/margins": 0.020149312913417816,
      "rewards/rejected": -0.028220700100064278,
      "step": 57
    },
    {
      "epoch": 0.08767951625094482,
      "epsilon_dpo/beta": 0.09696707874536514,
      "epsilon_dpo/beta_margin_grad_mean": -0.49949392676353455,
      "epsilon_dpo/beta_margin_grad_std": 0.013705207034945488,
      "epsilon_dpo/beta_margin_mean": 0.0020344173535704613,
      "epsilon_dpo/beta_margin_std": 0.054878149181604385,
      "epsilon_dpo/loss_margin_mean": 0.02534008026123047,
      "grad_norm": 22.337549209594727,
      "kl/avg_steps": -0.03125,
      "kl/beta": 0.09692709147930145,
      "kl/n_epsilon_steps": 0.515625,
      "kl/p_epsilon_steps": 0.484375,
      "learning_rate": 4.253731343283582e-07,
      "logits/chosen": 0.42537829279899597,
      "logits/rejected": 0.3799484670162201,
      "logps/chosen": -99.42530822753906,
      "logps/ref_chosen": -99.30342864990234,
      "logps/ref_rejected": -111.72179412841797,
      "logps/rejected": -111.8690185546875,
      "loss": 1.385,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": -0.012063508853316307,
      "rewards/margins": 0.002034379169344902,
      "rewards/rejected": -0.014097888022661209,
      "step": 58
    },
    {
      "epoch": 0.08919123204837491,
      "epsilon_dpo/beta": 0.09672466665506363,
      "epsilon_dpo/beta_margin_grad_mean": -0.4958970248699188,
      "epsilon_dpo/beta_margin_grad_std": 0.012421243824064732,
      "epsilon_dpo/beta_margin_mean": 0.016427284106612206,
      "epsilon_dpo/beta_margin_std": 0.04973134025931358,
      "epsilon_dpo/loss_margin_mean": 0.17345958948135376,
      "grad_norm": 19.432371139526367,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.09695739299058914,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.3283582089552234e-07,
      "logits/chosen": 0.4255552291870117,
      "logits/rejected": 0.2160358428955078,
      "logps/chosen": -71.87731170654297,
      "logps/ref_chosen": -71.85493469238281,
      "logps/ref_rejected": -97.24525451660156,
      "logps/rejected": -97.44110107421875,
      "loss": 1.3706,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.002300878055393696,
      "rewards/margins": 0.01642729714512825,
      "rewards/rejected": -0.01872817426919937,
      "step": 59
    },
    {
      "epoch": 0.09070294784580499,
      "epsilon_dpo/beta": 0.09657414257526398,
      "epsilon_dpo/beta_margin_grad_mean": -0.4995685815811157,
      "epsilon_dpo/beta_margin_grad_std": 0.012486227788031101,
      "epsilon_dpo/beta_margin_mean": 0.001717747189104557,
      "epsilon_dpo/beta_margin_std": 0.04999338835477829,
      "epsilon_dpo/loss_margin_mean": 0.021374523639678955,
      "grad_norm": 20.249168395996094,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.09671560674905777,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.4029850746268654e-07,
      "logits/chosen": 0.3270602822303772,
      "logits/rejected": 0.4148187041282654,
      "logps/chosen": -87.35511779785156,
      "logps/ref_chosen": -87.21815490722656,
      "logps/ref_rejected": -82.96167755126953,
      "logps/rejected": -83.12001037597656,
      "loss": 1.3852,
      "rewards/accuracies": 0.578125,
      "rewards/chosen": -0.013473069295287132,
      "rewards/margins": 0.0017178304260596633,
      "rewards/rejected": -0.01519089937210083,
      "step": 60
    },
    {
      "epoch": 0.09221466364323508,
      "epsilon_dpo/beta": 0.09663473814725876,
      "epsilon_dpo/beta_margin_grad_mean": -0.4984928071498871,
      "epsilon_dpo/beta_margin_grad_std": 0.013760149478912354,
      "epsilon_dpo/beta_margin_mean": 0.006036721635609865,
      "epsilon_dpo/beta_margin_std": 0.05508105829358101,
      "epsilon_dpo/loss_margin_mean": 0.06708309054374695,
      "grad_norm": 18.39585304260254,
      "kl/avg_steps": -0.0625,
      "kl/beta": 0.09656472504138947,
      "kl/n_epsilon_steps": 0.53125,
      "kl/p_epsilon_steps": 0.46875,
      "learning_rate": 4.4776119402985074e-07,
      "logits/chosen": 0.05619889125227928,
      "logits/rejected": 0.49303629994392395,
      "logps/chosen": -70.44015502929688,
      "logps/ref_chosen": -70.37516784667969,
      "logps/ref_rejected": -93.18226623535156,
      "logps/rejected": -93.3143310546875,
      "loss": 1.381,
      "rewards/accuracies": 0.453125,
      "rewards/chosen": -0.006510016042739153,
      "rewards/margins": 0.006036766339093447,
      "rewards/rejected": -0.012546783313155174,
      "step": 61
    },
    {
      "epoch": 0.09372637944066516,
      "epsilon_dpo/beta": 0.09655912220478058,
      "epsilon_dpo/beta_margin_grad_mean": -0.49866795539855957,
      "epsilon_dpo/beta_margin_grad_std": 0.014224527403712273,
      "epsilon_dpo/beta_margin_mean": 0.005336884409189224,
      "epsilon_dpo/beta_margin_std": 0.05695384740829468,
      "epsilon_dpo/loss_margin_mean": 0.059721678495407104,
      "grad_norm": 20.705806732177734,
      "kl/avg_steps": 0.078125,
      "kl/beta": 0.09662511199712753,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.53125,
      "learning_rate": 4.552238805970149e-07,
      "logits/chosen": 0.2867963910102844,
      "logits/rejected": 0.4729596972465515,
      "logps/chosen": -87.79457092285156,
      "logps/ref_chosen": -87.66146850585938,
      "logps/ref_rejected": -103.1637954711914,
      "logps/rejected": -103.35662078857422,
      "loss": 1.3818,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.013055233284831047,
      "rewards/margins": 0.0053368983790278435,
      "rewards/rejected": -0.018392130732536316,
      "step": 62
    },
    {
      "epoch": 0.09523809523809523,
      "epsilon_dpo/beta": 0.0963481143116951,
      "epsilon_dpo/beta_margin_grad_mean": -0.4944593906402588,
      "epsilon_dpo/beta_margin_grad_std": 0.014144835993647575,
      "epsilon_dpo/beta_margin_mean": 0.022186141461133957,
      "epsilon_dpo/beta_margin_std": 0.05663369968533516,
      "epsilon_dpo/loss_margin_mean": 0.2345995306968689,
      "grad_norm": 22.35875701904297,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.09654968231916428,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.626865671641791e-07,
      "logits/chosen": 0.3238663375377655,
      "logits/rejected": 0.24062295258045197,
      "logps/chosen": -83.79109954833984,
      "logps/ref_chosen": -83.74502563476562,
      "logps/ref_rejected": -107.7850341796875,
      "logps/rejected": -108.0656967163086,
      "loss": 1.365,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.004585597664117813,
      "rewards/margins": 0.02218601480126381,
      "rewards/rejected": -0.026771612465381622,
      "step": 63
    },
    {
      "epoch": 0.09674981103552532,
      "epsilon_dpo/beta": 0.09631846100091934,
      "epsilon_dpo/beta_margin_grad_mean": -0.4993444085121155,
      "epsilon_dpo/beta_margin_grad_std": 0.01322327833622694,
      "epsilon_dpo/beta_margin_mean": 0.0026313355192542076,
      "epsilon_dpo/beta_margin_std": 0.05294317379593849,
      "epsilon_dpo/loss_margin_mean": 0.03148919343948364,
      "grad_norm": 17.767595291137695,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.09633894264698029,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 4.701492537313433e-07,
      "logits/chosen": 0.24474751949310303,
      "logits/rejected": 0.3736121654510498,
      "logps/chosen": -74.78924560546875,
      "logps/ref_chosen": -74.63839721679688,
      "logps/ref_rejected": -88.96295166015625,
      "logps/rejected": -89.1452865600586,
      "loss": 1.3844,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.014759637415409088,
      "rewards/margins": 0.0026313296984881163,
      "rewards/rejected": -0.017390966415405273,
      "step": 64
    },
    {
      "epoch": 0.0982615268329554,
      "epsilon_dpo/beta": 0.09598737955093384,
      "epsilon_dpo/beta_margin_grad_mean": -0.4935208261013031,
      "epsilon_dpo/beta_margin_grad_std": 0.016076454892754555,
      "epsilon_dpo/beta_margin_mean": 0.02594032883644104,
      "epsilon_dpo/beta_margin_std": 0.06438540667295456,
      "epsilon_dpo/loss_margin_mean": 0.2748907208442688,
      "grad_norm": 21.29926872253418,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.09630884230136871,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.776119402985074e-07,
      "logits/chosen": 0.1908925324678421,
      "logits/rejected": 0.32332342863082886,
      "logps/chosen": -90.84135437011719,
      "logps/ref_chosen": -90.74732971191406,
      "logps/ref_rejected": -103.91267395019531,
      "logps/rejected": -104.28158569335938,
      "loss": 1.3616,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.009328576736152172,
      "rewards/margins": 0.025940343737602234,
      "rewards/rejected": -0.03526891767978668,
      "step": 65
    },
    {
      "epoch": 0.09977324263038549,
      "epsilon_dpo/beta": 0.09580853581428528,
      "epsilon_dpo/beta_margin_grad_mean": -0.49857693910598755,
      "epsilon_dpo/beta_margin_grad_std": 0.014432637952268124,
      "epsilon_dpo/beta_margin_mean": 0.005690241232514381,
      "epsilon_dpo/beta_margin_std": 0.05780486389994621,
      "epsilon_dpo/loss_margin_mean": 0.06374523043632507,
      "grad_norm": 20.495637893676758,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.09597891569137573,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.850746268656717e-07,
      "logits/chosen": 0.17386680841445923,
      "logits/rejected": 0.23358601331710815,
      "logps/chosen": -95.62905883789062,
      "logps/ref_chosen": -95.40287780761719,
      "logps/ref_rejected": -103.42100524902344,
      "logps/rejected": -103.7109375,
      "loss": 1.3814,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.02190394513309002,
      "rewards/margins": 0.005690223537385464,
      "rewards/rejected": -0.027594169601798058,
      "step": 66
    },
    {
      "epoch": 0.10128495842781557,
      "epsilon_dpo/beta": 0.09571905434131622,
      "epsilon_dpo/beta_margin_grad_mean": -0.49876338243484497,
      "epsilon_dpo/beta_margin_grad_std": 0.013045243918895721,
      "epsilon_dpo/beta_margin_mean": 0.004954536911100149,
      "epsilon_dpo/beta_margin_std": 0.052225664258003235,
      "epsilon_dpo/loss_margin_mean": 0.056086331605911255,
      "grad_norm": 22.02642059326172,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.09579929709434509,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 4.925373134328357e-07,
      "logits/chosen": 0.32260456681251526,
      "logits/rejected": 0.4844122529029846,
      "logps/chosen": -84.48126220703125,
      "logps/ref_chosen": -84.30548095703125,
      "logps/ref_rejected": -107.89781188964844,
      "logps/rejected": -108.12968444824219,
      "loss": 1.382,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.017115123569965363,
      "rewards/margins": 0.004954596050083637,
      "rewards/rejected": -0.022069718688726425,
      "step": 67
    },
    {
      "epoch": 0.10279667422524566,
      "epsilon_dpo/beta": 0.09562939405441284,
      "epsilon_dpo/beta_margin_grad_mean": -0.4978659451007843,
      "epsilon_dpo/beta_margin_grad_std": 0.014736686833202839,
      "epsilon_dpo/beta_margin_mean": 0.008545473217964172,
      "epsilon_dpo/beta_margin_std": 0.059007227420806885,
      "epsilon_dpo/loss_margin_mean": 0.09425216913223267,
      "grad_norm": 21.87251091003418,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.09570956230163574,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 5e-07,
      "logits/chosen": 0.45750993490219116,
      "logits/rejected": 0.4745368957519531,
      "logps/chosen": -107.02864074707031,
      "logps/ref_chosen": -106.75498962402344,
      "logps/ref_rejected": -122.26151275634766,
      "logps/rejected": -122.62940979003906,
      "loss": 1.3786,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": -0.02646515890955925,
      "rewards/margins": 0.008545521646738052,
      "rewards/rejected": -0.0350106805562973,
      "step": 68
    },
    {
      "epoch": 0.10430839002267574,
      "epsilon_dpo/beta": 0.09539041668176651,
      "epsilon_dpo/beta_margin_grad_mean": -0.4965830445289612,
      "epsilon_dpo/beta_margin_grad_std": 0.01738860085606575,
      "epsilon_dpo/beta_margin_mean": 0.013665448874235153,
      "epsilon_dpo/beta_margin_std": 0.06965122371912003,
      "epsilon_dpo/loss_margin_mean": 0.14875519275665283,
      "grad_norm": 21.092300415039062,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.09561992436647415,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.999965034812934e-07,
      "logits/chosen": 0.41545820236206055,
      "logits/rejected": 0.5641710758209229,
      "logps/chosen": -91.29107666015625,
      "logps/ref_chosen": -91.08714294433594,
      "logps/ref_rejected": -115.80476379394531,
      "logps/rejected": -116.15745544433594,
      "loss": 1.3739,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.01966257020831108,
      "rewards/margins": 0.013665375299751759,
      "rewards/rejected": -0.033327944576740265,
      "step": 69
    },
    {
      "epoch": 0.10582010582010581,
      "epsilon_dpo/beta": 0.09519709646701813,
      "epsilon_dpo/beta_margin_grad_mean": -0.4971049427986145,
      "epsilon_dpo/beta_margin_grad_std": 0.01725912094116211,
      "epsilon_dpo/beta_margin_mean": 0.011591562069952488,
      "epsilon_dpo/beta_margin_std": 0.06912032514810562,
      "epsilon_dpo/loss_margin_mean": 0.12734776735305786,
      "grad_norm": 20.18518829345703,
      "kl/avg_steps": 0.203125,
      "kl/beta": 0.0953814685344696,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.999860140229787e-07,
      "logits/chosen": 0.43227583169937134,
      "logits/rejected": 0.42724496126174927,
      "logps/chosen": -88.18643188476562,
      "logps/ref_chosen": -88.12431335449219,
      "logps/ref_rejected": -99.41046905517578,
      "logps/rejected": -99.59992980957031,
      "loss": 1.3759,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.006189244333654642,
      "rewards/margins": 0.011591588146984577,
      "rewards/rejected": -0.017780832946300507,
      "step": 70
    },
    {
      "epoch": 0.1073318216175359,
      "epsilon_dpo/beta": 0.09507863968610764,
      "epsilon_dpo/beta_margin_grad_mean": -0.5000821948051453,
      "epsilon_dpo/beta_margin_grad_std": 0.01915271021425724,
      "epsilon_dpo/beta_margin_mean": -0.00034239908563904464,
      "epsilon_dpo/beta_margin_std": 0.07671481370925903,
      "epsilon_dpo/loss_margin_mean": 0.002892807126045227,
      "grad_norm": 19.287860870361328,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.09518811851739883,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 4.999685319184688e-07,
      "logits/chosen": 0.10468558967113495,
      "logits/rejected": 0.2944987118244171,
      "logps/chosen": -87.19418334960938,
      "logps/ref_chosen": -86.94244384765625,
      "logps/ref_rejected": -92.30281829833984,
      "logps/rejected": -92.55744934082031,
      "loss": 1.3881,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.024355504661798477,
      "rewards/margins": -0.000342401210218668,
      "rewards/rejected": -0.024013103917241096,
      "step": 71
    },
    {
      "epoch": 0.10884353741496598,
      "epsilon_dpo/beta": 0.09478166699409485,
      "epsilon_dpo/beta_margin_grad_mean": -0.49388766288757324,
      "epsilon_dpo/beta_margin_grad_std": 0.017899975180625916,
      "epsilon_dpo/beta_margin_mean": 0.024477774277329445,
      "epsilon_dpo/beta_margin_std": 0.07170306146144867,
      "epsilon_dpo/loss_margin_mean": 0.26375046372413635,
      "grad_norm": 21.809051513671875,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.0950692817568779,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.999440576567755e-07,
      "logits/chosen": 0.31559497117996216,
      "logits/rejected": 0.34403306245803833,
      "logps/chosen": -77.3045654296875,
      "logps/ref_chosen": -77.19921112060547,
      "logps/ref_rejected": -105.40240478515625,
      "logps/rejected": -105.77151489257812,
      "loss": 1.3633,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.010216562077403069,
      "rewards/margins": 0.02447780966758728,
      "rewards/rejected": -0.0346943698823452,
      "step": 72
    },
    {
      "epoch": 0.11035525321239607,
      "epsilon_dpo/beta": 0.09469373524188995,
      "epsilon_dpo/beta_margin_grad_mean": -0.4983028471469879,
      "epsilon_dpo/beta_margin_grad_std": 0.019205359742045403,
      "epsilon_dpo/beta_margin_mean": 0.006816650275141001,
      "epsilon_dpo/beta_margin_std": 0.07694670557975769,
      "epsilon_dpo/loss_margin_mean": 0.0784463882446289,
      "grad_norm": 21.700668334960938,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.09477311372756958,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 4.999125919224965e-07,
      "logits/chosen": 0.45154646039009094,
      "logits/rejected": 0.4655054211616516,
      "logps/chosen": -95.38381958007812,
      "logps/ref_chosen": -95.17266845703125,
      "logps/ref_rejected": -104.91008758544922,
      "logps/rejected": -105.19967651367188,
      "loss": 1.381,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.02027609571814537,
      "rewards/margins": 0.006816700100898743,
      "rewards/rejected": -0.027092795819044113,
      "step": 73
    },
    {
      "epoch": 0.11186696900982615,
      "epsilon_dpo/beta": 0.0943978950381279,
      "epsilon_dpo/beta_margin_grad_mean": -0.4949096143245697,
      "epsilon_dpo/beta_margin_grad_std": 0.015205258503556252,
      "epsilon_dpo/beta_margin_mean": 0.02038200944662094,
      "epsilon_dpo/beta_margin_std": 0.06088561564683914,
      "epsilon_dpo/loss_margin_mean": 0.22069677710533142,
      "grad_norm": 19.499149322509766,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.09468434751033783,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.998741355957963e-07,
      "logits/chosen": 0.26441439986228943,
      "logits/rejected": 0.3638787865638733,
      "logps/chosen": -70.67909240722656,
      "logps/ref_chosen": -70.6429214477539,
      "logps/ref_rejected": -102.84405517578125,
      "logps/rejected": -103.1009292602539,
      "loss": 1.3669,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.003621622920036316,
      "rewards/margins": 0.020382072776556015,
      "rewards/rejected": -0.02400369569659233,
      "step": 74
    },
    {
      "epoch": 0.11337868480725624,
      "epsilon_dpo/beta": 0.09413331747055054,
      "epsilon_dpo/beta_margin_grad_mean": -0.4952016770839691,
      "epsilon_dpo/beta_margin_grad_std": 0.017589138820767403,
      "epsilon_dpo/beta_margin_mean": 0.019206415861845016,
      "epsilon_dpo/beta_margin_std": 0.0704592615365982,
      "epsilon_dpo/loss_margin_mean": 0.2095319628715515,
      "grad_norm": 18.27823829650879,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.09438937902450562,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.998286897523808e-07,
      "logits/chosen": 0.15772899985313416,
      "logits/rejected": 0.2349020391702652,
      "logps/chosen": -65.92100524902344,
      "logps/ref_chosen": -65.809326171875,
      "logps/ref_rejected": -87.82281494140625,
      "logps/rejected": -88.14402770996094,
      "loss": 1.3684,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.010867921635508537,
      "rewards/margins": 0.01920643448829651,
      "rewards/rejected": -0.030074356123805046,
      "step": 75
    },
    {
      "epoch": 0.11489040060468632,
      "epsilon_dpo/beta": 0.0939575582742691,
      "epsilon_dpo/beta_margin_grad_mean": -0.49271103739738464,
      "epsilon_dpo/beta_margin_grad_std": 0.019078722223639488,
      "epsilon_dpo/beta_margin_mean": 0.02922924794256687,
      "epsilon_dpo/beta_margin_std": 0.07647933810949326,
      "epsilon_dpo/loss_margin_mean": 0.3171220123767853,
      "grad_norm": 19.83082389831543,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.09412465244531631,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.997762556634679e-07,
      "logits/chosen": 0.2850942611694336,
      "logits/rejected": 0.3514692783355713,
      "logps/chosen": -75.77113342285156,
      "logps/ref_chosen": -75.6954574584961,
      "logps/ref_rejected": -104.007568359375,
      "logps/rejected": -104.40037536621094,
      "loss": 1.3587,
      "rewards/accuracies": 0.578125,
      "rewards/chosen": -0.007265619933605194,
      "rewards/margins": 0.029229262843728065,
      "rewards/rejected": -0.03649488091468811,
      "step": 76
    },
    {
      "epoch": 0.1164021164021164,
      "epsilon_dpo/beta": 0.09357620030641556,
      "epsilon_dpo/beta_margin_grad_mean": -0.491890013217926,
      "epsilon_dpo/beta_margin_grad_std": 0.020320318639278412,
      "epsilon_dpo/beta_margin_mean": 0.03250795230269432,
      "epsilon_dpo/beta_margin_std": 0.08144384622573853,
      "epsilon_dpo/loss_margin_mean": 0.3533650040626526,
      "grad_norm": 19.243268966674805,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.09394850581884384,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.99716834795752e-07,
      "logits/chosen": 0.18683543801307678,
      "logits/rejected": 0.33957868814468384,
      "logps/chosen": -78.54193878173828,
      "logps/ref_chosen": -78.43965148925781,
      "logps/ref_rejected": -101.59224700927734,
      "logps/rejected": -102.04789733886719,
      "loss": 1.3557,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.009762855246663094,
      "rewards/margins": 0.032507993280887604,
      "rewards/rejected": -0.04227085039019585,
      "step": 77
    },
    {
      "epoch": 0.11791383219954649,
      "epsilon_dpo/beta": 0.09346076846122742,
      "epsilon_dpo/beta_margin_grad_mean": -0.4948379695415497,
      "epsilon_dpo/beta_margin_grad_std": 0.02001882530748844,
      "epsilon_dpo/beta_margin_mean": 0.020674889907240868,
      "epsilon_dpo/beta_margin_std": 0.08018743991851807,
      "epsilon_dpo/loss_margin_mean": 0.22830542922019958,
      "grad_norm": 19.162540435791016,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.09356838464736938,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 4.996504288113623e-07,
      "logits/chosen": 0.30345967411994934,
      "logits/rejected": 0.33946341276168823,
      "logps/chosen": -90.49433898925781,
      "logps/ref_chosen": -90.30522155761719,
      "logps/ref_rejected": -101.54324340820312,
      "logps/rejected": -101.96066284179688,
      "loss": 1.3673,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.017984943464398384,
      "rewards/margins": 0.020674871280789375,
      "rewards/rejected": -0.03865981474518776,
      "step": 78
    },
    {
      "epoch": 0.11942554799697656,
      "epsilon_dpo/beta": 0.09302281588315964,
      "epsilon_dpo/beta_margin_grad_mean": -0.4897191822528839,
      "epsilon_dpo/beta_margin_grad_std": 0.02295556291937828,
      "epsilon_dpo/beta_margin_mean": 0.04126286879181862,
      "epsilon_dpo/beta_margin_std": 0.09222093224525452,
      "epsilon_dpo/loss_margin_mean": 0.44970422983169556,
      "grad_norm": 20.854549407958984,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.09345156699419022,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.995770395678171e-07,
      "logits/chosen": 0.2500825524330139,
      "logits/rejected": 0.3519323766231537,
      "logps/chosen": -71.46902465820312,
      "logps/ref_chosen": -71.36767578125,
      "logps/ref_rejected": -110.53877258300781,
      "logps/rejected": -111.08982849121094,
      "loss": 1.3476,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.009614603593945503,
      "rewards/margins": 0.04126285761594772,
      "rewards/rejected": -0.05087745934724808,
      "step": 79
    },
    {
      "epoch": 0.12093726379440665,
      "epsilon_dpo/beta": 0.09285043925046921,
      "epsilon_dpo/beta_margin_grad_mean": -0.49484947323799133,
      "epsilon_dpo/beta_margin_grad_std": 0.022218074649572372,
      "epsilon_dpo/beta_margin_mean": 0.02064460888504982,
      "epsilon_dpo/beta_margin_std": 0.08906199038028717,
      "epsilon_dpo/loss_margin_mean": 0.2295338213443756,
      "grad_norm": 19.17706871032715,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.09301555901765823,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.994966691179711e-07,
      "logits/chosen": 0.3593171536922455,
      "logits/rejected": 0.20739956200122833,
      "logps/chosen": -83.57428741455078,
      "logps/ref_chosen": -83.30223083496094,
      "logps/ref_rejected": -108.11129760742188,
      "logps/rejected": -108.61288452148438,
      "loss": 1.3677,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.025694172829389572,
      "rewards/margins": 0.02064460888504982,
      "rewards/rejected": -0.04633878171443939,
      "step": 80
    },
    {
      "epoch": 0.12244897959183673,
      "epsilon_dpo/beta": 0.0924445390701294,
      "epsilon_dpo/beta_margin_grad_mean": -0.4913916289806366,
      "epsilon_dpo/beta_margin_grad_std": 0.019580332562327385,
      "epsilon_dpo/beta_margin_mean": 0.03447142615914345,
      "epsilon_dpo/beta_margin_std": 0.07844133675098419,
      "epsilon_dpo/loss_margin_mean": 0.3787657916545868,
      "grad_norm": 19.880578994750977,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.09284147620201111,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.994093197099587e-07,
      "logits/chosen": 0.18329188227653503,
      "logits/rejected": 0.18096214532852173,
      "logps/chosen": -86.397705078125,
      "logps/ref_chosen": -86.2278060913086,
      "logps/ref_rejected": -102.50193786621094,
      "logps/rejected": -103.05059051513672,
      "loss": 1.3537,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.01595979556441307,
      "rewards/margins": 0.034471385180950165,
      "rewards/rejected": -0.05043117702007294,
      "step": 81
    },
    {
      "epoch": 0.12396069538926682,
      "epsilon_dpo/beta": 0.09207074344158173,
      "epsilon_dpo/beta_margin_grad_mean": -0.48697853088378906,
      "epsilon_dpo/beta_margin_grad_std": 0.02084416151046753,
      "epsilon_dpo/beta_margin_mean": 0.05220121517777443,
      "epsilon_dpo/beta_margin_std": 0.0836104229092598,
      "epsilon_dpo/loss_margin_mean": 0.5733838677406311,
      "grad_norm": 18.803672790527344,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.09243706613779068,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.993149937871306e-07,
      "logits/chosen": 0.18002444505691528,
      "logits/rejected": 0.2574087381362915,
      "logps/chosen": -68.62680053710938,
      "logps/ref_chosen": -68.67723846435547,
      "logps/ref_rejected": -93.22198486328125,
      "logps/rejected": -93.74493408203125,
      "loss": 1.3365,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.0043902089819312096,
      "rewards/margins": 0.05220119655132294,
      "rewards/rejected": -0.04781098663806915,
      "step": 82
    },
    {
      "epoch": 0.1254724111866969,
      "epsilon_dpo/beta": 0.09175577014684677,
      "epsilon_dpo/beta_margin_grad_mean": -0.48969775438308716,
      "epsilon_dpo/beta_margin_grad_std": 0.021411551162600517,
      "epsilon_dpo/beta_margin_mean": 0.04130866378545761,
      "epsilon_dpo/beta_margin_std": 0.0858808383345604,
      "epsilon_dpo/loss_margin_mean": 0.4566318392753601,
      "grad_norm": 19.55249786376953,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.0920630618929863,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.992136939879856e-07,
      "logits/chosen": 0.1689714640378952,
      "logits/rejected": 0.10307054221630096,
      "logps/chosen": -79.06647491455078,
      "logps/ref_chosen": -79.0299072265625,
      "logps/ref_rejected": -100.61555480957031,
      "logps/rejected": -101.10874938964844,
      "loss": 1.3473,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.0036600963212549686,
      "rewards/margins": 0.0413086824119091,
      "rewards/rejected": -0.044968780130147934,
      "step": 83
    },
    {
      "epoch": 0.12698412698412698,
      "epsilon_dpo/beta": 0.09152746200561523,
      "epsilon_dpo/beta_margin_grad_mean": -0.4955390691757202,
      "epsilon_dpo/beta_margin_grad_std": 0.024220502004027367,
      "epsilon_dpo/beta_margin_mean": 0.017843004316091537,
      "epsilon_dpo/beta_margin_std": 0.09716963768005371,
      "epsilon_dpo/loss_margin_mean": 0.20263022184371948,
      "grad_norm": 21.091583251953125,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.09174767881631851,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.991054231460969e-07,
      "logits/chosen": 0.3326488733291626,
      "logits/rejected": 0.47113683819770813,
      "logps/chosen": -87.80223083496094,
      "logps/ref_chosen": -87.48562622070312,
      "logps/ref_rejected": -107.02767944335938,
      "logps/rejected": -107.54692077636719,
      "loss": 1.3709,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.029343653470277786,
      "rewards/margins": 0.01784295029938221,
      "rewards/rejected": -0.04718660190701485,
      "step": 84
    },
    {
      "epoch": 0.12849584278155707,
      "epsilon_dpo/beta": 0.09109899401664734,
      "epsilon_dpo/beta_margin_grad_mean": -0.4891459047794342,
      "epsilon_dpo/beta_margin_grad_std": 0.023020127788186073,
      "epsilon_dpo/beta_margin_mean": 0.043475423008203506,
      "epsilon_dpo/beta_margin_std": 0.09232480823993683,
      "epsilon_dpo/loss_margin_mean": 0.4838877320289612,
      "grad_norm": 18.248411178588867,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.09151887893676758,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.989901842900325e-07,
      "logits/chosen": 0.1454373598098755,
      "logits/rejected": 0.0769268125295639,
      "logps/chosen": -74.00840759277344,
      "logps/ref_chosen": -73.96684265136719,
      "logps/ref_rejected": -91.79646301269531,
      "logps/rejected": -92.32191467285156,
      "loss": 1.3454,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.0040864236652851105,
      "rewards/margins": 0.04347540810704231,
      "rewards/rejected": -0.04756183177232742,
      "step": 85
    },
    {
      "epoch": 0.13000755857898716,
      "epsilon_dpo/beta": 0.09081630408763885,
      "epsilon_dpo/beta_margin_grad_mean": -0.49439719319343567,
      "epsilon_dpo/beta_margin_grad_std": 0.020537180826067924,
      "epsilon_dpo/beta_margin_mean": 0.02242736890912056,
      "epsilon_dpo/beta_margin_std": 0.08234351128339767,
      "epsilon_dpo/loss_margin_mean": 0.2534914016723633,
      "grad_norm": 18.693727493286133,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.09109188616275787,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.988679806432711e-07,
      "logits/chosen": -0.03249434754252434,
      "logits/rejected": -0.15132024884223938,
      "logps/chosen": -85.06571960449219,
      "logps/ref_chosen": -84.69587707519531,
      "logps/ref_rejected": -92.96185302734375,
      "logps/rejected": -93.58517456054688,
      "loss": 1.3657,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.03381003811955452,
      "rewards/margins": 0.02242732048034668,
      "rewards/rejected": -0.0562373585999012,
      "step": 86
    },
    {
      "epoch": 0.13151927437641722,
      "epsilon_dpo/beta": 0.09053338319063187,
      "epsilon_dpo/beta_margin_grad_mean": -0.48980429768562317,
      "epsilon_dpo/beta_margin_grad_std": 0.02582697570323944,
      "epsilon_dpo/beta_margin_mean": 0.040944747626781464,
      "epsilon_dpo/beta_margin_std": 0.10375433415174484,
      "epsilon_dpo/loss_margin_mean": 0.4599999487400055,
      "grad_norm": 19.913551330566406,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.09080810844898224,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.987388156241114e-07,
      "logits/chosen": 0.16291941702365875,
      "logits/rejected": 0.10878261923789978,
      "logps/chosen": -91.9239730834961,
      "logps/ref_chosen": -91.67546844482422,
      "logps/ref_rejected": -113.0994644165039,
      "logps/rejected": -113.80797576904297,
      "loss": 1.3485,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.022793810814619064,
      "rewards/margins": 0.040944769978523254,
      "rewards/rejected": -0.06373857706785202,
      "step": 87
    },
    {
      "epoch": 0.1330309901738473,
      "epsilon_dpo/beta": 0.09036451578140259,
      "epsilon_dpo/beta_margin_grad_mean": -0.49780169129371643,
      "epsilon_dpo/beta_margin_grad_std": 0.02381385862827301,
      "epsilon_dpo/beta_margin_mean": 0.008791986852884293,
      "epsilon_dpo/beta_margin_std": 0.09545378386974335,
      "epsilon_dpo/loss_margin_mean": 0.1055518090724945,
      "grad_norm": 19.109310150146484,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.09052521735429764,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.986026928455767e-07,
      "logits/chosen": 0.1496576964855194,
      "logits/rejected": 0.30393093824386597,
      "logps/chosen": -88.25755310058594,
      "logps/ref_chosen": -87.91022491455078,
      "logps/ref_rejected": -97.09971618652344,
      "logps/rejected": -97.55259704589844,
      "loss": 1.3798,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": -0.03172972798347473,
      "rewards/margins": 0.008791987784206867,
      "rewards/rejected": -0.04052171856164932,
      "step": 88
    },
    {
      "epoch": 0.1345427059712774,
      "epsilon_dpo/beta": 0.09002596884965897,
      "epsilon_dpo/beta_margin_grad_mean": -0.4860003590583801,
      "epsilon_dpo/beta_margin_grad_std": 0.02549973875284195,
      "epsilon_dpo/beta_margin_mean": 0.056202370673418045,
      "epsilon_dpo/beta_margin_std": 0.10239987820386887,
      "epsilon_dpo/loss_margin_mean": 0.6319105625152588,
      "grad_norm": 19.2386417388916,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.09035580605268478,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.984596161153135e-07,
      "logits/chosen": 0.22525973618030548,
      "logits/rejected": 0.19805394113063812,
      "logps/chosen": -61.868133544921875,
      "logps/ref_chosen": -61.86646270751953,
      "logps/ref_rejected": -109.35748291015625,
      "logps/rejected": -109.99107360839844,
      "loss": 1.3335,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.0005648209480568767,
      "rewards/margins": 0.05620241165161133,
      "rewards/rejected": -0.05676723271608353,
      "step": 89
    },
    {
      "epoch": 0.1360544217687075,
      "epsilon_dpo/beta": 0.08960523456335068,
      "epsilon_dpo/beta_margin_grad_mean": -0.48835986852645874,
      "epsilon_dpo/beta_margin_grad_std": 0.02396991290152073,
      "epsilon_dpo/beta_margin_mean": 0.0466950498521328,
      "epsilon_dpo/beta_margin_std": 0.09624534845352173,
      "epsilon_dpo/loss_margin_mean": 0.5280405282974243,
      "grad_norm": 20.006235122680664,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.09001823514699936,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.983095894354857e-07,
      "logits/chosen": 0.19097547233104706,
      "logits/rejected": 0.24766327440738678,
      "logps/chosen": -81.07152557373047,
      "logps/ref_chosen": -80.89642333984375,
      "logps/ref_rejected": -115.47240447998047,
      "logps/rejected": -116.17554473876953,
      "loss": 1.3425,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.01596353016793728,
      "rewards/margins": 0.046695053577423096,
      "rewards/rejected": -0.06265857815742493,
      "step": 90
    },
    {
      "epoch": 0.13756613756613756,
      "epsilon_dpo/beta": 0.08943919092416763,
      "epsilon_dpo/beta_margin_grad_mean": -0.4875122010707855,
      "epsilon_dpo/beta_margin_grad_std": 0.027514338493347168,
      "epsilon_dpo/beta_margin_mean": 0.05014381930232048,
      "epsilon_dpo/beta_margin_std": 0.11042484641075134,
      "epsilon_dpo/loss_margin_mean": 0.5706114172935486,
      "grad_norm": 17.863554000854492,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.08959824591875076,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.98152617002662e-07,
      "logits/chosen": 0.08106125146150589,
      "logits/rejected": 0.11380688101053238,
      "logps/chosen": -74.26775360107422,
      "logps/ref_chosen": -74.18321228027344,
      "logps/ref_rejected": -98.54835510253906,
      "logps/rejected": -99.20350646972656,
      "loss": 1.3398,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.008093868382275105,
      "rewards/margins": 0.0501437783241272,
      "rewards/rejected": -0.058237649500370026,
      "step": 91
    },
    {
      "epoch": 0.13907785336356765,
      "epsilon_dpo/beta": 0.08935565501451492,
      "epsilon_dpo/beta_margin_grad_mean": -0.49239903688430786,
      "epsilon_dpo/beta_margin_grad_std": 0.03181348368525505,
      "epsilon_dpo/beta_margin_mean": 0.03056526929140091,
      "epsilon_dpo/beta_margin_std": 0.1277761161327362,
      "epsilon_dpo/loss_margin_mean": 0.35297879576683044,
      "grad_norm": 19.929975509643555,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.0894305631518364,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 4.979887032076988e-07,
      "logits/chosen": 0.1341371238231659,
      "logits/rejected": 0.2965472638607025,
      "logps/chosen": -77.81866455078125,
      "logps/ref_chosen": -77.45124816894531,
      "logps/ref_rejected": -98.57789611816406,
      "logps/rejected": -99.29828643798828,
      "loss": 1.36,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.033338889479637146,
      "rewards/margins": 0.030565232038497925,
      "rewards/rejected": -0.06390412151813507,
      "step": 92
    },
    {
      "epoch": 0.14058956916099774,
      "epsilon_dpo/beta": 0.08924403786659241,
      "epsilon_dpo/beta_margin_grad_mean": -0.49429938197135925,
      "epsilon_dpo/beta_margin_grad_std": 0.03447291627526283,
      "epsilon_dpo/beta_margin_mean": 0.022963959723711014,
      "epsilon_dpo/beta_margin_std": 0.13867661356925964,
      "epsilon_dpo/loss_margin_mean": 0.2693634033203125,
      "grad_norm": 16.53227424621582,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.08934679627418518,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 4.978178526356172e-07,
      "logits/chosen": -0.046473048627376556,
      "logits/rejected": 0.13449837267398834,
      "logps/chosen": -67.758544921875,
      "logps/ref_chosen": -67.58747100830078,
      "logps/ref_rejected": -81.46165466308594,
      "logps/rejected": -81.90208435058594,
      "loss": 1.3683,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": -0.015754221007227898,
      "rewards/margins": 0.022963931784033775,
      "rewards/rejected": -0.03871815279126167,
      "step": 93
    },
    {
      "epoch": 0.1421012849584278,
      "epsilon_dpo/beta": 0.0889652892947197,
      "epsilon_dpo/beta_margin_grad_mean": -0.4860864281654358,
      "epsilon_dpo/beta_margin_grad_std": 0.038560304790735245,
      "epsilon_dpo/beta_margin_mean": 0.05622333660721779,
      "epsilon_dpo/beta_margin_std": 0.1561013162136078,
      "epsilon_dpo/loss_margin_mean": 0.6430590152740479,
      "grad_norm": 19.217985153198242,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.08923525363206863,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.976400700654751e-07,
      "logits/chosen": 0.021004803478717804,
      "logits/rejected": -0.012107852846384048,
      "logps/chosen": -85.24430847167969,
      "logps/ref_chosen": -85.1270523071289,
      "logps/ref_rejected": -102.55231475830078,
      "logps/rejected": -103.31263732910156,
      "loss": 1.3369,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.011046299710869789,
      "rewards/margins": 0.05622336268424988,
      "rewards/rejected": -0.06726966798305511,
      "step": 94
    },
    {
      "epoch": 0.1436130007558579,
      "epsilon_dpo/beta": 0.08874374628067017,
      "epsilon_dpo/beta_margin_grad_mean": -0.4858720302581787,
      "epsilon_dpo/beta_margin_grad_std": 0.036910202354192734,
      "epsilon_dpo/beta_margin_mean": 0.05691284313797951,
      "epsilon_dpo/beta_margin_std": 0.1488497406244278,
      "epsilon_dpo/loss_margin_mean": 0.6532175540924072,
      "grad_norm": 20.199888229370117,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.08895726501941681,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.974553604702332e-07,
      "logits/chosen": 0.19126339256763458,
      "logits/rejected": 0.3154688775539398,
      "logps/chosen": -85.20614624023438,
      "logps/ref_chosen": -84.89950561523438,
      "logps/ref_rejected": -116.98944091796875,
      "logps/rejected": -117.94930267333984,
      "loss": 1.3357,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.027601946145296097,
      "rewards/margins": 0.056912824511528015,
      "rewards/rejected": -0.08451476693153381,
      "step": 95
    },
    {
      "epoch": 0.14512471655328799,
      "epsilon_dpo/beta": 0.08855017274618149,
      "epsilon_dpo/beta_margin_grad_mean": -0.48356518149375916,
      "epsilon_dpo/beta_margin_grad_std": 0.038781337440013885,
      "epsilon_dpo/beta_margin_mean": 0.06643398851156235,
      "epsilon_dpo/beta_margin_std": 0.15669092535972595,
      "epsilon_dpo/loss_margin_mean": 0.7625659704208374,
      "grad_norm": 18.814918518066406,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.0887354239821434,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.972637290166157e-07,
      "logits/chosen": 0.20108938217163086,
      "logits/rejected": 0.3051398694515228,
      "logps/chosen": -83.65084838867188,
      "logps/ref_chosen": -83.37467956542969,
      "logps/ref_rejected": -109.94418334960938,
      "logps/rejected": -110.98291015625,
      "loss": 1.3271,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.02498556300997734,
      "rewards/margins": 0.06643401086330414,
      "rewards/rejected": -0.09141957014799118,
      "step": 96
    },
    {
      "epoch": 0.14663643235071808,
      "epsilon_dpo/beta": 0.08843990415334702,
      "epsilon_dpo/beta_margin_grad_mean": -0.49306973814964294,
      "epsilon_dpo/beta_margin_grad_std": 0.036061882972717285,
      "epsilon_dpo/beta_margin_mean": 0.027809543535113335,
      "epsilon_dpo/beta_margin_std": 0.14509648084640503,
      "epsilon_dpo/loss_margin_mean": 0.3274966776371002,
      "grad_norm": 19.39443016052246,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.08854173868894577,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 4.970651810649666e-07,
      "logits/chosen": 0.16154778003692627,
      "logits/rejected": 0.19344012439250946,
      "logps/chosen": -90.42269897460938,
      "logps/ref_chosen": -89.9176025390625,
      "logps/ref_rejected": -106.89872741699219,
      "logps/rejected": -107.73133087158203,
      "loss": 1.3639,
      "rewards/accuracies": 0.578125,
      "rewards/chosen": -0.04536530748009682,
      "rewards/margins": 0.02780964970588684,
      "rewards/rejected": -0.07317495346069336,
      "step": 97
    },
    {
      "epoch": 0.14814814814814814,
      "epsilon_dpo/beta": 0.08821894228458405,
      "epsilon_dpo/beta_margin_grad_mean": -0.4927210807800293,
      "epsilon_dpo/beta_margin_grad_std": 0.03449036926031113,
      "epsilon_dpo/beta_margin_mean": 0.029177924618124962,
      "epsilon_dpo/beta_margin_std": 0.13922542333602905,
      "epsilon_dpo/loss_margin_mean": 0.3414802551269531,
      "grad_norm": 18.046215057373047,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.08843120187520981,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.968597221690985e-07,
      "logits/chosen": 0.11720709502696991,
      "logits/rejected": 0.3440602421760559,
      "logps/chosen": -79.58689880371094,
      "logps/ref_chosen": -79.12416076660156,
      "logps/ref_rejected": -96.21038818359375,
      "logps/rejected": -97.01461029052734,
      "loss": 1.3622,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.04131529480218887,
      "rewards/margins": 0.02917795069515705,
      "rewards/rejected": -0.07049324363470078,
      "step": 98
    },
    {
      "epoch": 0.14965986394557823,
      "epsilon_dpo/beta": 0.08808165043592453,
      "epsilon_dpo/beta_margin_grad_mean": -0.48774832487106323,
      "epsilon_dpo/beta_margin_grad_std": 0.043416302651166916,
      "epsilon_dpo/beta_margin_mean": 0.049811091274023056,
      "epsilon_dpo/beta_margin_std": 0.1761889010667801,
      "epsilon_dpo/loss_margin_mean": 0.5790194272994995,
      "grad_norm": 19.454317092895508,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.08821067214012146,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.966473580761389e-07,
      "logits/chosen": 0.06653241068124771,
      "logits/rejected": 0.320538729429245,
      "logps/chosen": -82.21040344238281,
      "logps/ref_chosen": -81.95954895019531,
      "logps/ref_rejected": -110.06712341308594,
      "logps/rejected": -110.8969955444336,
      "loss": 1.3448,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.022646091878414154,
      "rewards/margins": 0.04981113597750664,
      "rewards/rejected": -0.07245723158121109,
      "step": 99
    },
    {
      "epoch": 0.15117157974300832,
      "epsilon_dpo/beta": 0.0878891870379448,
      "epsilon_dpo/beta_margin_grad_mean": -0.4885050058364868,
      "epsilon_dpo/beta_margin_grad_std": 0.04928956553339958,
      "epsilon_dpo/beta_margin_mean": 0.0469319149851799,
      "epsilon_dpo/beta_margin_std": 0.2008965015411377,
      "epsilon_dpo/loss_margin_mean": 0.5497127771377563,
      "grad_norm": 19.978914260864258,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.08807305991649628,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.964280947263676e-07,
      "logits/chosen": 0.1993522197008133,
      "logits/rejected": 0.27488380670547485,
      "logps/chosen": -106.40570068359375,
      "logps/ref_chosen": -105.83000183105469,
      "logps/ref_rejected": -113.98239135742188,
      "logps/rejected": -115.10780334472656,
      "loss": 1.3499,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.051381174474954605,
      "rewards/margins": 0.046931929886341095,
      "rewards/rejected": -0.098313108086586,
      "step": 100
    },
    {
      "epoch": 0.15117157974300832,
      "eval_epsilon_dpo/beta": 0.0876714214682579,
      "eval_epsilon_dpo/beta_margin_grad_mean": -0.48583516478538513,
      "eval_epsilon_dpo/beta_margin_grad_std": 0.04487954452633858,
      "eval_epsilon_dpo/beta_margin_mean": 0.05737590044736862,
      "eval_epsilon_dpo/beta_margin_std": 0.18227516114711761,
      "eval_epsilon_dpo/loss_margin_mean": 0.6693454384803772,
      "eval_kl/n_epsilon_steps": 0.37588027119636536,
      "eval_kl/p_epsilon_steps": 0.6241196990013123,
      "eval_logits/chosen": 0.11220303922891617,
      "eval_logits/rejected": 0.12404737621545792,
      "eval_logps/chosen": -87.733154296875,
      "eval_logps/ref_chosen": -87.42715454101562,
      "eval_logps/ref_rejected": -104.23548889160156,
      "eval_logps/rejected": -105.2108383178711,
      "eval_loss": 0.6692028045654297,
      "eval_rewards/accuracies": 0.6311619877815247,
      "eval_rewards/chosen": -0.027519822120666504,
      "eval_rewards/margins": 0.057375892996788025,
      "eval_rewards/rejected": -0.08489571511745453,
      "eval_runtime": 47.7081,
      "eval_samples_per_second": 48.273,
      "eval_steps_per_second": 1.509,
      "step": 100
    },
    {
      "epoch": 0.15268329554043839,
      "epsilon_dpo/beta": 0.08745016157627106,
      "epsilon_dpo/beta_margin_grad_mean": -0.48087450861930847,
      "epsilon_dpo/beta_margin_grad_std": 0.035437703132629395,
      "epsilon_dpo/beta_margin_mean": 0.07677444070577621,
      "epsilon_dpo/beta_margin_std": 0.142720028758049,
      "epsilon_dpo/loss_margin_mean": 0.8880053758621216,
      "grad_norm": 16.972736358642578,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.08788082003593445,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.96201938253052e-07,
      "logits/chosen": -0.2826855480670929,
      "logits/rejected": -0.02019951120018959,
      "logps/chosen": -74.92662048339844,
      "logps/ref_chosen": -74.89061737060547,
      "logps/ref_rejected": -96.2606430053711,
      "logps/rejected": -97.18464660644531,
      "loss": 1.3161,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.003706066869199276,
      "rewards/margins": 0.07677440345287323,
      "rewards/rejected": -0.08048047125339508,
      "step": 101
    },
    {
      "epoch": 0.15419501133786848,
      "epsilon_dpo/beta": 0.08742501586675644,
      "epsilon_dpo/beta_margin_grad_mean": -0.4893290102481842,
      "epsilon_dpo/beta_margin_grad_std": 0.0464043989777565,
      "epsilon_dpo/beta_margin_mean": 0.043459098786115646,
      "epsilon_dpo/beta_margin_std": 0.18812763690948486,
      "epsilon_dpo/loss_margin_mean": 0.5132368206977844,
      "grad_norm": 17.96248435974121,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.08744360506534576,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 4.959688949822748e-07,
      "logits/chosen": 0.18720057606697083,
      "logits/rejected": 0.10392941534519196,
      "logps/chosen": -87.634521484375,
      "logps/ref_chosen": -87.23258972167969,
      "logps/ref_rejected": -97.16923522949219,
      "logps/rejected": -98.08440399169922,
      "loss": 1.3521,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.035716209560632706,
      "rewards/margins": 0.04345907270908356,
      "rewards/rejected": -0.07917527854442596,
      "step": 102
    },
    {
      "epoch": 0.15570672713529857,
      "epsilon_dpo/beta": 0.08715182542800903,
      "epsilon_dpo/beta_margin_grad_mean": -0.47798603773117065,
      "epsilon_dpo/beta_margin_grad_std": 0.04238240793347359,
      "epsilon_dpo/beta_margin_mean": 0.08902125805616379,
      "epsilon_dpo/beta_margin_std": 0.17188170552253723,
      "epsilon_dpo/loss_margin_mean": 1.0343337059020996,
      "grad_norm": 20.719120025634766,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.08741628378629684,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.957289714327572e-07,
      "logits/chosen": -0.14926955103874207,
      "logits/rejected": 0.17941465973854065,
      "logps/chosen": -85.25311279296875,
      "logps/ref_chosen": -85.21533203125,
      "logps/ref_rejected": -105.19367980957031,
      "logps/rejected": -106.26579284667969,
      "loss": 1.3066,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.004020987544208765,
      "rewards/margins": 0.08902129530906677,
      "rewards/rejected": -0.0930422842502594,
      "step": 103
    },
    {
      "epoch": 0.15721844293272866,
      "epsilon_dpo/beta": 0.08682584762573242,
      "epsilon_dpo/beta_margin_grad_mean": -0.47746115922927856,
      "epsilon_dpo/beta_margin_grad_std": 0.05114719271659851,
      "epsilon_dpo/beta_margin_mean": 0.0918625146150589,
      "epsilon_dpo/beta_margin_std": 0.2094704508781433,
      "epsilon_dpo/loss_margin_mean": 1.073301911354065,
      "grad_norm": 19.88721466064453,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.08714395761489868,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.954821743156767e-07,
      "logits/chosen": 0.26530805230140686,
      "logits/rejected": 0.4458683133125305,
      "logps/chosen": -79.91732788085938,
      "logps/ref_chosen": -79.83291625976562,
      "logps/ref_rejected": -128.39964294433594,
      "logps/rejected": -129.557373046875,
      "loss": 1.3074,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.007967567071318626,
      "rewards/margins": 0.09186242520809174,
      "rewards/rejected": -0.09982998669147491,
      "step": 104
    },
    {
      "epoch": 0.15873015873015872,
      "epsilon_dpo/beta": 0.08647433668375015,
      "epsilon_dpo/beta_margin_grad_mean": -0.47956883907318115,
      "epsilon_dpo/beta_margin_grad_std": 0.05354239419102669,
      "epsilon_dpo/beta_margin_mean": 0.0827774778008461,
      "epsilon_dpo/beta_margin_std": 0.21730993688106537,
      "epsilon_dpo/loss_margin_mean": 0.9738532304763794,
      "grad_norm": 19.439359664916992,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.0868183895945549,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.952285105344791e-07,
      "logits/chosen": -0.006076253950595856,
      "logits/rejected": 0.18681776523590088,
      "logps/chosen": -95.36561584472656,
      "logps/ref_chosen": -94.9561767578125,
      "logps/ref_rejected": -119.2974853515625,
      "logps/rejected": -120.68077087402344,
      "loss": 1.3169,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.036373164504766464,
      "rewards/margins": 0.08277745544910431,
      "rewards/rejected": -0.11915061622858047,
      "step": 105
    },
    {
      "epoch": 0.1602418745275888,
      "epsilon_dpo/beta": 0.08609743416309357,
      "epsilon_dpo/beta_margin_grad_mean": -0.4816688895225525,
      "epsilon_dpo/beta_margin_grad_std": 0.04480881616473198,
      "epsilon_dpo/beta_margin_mean": 0.07405252009630203,
      "epsilon_dpo/beta_margin_std": 0.18235161900520325,
      "epsilon_dpo/loss_margin_mean": 0.8727110624313354,
      "grad_norm": 17.128450393676758,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.08646711707115173,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.949679871846857e-07,
      "logits/chosen": 0.1752602756023407,
      "logits/rejected": 0.12919707596302032,
      "logps/chosen": -81.92420959472656,
      "logps/ref_chosen": -82.00679016113281,
      "logps/ref_rejected": -89.77203369140625,
      "logps/rejected": -90.56216430664062,
      "loss": 1.3218,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": 0.0064305951818823814,
      "rewards/margins": 0.07405255734920502,
      "rewards/rejected": -0.06762196123600006,
      "step": 106
    },
    {
      "epoch": 0.1617535903250189,
      "epsilon_dpo/beta": 0.08591073751449585,
      "epsilon_dpo/beta_margin_grad_mean": -0.4878389835357666,
      "epsilon_dpo/beta_margin_grad_std": 0.04847485199570656,
      "epsilon_dpo/beta_margin_mean": 0.04896136373281479,
      "epsilon_dpo/beta_margin_std": 0.19566507637500763,
      "epsilon_dpo/loss_margin_mean": 0.5875788331031799,
      "grad_norm": 19.390165328979492,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.08609047532081604,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.947006115536947e-07,
      "logits/chosen": -0.012461047619581223,
      "logits/rejected": 0.23972593247890472,
      "logps/chosen": -102.92835998535156,
      "logps/ref_chosen": -102.37884521484375,
      "logps/ref_rejected": -113.08601379394531,
      "logps/rejected": -114.22311401367188,
      "loss": 1.3475,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.04798401519656181,
      "rewards/margins": 0.04896135628223419,
      "rewards/rejected": -0.0969453752040863,
      "step": 107
    },
    {
      "epoch": 0.16326530612244897,
      "epsilon_dpo/beta": 0.08564267307519913,
      "epsilon_dpo/beta_margin_grad_mean": -0.487002968788147,
      "epsilon_dpo/beta_margin_grad_std": 0.051084764301776886,
      "epsilon_dpo/beta_margin_mean": 0.05215499550104141,
      "epsilon_dpo/beta_margin_std": 0.20733976364135742,
      "epsilon_dpo/loss_margin_mean": 0.6264809966087341,
      "grad_norm": 17.811731338500977,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.08590255677700043,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.944263911205772e-07,
      "logits/chosen": 0.044683102518320084,
      "logits/rejected": -0.024815678596496582,
      "logps/chosen": -91.81816101074219,
      "logps/ref_chosen": -91.51483154296875,
      "logps/ref_rejected": -103.61510467529297,
      "logps/rejected": -104.544921875,
      "loss": 1.3455,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.02699120342731476,
      "rewards/margins": 0.052154965698719025,
      "rewards/rejected": -0.07914617657661438,
      "step": 108
    },
    {
      "epoch": 0.16477702191987906,
      "epsilon_dpo/beta": 0.08545617759227753,
      "epsilon_dpo/beta_margin_grad_mean": -0.48020491003990173,
      "epsilon_dpo/beta_margin_grad_std": 0.05594097077846527,
      "epsilon_dpo/beta_margin_mean": 0.08056788146495819,
      "epsilon_dpo/beta_margin_std": 0.22767306864261627,
      "epsilon_dpo/loss_margin_mean": 0.9629918336868286,
      "grad_norm": 17.902441024780273,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.08563495427370071,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.941453335558681e-07,
      "logits/chosen": 0.17891252040863037,
      "logits/rejected": -0.03912736475467682,
      "logps/chosen": -83.23799896240234,
      "logps/ref_chosen": -83.30767822265625,
      "logps/ref_rejected": -107.25475311279297,
      "logps/rejected": -108.14806365966797,
      "loss": 1.3202,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.005163197405636311,
      "rewards/margins": 0.08056794106960297,
      "rewards/rejected": -0.07540474086999893,
      "step": 109
    },
    {
      "epoch": 0.16628873771730915,
      "epsilon_dpo/beta": 0.08529634773731232,
      "epsilon_dpo/beta_margin_grad_mean": -0.496372789144516,
      "epsilon_dpo/beta_margin_grad_std": 0.0572805292904377,
      "epsilon_dpo/beta_margin_mean": 0.014190551824867725,
      "epsilon_dpo/beta_margin_std": 0.23347991704940796,
      "epsilon_dpo/loss_margin_mean": 0.18647027015686035,
      "grad_norm": 20.98345375061035,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.08544803410768509,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.938574467213517e-07,
      "logits/chosen": -0.02646956965327263,
      "logits/rejected": 0.17269685864448547,
      "logps/chosen": -104.24122619628906,
      "logps/ref_chosen": -103.62700653076172,
      "logps/ref_rejected": -98.57982635498047,
      "logps/rejected": -99.3805160522461,
      "loss": 1.3857,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.05344737321138382,
      "rewards/margins": 0.014190555550158024,
      "rewards/rejected": -0.06763792783021927,
      "step": 110
    },
    {
      "epoch": 0.16780045351473924,
      "epsilon_dpo/beta": 0.08516336977481842,
      "epsilon_dpo/beta_margin_grad_mean": -0.4857950806617737,
      "epsilon_dpo/beta_margin_grad_std": 0.05001501739025116,
      "epsilon_dpo/beta_margin_mean": 0.05734188109636307,
      "epsilon_dpo/beta_margin_std": 0.20205387473106384,
      "epsilon_dpo/loss_margin_mean": 0.691691517829895,
      "grad_norm": 16.861454010009766,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.08528811484575272,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.935627386698418e-07,
      "logits/chosen": -0.08613793551921844,
      "logits/rejected": -0.06585603207349777,
      "logps/chosen": -75.0671157836914,
      "logps/ref_chosen": -74.81674194335938,
      "logps/ref_rejected": -97.25582122802734,
      "logps/rejected": -98.19789123535156,
      "loss": 1.3399,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.022064432501792908,
      "rewards/margins": 0.05734182894229889,
      "rewards/rejected": -0.0794062614440918,
      "step": 111
    },
    {
      "epoch": 0.1693121693121693,
      "epsilon_dpo/beta": 0.08471114188432693,
      "epsilon_dpo/beta_margin_grad_mean": -0.46657806634902954,
      "epsilon_dpo/beta_margin_grad_std": 0.05246131494641304,
      "epsilon_dpo/beta_margin_mean": 0.13550089299678802,
      "epsilon_dpo/beta_margin_std": 0.21291041374206543,
      "epsilon_dpo/loss_margin_mean": 1.614532470703125,
      "grad_norm": 19.040555953979492,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.08515506237745285,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 4.932612176449559e-07,
      "logits/chosen": 0.0674043744802475,
      "logits/rejected": 0.21437129378318787,
      "logps/chosen": -82.7444839477539,
      "logps/ref_chosen": -83.02973175048828,
      "logps/ref_rejected": -130.31686401367188,
      "logps/rejected": -131.64613342285156,
      "loss": 1.2666,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": 0.02339255064725876,
      "rewards/margins": 0.13550087809562683,
      "rewards/rejected": -0.11210831254720688,
      "step": 112
    },
    {
      "epoch": 0.1708238851095994,
      "epsilon_dpo/beta": 0.08447527885437012,
      "epsilon_dpo/beta_margin_grad_mean": -0.48426786065101624,
      "epsilon_dpo/beta_margin_grad_std": 0.05853166803717613,
      "epsilon_dpo/beta_margin_mean": 0.06381077319383621,
      "epsilon_dpo/beta_margin_std": 0.2405387908220291,
      "epsilon_dpo/loss_margin_mean": 0.7743654251098633,
      "grad_norm": 17.83749771118164,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.08470506966114044,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.929528920808854e-07,
      "logits/chosen": -0.16013295948505402,
      "logits/rejected": 0.22781725227832794,
      "logps/chosen": -77.02522277832031,
      "logps/ref_chosen": -77.30706787109375,
      "logps/ref_rejected": -95.55424499511719,
      "logps/rejected": -96.04676818847656,
      "loss": 1.3378,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.022567156702280045,
      "rewards/margins": 0.06381077319383621,
      "rewards/rejected": -0.04124361276626587,
      "step": 113
    },
    {
      "epoch": 0.17233560090702948,
      "epsilon_dpo/beta": 0.08426475524902344,
      "epsilon_dpo/beta_margin_grad_mean": -0.4811977446079254,
      "epsilon_dpo/beta_margin_grad_std": 0.055794645100831985,
      "epsilon_dpo/beta_margin_mean": 0.07640980184078217,
      "epsilon_dpo/beta_margin_std": 0.226527139544487,
      "epsilon_dpo/loss_margin_mean": 0.9263713955879211,
      "grad_norm": 18.02449607849121,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.08446750044822693,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.92637770602159e-07,
      "logits/chosen": 0.0650443509221077,
      "logits/rejected": -0.15849056839942932,
      "logps/chosen": -91.77374267578125,
      "logps/ref_chosen": -91.81670379638672,
      "logps/ref_rejected": -104.94361877441406,
      "logps/rejected": -105.82703399658203,
      "loss": 1.3241,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.0029620258137583733,
      "rewards/margins": 0.07640981674194336,
      "rewards/rejected": -0.07344779372215271,
      "step": 114
    },
    {
      "epoch": 0.17384731670445955,
      "epsilon_dpo/beta": 0.08402828872203827,
      "epsilon_dpo/beta_margin_grad_mean": -0.4740752875804901,
      "epsilon_dpo/beta_margin_grad_std": 0.060490477830171585,
      "epsilon_dpo/beta_margin_mean": 0.1062379777431488,
      "epsilon_dpo/beta_margin_std": 0.24839532375335693,
      "epsilon_dpo/loss_margin_mean": 1.2842512130737305,
      "grad_norm": 17.447057723999023,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.0842568576335907,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.923158620234019e-07,
      "logits/chosen": -0.1340171992778778,
      "logits/rejected": -0.15569829940795898,
      "logps/chosen": -74.81437683105469,
      "logps/ref_chosen": -75.18775939941406,
      "logps/ref_rejected": -103.025634765625,
      "logps/rejected": -103.93651580810547,
      "loss": 1.2981,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.03075750358402729,
      "rewards/margins": 0.10623794794082642,
      "rewards/rejected": -0.07548044621944427,
      "step": 115
    },
    {
      "epoch": 0.17535903250188964,
      "epsilon_dpo/beta": 0.08368758857250214,
      "epsilon_dpo/beta_margin_grad_mean": -0.4681667387485504,
      "epsilon_dpo/beta_margin_grad_std": 0.0566866435110569,
      "epsilon_dpo/beta_margin_mean": 0.1293656975030899,
      "epsilon_dpo/beta_margin_std": 0.23101994395256042,
      "epsilon_dpo/loss_margin_mean": 1.5630714893341064,
      "grad_norm": 17.443565368652344,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.08402055501937866,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.91987175349089e-07,
      "logits/chosen": -0.1186988577246666,
      "logits/rejected": 0.14887529611587524,
      "logps/chosen": -68.73472595214844,
      "logps/ref_chosen": -69.45286560058594,
      "logps/ref_rejected": -96.70513153076172,
      "logps/rejected": -97.55005645751953,
      "loss": 1.2743,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": 0.05944942682981491,
      "rewards/margins": 0.1293656826019287,
      "rewards/rejected": -0.0699162632226944,
      "step": 116
    },
    {
      "epoch": 0.17687074829931973,
      "epsilon_dpo/beta": 0.08342743664979935,
      "epsilon_dpo/beta_margin_grad_mean": -0.47609150409698486,
      "epsilon_dpo/beta_margin_grad_std": 0.061780449002981186,
      "epsilon_dpo/beta_margin_mean": 0.09696857631206512,
      "epsilon_dpo/beta_margin_std": 0.2511261999607086,
      "epsilon_dpo/loss_margin_mean": 1.1843879222869873,
      "grad_norm": 16.713768005371094,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.08368059992790222,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.916517197732933e-07,
      "logits/chosen": -0.10571661591529846,
      "logits/rejected": -0.055547840893268585,
      "logps/chosen": -77.54984283447266,
      "logps/ref_chosen": -78.27897644042969,
      "logps/ref_rejected": -92.01252746582031,
      "logps/rejected": -92.46778869628906,
      "loss": 1.3073,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.05986713990569115,
      "rewards/margins": 0.09696857631206512,
      "rewards/rejected": -0.03710143640637398,
      "step": 117
    },
    {
      "epoch": 0.17838246409674982,
      "epsilon_dpo/beta": 0.08308932185173035,
      "epsilon_dpo/beta_margin_grad_mean": -0.47561800479888916,
      "epsilon_dpo/beta_margin_grad_std": 0.0590248741209507,
      "epsilon_dpo/beta_margin_mean": 0.09878911077976227,
      "epsilon_dpo/beta_margin_std": 0.23957890272140503,
      "epsilon_dpo/loss_margin_mean": 1.2095437049865723,
      "grad_norm": 16.629575729370117,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.08341991156339645,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.913095046794281e-07,
      "logits/chosen": 0.1436939239501953,
      "logits/rejected": 0.16659829020500183,
      "logps/chosen": -75.42961120605469,
      "logps/ref_chosen": -76.03759765625,
      "logps/ref_rejected": -103.52017211914062,
      "logps/rejected": -104.12173461914062,
      "loss": 1.3042,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.04961453005671501,
      "rewards/margins": 0.09878911077976227,
      "rewards/rejected": -0.049174580723047256,
      "step": 118
    },
    {
      "epoch": 0.17989417989417988,
      "epsilon_dpo/beta": 0.0829349085688591,
      "epsilon_dpo/beta_margin_grad_mean": -0.4789765775203705,
      "epsilon_dpo/beta_margin_grad_std": 0.07665891945362091,
      "epsilon_dpo/beta_margin_mean": 0.08581650257110596,
      "epsilon_dpo/beta_margin_std": 0.3149167597293854,
      "epsilon_dpo/loss_margin_mean": 1.063791036605835,
      "grad_norm": 17.0283260345459,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.08308239281177521,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.909605396399855e-07,
      "logits/chosen": -0.09115805476903915,
      "logits/rejected": -0.07789916545152664,
      "logps/chosen": -82.70497131347656,
      "logps/ref_chosen": -83.196044921875,
      "logps/ref_rejected": -103.79010009765625,
      "logps/rejected": -104.3628158569336,
      "loss": 1.3268,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": 0.03945211321115494,
      "rewards/margins": 0.08581651002168655,
      "rewards/rejected": -0.046364400535821915,
      "step": 119
    },
    {
      "epoch": 0.18140589569160998,
      "epsilon_dpo/beta": 0.08257235586643219,
      "epsilon_dpo/beta_margin_grad_mean": -0.46213966608047485,
      "epsilon_dpo/beta_margin_grad_std": 0.055534347891807556,
      "epsilon_dpo/beta_margin_mean": 0.15391035377979279,
      "epsilon_dpo/beta_margin_std": 0.22700658440589905,
      "epsilon_dpo/loss_margin_mean": 1.8817929029464722,
      "grad_norm": 18.204940795898438,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.082926906645298,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.906048344162676e-07,
      "logits/chosen": 0.13377802073955536,
      "logits/rejected": 0.15807950496673584,
      "logps/chosen": -83.54419708251953,
      "logps/ref_chosen": -84.4485855102539,
      "logps/ref_rejected": -110.97505950927734,
      "logps/rejected": -111.95246887207031,
      "loss": 1.251,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": 0.07405398041009903,
      "rewards/margins": 0.15391036868095398,
      "rewards/rejected": -0.07985638827085495,
      "step": 120
    },
    {
      "epoch": 0.18291761148904007,
      "epsilon_dpo/beta": 0.08239330351352692,
      "epsilon_dpo/beta_margin_grad_mean": -0.4695954918861389,
      "epsilon_dpo/beta_margin_grad_std": 0.07052520662546158,
      "epsilon_dpo/beta_margin_mean": 0.12572382390499115,
      "epsilon_dpo/beta_margin_std": 0.2914615273475647,
      "epsilon_dpo/loss_margin_mean": 1.5515471696853638,
      "grad_norm": 17.93547821044922,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.08256568014621735,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.902423989581143e-07,
      "logits/chosen": -0.16934193670749664,
      "logits/rejected": -0.0674653947353363,
      "logps/chosen": -76.95396423339844,
      "logps/ref_chosen": -77.71607971191406,
      "logps/ref_rejected": -127.38249969482422,
      "logps/rejected": -128.17193603515625,
      "loss": 1.2853,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.06202112138271332,
      "rewards/margins": 0.12572380900382996,
      "rewards/rejected": -0.06370268762111664,
      "step": 121
    },
    {
      "epoch": 0.18442932728647016,
      "epsilon_dpo/beta": 0.08221346139907837,
      "epsilon_dpo/beta_margin_grad_mean": -0.4754737317562103,
      "epsilon_dpo/beta_margin_grad_std": 0.07240553200244904,
      "epsilon_dpo/beta_margin_mean": 0.10210616141557693,
      "epsilon_dpo/beta_margin_std": 0.30038896203041077,
      "epsilon_dpo/loss_margin_mean": 1.2675155401229858,
      "grad_norm": 17.199674606323242,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.08238545805215836,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.898732434036243e-07,
      "logits/chosen": -0.11136245727539062,
      "logits/rejected": -0.16447636485099792,
      "logps/chosen": -83.83956909179688,
      "logps/ref_chosen": -84.65371704101562,
      "logps/ref_rejected": -98.93002319335938,
      "logps/rejected": -99.38338470458984,
      "loss": 1.3089,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.0658092126250267,
      "rewards/margins": 0.10210618376731873,
      "rewards/rejected": -0.03629697486758232,
      "step": 122
    },
    {
      "epoch": 0.18594104308390022,
      "epsilon_dpo/beta": 0.08200832456350327,
      "epsilon_dpo/beta_margin_grad_mean": -0.4690355062484741,
      "epsilon_dpo/beta_margin_grad_std": 0.06405887752771378,
      "epsilon_dpo/beta_margin_mean": 0.12652172148227692,
      "epsilon_dpo/beta_margin_std": 0.2623896598815918,
      "epsilon_dpo/loss_margin_mean": 1.5664746761322021,
      "grad_norm": 16.217876434326172,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.0822056382894516,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.894973780788722e-07,
      "logits/chosen": 0.05034737288951874,
      "logits/rejected": 0.10809341073036194,
      "logps/chosen": -76.31719970703125,
      "logps/ref_chosen": -77.63322448730469,
      "logps/ref_rejected": -101.60614013671875,
      "logps/rejected": -101.85658264160156,
      "loss": 1.2807,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.10695463418960571,
      "rewards/margins": 0.1265217363834381,
      "rewards/rejected": -0.019567107781767845,
      "step": 123
    },
    {
      "epoch": 0.1874527588813303,
      "epsilon_dpo/beta": 0.08172693103551865,
      "epsilon_dpo/beta_margin_grad_mean": -0.4645516574382782,
      "epsilon_dpo/beta_margin_grad_std": 0.08100985735654831,
      "epsilon_dpo/beta_margin_mean": 0.14648139476776123,
      "epsilon_dpo/beta_margin_std": 0.3352731168270111,
      "epsilon_dpo/loss_margin_mean": 1.8201913833618164,
      "grad_norm": 17.60958480834961,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.08200063556432724,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.89114813497619e-07,
      "logits/chosen": -0.2651221752166748,
      "logits/rejected": 0.24253401160240173,
      "logps/chosen": -75.82962036132812,
      "logps/ref_chosen": -77.36146545410156,
      "logps/ref_rejected": -116.55441284179688,
      "logps/rejected": -116.84275817871094,
      "loss": 1.2727,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.12411925941705704,
      "rewards/margins": 0.14648135006427765,
      "rewards/rejected": -0.02236209250986576,
      "step": 124
    },
    {
      "epoch": 0.1889644746787604,
      "epsilon_dpo/beta": 0.08131925761699677,
      "epsilon_dpo/beta_margin_grad_mean": -0.4667989909648895,
      "epsilon_dpo/beta_margin_grad_std": 0.07144204527139664,
      "epsilon_dpo/beta_margin_mean": 0.13624414801597595,
      "epsilon_dpo/beta_margin_std": 0.2952669858932495,
      "epsilon_dpo/loss_margin_mean": 1.698812484741211,
      "grad_norm": 18.070085525512695,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.08171971887350082,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.887255603610184e-07,
      "logits/chosen": 0.05429578199982643,
      "logits/rejected": -0.024670738726854324,
      "logps/chosen": -87.46548461914062,
      "logps/ref_chosen": -88.697998046875,
      "logps/ref_rejected": -121.4852523803711,
      "logps/rejected": -121.95155334472656,
      "loss": 1.2761,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": 0.09958155453205109,
      "rewards/margins": 0.13624417781829834,
      "rewards/rejected": -0.03666263073682785,
      "step": 125
    },
    {
      "epoch": 0.19047619047619047,
      "epsilon_dpo/beta": 0.08111798763275146,
      "epsilon_dpo/beta_margin_grad_mean": -0.4818357229232788,
      "epsilon_dpo/beta_margin_grad_std": 0.07389285415410995,
      "epsilon_dpo/beta_margin_mean": 0.07455535233020782,
      "epsilon_dpo/beta_margin_std": 0.3029778301715851,
      "epsilon_dpo/loss_margin_mean": 0.946387767791748,
      "grad_norm": 17.885833740234375,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.08131315559148788,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.883296295573176e-07,
      "logits/chosen": 0.0120854452252388,
      "logits/rejected": 0.16786350309848785,
      "logps/chosen": -90.82383728027344,
      "logps/ref_chosen": -92.44024658203125,
      "logps/ref_rejected": -91.21477508544922,
      "logps/rejected": -90.54476928710938,
      "loss": 1.3358,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.12970831990242004,
      "rewards/margins": 0.07455536723136902,
      "rewards/rejected": 0.05515296012163162,
      "step": 126
    },
    {
      "epoch": 0.19198790627362056,
      "epsilon_dpo/beta": 0.08101709187030792,
      "epsilon_dpo/beta_margin_grad_mean": -0.47536662220954895,
      "epsilon_dpo/beta_margin_grad_std": 0.06849021464586258,
      "epsilon_dpo/beta_margin_mean": 0.10088615864515305,
      "epsilon_dpo/beta_margin_std": 0.27939197421073914,
      "epsilon_dpo/loss_margin_mean": 1.2726936340332031,
      "grad_norm": 16.920372009277344,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.081110380589962,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 4.87927032161552e-07,
      "logits/chosen": -0.09022588282823563,
      "logits/rejected": 0.08949241042137146,
      "logps/chosen": -85.76454162597656,
      "logps/ref_chosen": -87.55062103271484,
      "logps/ref_rejected": -97.16986083984375,
      "logps/rejected": -96.65647888183594,
      "loss": 1.3072,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": 0.1433461308479309,
      "rewards/margins": 0.10088618844747543,
      "rewards/rejected": 0.04245994985103607,
      "step": 127
    },
    {
      "epoch": 0.19349962207105065,
      "epsilon_dpo/beta": 0.080764040350914,
      "epsilon_dpo/beta_margin_grad_mean": -0.47240495681762695,
      "epsilon_dpo/beta_margin_grad_std": 0.09400974959135056,
      "epsilon_dpo/beta_margin_mean": 0.11469464004039764,
      "epsilon_dpo/beta_margin_std": 0.39688840508461,
      "epsilon_dpo/loss_margin_mean": 1.452173113822937,
      "grad_norm": 18.381317138671875,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.08100911974906921,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.875177794352363e-07,
      "logits/chosen": 0.08796774595975876,
      "logits/rejected": -0.0978081226348877,
      "logps/chosen": -91.56980895996094,
      "logps/ref_chosen": -93.4315185546875,
      "logps/ref_rejected": -118.90899658203125,
      "logps/rejected": -118.49946594238281,
      "loss": 1.3131,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.14945122599601746,
      "rewards/margins": 0.11469465494155884,
      "rewards/rejected": 0.03475657105445862,
      "step": 128
    },
    {
      "epoch": 0.19501133786848074,
      "epsilon_dpo/beta": 0.08068910986185074,
      "epsilon_dpo/beta_margin_grad_mean": -0.48256921768188477,
      "epsilon_dpo/beta_margin_grad_std": 0.09119141101837158,
      "epsilon_dpo/beta_margin_mean": 0.07243873178958893,
      "epsilon_dpo/beta_margin_std": 0.3800669014453888,
      "epsilon_dpo/loss_margin_mean": 0.9343692064285278,
      "grad_norm": 18.35399627685547,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.08075675368309021,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 4.871018828260491e-07,
      "logits/chosen": -0.10205899924039841,
      "logits/rejected": -6.125867366790771e-05,
      "logps/chosen": -90.28472900390625,
      "logps/ref_chosen": -91.62559509277344,
      "logps/ref_rejected": -93.87883758544922,
      "logps/rejected": -93.47234344482422,
      "loss": 1.3505,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.10695064067840576,
      "rewards/margins": 0.07243867963552475,
      "rewards/rejected": 0.03451196104288101,
      "step": 129
    },
    {
      "epoch": 0.1965230536659108,
      "epsilon_dpo/beta": 0.08038660138845444,
      "epsilon_dpo/beta_margin_grad_mean": -0.4643150269985199,
      "epsilon_dpo/beta_margin_grad_std": 0.0763268992304802,
      "epsilon_dpo/beta_margin_mean": 0.1470072716474533,
      "epsilon_dpo/beta_margin_std": 0.31455495953559875,
      "epsilon_dpo/loss_margin_mean": 1.8561002016067505,
      "grad_norm": 17.832624435424805,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.08068111538887024,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.866793539675126e-07,
      "logits/chosen": -0.015426471829414368,
      "logits/rejected": 0.08768454939126968,
      "logps/chosen": -82.5267333984375,
      "logps/ref_chosen": -84.67145538330078,
      "logps/ref_rejected": -104.28001403808594,
      "logps/rejected": -103.99138641357422,
      "loss": 1.269,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.17135189473628998,
      "rewards/margins": 0.1470073014497757,
      "rewards/rejected": 0.024344589561223984,
      "step": 130
    },
    {
      "epoch": 0.1980347694633409,
      "epsilon_dpo/beta": 0.08016163855791092,
      "epsilon_dpo/beta_margin_grad_mean": -0.45822128653526306,
      "epsilon_dpo/beta_margin_grad_std": 0.08405930548906326,
      "epsilon_dpo/beta_margin_mean": 0.17493918538093567,
      "epsilon_dpo/beta_margin_std": 0.35443899035453796,
      "epsilon_dpo/loss_margin_mean": 2.212343692779541,
      "grad_norm": 16.11795997619629,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.08037969470024109,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.86250204678667e-07,
      "logits/chosen": -0.13019153475761414,
      "logits/rejected": -0.10465170443058014,
      "logps/chosen": -71.01728057861328,
      "logps/ref_chosen": -73.30256652832031,
      "logps/ref_rejected": -106.04727172851562,
      "logps/rejected": -105.97431945800781,
      "loss": 1.2494,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": 0.18196183443069458,
      "rewards/margins": 0.17493918538093567,
      "rewards/rejected": 0.00702265277504921,
      "step": 131
    },
    {
      "epoch": 0.19954648526077098,
      "epsilon_dpo/beta": 0.07983660697937012,
      "epsilon_dpo/beta_margin_grad_mean": -0.47368597984313965,
      "epsilon_dpo/beta_margin_grad_std": 0.07017382979393005,
      "epsilon_dpo/beta_margin_mean": 0.10802603513002396,
      "epsilon_dpo/beta_margin_std": 0.28927475214004517,
      "epsilon_dpo/loss_margin_mean": 1.3770979642868042,
      "grad_norm": 16.9212646484375,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.08015425503253937,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.858144469637408e-07,
      "logits/chosen": -0.0966077670454979,
      "logits/rejected": -0.13967418670654297,
      "logps/chosen": -85.59697723388672,
      "logps/ref_chosen": -87.72061920166016,
      "logps/ref_rejected": -96.35941314697266,
      "logps/rejected": -95.61286926269531,
      "loss": 1.3017,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.16805227100849152,
      "rewards/margins": 0.10802602767944336,
      "rewards/rejected": 0.06002624332904816,
      "step": 132
    },
    {
      "epoch": 0.20105820105820105,
      "epsilon_dpo/beta": 0.07968823611736298,
      "epsilon_dpo/beta_margin_grad_mean": -0.4730938971042633,
      "epsilon_dpo/beta_margin_grad_std": 0.07494788616895676,
      "epsilon_dpo/beta_margin_mean": 0.1104988306760788,
      "epsilon_dpo/beta_margin_std": 0.30697232484817505,
      "epsilon_dpo/loss_margin_mean": 1.416254997253418,
      "grad_norm": 17.388769149780273,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.07982994616031647,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.853720930118138e-07,
      "logits/chosen": -0.20302733778953552,
      "logits/rejected": 0.018017714843153954,
      "logps/chosen": -87.78105163574219,
      "logps/ref_chosen": -89.97994232177734,
      "logps/ref_rejected": -94.67462158203125,
      "logps/rejected": -93.89199829101562,
      "loss": 1.3021,
      "rewards/accuracies": 0.578125,
      "rewards/chosen": 0.17398512363433838,
      "rewards/margins": 0.11049885302782059,
      "rewards/rejected": 0.0634862631559372,
      "step": 133
    },
    {
      "epoch": 0.20256991685563114,
      "epsilon_dpo/beta": 0.07929006963968277,
      "epsilon_dpo/beta_margin_grad_mean": -0.45269298553466797,
      "epsilon_dpo/beta_margin_grad_std": 0.09166352450847626,
      "epsilon_dpo/beta_margin_mean": 0.19899149239063263,
      "epsilon_dpo/beta_margin_std": 0.3938853442668915,
      "epsilon_dpo/loss_margin_mean": 2.538273334503174,
      "grad_norm": 15.28248119354248,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.07968054711818695,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.849231551964771e-07,
      "logits/chosen": -0.2937329411506653,
      "logits/rejected": -0.21715673804283142,
      "logps/chosen": -73.44625854492188,
      "logps/ref_chosen": -76.23042297363281,
      "logps/ref_rejected": -100.25575256347656,
      "logps/rejected": -100.0098648071289,
      "loss": 1.2343,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": 0.21985578536987305,
      "rewards/margins": 0.19899150729179382,
      "rewards/rejected": 0.020864300429821014,
      "step": 134
    },
    {
      "epoch": 0.20408163265306123,
      "epsilon_dpo/beta": 0.07901948690414429,
      "epsilon_dpo/beta_margin_grad_mean": -0.4722573459148407,
      "epsilon_dpo/beta_margin_grad_std": 0.07582727819681168,
      "epsilon_dpo/beta_margin_mean": 0.11306837946176529,
      "epsilon_dpo/beta_margin_std": 0.3117622435092926,
      "epsilon_dpo/loss_margin_mean": 1.458648443222046,
      "grad_norm": 14.768946647644043,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.07928412407636642,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.844676460754862e-07,
      "logits/chosen": -0.32549357414245605,
      "logits/rejected": -0.032874904572963715,
      "logps/chosen": -67.47518920898438,
      "logps/ref_chosen": -70.47183990478516,
      "logps/ref_rejected": -87.71185302734375,
      "logps/rejected": -86.17385864257812,
      "loss": 1.3004,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": 0.23563119769096375,
      "rewards/margins": 0.11306841671466827,
      "rewards/rejected": 0.12256277352571487,
      "step": 135
    },
    {
      "epoch": 0.20559334845049132,
      "epsilon_dpo/beta": 0.07887225598096848,
      "epsilon_dpo/beta_margin_grad_mean": -0.45937618613243103,
      "epsilon_dpo/beta_margin_grad_std": 0.1084052100777626,
      "epsilon_dpo/beta_margin_mean": 0.17362651228904724,
      "epsilon_dpo/beta_margin_std": 0.46317678689956665,
      "epsilon_dpo/loss_margin_mean": 2.243760824203491,
      "grad_norm": 17.279558181762695,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.07901252061128616,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.840055783904106e-07,
      "logits/chosen": 0.16665995121002197,
      "logits/rejected": 0.0742223784327507,
      "logps/chosen": -88.99594116210938,
      "logps/ref_chosen": -91.05680847167969,
      "logps/ref_rejected": -125.45597839355469,
      "logps/rejected": -125.63887023925781,
      "loss": 1.2718,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": 0.1612674742937088,
      "rewards/margins": 0.17362651228904724,
      "rewards/rejected": -0.012359026819467545,
      "step": 136
    },
    {
      "epoch": 0.20710506424792138,
      "epsilon_dpo/beta": 0.07857676595449448,
      "epsilon_dpo/beta_margin_grad_mean": -0.46005377173423767,
      "epsilon_dpo/beta_margin_grad_std": 0.087054044008255,
      "epsilon_dpo/beta_margin_mean": 0.16871152818202972,
      "epsilon_dpo/beta_margin_std": 0.3723953664302826,
      "epsilon_dpo/loss_margin_mean": 2.1775121688842773,
      "grad_norm": 14.91991138458252,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.07886464893817902,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.835369650662767e-07,
      "logits/chosen": -0.2173086702823639,
      "logits/rejected": -0.10981342196464539,
      "logps/chosen": -73.40845489501953,
      "logps/ref_chosen": -76.20159912109375,
      "logps/ref_rejected": -95.24153137207031,
      "logps/rejected": -94.62591552734375,
      "loss": 1.258,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.2189544439315796,
      "rewards/margins": 0.1687115877866745,
      "rewards/rejected": 0.05024286359548569,
      "step": 137
    },
    {
      "epoch": 0.20861678004535147,
      "epsilon_dpo/beta": 0.07833231985569,
      "epsilon_dpo/beta_margin_grad_mean": -0.47148823738098145,
      "epsilon_dpo/beta_margin_grad_std": 0.09270107001066208,
      "epsilon_dpo/beta_margin_mean": 0.11959164589643478,
      "epsilon_dpo/beta_margin_std": 0.39499062299728394,
      "epsilon_dpo/loss_margin_mean": 1.5597786903381348,
      "grad_norm": 15.54496955871582,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.07857001572847366,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.830618192112065e-07,
      "logits/chosen": -0.33799946308135986,
      "logits/rejected": -0.5601421594619751,
      "logps/chosen": -76.29164123535156,
      "logps/ref_chosen": -78.20573425292969,
      "logps/ref_rejected": -90.40987396240234,
      "logps/rejected": -90.05555725097656,
      "loss": 1.3079,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": 0.14817282557487488,
      "rewards/margins": 0.11959163099527359,
      "rewards/rejected": 0.028581196442246437,
      "step": 138
    },
    {
      "epoch": 0.21012849584278157,
      "epsilon_dpo/beta": 0.07813724875450134,
      "epsilon_dpo/beta_margin_grad_mean": -0.4711554944515228,
      "epsilon_dpo/beta_margin_grad_std": 0.09030873328447342,
      "epsilon_dpo/beta_margin_mean": 0.11854705959558487,
      "epsilon_dpo/beta_margin_std": 0.3753059506416321,
      "epsilon_dpo/loss_margin_mean": 1.5526247024536133,
      "grad_norm": 18.069461822509766,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.07832524925470352,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.825801541160509e-07,
      "logits/chosen": -0.15845316648483276,
      "logits/rejected": -0.138905331492424,
      "logps/chosen": -93.53260803222656,
      "logps/ref_chosen": -94.67202758789062,
      "logps/ref_rejected": -108.34249114990234,
      "logps/rejected": -108.75569152832031,
      "loss": 1.3057,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.0876041054725647,
      "rewards/margins": 0.11854708194732666,
      "rewards/rejected": -0.030942972749471664,
      "step": 139
    },
    {
      "epoch": 0.21164021164021163,
      "epsilon_dpo/beta": 0.07794239372015,
      "epsilon_dpo/beta_margin_grad_mean": -0.4526260495185852,
      "epsilon_dpo/beta_margin_grad_std": 0.10473916679620743,
      "epsilon_dpo/beta_margin_mean": 0.19932420551776886,
      "epsilon_dpo/beta_margin_std": 0.4362448453903198,
      "epsilon_dpo/loss_margin_mean": 2.601555824279785,
      "grad_norm": 17.814538955688477,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.07812992483377457,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.820919832540181e-07,
      "logits/chosen": -0.23502662777900696,
      "logits/rejected": -0.037265803664922714,
      "logps/chosen": -85.66082763671875,
      "logps/ref_chosen": -87.68214416503906,
      "logps/ref_rejected": -109.80048370361328,
      "logps/rejected": -110.38072967529297,
      "loss": 1.2433,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.15537020564079285,
      "rewards/margins": 0.19932423532009125,
      "rewards/rejected": -0.0439540296792984,
      "step": 140
    },
    {
      "epoch": 0.21315192743764172,
      "epsilon_dpo/beta": 0.07765059173107147,
      "epsilon_dpo/beta_margin_grad_mean": -0.45165613293647766,
      "epsilon_dpo/beta_margin_grad_std": 0.09902672469615936,
      "epsilon_dpo/beta_margin_mean": 0.20049114525318146,
      "epsilon_dpo/beta_margin_std": 0.4123653471469879,
      "epsilon_dpo/loss_margin_mean": 2.6210761070251465,
      "grad_norm": 15.114709854125977,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.07793508470058441,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.815973202802966e-07,
      "logits/chosen": -0.14430901408195496,
      "logits/rejected": 0.059034563601017,
      "logps/chosen": -81.68354797363281,
      "logps/ref_chosen": -83.4490966796875,
      "logps/ref_rejected": -108.92472839355469,
      "logps/rejected": -109.78024291992188,
      "loss": 1.2373,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.13589784502983093,
      "rewards/margins": 0.20049111545085907,
      "rewards/rejected": -0.06459328532218933,
      "step": 141
    },
    {
      "epoch": 0.2146636432350718,
      "epsilon_dpo/beta": 0.07750608026981354,
      "epsilon_dpo/beta_margin_grad_mean": -0.46795445680618286,
      "epsilon_dpo/beta_margin_grad_std": 0.09034660458564758,
      "epsilon_dpo/beta_margin_mean": 0.13596083223819733,
      "epsilon_dpo/beta_margin_std": 0.3836277425289154,
      "epsilon_dpo/loss_margin_mean": 1.7882254123687744,
      "grad_norm": 16.118061065673828,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.07764391601085663,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.810961790316729e-07,
      "logits/chosen": -0.17365016043186188,
      "logits/rejected": -0.21685616672039032,
      "logps/chosen": -88.09880065917969,
      "logps/ref_chosen": -89.39207458496094,
      "logps/ref_rejected": -102.29324340820312,
      "logps/rejected": -102.78819274902344,
      "loss": 1.2905,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.09935611486434937,
      "rewards/margins": 0.13596080243587494,
      "rewards/rejected": -0.03660469129681587,
      "step": 142
    },
    {
      "epoch": 0.2161753590325019,
      "epsilon_dpo/beta": 0.07738525420427322,
      "epsilon_dpo/beta_margin_grad_mean": -0.47416046261787415,
      "epsilon_dpo/beta_margin_grad_std": 0.10722901672124863,
      "epsilon_dpo/beta_margin_mean": 0.10877778381109238,
      "epsilon_dpo/beta_margin_std": 0.4481872320175171,
      "epsilon_dpo/loss_margin_mean": 1.4516710042953491,
      "grad_norm": 18.209449768066406,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.07749860733747482,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.805885735261454e-07,
      "logits/chosen": -0.11513273417949677,
      "logits/rejected": -0.07733216881752014,
      "logps/chosen": -85.3487777709961,
      "logps/ref_chosen": -86.82447814941406,
      "logps/ref_rejected": -95.78927612304688,
      "logps/rejected": -95.76524353027344,
      "loss": 1.3295,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": 0.11212005466222763,
      "rewards/margins": 0.10877779126167297,
      "rewards/rejected": 0.0033422596752643585,
      "step": 143
    },
    {
      "epoch": 0.21768707482993196,
      "epsilon_dpo/beta": 0.07736126333475113,
      "epsilon_dpo/beta_margin_grad_mean": -0.47966277599334717,
      "epsilon_dpo/beta_margin_grad_std": 0.11707232892513275,
      "epsilon_dpo/beta_margin_mean": 0.09194158017635345,
      "epsilon_dpo/beta_margin_std": 0.4984915256500244,
      "epsilon_dpo/loss_margin_mean": 1.2385411262512207,
      "grad_norm": 18.285436630249023,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.07737770676612854,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 4.800745179625307e-07,
      "logits/chosen": -0.055155664682388306,
      "logits/rejected": 0.14967641234397888,
      "logps/chosen": -83.60164642333984,
      "logps/ref_chosen": -84.393310546875,
      "logps/ref_rejected": -98.43213653564453,
      "logps/rejected": -98.87901306152344,
      "loss": 1.3565,
      "rewards/accuracies": 0.484375,
      "rewards/chosen": 0.05851306766271591,
      "rewards/margins": 0.09194157272577286,
      "rewards/rejected": -0.03342851251363754,
      "step": 144
    },
    {
      "epoch": 0.21919879062736206,
      "epsilon_dpo/beta": 0.07711951434612274,
      "epsilon_dpo/beta_margin_grad_mean": -0.46541881561279297,
      "epsilon_dpo/beta_margin_grad_std": 0.1151481494307518,
      "epsilon_dpo/beta_margin_mean": 0.142835795879364,
      "epsilon_dpo/beta_margin_std": 0.48386383056640625,
      "epsilon_dpo/loss_margin_mean": 1.9007998704910278,
      "grad_norm": 19.481618881225586,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.07735353708267212,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.795540267200686e-07,
      "logits/chosen": -0.24535146355628967,
      "logits/rejected": -0.1606999635696411,
      "logps/chosen": -103.21641540527344,
      "logps/ref_chosen": -104.59640502929688,
      "logps/ref_rejected": -104.37325286865234,
      "logps/rejected": -104.8940658569336,
      "loss": 1.3056,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.10378595441579819,
      "rewards/margins": 0.14283576607704163,
      "rewards/rejected": -0.03904980793595314,
      "step": 145
    },
    {
      "epoch": 0.22071050642479215,
      "epsilon_dpo/beta": 0.07699976861476898,
      "epsilon_dpo/beta_margin_grad_mean": -0.4759455621242523,
      "epsilon_dpo/beta_margin_grad_std": 0.09818226099014282,
      "epsilon_dpo/beta_margin_mean": 0.09853528439998627,
      "epsilon_dpo/beta_margin_std": 0.4120221436023712,
      "epsilon_dpo/loss_margin_mean": 1.321279764175415,
      "grad_norm": 16.349870681762695,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.07711255550384521,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.790271143580173e-07,
      "logits/chosen": -0.3498794436454773,
      "logits/rejected": -0.0407610684633255,
      "logps/chosen": -80.74744415283203,
      "logps/ref_chosen": -82.47695922851562,
      "logps/ref_rejected": -88.8701171875,
      "logps/rejected": -88.46188354492188,
      "loss": 1.3316,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.13155552744865417,
      "rewards/margins": 0.09853523969650269,
      "rewards/rejected": 0.033020272850990295,
      "step": 146
    },
    {
      "epoch": 0.2222222222222222,
      "epsilon_dpo/beta": 0.07683151960372925,
      "epsilon_dpo/beta_margin_grad_mean": -0.4728216528892517,
      "epsilon_dpo/beta_margin_grad_std": 0.09379173070192337,
      "epsilon_dpo/beta_margin_mean": 0.11323540657758713,
      "epsilon_dpo/beta_margin_std": 0.3893289864063263,
      "epsilon_dpo/loss_margin_mean": 1.5119967460632324,
      "grad_norm": 17.558822631835938,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.07699225842952728,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.784937956152489e-07,
      "logits/chosen": -0.10575242340564728,
      "logits/rejected": -0.1430339366197586,
      "logps/chosen": -89.42935943603516,
      "logps/ref_chosen": -90.40057373046875,
      "logps/ref_rejected": -104.02629089355469,
      "logps/rejected": -104.56707763671875,
      "loss": 1.3134,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": 0.07326846569776535,
      "rewards/margins": 0.11323542892932892,
      "rewards/rejected": -0.03996695578098297,
      "step": 147
    },
    {
      "epoch": 0.2237339380196523,
      "epsilon_dpo/beta": 0.07651975750923157,
      "epsilon_dpo/beta_margin_grad_mean": -0.44731733202934265,
      "epsilon_dpo/beta_margin_grad_std": 0.11125834286212921,
      "epsilon_dpo/beta_margin_mean": 0.22437097132205963,
      "epsilon_dpo/beta_margin_std": 0.4774388372898102,
      "epsilon_dpo/loss_margin_mean": 2.9725053310394287,
      "grad_norm": 14.532676696777344,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.07682420313358307,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.779540854098347e-07,
      "logits/chosen": -0.1696356236934662,
      "logits/rejected": -0.37039607763290405,
      "logps/chosen": -69.44158935546875,
      "logps/ref_chosen": -71.44349670410156,
      "logps/ref_rejected": -97.9690170288086,
      "logps/rejected": -98.93960571289062,
      "loss": 1.229,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.15155889093875885,
      "rewards/margins": 0.22437095642089844,
      "rewards/rejected": -0.07281208038330078,
      "step": 148
    },
    {
      "epoch": 0.2252456538170824,
      "epsilon_dpo/beta": 0.07632970809936523,
      "epsilon_dpo/beta_margin_grad_mean": -0.46323949098587036,
      "epsilon_dpo/beta_margin_grad_std": 0.09741820394992828,
      "epsilon_dpo/beta_margin_mean": 0.15397903323173523,
      "epsilon_dpo/beta_margin_std": 0.40842485427856445,
      "epsilon_dpo/loss_margin_mean": 2.057908773422241,
      "grad_norm": 16.44625473022461,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.0765133649110794,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.774079988386296e-07,
      "logits/chosen": -0.2737234830856323,
      "logits/rejected": -0.3755127191543579,
      "logps/chosen": -77.71635437011719,
      "logps/ref_chosen": -78.34549713134766,
      "logps/ref_rejected": -90.18255615234375,
      "logps/rejected": -91.611328125,
      "loss": 1.2788,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.04624457284808159,
      "rewards/margins": 0.15397900342941284,
      "rewards/rejected": -0.10773443430662155,
      "step": 149
    },
    {
      "epoch": 0.22675736961451248,
      "epsilon_dpo/beta": 0.07602010667324066,
      "epsilon_dpo/beta_margin_grad_mean": -0.43523797392845154,
      "epsilon_dpo/beta_margin_grad_std": 0.12384334951639175,
      "epsilon_dpo/beta_margin_mean": 0.27531003952026367,
      "epsilon_dpo/beta_margin_std": 0.5316606163978577,
      "epsilon_dpo/loss_margin_mean": 3.6705944538116455,
      "grad_norm": 15.97244930267334,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.07632256299257278,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.768555511768486e-07,
      "logits/chosen": -0.1420062780380249,
      "logits/rejected": 0.244110107421875,
      "logps/chosen": -91.69683837890625,
      "logps/ref_chosen": -92.33145904541016,
      "logps/ref_rejected": -113.8134765625,
      "logps/rejected": -116.84945678710938,
      "loss": 1.1975,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": 0.045774202793836594,
      "rewards/margins": 0.27531006932258606,
      "rewards/rejected": -0.22953587770462036,
      "step": 150
    },
    {
      "epoch": 0.22826908541194255,
      "epsilon_dpo/beta": 0.07576003670692444,
      "epsilon_dpo/beta_margin_grad_mean": -0.44170546531677246,
      "epsilon_dpo/beta_margin_grad_std": 0.11201171576976776,
      "epsilon_dpo/beta_margin_mean": 0.25496262311935425,
      "epsilon_dpo/beta_margin_std": 0.4993725121021271,
      "epsilon_dpo/loss_margin_mean": 3.4074244499206543,
      "grad_norm": 15.150566101074219,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.07601375877857208,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.762967578776406e-07,
      "logits/chosen": -0.12288770824670792,
      "logits/rejected": -0.15996623039245605,
      "logps/chosen": -72.63821411132812,
      "logps/ref_chosen": -74.43798828125,
      "logps/ref_rejected": -96.98008728027344,
      "logps/rejected": -98.58773803710938,
      "loss": 1.2056,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.13440963625907898,
      "rewards/margins": 0.25496259331703186,
      "rewards/rejected": -0.12055293470621109,
      "step": 151
    },
    {
      "epoch": 0.22978080120937264,
      "epsilon_dpo/beta": 0.07542947679758072,
      "epsilon_dpo/beta_margin_grad_mean": -0.4548986852169037,
      "epsilon_dpo/beta_margin_grad_std": 0.11138568818569183,
      "epsilon_dpo/beta_margin_mean": 0.1916198432445526,
      "epsilon_dpo/beta_margin_std": 0.4807240962982178,
      "epsilon_dpo/loss_margin_mean": 2.5821034908294678,
      "grad_norm": 17.452680587768555,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.075753353536129,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.757316345716553e-07,
      "logits/chosen": -0.05779954418540001,
      "logits/rejected": -0.016348371282219887,
      "logps/chosen": -77.57362365722656,
      "logps/ref_chosen": -77.98807525634766,
      "logps/ref_rejected": -103.52305603027344,
      "logps/rejected": -105.69070434570312,
      "loss": 1.259,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.02951541170477867,
      "rewards/margins": 0.1916198581457138,
      "rewards/rejected": -0.16210445761680603,
      "step": 152
    },
    {
      "epoch": 0.23129251700680273,
      "epsilon_dpo/beta": 0.07521876692771912,
      "epsilon_dpo/beta_margin_grad_mean": -0.4610231816768646,
      "epsilon_dpo/beta_margin_grad_std": 0.09761146456003189,
      "epsilon_dpo/beta_margin_mean": 0.16371819376945496,
      "epsilon_dpo/beta_margin_std": 0.4085807204246521,
      "epsilon_dpo/loss_margin_mean": 2.216705560684204,
      "grad_norm": 16.362926483154297,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.07542337477207184,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.751601970666064e-07,
      "logits/chosen": -0.28554776310920715,
      "logits/rejected": -0.3057416081428528,
      "logps/chosen": -96.85858154296875,
      "logps/ref_chosen": -95.72166442871094,
      "logps/ref_rejected": -100.04006958007812,
      "logps/rejected": -103.39369201660156,
      "loss": 1.2699,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.08693025261163712,
      "rewards/margins": 0.16371814906597137,
      "rewards/rejected": -0.2506484091281891,
      "step": 153
    },
    {
      "epoch": 0.2328042328042328,
      "epsilon_dpo/beta": 0.07503131777048111,
      "epsilon_dpo/beta_margin_grad_mean": -0.4663916826248169,
      "epsilon_dpo/beta_margin_grad_std": 0.11229141056537628,
      "epsilon_dpo/beta_margin_mean": 0.1404602825641632,
      "epsilon_dpo/beta_margin_std": 0.47098055481910706,
      "epsilon_dpo/loss_margin_mean": 1.9198356866836548,
      "grad_norm": 16.172679901123047,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.07521184533834457,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.745824613468292e-07,
      "logits/chosen": -0.3652782440185547,
      "logits/rejected": -0.49543774127960205,
      "logps/chosen": -82.0560302734375,
      "logps/ref_chosen": -82.50230407714844,
      "logps/ref_rejected": -82.01812744140625,
      "logps/rejected": -83.49168395996094,
      "loss": 1.3048,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.03172617405653,
      "rewards/margins": 0.14046026766300201,
      "rewards/rejected": -0.10873409360647202,
      "step": 154
    },
    {
      "epoch": 0.23431594860166288,
      "epsilon_dpo/beta": 0.0747738629579544,
      "epsilon_dpo/beta_margin_grad_mean": -0.4513029158115387,
      "epsilon_dpo/beta_margin_grad_std": 0.11095915734767914,
      "epsilon_dpo/beta_margin_mean": 0.20482796430587769,
      "epsilon_dpo/beta_margin_std": 0.47065576910972595,
      "epsilon_dpo/loss_margin_mean": 2.7832815647125244,
      "grad_norm": 16.804861068725586,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.07502428442239761,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.7399844357283393e-07,
      "logits/chosen": -0.12051165103912354,
      "logits/rejected": -0.13051287829875946,
      "logps/chosen": -90.16606903076172,
      "logps/ref_chosen": -89.09580993652344,
      "logps/ref_rejected": -101.94664001464844,
      "logps/rejected": -105.80018615722656,
      "loss": 1.2454,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.08176211267709732,
      "rewards/margins": 0.20482799410820007,
      "rewards/rejected": -0.2865900993347168,
      "step": 155
    },
    {
      "epoch": 0.23582766439909297,
      "epsilon_dpo/beta": 0.07442424446344376,
      "epsilon_dpo/beta_margin_grad_mean": -0.42783400416374207,
      "epsilon_dpo/beta_margin_grad_std": 0.10387593507766724,
      "epsilon_dpo/beta_margin_mean": 0.3065878450870514,
      "epsilon_dpo/beta_margin_std": 0.44690340757369995,
      "epsilon_dpo/loss_margin_mean": 4.155799865722656,
      "grad_norm": 17.409835815429688,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.07476726919412613,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.7340816008085305e-07,
      "logits/chosen": -0.25020796060562134,
      "logits/rejected": 0.00035034865140914917,
      "logps/chosen": -95.34709930419922,
      "logps/ref_chosen": -94.36488342285156,
      "logps/ref_rejected": -119.41386413574219,
      "logps/rejected": -124.5518798828125,
      "loss": 1.1506,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.0749220922589302,
      "rewards/margins": 0.306587815284729,
      "rewards/rejected": -0.3815099000930786,
      "step": 156
    },
    {
      "epoch": 0.23733938019652306,
      "epsilon_dpo/beta": 0.07421655207872391,
      "epsilon_dpo/beta_margin_grad_mean": -0.4635535776615143,
      "epsilon_dpo/beta_margin_grad_std": 0.11218099296092987,
      "epsilon_dpo/beta_margin_mean": 0.15487989783287048,
      "epsilon_dpo/beta_margin_std": 0.4757290482521057,
      "epsilon_dpo/loss_margin_mean": 2.131009817123413,
      "grad_norm": 14.102409362792969,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.07441843301057816,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.728116273823847e-07,
      "logits/chosen": -0.44304779171943665,
      "logits/rejected": -0.15374836325645447,
      "logps/chosen": -75.98442077636719,
      "logps/ref_chosen": -75.79052734375,
      "logps/ref_rejected": -84.51288604736328,
      "logps/rejected": -86.83779907226562,
      "loss": 1.2921,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.016395092010498047,
      "rewards/margins": 0.15487989783287048,
      "rewards/rejected": -0.17127500474452972,
      "step": 157
    },
    {
      "epoch": 0.23885109599395313,
      "epsilon_dpo/beta": 0.07407798618078232,
      "epsilon_dpo/beta_margin_grad_mean": -0.46485766768455505,
      "epsilon_dpo/beta_margin_grad_std": 0.1275767832994461,
      "epsilon_dpo/beta_margin_mean": 0.15245291590690613,
      "epsilon_dpo/beta_margin_std": 0.5510936379432678,
      "epsilon_dpo/loss_margin_mean": 2.1135518550872803,
      "grad_norm": 15.723213195800781,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.07420971989631653,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.7220886216373085e-07,
      "logits/chosen": -0.3810843527317047,
      "logits/rejected": -0.2918296754360199,
      "logps/chosen": -87.435302734375,
      "logps/ref_chosen": -87.19711303710938,
      "logps/ref_rejected": -104.29627227783203,
      "logps/rejected": -106.64801025390625,
      "loss": 1.3123,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.0198702123016119,
      "rewards/margins": 0.1524529755115509,
      "rewards/rejected": -0.17232318222522736,
      "step": 158
    },
    {
      "epoch": 0.24036281179138322,
      "epsilon_dpo/beta": 0.07382359355688095,
      "epsilon_dpo/beta_margin_grad_mean": -0.4507734477519989,
      "epsilon_dpo/beta_margin_grad_std": 0.1107042208313942,
      "epsilon_dpo/beta_margin_mean": 0.20948928594589233,
      "epsilon_dpo/beta_margin_std": 0.4829138219356537,
      "epsilon_dpo/loss_margin_mean": 2.880485773086548,
      "grad_norm": 15.230273246765137,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.07407083362340927,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.715998812855304e-07,
      "logits/chosen": -0.2649644911289215,
      "logits/rejected": -0.09603996574878693,
      "logps/chosen": -78.42363739013672,
      "logps/ref_chosen": -78.20480346679688,
      "logps/ref_rejected": -95.99697875976562,
      "logps/rejected": -99.09629821777344,
      "loss": 1.2431,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.017694566398859024,
      "rewards/margins": 0.20948931574821472,
      "rewards/rejected": -0.22718387842178345,
      "step": 159
    },
    {
      "epoch": 0.2418745275888133,
      "epsilon_dpo/beta": 0.07359376549720764,
      "epsilon_dpo/beta_margin_grad_mean": -0.4621080160140991,
      "epsilon_dpo/beta_margin_grad_std": 0.10441814363002777,
      "epsilon_dpo/beta_margin_mean": 0.16097773611545563,
      "epsilon_dpo/beta_margin_std": 0.44186532497406006,
      "epsilon_dpo/loss_margin_mean": 2.2304911613464355,
      "grad_norm": 14.367936134338379,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.07381708920001984,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.7098470178228755e-07,
      "logits/chosen": -0.6084429621696472,
      "logits/rejected": -0.48294705152511597,
      "logps/chosen": -68.7177505493164,
      "logps/ref_chosen": -68.0002670288086,
      "logps/ref_rejected": -89.75141143798828,
      "logps/rejected": -92.69938659667969,
      "loss": 1.279,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.054394595324993134,
      "rewards/margins": 0.16097773611545563,
      "rewards/rejected": -0.21537232398986816,
      "step": 160
    },
    {
      "epoch": 0.24338624338624337,
      "epsilon_dpo/beta": 0.07329551130533218,
      "epsilon_dpo/beta_margin_grad_mean": -0.45018425583839417,
      "epsilon_dpo/beta_margin_grad_std": 0.1030559092760086,
      "epsilon_dpo/beta_margin_mean": 0.20355476438999176,
      "epsilon_dpo/beta_margin_std": 0.4396621882915497,
      "epsilon_dpo/loss_margin_mean": 2.8164150714874268,
      "grad_norm": 14.061203002929688,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.07358712702989578,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.703633408618955e-07,
      "logits/chosen": -0.23171710968017578,
      "logits/rejected": -0.12584610283374786,
      "logps/chosen": -75.62057495117188,
      "logps/ref_chosen": -75.69575500488281,
      "logps/ref_rejected": -87.35690307617188,
      "logps/rejected": -90.09814453125,
      "loss": 1.2397,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": 0.0038347430527210236,
      "rewards/margins": 0.20355473458766937,
      "rewards/rejected": -0.19971999526023865,
      "step": 161
    },
    {
      "epoch": 0.24489795918367346,
      "epsilon_dpo/beta": 0.07290733605623245,
      "epsilon_dpo/beta_margin_grad_mean": -0.42896273732185364,
      "epsilon_dpo/beta_margin_grad_std": 0.09937798976898193,
      "epsilon_dpo/beta_margin_mean": 0.2959074378013611,
      "epsilon_dpo/beta_margin_std": 0.4288792908191681,
      "epsilon_dpo/loss_margin_mean": 4.093425273895264,
      "grad_norm": 16.56731414794922,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.07328939437866211,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 4.697358159051549e-07,
      "logits/chosen": -0.42089033126831055,
      "logits/rejected": -0.21701693534851074,
      "logps/chosen": -98.07493591308594,
      "logps/ref_chosen": -97.45020294189453,
      "logps/ref_rejected": -125.35346984863281,
      "logps/rejected": -130.07164001464844,
      "loss": 1.156,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.04721798747777939,
      "rewards/margins": 0.2959074378013611,
      "rewards/rejected": -0.34312543272972107,
      "step": 162
    },
    {
      "epoch": 0.24640967498110355,
      "epsilon_dpo/beta": 0.0726131945848465,
      "epsilon_dpo/beta_margin_grad_mean": -0.4502231180667877,
      "epsilon_dpo/beta_margin_grad_std": 0.11270121484994888,
      "epsilon_dpo/beta_margin_mean": 0.209677055478096,
      "epsilon_dpo/beta_margin_std": 0.47489598393440247,
      "epsilon_dpo/loss_margin_mean": 2.9341773986816406,
      "grad_norm": 14.430964469909668,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.07290209829807281,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.691021444652876e-07,
      "logits/chosen": -0.2879790663719177,
      "logits/rejected": -0.12870730459690094,
      "logps/chosen": -73.21658325195312,
      "logps/ref_chosen": -73.41002655029297,
      "logps/ref_rejected": -95.94908905029297,
      "logps/rejected": -98.68983459472656,
      "loss": 1.2422,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.012572411447763443,
      "rewards/margins": 0.2096770703792572,
      "rewards/rejected": -0.19710463285446167,
      "step": 163
    },
    {
      "epoch": 0.24792139077853365,
      "epsilon_dpo/beta": 0.072319395840168,
      "epsilon_dpo/beta_margin_grad_mean": -0.4320308566093445,
      "epsilon_dpo/beta_margin_grad_std": 0.12061776965856552,
      "epsilon_dpo/beta_margin_mean": 0.2894424796104431,
      "epsilon_dpo/beta_margin_std": 0.5126076936721802,
      "epsilon_dpo/loss_margin_mean": 4.052041530609131,
      "grad_norm": 15.295539855957031,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.0726071298122406,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.6846234426744624e-07,
      "logits/chosen": -0.36896005272865295,
      "logits/rejected": -0.37494707107543945,
      "logps/chosen": -77.27589416503906,
      "logps/ref_chosen": -77.78909301757812,
      "logps/ref_rejected": -102.98542785644531,
      "logps/rejected": -106.52427673339844,
      "loss": 1.1808,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.034895047545433044,
      "rewards/margins": 0.2894425392150879,
      "rewards/rejected": -0.25454750657081604,
      "step": 164
    },
    {
      "epoch": 0.2494331065759637,
      "epsilon_dpo/beta": 0.071958988904953,
      "epsilon_dpo/beta_margin_grad_mean": -0.44002482295036316,
      "epsilon_dpo/beta_margin_grad_std": 0.1064891368150711,
      "epsilon_dpo/beta_margin_mean": 0.2548327147960663,
      "epsilon_dpo/beta_margin_std": 0.4599698781967163,
      "epsilon_dpo/loss_margin_mean": 3.5790834426879883,
      "grad_norm": 14.637746810913086,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.07231336086988449,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.678164332082175e-07,
      "logits/chosen": -0.16319644451141357,
      "logits/rejected": -0.32601630687713623,
      "logps/chosen": -72.39887237548828,
      "logps/ref_chosen": -72.82852935791016,
      "logps/ref_rejected": -91.37947845458984,
      "logps/rejected": -94.52890014648438,
      "loss": 1.198,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.029420459643006325,
      "rewards/margins": 0.2548326849937439,
      "rewards/rejected": -0.22541223466396332,
      "step": 165
    },
    {
      "epoch": 0.2509448223733938,
      "epsilon_dpo/beta": 0.07180337607860565,
      "epsilon_dpo/beta_margin_grad_mean": -0.45170384645462036,
      "epsilon_dpo/beta_margin_grad_std": 0.10957895219326019,
      "epsilon_dpo/beta_margin_mean": 0.20550940930843353,
      "epsilon_dpo/beta_margin_std": 0.46381574869155884,
      "epsilon_dpo/loss_margin_mean": 2.909606456756592,
      "grad_norm": 16.492395401000977,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.07195359468460083,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.6716442935512214e-07,
      "logits/chosen": -0.33589503169059753,
      "logits/rejected": -0.10567793250083923,
      "logps/chosen": -88.24549865722656,
      "logps/ref_chosen": -87.35054016113281,
      "logps/ref_rejected": -120.07548522949219,
      "logps/rejected": -123.88005065917969,
      "loss": 1.2433,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.06544095277786255,
      "rewards/margins": 0.20550942420959473,
      "rewards/rejected": -0.2709503769874573,
      "step": 166
    },
    {
      "epoch": 0.25245653817082386,
      "epsilon_dpo/beta": 0.07151201367378235,
      "epsilon_dpo/beta_margin_grad_mean": -0.4427506625652313,
      "epsilon_dpo/beta_margin_grad_std": 0.10156535357236862,
      "epsilon_dpo/beta_margin_mean": 0.24112164974212646,
      "epsilon_dpo/beta_margin_std": 0.43905171751976013,
      "epsilon_dpo/loss_margin_mean": 3.4096128940582275,
      "grad_norm": 14.830485343933105,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.07179653644561768,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.6650635094610966e-07,
      "logits/chosen": -0.08851295709609985,
      "logits/rejected": -0.012575190514326096,
      "logps/chosen": -80.8412857055664,
      "logps/ref_chosen": -81.61276245117188,
      "logps/ref_rejected": -98.826171875,
      "logps/rejected": -101.46430969238281,
      "loss": 1.2056,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.05437930300831795,
      "rewards/margins": 0.24112167954444885,
      "rewards/rejected": -0.18674236536026,
      "step": 167
    },
    {
      "epoch": 0.25396825396825395,
      "epsilon_dpo/beta": 0.07133441418409348,
      "epsilon_dpo/beta_margin_grad_mean": -0.4560203552246094,
      "epsilon_dpo/beta_margin_grad_std": 0.10047944635152817,
      "epsilon_dpo/beta_margin_mean": 0.18437211215496063,
      "epsilon_dpo/beta_margin_std": 0.4224920868873596,
      "epsilon_dpo/loss_margin_mean": 2.6299386024475098,
      "grad_norm": 15.255393028259277,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.07150604575872421,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.6584221638904767e-07,
      "logits/chosen": -0.13477931916713715,
      "logits/rejected": -0.16727593541145325,
      "logps/chosen": -88.02157592773438,
      "logps/ref_chosen": -88.23593139648438,
      "logps/ref_rejected": -101.98030090332031,
      "logps/rejected": -104.3958740234375,
      "loss": 1.2537,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.013690168038010597,
      "rewards/margins": 0.18437206745147705,
      "rewards/rejected": -0.1706819087266922,
      "step": 168
    },
    {
      "epoch": 0.25547996976568405,
      "epsilon_dpo/beta": 0.07113422453403473,
      "epsilon_dpo/beta_margin_grad_mean": -0.43899574875831604,
      "epsilon_dpo/beta_margin_grad_std": 0.13024921715259552,
      "epsilon_dpo/beta_margin_mean": 0.2600567936897278,
      "epsilon_dpo/beta_margin_std": 0.559776246547699,
      "epsilon_dpo/loss_margin_mean": 3.7151412963867188,
      "grad_norm": 17.136293411254883,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.07132772356271744,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.651720442612075e-07,
      "logits/chosen": -0.38109487295150757,
      "logits/rejected": -0.07364605367183685,
      "logps/chosen": -82.69183349609375,
      "logps/ref_chosen": -84.35769653320312,
      "logps/ref_rejected": -102.46897888183594,
      "logps/rejected": -104.51824951171875,
      "loss": 1.2181,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.11652705073356628,
      "rewards/margins": 0.2600567936897278,
      "rewards/rejected": -0.1435297578573227,
      "step": 169
    },
    {
      "epoch": 0.25699168556311414,
      "epsilon_dpo/beta": 0.07097917795181274,
      "epsilon_dpo/beta_margin_grad_mean": -0.46621790528297424,
      "epsilon_dpo/beta_margin_grad_std": 0.1452290415763855,
      "epsilon_dpo/beta_margin_mean": 0.14188018441200256,
      "epsilon_dpo/beta_margin_std": 0.6213703155517578,
      "epsilon_dpo/loss_margin_mean": 2.0687479972839355,
      "grad_norm": 15.189737319946289,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.07112767547369003,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.6449585330874425e-07,
      "logits/chosen": -0.35476934909820557,
      "logits/rejected": -0.16348260641098022,
      "logps/chosen": -76.20692443847656,
      "logps/ref_chosen": -77.82612609863281,
      "logps/ref_rejected": -79.933349609375,
      "logps/rejected": -80.38288879394531,
      "loss": 1.3426,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.11196792125701904,
      "rewards/margins": 0.14188018441200256,
      "rewards/rejected": -0.029912270605564117,
      "step": 170
    },
    {
      "epoch": 0.2585034013605442,
      "epsilon_dpo/beta": 0.0707133561372757,
      "epsilon_dpo/beta_margin_grad_mean": -0.44336846470832825,
      "epsilon_dpo/beta_margin_grad_std": 0.1324419528245926,
      "epsilon_dpo/beta_margin_mean": 0.23818716406822205,
      "epsilon_dpo/beta_margin_std": 0.5732473731040955,
      "epsilon_dpo/loss_margin_mean": 3.4258179664611816,
      "grad_norm": 14.782027244567871,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.07097242772579193,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.6381366244617224e-07,
      "logits/chosen": -0.18573148548603058,
      "logits/rejected": -0.20463137328624725,
      "logps/chosen": -83.67991638183594,
      "logps/ref_chosen": -85.51094055175781,
      "logps/ref_rejected": -104.4660873413086,
      "logps/rejected": -106.06088256835938,
      "loss": 1.2408,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.12760449945926666,
      "rewards/margins": 0.23818716406822205,
      "rewards/rejected": -0.11058266460895538,
      "step": 171
    },
    {
      "epoch": 0.2600151171579743,
      "epsilon_dpo/beta": 0.07049336284399033,
      "epsilon_dpo/beta_margin_grad_mean": -0.43517836928367615,
      "epsilon_dpo/beta_margin_grad_std": 0.11793834716081619,
      "epsilon_dpo/beta_margin_mean": 0.28078243136405945,
      "epsilon_dpo/beta_margin_std": 0.5101608037948608,
      "epsilon_dpo/loss_margin_mean": 4.030537128448486,
      "grad_norm": 14.178646087646484,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.0707072764635086,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.631254907558365e-07,
      "logits/chosen": -0.385815292596817,
      "logits/rejected": -0.11347803473472595,
      "logps/chosen": -72.62882995605469,
      "logps/ref_chosen": -74.67696380615234,
      "logps/ref_rejected": -106.31913757324219,
      "logps/rejected": -108.30152893066406,
      "loss": 1.1869,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.14252778887748718,
      "rewards/margins": 0.28078240156173706,
      "rewards/rejected": -0.13825464248657227,
      "step": 172
    },
    {
      "epoch": 0.2615268329554044,
      "epsilon_dpo/beta": 0.07025172561407089,
      "epsilon_dpo/beta_margin_grad_mean": -0.4308141767978668,
      "epsilon_dpo/beta_margin_grad_std": 0.14216464757919312,
      "epsilon_dpo/beta_margin_mean": 0.3127928078174591,
      "epsilon_dpo/beta_margin_std": 0.6453730463981628,
      "epsilon_dpo/loss_margin_mean": 4.508734703063965,
      "grad_norm": 14.488515853881836,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.07048700004816055,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.624313574873786e-07,
      "logits/chosen": -0.12725037336349487,
      "logits/rejected": -0.3742305636405945,
      "logps/chosen": -74.35868072509766,
      "logps/ref_chosen": -77.24813842773438,
      "logps/ref_rejected": -101.34745025634766,
      "logps/rejected": -102.96672058105469,
      "loss": 1.1937,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.20155192911624908,
      "rewards/margins": 0.31279274821281433,
      "rewards/rejected": -0.11124081909656525,
      "step": 173
    },
    {
      "epoch": 0.26303854875283444,
      "epsilon_dpo/beta": 0.06990130245685577,
      "epsilon_dpo/beta_margin_grad_mean": -0.4227776527404785,
      "epsilon_dpo/beta_margin_grad_std": 0.11874371767044067,
      "epsilon_dpo/beta_margin_mean": 0.3279976546764374,
      "epsilon_dpo/beta_margin_std": 0.512088418006897,
      "epsilon_dpo/loss_margin_mean": 4.736902713775635,
      "grad_norm": 15.51446533203125,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.07024553418159485,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.61731282057198e-07,
      "logits/chosen": -0.1422739326953888,
      "logits/rejected": -0.12538309395313263,
      "logps/chosen": -80.69758605957031,
      "logps/ref_chosen": -83.51936340332031,
      "logps/ref_rejected": -115.21611785888672,
      "logps/rejected": -117.13125610351562,
      "loss": 1.1474,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": 0.19557899236679077,
      "rewards/margins": 0.3279976546764374,
      "rewards/rejected": -0.1324186623096466,
      "step": 174
    },
    {
      "epoch": 0.26455026455026454,
      "epsilon_dpo/beta": 0.0696190595626831,
      "epsilon_dpo/beta_margin_grad_mean": -0.434917688369751,
      "epsilon_dpo/beta_margin_grad_std": 0.13931988179683685,
      "epsilon_dpo/beta_margin_mean": 0.28064674139022827,
      "epsilon_dpo/beta_margin_std": 0.6153140664100647,
      "epsilon_dpo/loss_margin_mean": 4.090598106384277,
      "grad_norm": 15.483464241027832,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.06989604979753494,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.6102528404790965e-07,
      "logits/chosen": -0.2617759704589844,
      "logits/rejected": -0.09605693817138672,
      "logps/chosen": -89.11908721923828,
      "logps/ref_chosen": -91.16883087158203,
      "logps/ref_rejected": -111.66940307617188,
      "logps/rejected": -113.71025085449219,
      "loss": 1.2145,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.14015761017799377,
      "rewards/margins": 0.28064674139022827,
      "rewards/rejected": -0.1404891163110733,
      "step": 175
    },
    {
      "epoch": 0.2660619803476946,
      "epsilon_dpo/beta": 0.06951142847537994,
      "epsilon_dpo/beta_margin_grad_mean": -0.4651341736316681,
      "epsilon_dpo/beta_margin_grad_std": 0.12862923741340637,
      "epsilon_dpo/beta_margin_mean": 0.15457409620285034,
      "epsilon_dpo/beta_margin_std": 0.5627329349517822,
      "epsilon_dpo/loss_margin_mean": 2.2834548950195312,
      "grad_norm": 19.406566619873047,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.0696132481098175,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.603133832077953e-07,
      "logits/chosen": -0.01893380470573902,
      "logits/rejected": -0.12899591028690338,
      "logps/chosen": -104.9249038696289,
      "logps/ref_chosen": -105.64872741699219,
      "logps/ref_rejected": -114.00184631347656,
      "logps/rejected": -115.56147766113281,
      "loss": 1.3128,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.04813181981444359,
      "rewards/margins": 0.15457406640052795,
      "rewards/rejected": -0.10644224286079407,
      "step": 176
    },
    {
      "epoch": 0.2675736961451247,
      "epsilon_dpo/beta": 0.0691530704498291,
      "epsilon_dpo/beta_margin_grad_mean": -0.3918303847312927,
      "epsilon_dpo/beta_margin_grad_std": 0.128380686044693,
      "epsilon_dpo/beta_margin_mean": 0.47757646441459656,
      "epsilon_dpo/beta_margin_std": 0.5835011601448059,
      "epsilon_dpo/loss_margin_mean": 6.953527450561523,
      "grad_norm": 16.948251724243164,
      "kl/avg_steps": 0.515625,
      "kl/beta": 0.06950464844703674,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.5959559945025183e-07,
      "logits/chosen": -0.1951223760843277,
      "logits/rejected": -0.2172054648399353,
      "logps/chosen": -79.67855834960938,
      "logps/ref_chosen": -83.90325927734375,
      "logps/ref_rejected": -124.04662322998047,
      "logps/rejected": -126.77544403076172,
      "loss": 1.0424,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": 0.29099011421203613,
      "rewards/margins": 0.4775765538215637,
      "rewards/rejected": -0.1865864247083664,
      "step": 177
    },
    {
      "epoch": 0.2690854119425548,
      "epsilon_dpo/beta": 0.06880924105644226,
      "epsilon_dpo/beta_margin_grad_mean": -0.4229211211204529,
      "epsilon_dpo/beta_margin_grad_std": 0.1171141117811203,
      "epsilon_dpo/beta_margin_mean": 0.3316393792629242,
      "epsilon_dpo/beta_margin_std": 0.5047381520271301,
      "epsilon_dpo/loss_margin_mean": 4.8638386726379395,
      "grad_norm": 14.013132095336914,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.06914810091257095,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.588719528532341e-07,
      "logits/chosen": -0.4607001841068268,
      "logits/rejected": -0.426104336977005,
      "logps/chosen": -82.26956176757812,
      "logps/ref_chosen": -85.39146423339844,
      "logps/ref_rejected": -104.77743530273438,
      "logps/rejected": -106.51936340332031,
      "loss": 1.1424,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.2138364315032959,
      "rewards/margins": 0.3316393494606018,
      "rewards/rejected": -0.1178029254078865,
      "step": 178
    },
    {
      "epoch": 0.2705971277399849,
      "epsilon_dpo/beta": 0.06862808018922806,
      "epsilon_dpo/beta_margin_grad_mean": -0.4533432424068451,
      "epsilon_dpo/beta_margin_grad_std": 0.13178221881389618,
      "epsilon_dpo/beta_margin_mean": 0.20900557935237885,
      "epsilon_dpo/beta_margin_std": 0.5843459367752075,
      "epsilon_dpo/loss_margin_mean": 3.102468729019165,
      "grad_norm": 15.235759735107422,
      "kl/avg_steps": 0.265625,
      "kl/beta": 0.06880408525466919,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.581424636586928e-07,
      "logits/chosen": -0.32951170206069946,
      "logits/rejected": -0.2942412495613098,
      "logps/chosen": -86.45561218261719,
      "logps/ref_chosen": -89.62193298339844,
      "logps/ref_rejected": -100.42626190185547,
      "logps/rejected": -100.36241149902344,
      "loss": 1.2684,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.21552518010139465,
      "rewards/margins": 0.20900559425354004,
      "rewards/rejected": 0.00651959702372551,
      "step": 179
    },
    {
      "epoch": 0.272108843537415,
      "epsilon_dpo/beta": 0.06852144002914429,
      "epsilon_dpo/beta_margin_grad_mean": -0.45583203434944153,
      "epsilon_dpo/beta_margin_grad_std": 0.12107642740011215,
      "epsilon_dpo/beta_margin_mean": 0.1924244910478592,
      "epsilon_dpo/beta_margin_std": 0.5202975869178772,
      "epsilon_dpo/loss_margin_mean": 2.8651440143585205,
      "grad_norm": 13.99289321899414,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.06862180680036545,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.5740715227200897e-07,
      "logits/chosen": -0.39849674701690674,
      "logits/rejected": -0.4529162645339966,
      "logps/chosen": -77.17005920410156,
      "logps/ref_chosen": -80.65196228027344,
      "logps/ref_rejected": -86.28292083740234,
      "logps/rejected": -85.66616821289062,
      "loss": 1.2679,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": 0.2367887794971466,
      "rewards/margins": 0.19242452085018158,
      "rewards/rejected": 0.044364262372255325,
      "step": 180
    },
    {
      "epoch": 0.273620559334845,
      "epsilon_dpo/beta": 0.06830747425556183,
      "epsilon_dpo/beta_margin_grad_mean": -0.43471357226371765,
      "epsilon_dpo/beta_margin_grad_std": 0.12012200802564621,
      "epsilon_dpo/beta_margin_mean": 0.28556621074676514,
      "epsilon_dpo/beta_margin_std": 0.5327972173690796,
      "epsilon_dpo/loss_margin_mean": 4.225931644439697,
      "grad_norm": 13.607747077941895,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.06851474940776825,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.566660392614228e-07,
      "logits/chosen": -0.5008969902992249,
      "logits/rejected": -0.48943305015563965,
      "logps/chosen": -79.5925521850586,
      "logps/ref_chosen": -83.57610321044922,
      "logps/ref_rejected": -103.83201599121094,
      "logps/rejected": -104.07440185546875,
      "loss": 1.1874,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.2721977233886719,
      "rewards/margins": 0.28556621074676514,
      "rewards/rejected": -0.0133685152977705,
      "step": 181
    },
    {
      "epoch": 0.2751322751322751,
      "epsilon_dpo/beta": 0.06791312992572784,
      "epsilon_dpo/beta_margin_grad_mean": -0.40505993366241455,
      "epsilon_dpo/beta_margin_grad_std": 0.126272052526474,
      "epsilon_dpo/beta_margin_mean": 0.410185843706131,
      "epsilon_dpo/beta_margin_std": 0.5547764897346497,
      "epsilon_dpo/loss_margin_mean": 6.082636833190918,
      "grad_norm": 14.72663688659668,
      "kl/avg_steps": 0.578125,
      "kl/beta": 0.06830131262540817,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 4.5591914535745817e-07,
      "logits/chosen": -0.30189305543899536,
      "logits/rejected": -0.24266424775123596,
      "logps/chosen": -77.49993896484375,
      "logps/ref_chosen": -80.32312774658203,
      "logps/ref_rejected": -116.65972900390625,
      "logps/rejected": -119.919189453125,
      "loss": 1.0898,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": 0.1906897872686386,
      "rewards/margins": 0.4101858139038086,
      "rewards/rejected": -0.21949604153633118,
      "step": 182
    },
    {
      "epoch": 0.2766439909297052,
      "epsilon_dpo/beta": 0.06780938804149628,
      "epsilon_dpo/beta_margin_grad_mean": -0.45864781737327576,
      "epsilon_dpo/beta_margin_grad_std": 0.1380331814289093,
      "epsilon_dpo/beta_margin_mean": 0.18381942808628082,
      "epsilon_dpo/beta_margin_std": 0.6125727295875549,
      "epsilon_dpo/loss_margin_mean": 2.769742727279663,
      "grad_norm": 15.092416763305664,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.06790871173143387,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.551664914523433e-07,
      "logits/chosen": -0.32394590973854065,
      "logits/rejected": -0.3546528220176697,
      "logps/chosen": -86.47184753417969,
      "logps/ref_chosen": -86.9107437133789,
      "logps/ref_rejected": -97.24227142333984,
      "logps/rejected": -99.5731201171875,
      "loss": 1.2994,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.028311219066381454,
      "rewards/margins": 0.18381944298744202,
      "rewards/rejected": -0.15550823509693146,
      "step": 183
    },
    {
      "epoch": 0.2781557067271353,
      "epsilon_dpo/beta": 0.06756575405597687,
      "epsilon_dpo/beta_margin_grad_mean": -0.4294240176677704,
      "epsilon_dpo/beta_margin_grad_std": 0.10901588201522827,
      "epsilon_dpo/beta_margin_mean": 0.30280619859695435,
      "epsilon_dpo/beta_margin_std": 0.47133877873420715,
      "epsilon_dpo/loss_margin_mean": 4.52435302734375,
      "grad_norm": 13.005136489868164,
      "kl/avg_steps": 0.359375,
      "kl/beta": 0.06780277192592621,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.544080985994258e-07,
      "logits/chosen": -0.3298104405403137,
      "logits/rejected": -0.39565029740333557,
      "logps/chosen": -62.69512176513672,
      "logps/ref_chosen": -66.81837463378906,
      "logps/ref_rejected": -89.96328735351562,
      "logps/rejected": -90.36438751220703,
      "loss": 1.159,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.2780967354774475,
      "rewards/margins": 0.30280619859695435,
      "rewards/rejected": -0.02470945566892624,
      "step": 184
    },
    {
      "epoch": 0.2796674225245654,
      "epsilon_dpo/beta": 0.06735558807849884,
      "epsilon_dpo/beta_margin_grad_mean": -0.43276891112327576,
      "epsilon_dpo/beta_margin_grad_std": 0.15843592584133148,
      "epsilon_dpo/beta_margin_mean": 0.29344916343688965,
      "epsilon_dpo/beta_margin_std": 0.6960242986679077,
      "epsilon_dpo/loss_margin_mean": 4.434981822967529,
      "grad_norm": 13.516302108764648,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.06755997985601425,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.5364398801258394e-07,
      "logits/chosen": -0.15685924887657166,
      "logits/rejected": -0.12343727052211761,
      "logps/chosen": -69.97865295410156,
      "logps/ref_chosen": -72.42000579833984,
      "logps/ref_rejected": -92.72299194335938,
      "logps/rejected": -94.71661376953125,
      "loss": 1.2288,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.16207338869571686,
      "rewards/margins": 0.29344916343688965,
      "rewards/rejected": -0.13137578964233398,
      "step": 185
    },
    {
      "epoch": 0.2811791383219955,
      "epsilon_dpo/beta": 0.06727205216884613,
      "epsilon_dpo/beta_margin_grad_mean": -0.43999987840652466,
      "epsilon_dpo/beta_margin_grad_std": 0.1602521687746048,
      "epsilon_dpo/beta_margin_mean": 0.27616897225379944,
      "epsilon_dpo/beta_margin_std": 0.7218940258026123,
      "epsilon_dpo/loss_margin_mean": 4.186398506164551,
      "grad_norm": 15.226896286010742,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.06734950840473175,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 4.5287418106563354e-07,
      "logits/chosen": -0.2841332256793976,
      "logits/rejected": -0.4217131435871124,
      "logps/chosen": -90.3564453125,
      "logps/ref_chosen": -92.51256561279297,
      "logps/ref_rejected": -108.71427917480469,
      "logps/rejected": -110.74455261230469,
      "loss": 1.2501,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.14296837151050568,
      "rewards/margins": 0.2761688828468323,
      "rewards/rejected": -0.1332004964351654,
      "step": 186
    },
    {
      "epoch": 0.28269085411942557,
      "epsilon_dpo/beta": 0.06704090535640717,
      "epsilon_dpo/beta_margin_grad_mean": -0.43871721625328064,
      "epsilon_dpo/beta_margin_grad_std": 0.14905036985874176,
      "epsilon_dpo/beta_margin_mean": 0.2713906466960907,
      "epsilon_dpo/beta_margin_std": 0.6602548956871033,
      "epsilon_dpo/loss_margin_mean": 4.1168060302734375,
      "grad_norm": 18.4125919342041,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06726542860269547,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.520986992917297e-07,
      "logits/chosen": -0.38112810254096985,
      "logits/rejected": -0.31438401341438293,
      "logps/chosen": -102.13521575927734,
      "logps/ref_chosen": -102.16010284423828,
      "logps/ref_rejected": -131.34176635742188,
      "logps/rejected": -135.43368530273438,
      "loss": 1.2356,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.0005250517278909683,
      "rewards/margins": 0.2713906168937683,
      "rewards/rejected": -0.27191564440727234,
      "step": 187
    },
    {
      "epoch": 0.2842025699168556,
      "epsilon_dpo/beta": 0.06681124120950699,
      "epsilon_dpo/beta_margin_grad_mean": -0.4493896961212158,
      "epsilon_dpo/beta_margin_grad_std": 0.12668459117412567,
      "epsilon_dpo/beta_margin_mean": 0.21586395800113678,
      "epsilon_dpo/beta_margin_std": 0.5554152131080627,
      "epsilon_dpo/loss_margin_mean": 3.2856740951538086,
      "grad_norm": 14.878789901733398,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.0670349970459938,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.5131756438276466e-07,
      "logits/chosen": -0.021359417587518692,
      "logits/rejected": -0.14762753248214722,
      "logps/chosen": -93.09027099609375,
      "logps/ref_chosen": -93.75152587890625,
      "logps/ref_rejected": -102.76870727539062,
      "logps/rejected": -105.39312744140625,
      "loss": 1.2552,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.04284510016441345,
      "rewards/margins": 0.21586398780345917,
      "rewards/rejected": -0.17301888763904572,
      "step": 188
    },
    {
      "epoch": 0.2857142857142857,
      "epsilon_dpo/beta": 0.06656148284673691,
      "epsilon_dpo/beta_margin_grad_mean": -0.453034371137619,
      "epsilon_dpo/beta_margin_grad_std": 0.1559441089630127,
      "epsilon_dpo/beta_margin_mean": 0.1943899691104889,
      "epsilon_dpo/beta_margin_std": 0.6854284405708313,
      "epsilon_dpo/loss_margin_mean": 2.9941604137420654,
      "grad_norm": 18.815332412719727,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.06680534780025482,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.5053079818876096e-07,
      "logits/chosen": 0.04337912052869797,
      "logits/rejected": -0.1719989776611328,
      "logps/chosen": -99.32482147216797,
      "logps/ref_chosen": -100.89401245117188,
      "logps/ref_rejected": -95.35594177246094,
      "logps/rejected": -96.78091430664062,
      "loss": 1.313,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.10105113685131073,
      "rewards/margins": 0.19438999891281128,
      "rewards/rejected": -0.09333885461091995,
      "step": 189
    },
    {
      "epoch": 0.2872260015117158,
      "epsilon_dpo/beta": 0.06612560898065567,
      "epsilon_dpo/beta_margin_grad_mean": -0.3864794373512268,
      "epsilon_dpo/beta_margin_grad_std": 0.1279653012752533,
      "epsilon_dpo/beta_margin_mean": 0.49961158633232117,
      "epsilon_dpo/beta_margin_std": 0.5765790939331055,
      "epsilon_dpo/loss_margin_mean": 7.596166610717773,
      "grad_norm": 17.344053268432617,
      "kl/avg_steps": 0.65625,
      "kl/beta": 0.06655576825141907,
      "kl/n_epsilon_steps": 0.171875,
      "kl/p_epsilon_steps": 0.828125,
      "learning_rate": 4.4973842271726024e-07,
      "logits/chosen": -0.1879466474056244,
      "logits/rejected": -0.4992838501930237,
      "logps/chosen": -74.71463012695312,
      "logps/ref_chosen": -77.86666107177734,
      "logps/ref_rejected": -133.73385620117188,
      "logps/rejected": -138.17800903320312,
      "loss": 1.0242,
      "rewards/accuracies": 0.859375,
      "rewards/chosen": 0.20717057585716248,
      "rewards/margins": 0.49961161613464355,
      "rewards/rejected": -0.2924410402774811,
      "step": 190
    },
    {
      "epoch": 0.2887377173091459,
      "epsilon_dpo/beta": 0.06591136008501053,
      "epsilon_dpo/beta_margin_grad_mean": -0.44709786772727966,
      "epsilon_dpo/beta_margin_grad_std": 0.1486724317073822,
      "epsilon_dpo/beta_margin_mean": 0.2231767624616623,
      "epsilon_dpo/beta_margin_std": 0.6557565331459045,
      "epsilon_dpo/loss_margin_mean": 3.4554193019866943,
      "grad_norm": 18.134878158569336,
      "kl/avg_steps": 0.328125,
      "kl/beta": 0.06612183898687363,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.48940460132708e-07,
      "logits/chosen": -0.19696848094463348,
      "logits/rejected": -0.2077828049659729,
      "logps/chosen": -100.0518798828125,
      "logps/ref_chosen": -99.27456665039062,
      "logps/ref_rejected": -114.78330993652344,
      "logps/rejected": -119.01604461669922,
      "loss": 1.2774,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.053733598440885544,
      "rewards/margins": 0.22317670285701752,
      "rewards/rejected": -0.27691030502319336,
      "step": 191
    },
    {
      "epoch": 0.29024943310657597,
      "epsilon_dpo/beta": 0.06574740260839462,
      "epsilon_dpo/beta_margin_grad_mean": -0.4566137194633484,
      "epsilon_dpo/beta_margin_grad_std": 0.1029883325099945,
      "epsilon_dpo/beta_margin_mean": 0.18087714910507202,
      "epsilon_dpo/beta_margin_std": 0.4595518112182617,
      "epsilon_dpo/loss_margin_mean": 2.7946553230285645,
      "grad_norm": 12.453398704528809,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.06590559333562851,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.481369327558329e-07,
      "logits/chosen": -0.22844773530960083,
      "logits/rejected": -0.22785356640815735,
      "logps/chosen": -73.15953063964844,
      "logps/ref_chosen": -73.45323181152344,
      "logps/ref_rejected": -85.39158630371094,
      "logps/rejected": -87.8925552368164,
      "loss": 1.2632,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.018335875123739243,
      "rewards/margins": 0.18087714910507202,
      "rewards/rejected": -0.16254127025604248,
      "step": 192
    },
    {
      "epoch": 0.29176114890400606,
      "epsilon_dpo/beta": 0.06554234027862549,
      "epsilon_dpo/beta_margin_grad_mean": -0.42428433895111084,
      "epsilon_dpo/beta_margin_grad_std": 0.13521122932434082,
      "epsilon_dpo/beta_margin_mean": 0.3341885805130005,
      "epsilon_dpo/beta_margin_std": 0.594582200050354,
      "epsilon_dpo/loss_margin_mean": 5.164119243621826,
      "grad_norm": 16.317344665527344,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.06574123352766037,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.47327863063023e-07,
      "logits/chosen": -0.36393725872039795,
      "logits/rejected": -0.1663932204246521,
      "logps/chosen": -78.37696838378906,
      "logps/ref_chosen": -79.29537963867188,
      "logps/ref_rejected": -85.5767822265625,
      "logps/rejected": -89.82247924804688,
      "loss": 1.1627,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": 0.05778112635016441,
      "rewards/margins": 0.3341885805130005,
      "rewards/rejected": -0.27640748023986816,
      "step": 193
    },
    {
      "epoch": 0.29327286470143615,
      "epsilon_dpo/beta": 0.0654405727982521,
      "epsilon_dpo/beta_margin_grad_mean": -0.46438026428222656,
      "epsilon_dpo/beta_margin_grad_std": 0.15544359385967255,
      "epsilon_dpo/beta_margin_mean": 0.15591198205947876,
      "epsilon_dpo/beta_margin_std": 0.6794894933700562,
      "epsilon_dpo/loss_margin_mean": 2.4638755321502686,
      "grad_norm": 15.462275505065918,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.06553643196821213,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 4.4651327368569684e-07,
      "logits/chosen": -0.4889231324195862,
      "logits/rejected": -0.4247901439666748,
      "logps/chosen": -84.87882995605469,
      "logps/ref_chosen": -83.12230682373047,
      "logps/ref_rejected": -93.76890563964844,
      "logps/rejected": -97.98930358886719,
      "loss": 1.3464,
      "rewards/accuracies": 0.578125,
      "rewards/chosen": -0.11777393519878387,
      "rewards/margins": 0.15591202676296234,
      "rewards/rejected": -0.2736859619617462,
      "step": 194
    },
    {
      "epoch": 0.2947845804988662,
      "epsilon_dpo/beta": 0.06521578133106232,
      "epsilon_dpo/beta_margin_grad_mean": -0.4328776001930237,
      "epsilon_dpo/beta_margin_grad_std": 0.13702498376369476,
      "epsilon_dpo/beta_margin_mean": 0.29111796617507935,
      "epsilon_dpo/beta_margin_std": 0.5985894799232483,
      "epsilon_dpo/loss_margin_mean": 4.530278205871582,
      "grad_norm": 16.169418334960938,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06543419510126114,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.4569318740967043e-07,
      "logits/chosen": -0.3322495222091675,
      "logits/rejected": -0.22550660371780396,
      "logps/chosen": -99.76998901367188,
      "logps/ref_chosen": -96.8233642578125,
      "logps/ref_rejected": -96.86456298828125,
      "logps/rejected": -104.34147644042969,
      "loss": 1.201,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.19461895525455475,
      "rewards/margins": 0.2911180257797241,
      "rewards/rejected": -0.4857369661331177,
      "step": 195
    },
    {
      "epoch": 0.2962962962962963,
      "epsilon_dpo/beta": 0.06499237567186356,
      "epsilon_dpo/beta_margin_grad_mean": -0.4459932744503021,
      "epsilon_dpo/beta_margin_grad_std": 0.1392875760793686,
      "epsilon_dpo/beta_margin_mean": 0.23551592230796814,
      "epsilon_dpo/beta_margin_std": 0.6242495179176331,
      "epsilon_dpo/loss_margin_mean": 3.6898553371429443,
      "grad_norm": 14.393869400024414,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06521003693342209,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.448676271745197e-07,
      "logits/chosen": -0.21855714917182922,
      "logits/rejected": -0.2570999562740326,
      "logps/chosen": -83.5823974609375,
      "logps/ref_chosen": -82.40751647949219,
      "logps/ref_rejected": -105.99139404296875,
      "logps/rejected": -110.85611724853516,
      "loss": 1.2556,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.07799074053764343,
      "rewards/margins": 0.23551593720912933,
      "rewards/rejected": -0.3135066628456116,
      "step": 196
    },
    {
      "epoch": 0.29780801209372637,
      "epsilon_dpo/beta": 0.06474941968917847,
      "epsilon_dpo/beta_margin_grad_mean": -0.42367416620254517,
      "epsilon_dpo/beta_margin_grad_std": 0.16686230897903442,
      "epsilon_dpo/beta_margin_mean": 0.32828599214553833,
      "epsilon_dpo/beta_margin_std": 0.7847632169723511,
      "epsilon_dpo/loss_margin_mean": 5.152902603149414,
      "grad_norm": 15.43140697479248,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.06498664617538452,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.440366160729392e-07,
      "logits/chosen": -0.6653242111206055,
      "logits/rejected": -0.3941148519515991,
      "logps/chosen": -75.24999237060547,
      "logps/ref_chosen": -74.27025604248047,
      "logps/ref_rejected": -95.37227630615234,
      "logps/rejected": -101.50491333007812,
      "loss": 1.2247,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.06569398194551468,
      "rewards/margins": 0.32828599214553833,
      "rewards/rejected": -0.3939799666404724,
      "step": 197
    },
    {
      "epoch": 0.29931972789115646,
      "epsilon_dpo/beta": 0.06440634280443192,
      "epsilon_dpo/beta_margin_grad_mean": -0.41623857617378235,
      "epsilon_dpo/beta_margin_grad_std": 0.12911495566368103,
      "epsilon_dpo/beta_margin_mean": 0.358009934425354,
      "epsilon_dpo/beta_margin_std": 0.5631706118583679,
      "epsilon_dpo/loss_margin_mean": 5.61193323135376,
      "grad_norm": 14.343767166137695,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.06474385410547256,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 4.432001773500957e-07,
      "logits/chosen": -0.4333815574645996,
      "logits/rejected": -0.5065505504608154,
      "logps/chosen": -86.1900634765625,
      "logps/ref_chosen": -84.80004119873047,
      "logps/ref_rejected": -98.13402557373047,
      "logps/rejected": -105.135986328125,
      "loss": 1.1351,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.09154967963695526,
      "rewards/margins": 0.358009934425354,
      "rewards/rejected": -0.44955962896347046,
      "step": 198
    },
    {
      "epoch": 0.30083144368858655,
      "epsilon_dpo/beta": 0.06418675184249878,
      "epsilon_dpo/beta_margin_grad_mean": -0.4314574897289276,
      "epsilon_dpo/beta_margin_grad_std": 0.16398167610168457,
      "epsilon_dpo/beta_margin_mean": 0.2991337776184082,
      "epsilon_dpo/beta_margin_std": 0.7234874963760376,
      "epsilon_dpo/loss_margin_mean": 4.746899127960205,
      "grad_norm": 14.889111518859863,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06440171599388123,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.4235833440297856e-07,
      "logits/chosen": -0.2650891840457916,
      "logits/rejected": -0.44171708822250366,
      "logps/chosen": -82.58881378173828,
      "logps/ref_chosen": -79.6952133178711,
      "logps/ref_rejected": -110.57994079589844,
      "logps/rejected": -118.2204360961914,
      "loss": 1.2329,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.18838956952095032,
      "rewards/margins": 0.2991337776184082,
      "rewards/rejected": -0.48752331733703613,
      "step": 199
    },
    {
      "epoch": 0.30234315948601664,
      "epsilon_dpo/beta": 0.06386657804250717,
      "epsilon_dpo/beta_margin_grad_mean": -0.4096252918243408,
      "epsilon_dpo/beta_margin_grad_std": 0.14677537977695465,
      "epsilon_dpo/beta_margin_mean": 0.39887288212776184,
      "epsilon_dpo/beta_margin_std": 0.6632535457611084,
      "epsilon_dpo/loss_margin_mean": 6.310145378112793,
      "grad_norm": 13.412322044372559,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.06418109685182571,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.415111107797445e-07,
      "logits/chosen": -0.3834892511367798,
      "logits/rejected": -0.5493783354759216,
      "logps/chosen": -81.55732727050781,
      "logps/ref_chosen": -80.175048828125,
      "logps/ref_rejected": -113.15582275390625,
      "logps/rejected": -120.84823608398438,
      "loss": 1.1283,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.09074348211288452,
      "rewards/margins": 0.39887285232543945,
      "rewards/rejected": -0.489616334438324,
      "step": 200
    },
    {
      "epoch": 0.30234315948601664,
      "eval_epsilon_dpo/beta": 0.06364637613296509,
      "eval_epsilon_dpo/beta_margin_grad_mean": -0.4313368499279022,
      "eval_epsilon_dpo/beta_margin_grad_std": 0.14446234703063965,
      "eval_epsilon_dpo/beta_margin_mean": 0.3057065010070801,
      "eval_epsilon_dpo/beta_margin_std": 0.6497645974159241,
      "eval_epsilon_dpo/loss_margin_mean": 4.874636650085449,
      "eval_kl/n_epsilon_steps": 0.32614436745643616,
      "eval_kl/p_epsilon_steps": 0.67341548204422,
      "eval_logits/chosen": -0.338997483253479,
      "eval_logits/rejected": -0.38629141449928284,
      "eval_logps/chosen": -89.90900421142578,
      "eval_logps/ref_chosen": -87.42715454101562,
      "eval_logps/ref_rejected": -104.23548889160156,
      "eval_logps/rejected": -111.59197998046875,
      "eval_loss": 0.6027774810791016,
      "eval_rewards/accuracies": 0.6830986142158508,
      "eval_rewards/chosen": -0.16038812696933746,
      "eval_rewards/margins": 0.30570653080940247,
      "eval_rewards/rejected": -0.46609458327293396,
      "eval_runtime": 47.6329,
      "eval_samples_per_second": 48.349,
      "eval_steps_per_second": 1.512,
      "step": 200
    },
    {
      "epoch": 0.30385487528344673,
      "epsilon_dpo/beta": 0.06350891292095184,
      "epsilon_dpo/beta_margin_grad_mean": -0.42801105976104736,
      "epsilon_dpo/beta_margin_grad_std": 0.1340506672859192,
      "epsilon_dpo/beta_margin_mean": 0.3071900010108948,
      "epsilon_dpo/beta_margin_std": 0.5804892182350159,
      "epsilon_dpo/loss_margin_mean": 4.892658710479736,
      "grad_norm": 16.5773983001709,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.06386178731918335,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 4.4065853017905953e-07,
      "logits/chosen": -0.4059561491012573,
      "logits/rejected": -0.4868294894695282,
      "logps/chosen": -88.41127014160156,
      "logps/ref_chosen": -84.85652160644531,
      "logps/ref_rejected": -108.86637878417969,
      "logps/rejected": -117.31378173828125,
      "loss": 1.1828,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.22657953202724457,
      "rewards/margins": 0.30719006061553955,
      "rewards/rejected": -0.5337696075439453,
      "step": 201
    },
    {
      "epoch": 0.30536659108087677,
      "epsilon_dpo/beta": 0.06325291842222214,
      "epsilon_dpo/beta_margin_grad_mean": -0.4159080684185028,
      "epsilon_dpo/beta_margin_grad_std": 0.12844079732894897,
      "epsilon_dpo/beta_margin_mean": 0.3679068386554718,
      "epsilon_dpo/beta_margin_std": 0.5659220218658447,
      "epsilon_dpo/loss_margin_mean": 5.876931667327881,
      "grad_norm": 13.360112190246582,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.06350457668304443,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.3980061644943575e-07,
      "logits/chosen": -0.134725883603096,
      "logits/rejected": -0.25779327750205994,
      "logps/chosen": -71.51620483398438,
      "logps/ref_chosen": -70.20382690429688,
      "logps/ref_rejected": -100.29237365722656,
      "logps/rejected": -107.481689453125,
      "loss": 1.1267,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.08515353500843048,
      "rewards/margins": 0.3679068088531494,
      "rewards/rejected": -0.4530603289604187,
      "step": 202
    },
    {
      "epoch": 0.30687830687830686,
      "epsilon_dpo/beta": 0.06301674991846085,
      "epsilon_dpo/beta_margin_grad_mean": -0.42265784740448,
      "epsilon_dpo/beta_margin_grad_std": 0.12643830478191376,
      "epsilon_dpo/beta_margin_mean": 0.3411216735839844,
      "epsilon_dpo/beta_margin_std": 0.5760848522186279,
      "epsilon_dpo/loss_margin_mean": 5.4683661460876465,
      "grad_norm": 14.424872398376465,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.06324762850999832,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.3893739358856455e-07,
      "logits/chosen": -0.09640582650899887,
      "logits/rejected": -0.22556348145008087,
      "logps/chosen": -89.77330017089844,
      "logps/ref_chosen": -86.46543884277344,
      "logps/ref_rejected": -120.1901626586914,
      "logps/rejected": -128.9663848876953,
      "loss": 1.1503,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.21057778596878052,
      "rewards/margins": 0.3411216735839844,
      "rewards/rejected": -0.5516994595527649,
      "step": 203
    },
    {
      "epoch": 0.30839002267573695,
      "epsilon_dpo/beta": 0.06276163458824158,
      "epsilon_dpo/beta_margin_grad_mean": -0.4142310321331024,
      "epsilon_dpo/beta_margin_grad_std": 0.14236027002334595,
      "epsilon_dpo/beta_margin_mean": 0.38478749990463257,
      "epsilon_dpo/beta_margin_std": 0.6492862701416016,
      "epsilon_dpo/loss_margin_mean": 6.1974663734436035,
      "grad_norm": 16.044506072998047,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.06301134079694748,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.380688857426449e-07,
      "logits/chosen": -0.31047549843788147,
      "logits/rejected": -0.5649159550666809,
      "logps/chosen": -71.08057403564453,
      "logps/ref_chosen": -70.06008911132812,
      "logps/ref_rejected": -91.88562774658203,
      "logps/rejected": -99.10357666015625,
      "loss": 1.1346,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.0658092349767685,
      "rewards/margins": 0.38478749990463257,
      "rewards/rejected": -0.4505966901779175,
      "step": 204
    },
    {
      "epoch": 0.30990173847316704,
      "epsilon_dpo/beta": 0.06256653368473053,
      "epsilon_dpo/beta_margin_grad_mean": -0.442996084690094,
      "epsilon_dpo/beta_margin_grad_std": 0.1461196094751358,
      "epsilon_dpo/beta_margin_mean": 0.26055842638015747,
      "epsilon_dpo/beta_margin_std": 0.6598429679870605,
      "epsilon_dpo/loss_margin_mean": 4.234796047210693,
      "grad_norm": 17.57623863220215,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.06275638937950134,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.3719511720570814e-07,
      "logits/chosen": -0.4885348081588745,
      "logits/rejected": -0.2372855544090271,
      "logps/chosen": -96.82564544677734,
      "logps/ref_chosen": -93.42503356933594,
      "logps/ref_rejected": -119.87689208984375,
      "logps/rejected": -127.51229858398438,
      "loss": 1.2436,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.21464571356773376,
      "rewards/margins": 0.26055842638015747,
      "rewards/rejected": -0.47520411014556885,
      "step": 205
    },
    {
      "epoch": 0.31141345427059713,
      "epsilon_dpo/beta": 0.06254758685827255,
      "epsilon_dpo/beta_margin_grad_mean": -0.47024503350257874,
      "epsilon_dpo/beta_margin_grad_std": 0.14309605956077576,
      "epsilon_dpo/beta_margin_mean": 0.14074857532978058,
      "epsilon_dpo/beta_margin_std": 0.6398756504058838,
      "epsilon_dpo/loss_margin_mean": 2.3255414962768555,
      "grad_norm": 17.001508712768555,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.06256088614463806,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 4.363161124189387e-07,
      "logits/chosen": -0.2468646764755249,
      "logits/rejected": -0.2657265067100525,
      "logps/chosen": -98.77737426757812,
      "logps/ref_chosen": -95.62075805664062,
      "logps/ref_rejected": -104.67747497558594,
      "logps/rejected": -110.15962982177734,
      "loss": 1.3465,
      "rewards/accuracies": 0.515625,
      "rewards/chosen": -0.2001642882823944,
      "rewards/margins": 0.14074859023094177,
      "rewards/rejected": -0.3409128785133362,
      "step": 206
    },
    {
      "epoch": 0.3129251700680272,
      "epsilon_dpo/beta": 0.06233258917927742,
      "epsilon_dpo/beta_margin_grad_mean": -0.4336472749710083,
      "epsilon_dpo/beta_margin_grad_std": 0.1409965306520462,
      "epsilon_dpo/beta_margin_mean": 0.3076102137565613,
      "epsilon_dpo/beta_margin_std": 0.657630980014801,
      "epsilon_dpo/loss_margin_mean": 5.001832008361816,
      "grad_norm": 15.785563468933105,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06254134327173233,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.3543189596998986e-07,
      "logits/chosen": -0.29754525423049927,
      "logits/rejected": -0.6161233186721802,
      "logps/chosen": -96.36029815673828,
      "logps/ref_chosen": -92.49520111083984,
      "logps/ref_rejected": -111.60420989990234,
      "logps/rejected": -120.47113800048828,
      "loss": 1.1999,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.2434634566307068,
      "rewards/margins": 0.3076101243495941,
      "rewards/rejected": -0.5510735511779785,
      "step": 207
    },
    {
      "epoch": 0.3144368858654573,
      "epsilon_dpo/beta": 0.06219697371125221,
      "epsilon_dpo/beta_margin_grad_mean": -0.46254763007164,
      "epsilon_dpo/beta_margin_grad_std": 0.1336774379014969,
      "epsilon_dpo/beta_margin_mean": 0.16149714589118958,
      "epsilon_dpo/beta_margin_std": 0.579863429069519,
      "epsilon_dpo/loss_margin_mean": 2.666975736618042,
      "grad_norm": 18.437692642211914,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.06232709437608719,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.3454249259229664e-07,
      "logits/chosen": -0.3396787643432617,
      "logits/rejected": -0.10453208535909653,
      "logps/chosen": -84.78022003173828,
      "logps/ref_chosen": -83.91865539550781,
      "logps/ref_rejected": -95.56571960449219,
      "logps/rejected": -99.09425354003906,
      "loss": 1.3116,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.05670349299907684,
      "rewards/margins": 0.16149716079235077,
      "rewards/rejected": -0.21820063889026642,
      "step": 208
    },
    {
      "epoch": 0.31594860166288735,
      "epsilon_dpo/beta": 0.061905719339847565,
      "epsilon_dpo/beta_margin_grad_mean": -0.4098471403121948,
      "epsilon_dpo/beta_margin_grad_std": 0.1620044857263565,
      "epsilon_dpo/beta_margin_mean": 0.41248318552970886,
      "epsilon_dpo/beta_margin_std": 0.7426827549934387,
      "epsilon_dpo/loss_margin_mean": 6.738656044006348,
      "grad_norm": 15.497926712036133,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.06219105049967766,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.336479271643833e-07,
      "logits/chosen": -0.33976924419403076,
      "logits/rejected": -0.22139668464660645,
      "logps/chosen": -94.8779296875,
      "logps/ref_chosen": -93.79673767089844,
      "logps/ref_rejected": -115.7679672241211,
      "logps/rejected": -123.58782196044922,
      "loss": 1.1415,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.06984454393386841,
      "rewards/margins": 0.4124831557273865,
      "rewards/rejected": -0.4823276996612549,
      "step": 209
    },
    {
      "epoch": 0.31746031746031744,
      "epsilon_dpo/beta": 0.06163623929023743,
      "epsilon_dpo/beta_margin_grad_mean": -0.40473130345344543,
      "epsilon_dpo/beta_margin_grad_std": 0.14648084342479706,
      "epsilon_dpo/beta_margin_mean": 0.42534348368644714,
      "epsilon_dpo/beta_margin_std": 0.6573120951652527,
      "epsilon_dpo/loss_margin_mean": 6.972165584564209,
      "grad_norm": 13.21751594543457,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.06190089136362076,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.327482247091679e-07,
      "logits/chosen": -0.26555413007736206,
      "logits/rejected": -0.3610946238040924,
      "logps/chosen": -78.26898956298828,
      "logps/ref_chosen": -77.90690612792969,
      "logps/ref_rejected": -110.78089141845703,
      "logps/rejected": -118.11514282226562,
      "loss": 1.1053,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.024063657969236374,
      "rewards/margins": 0.42534342408180237,
      "rewards/rejected": -0.44940710067749023,
      "step": 210
    },
    {
      "epoch": 0.31897203325774753,
      "epsilon_dpo/beta": 0.06140627712011337,
      "epsilon_dpo/beta_margin_grad_mean": -0.4172515571117401,
      "epsilon_dpo/beta_margin_grad_std": 0.14959876239299774,
      "epsilon_dpo/beta_margin_mean": 0.3725390136241913,
      "epsilon_dpo/beta_margin_std": 0.6855942010879517,
      "epsilon_dpo/loss_margin_mean": 6.1406707763671875,
      "grad_norm": 14.281912803649902,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.061631254851818085,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.3184341039326217e-07,
      "logits/chosen": -0.3958446979522705,
      "logits/rejected": -0.356731116771698,
      "logps/chosen": -76.7433853149414,
      "logps/ref_chosen": -76.47579956054688,
      "logps/ref_rejected": -117.42009735107422,
      "logps/rejected": -123.82835388183594,
      "loss": 1.1555,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.0184482354670763,
      "rewards/margins": 0.37253904342651367,
      "rewards/rejected": -0.3909872770309448,
      "step": 211
    },
    {
      "epoch": 0.3204837490551776,
      "epsilon_dpo/beta": 0.06119605526328087,
      "epsilon_dpo/beta_margin_grad_mean": -0.4093796908855438,
      "epsilon_dpo/beta_margin_grad_std": 0.16085541248321533,
      "epsilon_dpo/beta_margin_mean": 0.406364768743515,
      "epsilon_dpo/beta_margin_std": 0.7220626473426819,
      "epsilon_dpo/loss_margin_mean": 6.725605487823486,
      "grad_norm": 12.916170120239258,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06140100210905075,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.309335095262675e-07,
      "logits/chosen": -0.272196501493454,
      "logits/rejected": -0.17583511769771576,
      "logps/chosen": -78.71829223632812,
      "logps/ref_chosen": -79.10020446777344,
      "logps/ref_rejected": -103.29380798339844,
      "logps/rejected": -109.63749694824219,
      "loss": 1.1416,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": 0.020729010924696922,
      "rewards/margins": 0.4063647985458374,
      "rewards/rejected": -0.3856357932090759,
      "step": 212
    },
    {
      "epoch": 0.3219954648526077,
      "epsilon_dpo/beta": 0.06105324998497963,
      "epsilon_dpo/beta_margin_grad_mean": -0.4278034269809723,
      "epsilon_dpo/beta_margin_grad_std": 0.156768798828125,
      "epsilon_dpo/beta_margin_mean": 0.32995331287384033,
      "epsilon_dpo/beta_margin_std": 0.7114284038543701,
      "epsilon_dpo/loss_margin_mean": 5.488154888153076,
      "grad_norm": 14.290369033813477,
      "kl/avg_steps": 0.234375,
      "kl/beta": 0.0611906573176384,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 4.3001854756006724e-07,
      "logits/chosen": -0.6963798999786377,
      "logits/rejected": -0.4149439334869385,
      "logps/chosen": -86.58470153808594,
      "logps/ref_chosen": -86.60450744628906,
      "logps/ref_rejected": -100.115478515625,
      "logps/rejected": -105.58381652832031,
      "loss": 1.2,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.0002879425883293152,
      "rewards/margins": 0.32995331287384033,
      "rewards/rejected": -0.33024126291275024,
      "step": 213
    },
    {
      "epoch": 0.3235071806500378,
      "epsilon_dpo/beta": 0.060843806713819504,
      "epsilon_dpo/beta_margin_grad_mean": -0.4379379451274872,
      "epsilon_dpo/beta_margin_grad_std": 0.14877955615520477,
      "epsilon_dpo/beta_margin_mean": 0.28321486711502075,
      "epsilon_dpo/beta_margin_std": 0.6764626502990723,
      "epsilon_dpo/loss_margin_mean": 4.7317938804626465,
      "grad_norm": 15.292763710021973,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06104757636785507,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.290985500881143e-07,
      "logits/chosen": -0.5268340706825256,
      "logits/rejected": -0.4801381826400757,
      "logps/chosen": -83.47666931152344,
      "logps/ref_chosen": -83.61862182617188,
      "logps/ref_rejected": -89.77618408203125,
      "logps/rejected": -94.36602020263672,
      "loss": 1.2285,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": 0.005956515669822693,
      "rewards/margins": 0.28321486711502075,
      "rewards/rejected": -0.27725836634635925,
      "step": 214
    },
    {
      "epoch": 0.3250188964474679,
      "epsilon_dpo/beta": 0.06063537672162056,
      "epsilon_dpo/beta_margin_grad_mean": -0.406276673078537,
      "epsilon_dpo/beta_margin_grad_std": 0.15514318645000458,
      "epsilon_dpo/beta_margin_mean": 0.43478846549987793,
      "epsilon_dpo/beta_margin_std": 0.7331078052520752,
      "epsilon_dpo/loss_margin_mean": 7.249897480010986,
      "grad_norm": 13.326201438903809,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.06083844602108002,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.281735428447157e-07,
      "logits/chosen": -0.2851630747318268,
      "logits/rejected": -0.34577393531799316,
      "logps/chosen": -87.37467956542969,
      "logps/ref_chosen": -85.94501495361328,
      "logps/ref_rejected": -125.37004089355469,
      "logps/rejected": -134.0496063232422,
      "loss": 1.1174,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.08875112980604172,
      "rewards/margins": 0.43478846549987793,
      "rewards/rejected": -0.5235395431518555,
      "step": 215
    },
    {
      "epoch": 0.32653061224489793,
      "epsilon_dpo/beta": 0.060351863503456116,
      "epsilon_dpo/beta_margin_grad_mean": -0.421083003282547,
      "epsilon_dpo/beta_margin_grad_std": 0.1392340064048767,
      "epsilon_dpo/beta_margin_mean": 0.34421074390411377,
      "epsilon_dpo/beta_margin_std": 0.6146458983421326,
      "epsilon_dpo/loss_margin_mean": 5.769903659820557,
      "grad_norm": 13.648635864257812,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.06063003093004227,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.2724355170431247e-07,
      "logits/chosen": -0.1937863528728485,
      "logits/rejected": -0.4227668046951294,
      "logps/chosen": -91.29273986816406,
      "logps/ref_chosen": -90.13990783691406,
      "logps/ref_rejected": -121.17817687988281,
      "logps/rejected": -128.10092163085938,
      "loss": 1.1601,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.07172287255525589,
      "rewards/margins": 0.34421074390411377,
      "rewards/rejected": -0.41593360900878906,
      "step": 216
    },
    {
      "epoch": 0.328042328042328,
      "epsilon_dpo/beta": 0.060070279985666275,
      "epsilon_dpo/beta_margin_grad_mean": -0.41138482093811035,
      "epsilon_dpo/beta_margin_grad_std": 0.16538718342781067,
      "epsilon_dpo/beta_margin_mean": 0.41298747062683105,
      "epsilon_dpo/beta_margin_std": 0.7676586508750916,
      "epsilon_dpo/loss_margin_mean": 6.954167366027832,
      "grad_norm": 13.441911697387695,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.06034715101122856,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.26308602680756e-07,
      "logits/chosen": -0.3257751166820526,
      "logits/rejected": -0.33707672357559204,
      "logps/chosen": -84.47261047363281,
      "logps/ref_chosen": -83.57549285888672,
      "logps/ref_rejected": -116.87894439697266,
      "logps/rejected": -124.73023986816406,
      "loss": 1.1486,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.056727487593889236,
      "rewards/margins": 0.41298750042915344,
      "rewards/rejected": -0.4697149693965912,
      "step": 217
    },
    {
      "epoch": 0.3295540438397581,
      "epsilon_dpo/beta": 0.059921421110630035,
      "epsilon_dpo/beta_margin_grad_mean": -0.44961637258529663,
      "epsilon_dpo/beta_margin_grad_std": 0.1689213514328003,
      "epsilon_dpo/beta_margin_mean": 0.22604058682918549,
      "epsilon_dpo/beta_margin_std": 0.7624347805976868,
      "epsilon_dpo/loss_margin_mean": 3.869769334793091,
      "grad_norm": 15.115226745605469,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.06006559357047081,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.253687219265803e-07,
      "logits/chosen": -0.2687731981277466,
      "logits/rejected": -0.22891171276569366,
      "logps/chosen": -101.1187744140625,
      "logps/ref_chosen": -98.71665954589844,
      "logps/ref_rejected": -97.98152160644531,
      "logps/rejected": -104.25341033935547,
      "loss": 1.3086,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.1473478376865387,
      "rewards/margins": 0.22604063153266907,
      "rewards/rejected": -0.37338846921920776,
      "step": 218
    },
    {
      "epoch": 0.3310657596371882,
      "epsilon_dpo/beta": 0.05964091420173645,
      "epsilon_dpo/beta_margin_grad_mean": -0.4214233458042145,
      "epsilon_dpo/beta_margin_grad_std": 0.13727082312107086,
      "epsilon_dpo/beta_margin_mean": 0.34724336862564087,
      "epsilon_dpo/beta_margin_std": 0.6157631278038025,
      "epsilon_dpo/loss_margin_mean": 5.883241176605225,
      "grad_norm": 12.273845672607422,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.05991580709815025,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.2442393573227043e-07,
      "logits/chosen": -0.3256840109825134,
      "logits/rejected": -0.4019477367401123,
      "logps/chosen": -83.3803939819336,
      "logps/ref_chosen": -82.32901000976562,
      "logps/ref_rejected": -97.81877136230469,
      "logps/rejected": -104.7533950805664,
      "loss": 1.1569,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.06491810828447342,
      "rewards/margins": 0.34724336862564087,
      "rewards/rejected": -0.4121614992618561,
      "step": 219
    },
    {
      "epoch": 0.3325774754346183,
      "epsilon_dpo/beta": 0.05939992889761925,
      "epsilon_dpo/beta_margin_grad_mean": -0.43890780210494995,
      "epsilon_dpo/beta_margin_grad_std": 0.1572382152080536,
      "epsilon_dpo/beta_margin_mean": 0.27426812052726746,
      "epsilon_dpo/beta_margin_std": 0.6969823837280273,
      "epsilon_dpo/loss_margin_mean": 4.694486618041992,
      "grad_norm": 12.854135513305664,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.05963626131415367,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.234742705255272e-07,
      "logits/chosen": -0.30797386169433594,
      "logits/rejected": -0.4406611919403076,
      "logps/chosen": -67.56721496582031,
      "logps/ref_chosen": -67.49903106689453,
      "logps/ref_rejected": -90.92181396484375,
      "logps/rejected": -95.68447875976562,
      "loss": 1.2449,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.006153309717774391,
      "rewards/margins": 0.27426809072494507,
      "rewards/rejected": -0.2804214358329773,
      "step": 220
    },
    {
      "epoch": 0.3340891912320484,
      "epsilon_dpo/beta": 0.059215277433395386,
      "epsilon_dpo/beta_margin_grad_mean": -0.41286715865135193,
      "epsilon_dpo/beta_margin_grad_std": 0.1621244251728058,
      "epsilon_dpo/beta_margin_mean": 0.4035100042819977,
      "epsilon_dpo/beta_margin_std": 0.7555540800094604,
      "epsilon_dpo/loss_margin_mean": 6.901949405670166,
      "grad_norm": 13.689908027648926,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.059394966810941696,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.22519752870528e-07,
      "logits/chosen": -0.3771657347679138,
      "logits/rejected": -0.46649545431137085,
      "logps/chosen": -83.02555847167969,
      "logps/ref_chosen": -83.29585266113281,
      "logps/ref_rejected": -115.793701171875,
      "logps/rejected": -122.42536163330078,
      "loss": 1.1516,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.014289181679487228,
      "rewards/margins": 0.4035099744796753,
      "rewards/rejected": -0.38922083377838135,
      "step": 221
    },
    {
      "epoch": 0.3356009070294785,
      "epsilon_dpo/beta": 0.05888276919722557,
      "epsilon_dpo/beta_margin_grad_mean": -0.3975021243095398,
      "epsilon_dpo/beta_margin_grad_std": 0.1471283882856369,
      "epsilon_dpo/beta_margin_mean": 0.46506625413894653,
      "epsilon_dpo/beta_margin_std": 0.6867176294326782,
      "epsilon_dpo/loss_margin_mean": 7.96026086807251,
      "grad_norm": 15.286458015441895,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.05920993909239769,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 4.2156040946718343e-07,
      "logits/chosen": -0.23492375016212463,
      "logits/rejected": -0.323875367641449,
      "logps/chosen": -82.56568908691406,
      "logps/ref_chosen": -82.82731628417969,
      "logps/ref_rejected": -131.6415252685547,
      "logps/rejected": -139.34014892578125,
      "loss": 1.0802,
      "rewards/accuracies": 0.75,
      "rewards/chosen": 0.012266065925359726,
      "rewards/margins": 0.4650663137435913,
      "rewards/rejected": -0.4528002142906189,
      "step": 222
    },
    {
      "epoch": 0.3371126228269085,
      "epsilon_dpo/beta": 0.05864541232585907,
      "epsilon_dpo/beta_margin_grad_mean": -0.40855130553245544,
      "epsilon_dpo/beta_margin_grad_std": 0.1467190384864807,
      "epsilon_dpo/beta_margin_mean": 0.4161318242549896,
      "epsilon_dpo/beta_margin_std": 0.6898074150085449,
      "epsilon_dpo/loss_margin_mean": 7.165706634521484,
      "grad_norm": 13.55103588104248,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.05887874588370323,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.2059626715039065e-07,
      "logits/chosen": -0.4994870126247406,
      "logits/rejected": -0.4353235363960266,
      "logps/chosen": -86.22138977050781,
      "logps/ref_chosen": -84.97879028320312,
      "logps/ref_rejected": -108.11666870117188,
      "logps/rejected": -116.52497100830078,
      "loss": 1.1195,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.07521222531795502,
      "rewards/margins": 0.41613179445266724,
      "rewards/rejected": -0.49134403467178345,
      "step": 223
    },
    {
      "epoch": 0.3386243386243386,
      "epsilon_dpo/beta": 0.05844460055232048,
      "epsilon_dpo/beta_margin_grad_mean": -0.4341517984867096,
      "epsilon_dpo/beta_margin_grad_std": 0.1348971128463745,
      "epsilon_dpo/beta_margin_mean": 0.29352185130119324,
      "epsilon_dpo/beta_margin_std": 0.5995880365371704,
      "epsilon_dpo/loss_margin_mean": 5.082745552062988,
      "grad_norm": 14.399001121520996,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.05864051729440689,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.1962735288928304e-07,
      "logits/chosen": -0.36321765184402466,
      "logits/rejected": -0.22956721484661102,
      "logps/chosen": -86.13837432861328,
      "logps/ref_chosen": -85.66336822509766,
      "logps/ref_rejected": -96.351806640625,
      "logps/rejected": -101.90956115722656,
      "loss": 1.1982,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.02954496629536152,
      "rewards/margins": 0.29352182149887085,
      "rewards/rejected": -0.3230668306350708,
      "step": 224
    },
    {
      "epoch": 0.3401360544217687,
      "epsilon_dpo/beta": 0.05822630599141121,
      "epsilon_dpo/beta_margin_grad_mean": -0.4103103578090668,
      "epsilon_dpo/beta_margin_grad_std": 0.17332212626934052,
      "epsilon_dpo/beta_margin_mean": 0.42294129729270935,
      "epsilon_dpo/beta_margin_std": 0.8520727753639221,
      "epsilon_dpo/loss_margin_mean": 7.355151176452637,
      "grad_norm": 16.749916076660156,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.05843963101506233,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.186536937864752e-07,
      "logits/chosen": -0.5275589823722839,
      "logits/rejected": -0.610481321811676,
      "logps/chosen": -96.54353332519531,
      "logps/ref_chosen": -95.25894165039062,
      "logps/ref_rejected": -144.3567657470703,
      "logps/rejected": -152.99652099609375,
      "loss": 1.1655,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.07754532992839813,
      "rewards/margins": 0.4229414165019989,
      "rewards/rejected": -0.5004867315292358,
      "step": 225
    },
    {
      "epoch": 0.3416477702191988,
      "epsilon_dpo/beta": 0.0580451674759388,
      "epsilon_dpo/beta_margin_grad_mean": -0.41428643465042114,
      "epsilon_dpo/beta_margin_grad_std": 0.1615278124809265,
      "epsilon_dpo/beta_margin_mean": 0.3854374289512634,
      "epsilon_dpo/beta_margin_std": 0.7472378015518188,
      "epsilon_dpo/loss_margin_mean": 6.730257034301758,
      "grad_norm": 12.055249214172363,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.058221302926540375,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.176753170773052e-07,
      "logits/chosen": -0.4694936275482178,
      "logits/rejected": -0.34530776739120483,
      "logps/chosen": -71.9918441772461,
      "logps/ref_chosen": -73.33175659179688,
      "logps/ref_rejected": -90.7422103881836,
      "logps/rejected": -96.13255310058594,
      "loss": 1.1649,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": 0.07627996802330017,
      "rewards/margins": 0.3854374289512634,
      "rewards/rejected": -0.30915746092796326,
      "step": 226
    },
    {
      "epoch": 0.3431594860166289,
      "epsilon_dpo/beta": 0.057846199721097946,
      "epsilon_dpo/beta_margin_grad_mean": -0.4127749502658844,
      "epsilon_dpo/beta_margin_grad_std": 0.1692953109741211,
      "epsilon_dpo/beta_margin_mean": 0.39249685406684875,
      "epsilon_dpo/beta_margin_std": 0.7651001811027527,
      "epsilon_dpo/loss_margin_mean": 6.88016939163208,
      "grad_norm": 18.469097137451172,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.05803992599248886,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.166922501290729e-07,
      "logits/chosen": -0.22612786293029785,
      "logits/rejected": -0.5124807953834534,
      "logps/chosen": -81.56525421142578,
      "logps/ref_chosen": -80.71163940429688,
      "logps/ref_rejected": -98.33506774902344,
      "logps/rejected": -106.06885528564453,
      "loss": 1.1672,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.0527096726000309,
      "rewards/margins": 0.39249682426452637,
      "rewards/rejected": -0.44520652294158936,
      "step": 227
    },
    {
      "epoch": 0.34467120181405897,
      "epsilon_dpo/beta": 0.0576661080121994,
      "epsilon_dpo/beta_margin_grad_mean": -0.42423751950263977,
      "epsilon_dpo/beta_margin_grad_std": 0.15881699323654175,
      "epsilon_dpo/beta_margin_mean": 0.34450241923332214,
      "epsilon_dpo/beta_margin_std": 0.7262859344482422,
      "epsilon_dpo/loss_margin_mean": 6.062928199768066,
      "grad_norm": 13.120859146118164,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.05784109607338905,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.1570452044027405e-07,
      "logits/chosen": -0.4411237835884094,
      "logits/rejected": -0.3885362148284912,
      "logps/chosen": -78.60456848144531,
      "logps/ref_chosen": -77.68626403808594,
      "logps/ref_rejected": -103.74498748779297,
      "logps/rejected": -110.7262191772461,
      "loss": 1.192,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.05492077395319939,
      "rewards/margins": 0.34450244903564453,
      "rewards/rejected": -0.39942318201065063,
      "step": 228
    },
    {
      "epoch": 0.34618291761148906,
      "epsilon_dpo/beta": 0.05747736245393753,
      "epsilon_dpo/beta_margin_grad_mean": -0.41639742255210876,
      "epsilon_dpo/beta_margin_grad_std": 0.15416114032268524,
      "epsilon_dpo/beta_margin_mean": 0.37639522552490234,
      "epsilon_dpo/beta_margin_std": 0.7211213707923889,
      "epsilon_dpo/loss_margin_mean": 6.627065658569336,
      "grad_norm": 12.59009075164795,
      "kl/avg_steps": 0.328125,
      "kl/beta": 0.05766090750694275,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.147121556398312e-07,
      "logits/chosen": -0.40707510709762573,
      "logits/rejected": -0.6244249939918518,
      "logps/chosen": -69.3853759765625,
      "logps/ref_chosen": -70.9267349243164,
      "logps/ref_rejected": -98.36172485351562,
      "logps/rejected": -103.44743347167969,
      "loss": 1.1623,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": 0.08642115443944931,
      "rewards/margins": 0.37639522552490234,
      "rewards/rejected": -0.28997403383255005,
      "step": 229
    },
    {
      "epoch": 0.3476946334089191,
      "epsilon_dpo/beta": 0.057280492037534714,
      "epsilon_dpo/beta_margin_grad_mean": -0.4393908679485321,
      "epsilon_dpo/beta_margin_grad_std": 0.16247457265853882,
      "epsilon_dpo/beta_margin_mean": 0.2653091549873352,
      "epsilon_dpo/beta_margin_std": 0.730425238609314,
      "epsilon_dpo/loss_margin_mean": 4.723310947418213,
      "grad_norm": 15.629660606384277,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.05747232586145401,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.137151834863213e-07,
      "logits/chosen": -0.46971315145492554,
      "logits/rejected": -0.42873239517211914,
      "logps/chosen": -83.31251525878906,
      "logps/ref_chosen": -82.13688659667969,
      "logps/ref_rejected": -83.09074401855469,
      "logps/rejected": -88.98968505859375,
      "loss": 1.2632,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.07054407149553299,
      "rewards/margins": 0.26530909538269043,
      "rewards/rejected": -0.3358531594276428,
      "step": 230
    },
    {
      "epoch": 0.3492063492063492,
      "epsilon_dpo/beta": 0.05708426609635353,
      "epsilon_dpo/beta_margin_grad_mean": -0.4018583595752716,
      "epsilon_dpo/beta_margin_grad_std": 0.16053839027881622,
      "epsilon_dpo/beta_margin_mean": 0.4458349347114563,
      "epsilon_dpo/beta_margin_std": 0.7325132489204407,
      "epsilon_dpo/loss_margin_mean": 7.896346569061279,
      "grad_norm": 13.946819305419922,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.05727544054389,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.1271363186719835e-07,
      "logits/chosen": -0.14116013050079346,
      "logits/rejected": -0.1820957064628601,
      "logps/chosen": -102.2670669555664,
      "logps/ref_chosen": -100.55113220214844,
      "logps/ref_rejected": -107.24727630615234,
      "logps/rejected": -116.85955810546875,
      "loss": 1.1122,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.09947866201400757,
      "rewards/margins": 0.4458349943161011,
      "rewards/rejected": -0.5453135967254639,
      "step": 231
    },
    {
      "epoch": 0.3507180650037793,
      "epsilon_dpo/beta": 0.056933220475912094,
      "epsilon_dpo/beta_margin_grad_mean": -0.43795302510261536,
      "epsilon_dpo/beta_margin_grad_std": 0.1754705309867859,
      "epsilon_dpo/beta_margin_mean": 0.3054857552051544,
      "epsilon_dpo/beta_margin_std": 0.8524812459945679,
      "epsilon_dpo/loss_margin_mean": 5.4650115966796875,
      "grad_norm": 14.59022331237793,
      "kl/avg_steps": 0.265625,
      "kl/beta": 0.05707923322916031,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.1170752879801436e-07,
      "logits/chosen": -0.5040056109428406,
      "logits/rejected": -0.3939094543457031,
      "logps/chosen": -94.84075927734375,
      "logps/ref_chosen": -92.6686019897461,
      "logps/ref_rejected": -103.07643127441406,
      "logps/rejected": -110.71360778808594,
      "loss": 1.2637,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.1255113184452057,
      "rewards/margins": 0.3054857552051544,
      "rewards/rejected": -0.4309970736503601,
      "step": 232
    },
    {
      "epoch": 0.35222978080120937,
      "epsilon_dpo/beta": 0.05688033252954483,
      "epsilon_dpo/beta_margin_grad_mean": -0.4426785707473755,
      "epsilon_dpo/beta_margin_grad_std": 0.17714117467403412,
      "epsilon_dpo/beta_margin_mean": 0.270059198141098,
      "epsilon_dpo/beta_margin_std": 0.8090578317642212,
      "epsilon_dpo/loss_margin_mean": 4.856035232543945,
      "grad_norm": 15.220146179199219,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.05692801624536514,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 4.106969024216348e-07,
      "logits/chosen": -0.33606183528900146,
      "logits/rejected": -0.13796675205230713,
      "logps/chosen": -80.97364807128906,
      "logps/ref_chosen": -78.285400390625,
      "logps/ref_rejected": -90.51336669921875,
      "logps/rejected": -98.05764770507812,
      "loss": 1.2856,
      "rewards/accuracies": 0.546875,
      "rewards/chosen": -0.15561680495738983,
      "rewards/margins": 0.27005913853645325,
      "rewards/rejected": -0.4256759583950043,
      "step": 233
    },
    {
      "epoch": 0.35374149659863946,
      "epsilon_dpo/beta": 0.05677373334765434,
      "epsilon_dpo/beta_margin_grad_mean": -0.4403716027736664,
      "epsilon_dpo/beta_margin_grad_std": 0.1675698608160019,
      "epsilon_dpo/beta_margin_mean": 0.2730686664581299,
      "epsilon_dpo/beta_margin_std": 0.7592166662216187,
      "epsilon_dpo/loss_margin_mean": 4.9085798263549805,
      "grad_norm": 13.379533767700195,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.05687469616532326,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 4.09681781007452e-07,
      "logits/chosen": -0.3230592608451843,
      "logits/rejected": -0.5212711095809937,
      "logps/chosen": -77.2991943359375,
      "logps/ref_chosen": -75.72798919677734,
      "logps/ref_rejected": -74.1239013671875,
      "logps/rejected": -80.60368347167969,
      "loss": 1.2656,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.09258531779050827,
      "rewards/margins": 0.2730686664581299,
      "rewards/rejected": -0.36565396189689636,
      "step": 234
    },
    {
      "epoch": 0.35525321239606955,
      "epsilon_dpo/beta": 0.05650780722498894,
      "epsilon_dpo/beta_margin_grad_mean": -0.3946419656276703,
      "epsilon_dpo/beta_margin_grad_std": 0.15039300918579102,
      "epsilon_dpo/beta_margin_mean": 0.47568777203559875,
      "epsilon_dpo/beta_margin_std": 0.6847705841064453,
      "epsilon_dpo/loss_margin_mean": 8.491762161254883,
      "grad_norm": 13.730059623718262,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.056768257170915604,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.08662192950594e-07,
      "logits/chosen": -0.6019189953804016,
      "logits/rejected": -0.48583483695983887,
      "logps/chosen": -91.29273223876953,
      "logps/ref_chosen": -91.11560821533203,
      "logps/ref_rejected": -103.01558685302734,
      "logps/rejected": -111.68447875976562,
      "loss": 1.0732,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.011768028140068054,
      "rewards/margins": 0.47568780183792114,
      "rewards/rejected": -0.4874558448791504,
      "step": 235
    },
    {
      "epoch": 0.35676492819349964,
      "epsilon_dpo/beta": 0.05634119734168053,
      "epsilon_dpo/beta_margin_grad_mean": -0.4295879006385803,
      "epsilon_dpo/beta_margin_grad_std": 0.17866680026054382,
      "epsilon_dpo/beta_margin_mean": 0.31965601444244385,
      "epsilon_dpo/beta_margin_std": 0.8243433833122253,
      "epsilon_dpo/loss_margin_mean": 5.777461051940918,
      "grad_norm": 16.337749481201172,
      "kl/avg_steps": 0.296875,
      "kl/beta": 0.05650339648127556,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.076381667711306e-07,
      "logits/chosen": -0.3394400179386139,
      "logits/rejected": -0.1670718789100647,
      "logps/chosen": -101.02061462402344,
      "logps/ref_chosen": -97.19900512695312,
      "logps/ref_rejected": -106.0116958618164,
      "logps/rejected": -115.61076354980469,
      "loss": 1.2481,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.21814972162246704,
      "rewards/margins": 0.31965598464012146,
      "rewards/rejected": -0.5378056764602661,
      "step": 236
    },
    {
      "epoch": 0.35827664399092973,
      "epsilon_dpo/beta": 0.05618331953883171,
      "epsilon_dpo/beta_margin_grad_mean": -0.4469059705734253,
      "epsilon_dpo/beta_margin_grad_std": 0.15130768716335297,
      "epsilon_dpo/beta_margin_mean": 0.23650793731212616,
      "epsilon_dpo/beta_margin_std": 0.6870653629302979,
      "epsilon_dpo/loss_margin_mean": 4.295169830322266,
      "grad_norm": 15.307231903076172,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.05633614957332611,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.066097311132753e-07,
      "logits/chosen": -0.24095523357391357,
      "logits/rejected": -0.20215189456939697,
      "logps/chosen": -102.84424591064453,
      "logps/ref_chosen": -99.35952758789062,
      "logps/ref_rejected": -105.2945327758789,
      "logps/rejected": -113.07442474365234,
      "loss": 1.2733,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.19786638021469116,
      "rewards/margins": 0.23650792241096497,
      "rewards/rejected": -0.43437430262565613,
      "step": 237
    },
    {
      "epoch": 0.35978835978835977,
      "epsilon_dpo/beta": 0.055964212864637375,
      "epsilon_dpo/beta_margin_grad_mean": -0.41444844007492065,
      "epsilon_dpo/beta_margin_grad_std": 0.14432160556316376,
      "epsilon_dpo/beta_margin_mean": 0.38114407658576965,
      "epsilon_dpo/beta_margin_std": 0.648664116859436,
      "epsilon_dpo/loss_margin_mean": 6.885334014892578,
      "grad_norm": 13.714317321777344,
      "kl/avg_steps": 0.390625,
      "kl/beta": 0.056178148835897446,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 4.0557691474458414e-07,
      "logits/chosen": -0.6821615695953369,
      "logits/rejected": -0.23968210816383362,
      "logps/chosen": -89.17784118652344,
      "logps/ref_chosen": -87.75861358642578,
      "logps/ref_rejected": -99.15858459472656,
      "logps/rejected": -107.46315002441406,
      "loss": 1.1384,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.08189239352941513,
      "rewards/margins": 0.38114404678344727,
      "rewards/rejected": -0.463036447763443,
      "step": 238
    },
    {
      "epoch": 0.36130007558578986,
      "epsilon_dpo/beta": 0.055807750672101974,
      "epsilon_dpo/beta_margin_grad_mean": -0.4088519811630249,
      "epsilon_dpo/beta_margin_grad_std": 0.1726495772600174,
      "epsilon_dpo/beta_margin_mean": 0.43803808093070984,
      "epsilon_dpo/beta_margin_std": 0.8137868046760559,
      "epsilon_dpo/loss_margin_mean": 7.949976921081543,
      "grad_norm": 14.539200782775879,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.0559595562517643,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 4.045397465551513e-07,
      "logits/chosen": -0.19835148751735687,
      "logits/rejected": -0.379528284072876,
      "logps/chosen": -84.41043090820312,
      "logps/ref_chosen": -81.45846557617188,
      "logps/ref_rejected": -142.63424682617188,
      "logps/rejected": -153.5361785888672,
      "loss": 1.1434,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.16567102074623108,
      "rewards/margins": 0.4380381107330322,
      "rewards/rejected": -0.6037091016769409,
      "step": 239
    },
    {
      "epoch": 0.36281179138321995,
      "epsilon_dpo/beta": 0.0555814690887928,
      "epsilon_dpo/beta_margin_grad_mean": -0.38159269094467163,
      "epsilon_dpo/beta_margin_grad_std": 0.17057675123214722,
      "epsilon_dpo/beta_margin_mean": 0.5508310198783875,
      "epsilon_dpo/beta_margin_std": 0.7981725335121155,
      "epsilon_dpo/loss_margin_mean": 10.005630493164062,
      "grad_norm": 25.778789520263672,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.0558026097714901,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.0349825555680045e-07,
      "logits/chosen": -0.7162148356437683,
      "logits/rejected": -0.9191411137580872,
      "logps/chosen": -79.45154571533203,
      "logps/ref_chosen": -77.65220642089844,
      "logps/ref_rejected": -106.41152954101562,
      "logps/rejected": -118.21650695800781,
      "loss": 1.0521,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.10241775214672089,
      "rewards/margins": 0.5508310198783875,
      "rewards/rejected": -0.6532487869262695,
      "step": 240
    },
    {
      "epoch": 0.36432350718065004,
      "epsilon_dpo/beta": 0.055408693850040436,
      "epsilon_dpo/beta_margin_grad_mean": -0.43489742279052734,
      "epsilon_dpo/beta_margin_grad_std": 0.15454921126365662,
      "epsilon_dpo/beta_margin_mean": 0.293466180562973,
      "epsilon_dpo/beta_margin_std": 0.6903191208839417,
      "epsilon_dpo/loss_margin_mean": 5.379743576049805,
      "grad_norm": 17.371824264526367,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.055576831102371216,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.0245247088227377e-07,
      "logits/chosen": -0.3993128538131714,
      "logits/rejected": -0.2141626477241516,
      "logps/chosen": -97.16249084472656,
      "logps/ref_chosen": -93.96852111816406,
      "logps/ref_rejected": -109.2076187133789,
      "logps/rejected": -117.78132629394531,
      "loss": 1.2255,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.18001073598861694,
      "rewards/margins": 0.2934662103652954,
      "rewards/rejected": -0.47347694635391235,
      "step": 241
    },
    {
      "epoch": 0.36583522297808013,
      "epsilon_dpo/beta": 0.0550975576043129,
      "epsilon_dpo/beta_margin_grad_mean": -0.3924703598022461,
      "epsilon_dpo/beta_margin_grad_std": 0.15534812211990356,
      "epsilon_dpo/beta_margin_mean": 0.4966244101524353,
      "epsilon_dpo/beta_margin_std": 0.757362961769104,
      "epsilon_dpo/loss_margin_mean": 9.078295707702637,
      "grad_norm": 13.380143165588379,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.0554036945104599,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 4.0140242178441665e-07,
      "logits/chosen": -0.2932683825492859,
      "logits/rejected": -0.4387352466583252,
      "logps/chosen": -82.11421966552734,
      "logps/ref_chosen": -82.18013000488281,
      "logps/ref_rejected": -89.77629089355469,
      "logps/rejected": -98.78868103027344,
      "loss": 1.0755,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": 0.0007947832345962524,
      "rewards/margins": 0.4966244101524353,
      "rewards/rejected": -0.49582961201667786,
      "step": 242
    },
    {
      "epoch": 0.3673469387755102,
      "epsilon_dpo/beta": 0.05490989610552788,
      "epsilon_dpo/beta_margin_grad_mean": -0.4230293035507202,
      "epsilon_dpo/beta_margin_grad_std": 0.14619854092597961,
      "epsilon_dpo/beta_margin_mean": 0.341871976852417,
      "epsilon_dpo/beta_margin_std": 0.6454283595085144,
      "epsilon_dpo/loss_margin_mean": 6.306732177734375,
      "grad_norm": 13.98609447479248,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.05509379133582115,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.003481376353596e-07,
      "logits/chosen": -0.4575600028038025,
      "logits/rejected": -0.29443448781967163,
      "logps/chosen": -104.02952575683594,
      "logps/ref_chosen": -100.66859436035156,
      "logps/ref_rejected": -97.08364868164062,
      "logps/rejected": -106.75131225585938,
      "loss": 1.1711,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.1866493821144104,
      "rewards/margins": 0.341871976852417,
      "rewards/rejected": -0.5285213589668274,
      "step": 243
    },
    {
      "epoch": 0.3688586545729403,
      "epsilon_dpo/beta": 0.054567355662584305,
      "epsilon_dpo/beta_margin_grad_mean": -0.3743903636932373,
      "epsilon_dpo/beta_margin_grad_std": 0.12789258360862732,
      "epsilon_dpo/beta_margin_mean": 0.5610859990119934,
      "epsilon_dpo/beta_margin_std": 0.6003847718238831,
      "epsilon_dpo/loss_margin_mean": 10.335267066955566,
      "grad_norm": 12.74890422821045,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.05490505322813988,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 3.9928964792569654e-07,
      "logits/chosen": -0.606474757194519,
      "logits/rejected": -0.2900153696537018,
      "logps/chosen": -76.24296569824219,
      "logps/ref_chosen": -74.50486755371094,
      "logps/ref_rejected": -102.90267944335938,
      "logps/rejected": -114.97602844238281,
      "loss": 0.9822,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.09550999850034714,
      "rewards/margins": 0.5610860586166382,
      "rewards/rejected": -0.6565960049629211,
      "step": 244
    },
    {
      "epoch": 0.37037037037037035,
      "epsilon_dpo/beta": 0.05426253378391266,
      "epsilon_dpo/beta_margin_grad_mean": -0.3894428312778473,
      "epsilon_dpo/beta_margin_grad_std": 0.14738242328166962,
      "epsilon_dpo/beta_margin_mean": 0.49736499786376953,
      "epsilon_dpo/beta_margin_std": 0.6809157729148865,
      "epsilon_dpo/loss_margin_mean": 9.233874320983887,
      "grad_norm": 13.286972999572754,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.05456402897834778,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 3.982269822636601e-07,
      "logits/chosen": -0.4713189899921417,
      "logits/rejected": -0.4719199240207672,
      "logps/chosen": -97.87628173828125,
      "logps/ref_chosen": -94.40076446533203,
      "logps/ref_rejected": -102.05097961425781,
      "logps/rejected": -114.7603759765625,
      "loss": 1.0545,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.18986687064170837,
      "rewards/margins": 0.49736499786376953,
      "rewards/rejected": -0.6872318387031555,
      "step": 245
    },
    {
      "epoch": 0.37188208616780044,
      "epsilon_dpo/beta": 0.05406075716018677,
      "epsilon_dpo/beta_margin_grad_mean": -0.39330968260765076,
      "epsilon_dpo/beta_margin_grad_std": 0.16505803167819977,
      "epsilon_dpo/beta_margin_mean": 0.4963088929653168,
      "epsilon_dpo/beta_margin_std": 0.7848570942878723,
      "epsilon_dpo/loss_margin_mean": 9.28079891204834,
      "grad_norm": 13.795814514160156,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.05425882339477539,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.971601703742932e-07,
      "logits/chosen": -0.3446606397628784,
      "logits/rejected": -0.3383950889110565,
      "logps/chosen": -103.00593566894531,
      "logps/ref_chosen": -97.2917709350586,
      "logps/ref_rejected": -120.41865539550781,
      "logps/rejected": -135.4136199951172,
      "loss": 1.0868,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.31252914667129517,
      "rewards/margins": 0.4963088631629944,
      "rewards/rejected": -0.8088380098342896,
      "step": 246
    },
    {
      "epoch": 0.37339380196523053,
      "epsilon_dpo/beta": 0.05397704616189003,
      "epsilon_dpo/beta_margin_grad_mean": -0.4514468312263489,
      "epsilon_dpo/beta_margin_grad_std": 0.15922772884368896,
      "epsilon_dpo/beta_margin_mean": 0.2120126336812973,
      "epsilon_dpo/beta_margin_std": 0.7254804968833923,
      "epsilon_dpo/loss_margin_mean": 4.027678489685059,
      "grad_norm": 14.925507545471191,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.054056111723184586,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 3.960892420986177e-07,
      "logits/chosen": -0.24031943082809448,
      "logits/rejected": -0.38908693194389343,
      "logps/chosen": -117.6470718383789,
      "logps/ref_chosen": -110.15637969970703,
      "logps/ref_rejected": -112.48919677734375,
      "logps/rejected": -124.007568359375,
      "loss": 1.3077,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.40666455030441284,
      "rewards/margins": 0.2120126336812973,
      "rewards/rejected": -0.6186771988868713,
      "step": 247
    },
    {
      "epoch": 0.3749055177626606,
      "epsilon_dpo/beta": 0.053825367242097855,
      "epsilon_dpo/beta_margin_grad_mean": -0.4187783896923065,
      "epsilon_dpo/beta_margin_grad_std": 0.18080280721187592,
      "epsilon_dpo/beta_margin_mean": 0.3853808343410492,
      "epsilon_dpo/beta_margin_std": 0.8422636389732361,
      "epsilon_dpo/loss_margin_mean": 7.274705410003662,
      "grad_norm": 17.41864585876465,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.0539717823266983,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 3.9501422739279953e-07,
      "logits/chosen": -0.3880723714828491,
      "logits/rejected": -0.15597718954086304,
      "logps/chosen": -99.78440856933594,
      "logps/ref_chosen": -94.91020202636719,
      "logps/ref_rejected": -79.59503936767578,
      "logps/rejected": -91.74395751953125,
      "loss": 1.1984,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.2669926881790161,
      "rewards/margins": 0.3853807747364044,
      "rewards/rejected": -0.6523734927177429,
      "step": 248
    },
    {
      "epoch": 0.3764172335600907,
      "epsilon_dpo/beta": 0.053691230714321136,
      "epsilon_dpo/beta_margin_grad_mean": -0.473882794380188,
      "epsilon_dpo/beta_margin_grad_std": 0.17014063894748688,
      "epsilon_dpo/beta_margin_mean": 0.1020829826593399,
      "epsilon_dpo/beta_margin_std": 0.773459255695343,
      "epsilon_dpo/loss_margin_mean": 2.0085771083831787,
      "grad_norm": 16.087905883789062,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.0538204126060009,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 3.9393515632731094e-07,
      "logits/chosen": -0.12361128628253937,
      "logits/rejected": -0.1863332986831665,
      "logps/chosen": -108.78060913085938,
      "logps/ref_chosen": -99.98388671875,
      "logps/ref_rejected": -84.525146484375,
      "logps/rejected": -95.33045959472656,
      "loss": 1.4266,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.47448354959487915,
      "rewards/margins": 0.10208297520875931,
      "rewards/rejected": -0.5765665173530579,
      "step": 249
    },
    {
      "epoch": 0.3779289493575208,
      "epsilon_dpo/beta": 0.05338955298066139,
      "epsilon_dpo/beta_margin_grad_mean": -0.3938334286212921,
      "epsilon_dpo/beta_margin_grad_std": 0.16821321845054626,
      "epsilon_dpo/beta_margin_mean": 0.49425008893013,
      "epsilon_dpo/beta_margin_std": 0.8463791012763977,
      "epsilon_dpo/loss_margin_mean": 9.341487884521484,
      "grad_norm": 15.321463584899902,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.053686197847127914,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 3.9285205908608934e-07,
      "logits/chosen": -0.49714428186416626,
      "logits/rejected": -0.35792437195777893,
      "logps/chosen": -92.70217895507812,
      "logps/ref_chosen": -85.96786499023438,
      "logps/ref_rejected": -103.78448486328125,
      "logps/rejected": -119.86029052734375,
      "loss": 1.1055,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.36198216676712036,
      "rewards/margins": 0.4942500591278076,
      "rewards/rejected": -0.8562322854995728,
      "step": 250
    },
    {
      "epoch": 0.3794406651549509,
      "epsilon_dpo/beta": 0.05319102481007576,
      "epsilon_dpo/beta_margin_grad_mean": -0.4373018145561218,
      "epsilon_dpo/beta_margin_grad_std": 0.1502736508846283,
      "epsilon_dpo/beta_margin_mean": 0.2876345217227936,
      "epsilon_dpo/beta_margin_std": 0.6921249032020569,
      "epsilon_dpo/loss_margin_mean": 5.49154806137085,
      "grad_norm": 12.652570724487305,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.0533859021961689,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.9176496596569265e-07,
      "logits/chosen": -0.245034322142601,
      "logits/rejected": -0.4103810787200928,
      "logps/chosen": -95.39961242675781,
      "logps/ref_chosen": -88.44728088378906,
      "logps/ref_rejected": -106.68901062011719,
      "logps/rejected": -119.13288879394531,
      "loss": 1.2288,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.3725150227546692,
      "rewards/margins": 0.2876344919204712,
      "rewards/rejected": -0.6601495146751404,
      "step": 251
    },
    {
      "epoch": 0.38095238095238093,
      "epsilon_dpo/beta": 0.05302554741501808,
      "epsilon_dpo/beta_margin_grad_mean": -0.44193825125694275,
      "epsilon_dpo/beta_margin_grad_std": 0.16500435769557953,
      "epsilon_dpo/beta_margin_mean": 0.2562330365180969,
      "epsilon_dpo/beta_margin_std": 0.7518081068992615,
      "epsilon_dpo/loss_margin_mean": 4.933616638183594,
      "grad_norm": 14.027976989746094,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.05318645387887955,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 3.9067390737445254e-07,
      "logits/chosen": -0.49801328778266907,
      "logits/rejected": -0.47471922636032104,
      "logps/chosen": -83.92085266113281,
      "logps/ref_chosen": -78.4037857055664,
      "logps/ref_rejected": -99.27626037597656,
      "logps/rejected": -109.7269515991211,
      "loss": 1.2776,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.2952837347984314,
      "rewards/margins": 0.2562330365180969,
      "rewards/rejected": -0.5515167713165283,
      "step": 252
    },
    {
      "epoch": 0.382464096749811,
      "epsilon_dpo/beta": 0.052843790501356125,
      "epsilon_dpo/beta_margin_grad_mean": -0.44336360692977905,
      "epsilon_dpo/beta_margin_grad_std": 0.16899996995925903,
      "epsilon_dpo/beta_margin_mean": 0.265546977519989,
      "epsilon_dpo/beta_margin_std": 0.7731426358222961,
      "epsilon_dpo/loss_margin_mean": 5.125515937805176,
      "grad_norm": 12.717514038085938,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.05302076414227486,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.8957891383162304e-07,
      "logits/chosen": -0.19492888450622559,
      "logits/rejected": -0.21276481449604034,
      "logps/chosen": -78.13883972167969,
      "logps/ref_chosen": -72.53218078613281,
      "logps/ref_rejected": -83.00707244873047,
      "logps/rejected": -93.7392578125,
      "loss": 1.276,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.298408180475235,
      "rewards/margins": 0.265546977519989,
      "rewards/rejected": -0.5639551877975464,
      "step": 253
    },
    {
      "epoch": 0.3839758125472411,
      "epsilon_dpo/beta": 0.05267927050590515,
      "epsilon_dpo/beta_margin_grad_mean": -0.41813188791275024,
      "epsilon_dpo/beta_margin_grad_std": 0.15210388600826263,
      "epsilon_dpo/beta_margin_mean": 0.374256432056427,
      "epsilon_dpo/beta_margin_std": 0.6875149011611938,
      "epsilon_dpo/loss_margin_mean": 7.195765495300293,
      "grad_norm": 14.125617980957031,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.05283912643790245,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 3.884800159665276e-07,
      "logits/chosen": -0.7151049971580505,
      "logits/rejected": -0.4137560725212097,
      "logps/chosen": -93.89373779296875,
      "logps/ref_chosen": -87.48554229736328,
      "logps/ref_rejected": -109.45498657226562,
      "logps/rejected": -123.0589370727539,
      "loss": 1.1554,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.3408017158508301,
      "rewards/margins": 0.374256432056427,
      "rewards/rejected": -0.7150582075119019,
      "step": 254
    },
    {
      "epoch": 0.3854875283446712,
      "epsilon_dpo/beta": 0.052449315786361694,
      "epsilon_dpo/beta_margin_grad_mean": -0.41780516505241394,
      "epsilon_dpo/beta_margin_grad_std": 0.16844013333320618,
      "epsilon_dpo/beta_margin_mean": 0.3639993667602539,
      "epsilon_dpo/beta_margin_std": 0.7727952599525452,
      "epsilon_dpo/loss_margin_mean": 7.039681434631348,
      "grad_norm": 18.118885040283203,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.05267452076077461,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.873772445177015e-07,
      "logits/chosen": -0.3129955530166626,
      "logits/rejected": -0.255302757024765,
      "logps/chosen": -103.62435913085938,
      "logps/ref_chosen": -96.826171875,
      "logps/ref_rejected": -111.32254028320312,
      "logps/rejected": -125.16040802001953,
      "loss": 1.1925,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.35845616459846497,
      "rewards/margins": 0.3639993667602539,
      "rewards/rejected": -0.7224555015563965,
      "step": 255
    },
    {
      "epoch": 0.3869992441421013,
      "epsilon_dpo/beta": 0.05220445618033409,
      "epsilon_dpo/beta_margin_grad_mean": -0.3989577889442444,
      "epsilon_dpo/beta_margin_grad_std": 0.1645575612783432,
      "epsilon_dpo/beta_margin_mean": 0.45816469192504883,
      "epsilon_dpo/beta_margin_std": 0.7570108771324158,
      "epsilon_dpo/loss_margin_mean": 8.868268966674805,
      "grad_norm": 14.02364730834961,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.052445072680711746,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.862706303320329e-07,
      "logits/chosen": -0.24509219825267792,
      "logits/rejected": -0.5150793790817261,
      "logps/chosen": -93.66590881347656,
      "logps/ref_chosen": -87.4644546508789,
      "logps/ref_rejected": -119.12409973144531,
      "logps/rejected": -134.19381713867188,
      "loss": 1.1105,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.3268200755119324,
      "rewards/margins": 0.45816469192504883,
      "rewards/rejected": -0.7849847674369812,
      "step": 256
    },
    {
      "epoch": 0.3885109599395314,
      "epsilon_dpo/beta": 0.05199351906776428,
      "epsilon_dpo/beta_margin_grad_mean": -0.42260271310806274,
      "epsilon_dpo/beta_margin_grad_std": 0.1644478142261505,
      "epsilon_dpo/beta_margin_mean": 0.3578561544418335,
      "epsilon_dpo/beta_margin_std": 0.7562159895896912,
      "epsilon_dpo/loss_margin_mean": 6.97858190536499,
      "grad_norm": 15.032943725585938,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.05220038443803787,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.851602043638994e-07,
      "logits/chosen": -0.33761709928512573,
      "logits/rejected": -0.21985182166099548,
      "logps/chosen": -104.91004943847656,
      "logps/ref_chosen": -96.47090148925781,
      "logps/ref_rejected": -131.09390258789062,
      "logps/rejected": -146.5116424560547,
      "loss": 1.1908,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.4412100315093994,
      "rewards/margins": 0.3578561544418335,
      "rewards/rejected": -0.7990661859512329,
      "step": 257
    },
    {
      "epoch": 0.3900226757369615,
      "epsilon_dpo/beta": 0.051734402775764465,
      "epsilon_dpo/beta_margin_grad_mean": -0.4043444097042084,
      "epsilon_dpo/beta_margin_grad_std": 0.14780445396900177,
      "epsilon_dpo/beta_margin_mean": 0.42923951148986816,
      "epsilon_dpo/beta_margin_std": 0.6817684769630432,
      "epsilon_dpo/loss_margin_mean": 8.37546443939209,
      "grad_norm": 12.723770141601562,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.051989179104566574,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.840459976743023e-07,
      "logits/chosen": -0.3793967664241791,
      "logits/rejected": -0.5327560305595398,
      "logps/chosen": -90.00956726074219,
      "logps/ref_chosen": -82.86605834960938,
      "logps/ref_rejected": -104.7901611328125,
      "logps/rejected": -120.30911254882812,
      "loss": 1.108,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.37101608514785767,
      "rewards/margins": 0.42923951148986816,
      "rewards/rejected": -0.800255537033081,
      "step": 258
    },
    {
      "epoch": 0.3915343915343915,
      "epsilon_dpo/beta": 0.05148502439260483,
      "epsilon_dpo/beta_margin_grad_mean": -0.3661709129810333,
      "epsilon_dpo/beta_margin_grad_std": 0.1684502214193344,
      "epsilon_dpo/beta_margin_mean": 0.6204766035079956,
      "epsilon_dpo/beta_margin_std": 0.8014397025108337,
      "epsilon_dpo/loss_margin_mean": 12.153839111328125,
      "grad_norm": 13.267675399780273,
      "kl/avg_steps": 0.484375,
      "kl/beta": 0.051730524748563766,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.8292804142999796e-07,
      "logits/chosen": -0.2220759093761444,
      "logits/rejected": -0.213998943567276,
      "logps/chosen": -90.79042053222656,
      "logps/ref_chosen": -87.27658081054688,
      "logps/ref_rejected": -124.00669860839844,
      "logps/rejected": -139.67437744140625,
      "loss": 1.0011,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.18326766788959503,
      "rewards/margins": 0.6204766035079956,
      "rewards/rejected": -0.8037442564964294,
      "step": 259
    },
    {
      "epoch": 0.3930461073318216,
      "epsilon_dpo/beta": 0.05129323527216911,
      "epsilon_dpo/beta_margin_grad_mean": -0.41950494050979614,
      "epsilon_dpo/beta_margin_grad_std": 0.18360477685928345,
      "epsilon_dpo/beta_margin_mean": 0.36882904171943665,
      "epsilon_dpo/beta_margin_std": 0.8418011665344238,
      "epsilon_dpo/loss_margin_mean": 7.304553508758545,
      "grad_norm": 17.972858428955078,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.05148116126656532,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.818063669026256e-07,
      "logits/chosen": -0.2347155511379242,
      "logits/rejected": -0.4262591600418091,
      "logps/chosen": -75.84223937988281,
      "logps/ref_chosen": -69.40695190429688,
      "logps/ref_rejected": -104.71173858642578,
      "logps/rejected": -118.45158386230469,
      "loss": 1.2141,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.3316202461719513,
      "rewards/margins": 0.36882901191711426,
      "rewards/rejected": -0.7004492282867432,
      "step": 260
    },
    {
      "epoch": 0.3945578231292517,
      "epsilon_dpo/beta": 0.05108557641506195,
      "epsilon_dpo/beta_margin_grad_mean": -0.43297892808914185,
      "epsilon_dpo/beta_margin_grad_std": 0.16156414151191711,
      "epsilon_dpo/beta_margin_mean": 0.3018457889556885,
      "epsilon_dpo/beta_margin_std": 0.7393701076507568,
      "epsilon_dpo/loss_margin_mean": 5.996203899383545,
      "grad_norm": 15.106315612792969,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.05128882825374603,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.806810054678331e-07,
      "logits/chosen": -0.23062291741371155,
      "logits/rejected": -0.158943310379982,
      "logps/chosen": -101.55908203125,
      "logps/ref_chosen": -96.14897918701172,
      "logps/ref_rejected": -89.86795043945312,
      "logps/rejected": -101.27426147460938,
      "loss": 1.2331,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.27909642457962036,
      "rewards/margins": 0.3018457889556885,
      "rewards/rejected": -0.5809422135353088,
      "step": 261
    },
    {
      "epoch": 0.3960695389266818,
      "epsilon_dpo/beta": 0.050894845277071,
      "epsilon_dpo/beta_margin_grad_mean": -0.41639986634254456,
      "epsilon_dpo/beta_margin_grad_std": 0.14282996952533722,
      "epsilon_dpo/beta_margin_mean": 0.3772861957550049,
      "epsilon_dpo/beta_margin_std": 0.6464824080467224,
      "epsilon_dpo/loss_margin_mean": 7.49705696105957,
      "grad_norm": 13.404562950134277,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.05108131095767021,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.7955198860439887e-07,
      "logits/chosen": -0.32994481921195984,
      "logits/rejected": -0.48363634943962097,
      "logps/chosen": -96.72879028320312,
      "logps/ref_chosen": -93.10110473632812,
      "logps/ref_rejected": -114.30091094970703,
      "logps/rejected": -125.42564392089844,
      "loss": 1.1403,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.18662017583847046,
      "rewards/margins": 0.3772861361503601,
      "rewards/rejected": -0.5639063119888306,
      "step": 262
    },
    {
      "epoch": 0.3975812547241119,
      "epsilon_dpo/beta": 0.050704702734947205,
      "epsilon_dpo/beta_margin_grad_mean": -0.40215593576431274,
      "epsilon_dpo/beta_margin_grad_std": 0.1577272117137909,
      "epsilon_dpo/beta_margin_mean": 0.4436055123806,
      "epsilon_dpo/beta_margin_std": 0.7195088863372803,
      "epsilon_dpo/loss_margin_mean": 8.844717025756836,
      "grad_norm": 12.462885856628418,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.050890471786260605,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.784193478933516e-07,
      "logits/chosen": 0.021105842664837837,
      "logits/rejected": -0.2684558629989624,
      "logps/chosen": -80.36698913574219,
      "logps/ref_chosen": -78.04122924804688,
      "logps/ref_rejected": -112.9266357421875,
      "logps/rejected": -124.09712219238281,
      "loss": 1.1095,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.12003128975629807,
      "rewards/margins": 0.4436054825782776,
      "rewards/rejected": -0.5636367797851562,
      "step": 263
    },
    {
      "epoch": 0.39909297052154197,
      "epsilon_dpo/beta": 0.05043604597449303,
      "epsilon_dpo/beta_margin_grad_mean": -0.40311023592948914,
      "epsilon_dpo/beta_margin_grad_std": 0.1410197913646698,
      "epsilon_dpo/beta_margin_mean": 0.44368380308151245,
      "epsilon_dpo/beta_margin_std": 0.6663060784339905,
      "epsilon_dpo/loss_margin_mean": 8.859919548034668,
      "grad_norm": 13.311902046203613,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.0507003478705883,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 3.7728311501708674e-07,
      "logits/chosen": -0.061237186193466187,
      "logits/rejected": -0.34207794070243835,
      "logps/chosen": -115.7286376953125,
      "logps/ref_chosen": -110.40716552734375,
      "logps/ref_rejected": -123.06625366210938,
      "logps/rejected": -137.24765014648438,
      "loss": 1.0897,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.2699325680732727,
      "rewards/margins": 0.44368380308151245,
      "rewards/rejected": -0.7136163711547852,
      "step": 264
    },
    {
      "epoch": 0.40060468631897206,
      "epsilon_dpo/beta": 0.05024832859635353,
      "epsilon_dpo/beta_margin_grad_mean": -0.37590134143829346,
      "epsilon_dpo/beta_margin_grad_std": 0.17293144762516022,
      "epsilon_dpo/beta_margin_mean": 0.577387273311615,
      "epsilon_dpo/beta_margin_std": 0.7995150685310364,
      "epsilon_dpo/loss_margin_mean": 11.604867935180664,
      "grad_norm": 11.446062088012695,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.05043242499232292,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.7614332175848027e-07,
      "logits/chosen": -0.3174809217453003,
      "logits/rejected": -0.14214938879013062,
      "logps/chosen": -71.36831665039062,
      "logps/ref_chosen": -69.86323547363281,
      "logps/ref_rejected": -96.0023422241211,
      "logps/rejected": -109.11228942871094,
      "loss": 1.034,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.07832078635692596,
      "rewards/margins": 0.577387273311615,
      "rewards/rejected": -0.6557080745697021,
      "step": 265
    },
    {
      "epoch": 0.4021164021164021,
      "epsilon_dpo/beta": 0.05004489794373512,
      "epsilon_dpo/beta_margin_grad_mean": -0.4076755940914154,
      "epsilon_dpo/beta_margin_grad_std": 0.14683128893375397,
      "epsilon_dpo/beta_margin_mean": 0.4213828146457672,
      "epsilon_dpo/beta_margin_std": 0.6990775465965271,
      "epsilon_dpo/loss_margin_mean": 8.499797821044922,
      "grad_norm": 14.631038665771484,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.050244010984897614,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.75e-07,
      "logits/chosen": -0.42971134185791016,
      "logits/rejected": -0.3967766761779785,
      "logps/chosen": -82.61585998535156,
      "logps/ref_chosen": -80.44645690917969,
      "logps/ref_rejected": -106.86441040039062,
      "logps/rejected": -117.53360748291016,
      "loss": 1.1174,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.11023129522800446,
      "rewards/margins": 0.4213827848434448,
      "rewards/rejected": -0.5316140651702881,
      "step": 266
    },
    {
      "epoch": 0.4036281179138322,
      "epsilon_dpo/beta": 0.049842409789562225,
      "epsilon_dpo/beta_margin_grad_mean": -0.4204915165901184,
      "epsilon_dpo/beta_margin_grad_std": 0.15187732875347137,
      "epsilon_dpo/beta_margin_mean": 0.34613466262817383,
      "epsilon_dpo/beta_margin_std": 0.6853890419006348,
      "epsilon_dpo/loss_margin_mean": 7.032593250274658,
      "grad_norm": 11.797558784484863,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.05004071816802025,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.738531817228131e-07,
      "logits/chosen": -0.24386508762836456,
      "logits/rejected": -0.20387592911720276,
      "logps/chosen": -79.5445785522461,
      "logps/ref_chosen": -78.45423889160156,
      "logps/ref_rejected": -87.34706115722656,
      "logps/rejected": -95.46998596191406,
      "loss": 1.179,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.0566682331264019,
      "rewards/margins": 0.34613466262817383,
      "rewards/rejected": -0.40280288457870483,
      "step": 267
    },
    {
      "epoch": 0.4051398337112623,
      "epsilon_dpo/beta": 0.04979650676250458,
      "epsilon_dpo/beta_margin_grad_mean": -0.44863420724868774,
      "epsilon_dpo/beta_margin_grad_std": 0.15428493916988373,
      "epsilon_dpo/beta_margin_mean": 0.24120254814624786,
      "epsilon_dpo/beta_margin_std": 0.7026641368865967,
      "epsilon_dpo/loss_margin_mean": 4.94456672668457,
      "grad_norm": 13.300025939941406,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.04983825236558914,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 3.7270289900589204e-07,
      "logits/chosen": -0.17837348580360413,
      "logits/rejected": -0.3197210431098938,
      "logps/chosen": -88.37873840332031,
      "logps/ref_chosen": -86.03016662597656,
      "logps/ref_rejected": -93.82392883300781,
      "logps/rejected": -101.1170654296875,
      "loss": 1.2736,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.11823678016662598,
      "rewards/margins": 0.24120256304740906,
      "rewards/rejected": -0.35943934321403503,
      "step": 268
    },
    {
      "epoch": 0.40665154950869237,
      "epsilon_dpo/beta": 0.049594249576330185,
      "epsilon_dpo/beta_margin_grad_mean": -0.4113570749759674,
      "epsilon_dpo/beta_margin_grad_std": 0.15016913414001465,
      "epsilon_dpo/beta_margin_mean": 0.3949938714504242,
      "epsilon_dpo/beta_margin_std": 0.6843596696853638,
      "epsilon_dpo/loss_margin_mean": 8.056325912475586,
      "grad_norm": 13.268834114074707,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.049791570752859116,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.7154918402511714e-07,
      "logits/chosen": -0.5032898783683777,
      "logits/rejected": -0.3996407091617584,
      "logps/chosen": -92.66567993164062,
      "logps/ref_chosen": -89.21455383300781,
      "logps/ref_rejected": -108.29411315917969,
      "logps/rejected": -119.80156707763672,
      "loss": 1.1374,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.17236913740634918,
      "rewards/margins": 0.3949939012527466,
      "rewards/rejected": -0.5673630237579346,
      "step": 269
    },
    {
      "epoch": 0.40816326530612246,
      "epsilon_dpo/beta": 0.049409087747335434,
      "epsilon_dpo/beta_margin_grad_mean": -0.41228729486465454,
      "epsilon_dpo/beta_margin_grad_std": 0.15867401659488678,
      "epsilon_dpo/beta_margin_mean": 0.39783310890197754,
      "epsilon_dpo/beta_margin_std": 0.7360899448394775,
      "epsilon_dpo/loss_margin_mean": 8.144404411315918,
      "grad_norm": 14.946235656738281,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.049590110778808594,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.7039206905237656e-07,
      "logits/chosen": -0.25615566968917847,
      "logits/rejected": -0.4289062023162842,
      "logps/chosen": -92.39274597167969,
      "logps/ref_chosen": -90.55712890625,
      "logps/ref_rejected": -115.80068969726562,
      "logps/rejected": -125.78071594238281,
      "loss": 1.1507,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.09273286163806915,
      "rewards/margins": 0.3978331387042999,
      "rewards/rejected": -0.4905660152435303,
      "step": 270
    },
    {
      "epoch": 0.40967498110355255,
      "epsilon_dpo/beta": 0.04927081987261772,
      "epsilon_dpo/beta_margin_grad_mean": -0.43657374382019043,
      "epsilon_dpo/beta_margin_grad_std": 0.1702788919210434,
      "epsilon_dpo/beta_margin_mean": 0.31171828508377075,
      "epsilon_dpo/beta_margin_std": 0.8074589371681213,
      "epsilon_dpo/loss_margin_mean": 6.436391830444336,
      "grad_norm": 14.009123802185059,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.049404844641685486,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 3.692315864546635e-07,
      "logits/chosen": -0.6964402198791504,
      "logits/rejected": -0.5896154046058655,
      "logps/chosen": -96.97105407714844,
      "logps/ref_chosen": -93.89132690429688,
      "logps/ref_rejected": -117.50479888916016,
      "logps/rejected": -127.02091979980469,
      "loss": 1.2445,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.15434423089027405,
      "rewards/margins": 0.3117183446884155,
      "rewards/rejected": -0.4660625457763672,
      "step": 271
    },
    {
      "epoch": 0.41118669690098264,
      "epsilon_dpo/beta": 0.04907866567373276,
      "epsilon_dpo/beta_margin_grad_mean": -0.3888590931892395,
      "epsilon_dpo/beta_margin_grad_std": 0.1553007811307907,
      "epsilon_dpo/beta_margin_mean": 0.5047463178634644,
      "epsilon_dpo/beta_margin_std": 0.733122706413269,
      "epsilon_dpo/loss_margin_mean": 10.382523536682129,
      "grad_norm": 11.808626174926758,
      "kl/avg_steps": 0.390625,
      "kl/beta": 0.049266282469034195,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.6806776869317067e-07,
      "logits/chosen": -0.2755691111087799,
      "logits/rejected": -0.2921621799468994,
      "logps/chosen": -94.78904724121094,
      "logps/ref_chosen": -92.47999572753906,
      "logps/ref_rejected": -88.89102935791016,
      "logps/rejected": -101.58261108398438,
      "loss": 1.0642,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.11534780263900757,
      "rewards/margins": 0.5047463774681091,
      "rewards/rejected": -0.6200941801071167,
      "step": 272
    },
    {
      "epoch": 0.4126984126984127,
      "epsilon_dpo/beta": 0.04888010397553444,
      "epsilon_dpo/beta_margin_grad_mean": -0.40970584750175476,
      "epsilon_dpo/beta_margin_grad_std": 0.18216586112976074,
      "epsilon_dpo/beta_margin_mean": 0.4161832630634308,
      "epsilon_dpo/beta_margin_std": 0.8531363010406494,
      "epsilon_dpo/loss_margin_mean": 8.62950611114502,
      "grad_norm": 13.147982597351074,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.04907458275556564,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.669006483223828e-07,
      "logits/chosen": -0.46416640281677246,
      "logits/rejected": -0.3967297077178955,
      "logps/chosen": -85.8084716796875,
      "logps/ref_chosen": -81.18894958496094,
      "logps/ref_rejected": -108.72845458984375,
      "logps/rejected": -121.97748565673828,
      "loss": 1.1775,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.22839263081550598,
      "rewards/margins": 0.4161832332611084,
      "rewards/rejected": -0.644575834274292,
      "step": 273
    },
    {
      "epoch": 0.41421012849584277,
      "epsilon_dpo/beta": 0.048682332038879395,
      "epsilon_dpo/beta_margin_grad_mean": -0.3969515562057495,
      "epsilon_dpo/beta_margin_grad_std": 0.17456351220607758,
      "epsilon_dpo/beta_margin_mean": 0.48041605949401855,
      "epsilon_dpo/beta_margin_std": 0.8032354712486267,
      "epsilon_dpo/loss_margin_mean": 9.981647491455078,
      "grad_norm": 13.131810188293457,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.04887602478265762,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.657302579891656e-07,
      "logits/chosen": -0.1510474979877472,
      "logits/rejected": 0.027453720569610596,
      "logps/chosen": -89.0546875,
      "logps/ref_chosen": -85.10057067871094,
      "logps/ref_rejected": -93.13152313232422,
      "logps/rejected": -107.06729125976562,
      "loss": 1.1088,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.19507646560668945,
      "rewards/margins": 0.4804159998893738,
      "rewards/rejected": -0.675492525100708,
      "step": 274
    },
    {
      "epoch": 0.41572184429327286,
      "epsilon_dpo/beta": 0.04848536476492882,
      "epsilon_dpo/beta_margin_grad_mean": -0.39388585090637207,
      "epsilon_dpo/beta_margin_grad_std": 0.16643884778022766,
      "epsilon_dpo/beta_margin_mean": 0.4981761872768402,
      "epsilon_dpo/beta_margin_std": 0.779217004776001,
      "epsilon_dpo/loss_margin_mean": 10.376490592956543,
      "grad_norm": 13.791189193725586,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.04867827147245407,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.645566304318526e-07,
      "logits/chosen": -0.30934983491897583,
      "logits/rejected": -0.3148757815361023,
      "logps/chosen": -79.164794921875,
      "logps/ref_chosen": -76.37564849853516,
      "logps/ref_rejected": -106.13626098632812,
      "logps/rejected": -119.30189514160156,
      "loss": 1.0848,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.13688188791275024,
      "rewards/margins": 0.4981761574745178,
      "rewards/rejected": -0.6350580453872681,
      "step": 275
    },
    {
      "epoch": 0.41723356009070295,
      "epsilon_dpo/beta": 0.04830434173345566,
      "epsilon_dpo/beta_margin_grad_mean": -0.3967646062374115,
      "epsilon_dpo/beta_margin_grad_std": 0.16201792657375336,
      "epsilon_dpo/beta_margin_mean": 0.4819945991039276,
      "epsilon_dpo/beta_margin_std": 0.759510338306427,
      "epsilon_dpo/loss_margin_mean": 10.081770896911621,
      "grad_norm": 14.872968673706055,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.04848131537437439,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.633797984793294e-07,
      "logits/chosen": -0.2909401059150696,
      "logits/rejected": -0.37871038913726807,
      "logps/chosen": -77.63218688964844,
      "logps/ref_chosen": -75.13729858398438,
      "logps/ref_rejected": -84.94955444335938,
      "logps/rejected": -97.52620697021484,
      "loss": 1.0904,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.12274863570928574,
      "rewards/margins": 0.48199462890625,
      "rewards/rejected": -0.604743242263794,
      "step": 276
    },
    {
      "epoch": 0.41874527588813304,
      "epsilon_dpo/beta": 0.04825973138213158,
      "epsilon_dpo/beta_margin_grad_mean": -0.45563802123069763,
      "epsilon_dpo/beta_margin_grad_std": 0.16357764601707458,
      "epsilon_dpo/beta_margin_mean": 0.20618519186973572,
      "epsilon_dpo/beta_margin_std": 0.7475130558013916,
      "epsilon_dpo/loss_margin_mean": 4.387637138366699,
      "grad_norm": 15.38239574432373,
      "kl/avg_steps": 0.09375,
      "kl/beta": 0.04830018803477287,
      "kl/n_epsilon_steps": 0.453125,
      "kl/p_epsilon_steps": 0.546875,
      "learning_rate": 3.6219979505011555e-07,
      "logits/chosen": -0.44621503353118896,
      "logits/rejected": -0.3698129653930664,
      "logps/chosen": -106.39254760742188,
      "logps/ref_chosen": -100.04359436035156,
      "logps/ref_rejected": -93.42616271972656,
      "logps/rejected": -104.16275024414062,
      "loss": 1.3199,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.3093889653682709,
      "rewards/margins": 0.2061852216720581,
      "rewards/rejected": -0.5155741572380066,
      "step": 277
    },
    {
      "epoch": 0.42025699168556313,
      "epsilon_dpo/beta": 0.04813912510871887,
      "epsilon_dpo/beta_margin_grad_mean": -0.4347180724143982,
      "epsilon_dpo/beta_margin_grad_std": 0.16170021891593933,
      "epsilon_dpo/beta_margin_mean": 0.2943073511123657,
      "epsilon_dpo/beta_margin_std": 0.7458164691925049,
      "epsilon_dpo/loss_margin_mean": 6.219837665557861,
      "grad_norm": 20.242382049560547,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.04825494810938835,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 3.6101665315144353e-07,
      "logits/chosen": -0.3930334448814392,
      "logits/rejected": -0.5034379363059998,
      "logps/chosen": -99.878173828125,
      "logps/ref_chosen": -93.06146240234375,
      "logps/ref_rejected": -109.71893310546875,
      "logps/rejected": -122.75547790527344,
      "loss": 1.2412,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.33152052760124207,
      "rewards/margins": 0.2943073511123657,
      "rewards/rejected": -0.6258278489112854,
      "step": 278
    },
    {
      "epoch": 0.4217687074829932,
      "epsilon_dpo/beta": 0.047846004366874695,
      "epsilon_dpo/beta_margin_grad_mean": -0.36673232913017273,
      "epsilon_dpo/beta_margin_grad_std": 0.15627476572990417,
      "epsilon_dpo/beta_margin_mean": 0.6086516976356506,
      "epsilon_dpo/beta_margin_std": 0.7403861880302429,
      "epsilon_dpo/loss_margin_mean": 12.797127723693848,
      "grad_norm": 15.089401245117188,
      "kl/avg_steps": 0.609375,
      "kl/beta": 0.048134613782167435,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 3.5983040587833563e-07,
      "logits/chosen": -0.570505678653717,
      "logits/rejected": -0.41417115926742554,
      "logps/chosen": -75.94297790527344,
      "logps/ref_chosen": -76.2708740234375,
      "logps/ref_rejected": -91.68028259277344,
      "logps/rejected": -104.14952087402344,
      "loss": 0.9893,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": 0.014286134392023087,
      "rewards/margins": 0.6086516976356506,
      "rewards/rejected": -0.5943655371665955,
      "step": 279
    },
    {
      "epoch": 0.42328042328042326,
      "epsilon_dpo/beta": 0.0475488044321537,
      "epsilon_dpo/beta_margin_grad_mean": -0.3657313883304596,
      "epsilon_dpo/beta_margin_grad_std": 0.15367664396762848,
      "epsilon_dpo/beta_margin_mean": 0.6205738186836243,
      "epsilon_dpo/beta_margin_std": 0.7367238998413086,
      "epsilon_dpo/loss_margin_mean": 13.125807762145996,
      "grad_norm": 11.988062858581543,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.04784306883811951,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 3.586410864126781e-07,
      "logits/chosen": -0.3317842483520508,
      "logits/rejected": -0.27062520384788513,
      "logps/chosen": -84.26651000976562,
      "logps/ref_chosen": -81.00099182128906,
      "logps/ref_rejected": -101.10507202148438,
      "logps/rejected": -117.49641418457031,
      "loss": 0.9782,
      "rewards/accuracies": 0.875,
      "rewards/chosen": -0.15623867511749268,
      "rewards/margins": 0.6205738186836243,
      "rewards/rejected": -0.7768125534057617,
      "step": 280
    },
    {
      "epoch": 0.42479213907785335,
      "epsilon_dpo/beta": 0.04734262824058533,
      "epsilon_dpo/beta_margin_grad_mean": -0.39394766092300415,
      "epsilon_dpo/beta_margin_grad_std": 0.1553352326154709,
      "epsilon_dpo/beta_margin_mean": 0.48012128472328186,
      "epsilon_dpo/beta_margin_std": 0.7120420336723328,
      "epsilon_dpo/loss_margin_mean": 10.240545272827148,
      "grad_norm": 10.852984428405762,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.0475459061563015,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.574487280222929e-07,
      "logits/chosen": -0.39467182755470276,
      "logits/rejected": -0.2560708224773407,
      "logps/chosen": -85.82388305664062,
      "logps/ref_chosen": -82.56924438476562,
      "logps/ref_rejected": -83.29329681396484,
      "logps/rejected": -96.78848266601562,
      "loss": 1.0782,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.15683114528656006,
      "rewards/margins": 0.48012131452560425,
      "rewards/rejected": -0.6369524598121643,
      "step": 281
    },
    {
      "epoch": 0.42630385487528344,
      "epsilon_dpo/beta": 0.04717332124710083,
      "epsilon_dpo/beta_margin_grad_mean": -0.3915327191352844,
      "epsilon_dpo/beta_margin_grad_std": 0.17149022221565247,
      "epsilon_dpo/beta_margin_mean": 0.5186943411827087,
      "epsilon_dpo/beta_margin_std": 0.8207116723060608,
      "epsilon_dpo/loss_margin_mean": 11.112959861755371,
      "grad_norm": 12.8870210647583,
      "kl/avg_steps": 0.359375,
      "kl/beta": 0.04733880236744881,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.562533640600075e-07,
      "logits/chosen": -0.5010318756103516,
      "logits/rejected": -0.29100731015205383,
      "logps/chosen": -89.39044189453125,
      "logps/ref_chosen": -84.25967407226562,
      "logps/ref_rejected": -91.84246826171875,
      "logps/rejected": -108.08619689941406,
      "loss": 1.0816,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.24529683589935303,
      "rewards/margins": 0.5186943411827087,
      "rewards/rejected": -0.7639911770820618,
      "step": 282
    },
    {
      "epoch": 0.42781557067271353,
      "epsilon_dpo/beta": 0.04699710011482239,
      "epsilon_dpo/beta_margin_grad_mean": -0.4094040095806122,
      "epsilon_dpo/beta_margin_grad_std": 0.1825781762599945,
      "epsilon_dpo/beta_margin_mean": 0.4076806604862213,
      "epsilon_dpo/beta_margin_std": 0.8461658954620361,
      "epsilon_dpo/loss_margin_mean": 8.805419921875,
      "grad_norm": 13.878371238708496,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.047169286757707596,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.550550279627215e-07,
      "logits/chosen": -0.17757020890712738,
      "logits/rejected": -0.5387111306190491,
      "logps/chosen": -96.1683120727539,
      "logps/ref_chosen": -87.76092529296875,
      "logps/ref_rejected": -123.7965087890625,
      "logps/rejected": -141.00930786132812,
      "loss": 1.1833,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.3980576992034912,
      "rewards/margins": 0.4076806604862213,
      "rewards/rejected": -0.8057383894920349,
      "step": 283
    },
    {
      "epoch": 0.4293272864701436,
      "epsilon_dpo/beta": 0.04679214581847191,
      "epsilon_dpo/beta_margin_grad_mean": -0.39326760172843933,
      "epsilon_dpo/beta_margin_grad_std": 0.15346704423427582,
      "epsilon_dpo/beta_margin_mean": 0.4926775395870209,
      "epsilon_dpo/beta_margin_std": 0.7348820567131042,
      "epsilon_dpo/loss_margin_mean": 10.620450973510742,
      "grad_norm": 12.785466194152832,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.046993061900138855,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.5385375325047163e-07,
      "logits/chosen": -0.2667410373687744,
      "logits/rejected": -0.15663591027259827,
      "logps/chosen": -87.38343811035156,
      "logps/ref_chosen": -81.22230529785156,
      "logps/ref_rejected": -111.91586303710938,
      "logps/rejected": -128.69744873046875,
      "loss": 1.0723,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.28914564847946167,
      "rewards/margins": 0.4926775097846985,
      "rewards/rejected": -0.7818231582641602,
      "step": 284
    },
    {
      "epoch": 0.4308390022675737,
      "epsilon_dpo/beta": 0.046661436557769775,
      "epsilon_dpo/beta_margin_grad_mean": -0.43417391180992126,
      "epsilon_dpo/beta_margin_grad_std": 0.17843972146511078,
      "epsilon_dpo/beta_margin_mean": 0.311433881521225,
      "epsilon_dpo/beta_margin_std": 0.8369320034980774,
      "epsilon_dpo/loss_margin_mean": 6.8021392822265625,
      "grad_norm": 13.3861722946167,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.04678836464881897,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 3.5264957352549375e-07,
      "logits/chosen": -0.33967140316963196,
      "logits/rejected": -0.2970946431159973,
      "logps/chosen": -104.98358154296875,
      "logps/ref_chosen": -92.5496597290039,
      "logps/ref_rejected": -102.47920227050781,
      "logps/rejected": -121.71526336669922,
      "loss": 1.2579,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.5837230086326599,
      "rewards/margins": 0.31143394112586975,
      "rewards/rejected": -0.895156979560852,
      "step": 285
    },
    {
      "epoch": 0.4323507180650038,
      "epsilon_dpo/beta": 0.04642849788069725,
      "epsilon_dpo/beta_margin_grad_mean": -0.3862251043319702,
      "epsilon_dpo/beta_margin_grad_std": 0.1750505268573761,
      "epsilon_dpo/beta_margin_mean": 0.532291054725647,
      "epsilon_dpo/beta_margin_std": 0.8362828493118286,
      "epsilon_dpo/loss_margin_mean": 11.574300765991211,
      "grad_norm": 12.411300659179688,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.04665714129805565,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.514425224712835e-07,
      "logits/chosen": -0.36736541986465454,
      "logits/rejected": -0.30413326621055603,
      "logps/chosen": -93.0548324584961,
      "logps/ref_chosen": -83.22084045410156,
      "logps/ref_rejected": -118.17338562011719,
      "logps/rejected": -139.58168029785156,
      "loss": 1.0783,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.4587008059024811,
      "rewards/margins": 0.532291054725647,
      "rewards/rejected": -0.9909918308258057,
      "step": 286
    },
    {
      "epoch": 0.43386243386243384,
      "epsilon_dpo/beta": 0.04615398123860359,
      "epsilon_dpo/beta_margin_grad_mean": -0.3491361737251282,
      "epsilon_dpo/beta_margin_grad_std": 0.1557650864124298,
      "epsilon_dpo/beta_margin_mean": 0.7086184024810791,
      "epsilon_dpo/beta_margin_std": 0.7686328291893005,
      "epsilon_dpo/loss_margin_mean": 15.439092636108398,
      "grad_norm": 11.449210166931152,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.04642501473426819,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 3.502326338516534e-07,
      "logits/chosen": -0.6256189942359924,
      "logits/rejected": -0.15066683292388916,
      "logps/chosen": -73.1468734741211,
      "logps/ref_chosen": -67.09947204589844,
      "logps/ref_rejected": -83.48188781738281,
      "logps/rejected": -104.9683837890625,
      "loss": 0.9251,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.2815502882003784,
      "rewards/margins": 0.7086184024810791,
      "rewards/rejected": -0.9901686906814575,
      "step": 287
    },
    {
      "epoch": 0.43537414965986393,
      "epsilon_dpo/beta": 0.04596810042858124,
      "epsilon_dpo/beta_margin_grad_mean": -0.4099636971950531,
      "epsilon_dpo/beta_margin_grad_std": 0.15979404747486115,
      "epsilon_dpo/beta_margin_mean": 0.41960883140563965,
      "epsilon_dpo/beta_margin_std": 0.7552804946899414,
      "epsilon_dpo/loss_margin_mean": 9.233772277832031,
      "grad_norm": 15.480899810791016,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.046150993555784225,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.490199415097892e-07,
      "logits/chosen": -0.570479154586792,
      "logits/rejected": -0.406840443611145,
      "logps/chosen": -101.13687133789062,
      "logps/ref_chosen": -89.84213256835938,
      "logps/ref_rejected": -113.61448669433594,
      "logps/rejected": -134.14300537109375,
      "loss": 1.1375,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5225342512130737,
      "rewards/margins": 0.41960883140563965,
      "rewards/rejected": -0.9421430826187134,
      "step": 288
    },
    {
      "epoch": 0.436885865457294,
      "epsilon_dpo/beta": 0.045825205743312836,
      "epsilon_dpo/beta_margin_grad_mean": -0.3875085413455963,
      "epsilon_dpo/beta_margin_grad_std": 0.15026584267616272,
      "epsilon_dpo/beta_margin_mean": 0.5227700471878052,
      "epsilon_dpo/beta_margin_std": 0.7179827094078064,
      "epsilon_dpo/loss_margin_mean": 11.50743293762207,
      "grad_norm": 10.489204406738281,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.04596426337957382,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 3.4780447936730247e-07,
      "logits/chosen": -0.22212985157966614,
      "logits/rejected": -0.39587292075157166,
      "logps/chosen": -88.09721374511719,
      "logps/ref_chosen": -78.14584350585938,
      "logps/ref_rejected": -93.63999938964844,
      "logps/rejected": -115.09880065917969,
      "loss": 1.0435,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.4589444398880005,
      "rewards/margins": 0.5227700471878052,
      "rewards/rejected": -0.9817144870758057,
      "step": 289
    },
    {
      "epoch": 0.4383975812547241,
      "epsilon_dpo/beta": 0.04562516510486603,
      "epsilon_dpo/beta_margin_grad_mean": -0.4058082401752472,
      "epsilon_dpo/beta_margin_grad_std": 0.15972179174423218,
      "epsilon_dpo/beta_margin_mean": 0.4322429895401001,
      "epsilon_dpo/beta_margin_std": 0.7467278242111206,
      "epsilon_dpo/loss_margin_mean": 9.5772705078125,
      "grad_norm": 12.687112808227539,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.04582107067108154,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.465862814232821e-07,
      "logits/chosen": -0.34867942333221436,
      "logits/rejected": -0.12253440916538239,
      "logps/chosen": -97.43099975585938,
      "logps/ref_chosen": -83.81318664550781,
      "logps/ref_rejected": -114.35005187988281,
      "logps/rejected": -137.54513549804688,
      "loss": 1.1257,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.6226211190223694,
      "rewards/margins": 0.4322430491447449,
      "rewards/rejected": -1.0548641681671143,
      "step": 290
    },
    {
      "epoch": 0.4399092970521542,
      "epsilon_dpo/beta": 0.04536939039826393,
      "epsilon_dpo/beta_margin_grad_mean": -0.38911837339401245,
      "epsilon_dpo/beta_margin_grad_std": 0.17643475532531738,
      "epsilon_dpo/beta_margin_mean": 0.5134809613227844,
      "epsilon_dpo/beta_margin_std": 0.8421981334686279,
      "epsilon_dpo/loss_margin_mean": 11.423192977905273,
      "grad_norm": 12.824395179748535,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.04562147706747055,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 3.4536538175334343e-07,
      "logits/chosen": -0.41977792978286743,
      "logits/rejected": -0.8858845233917236,
      "logps/chosen": -83.10356140136719,
      "logps/ref_chosen": -71.1761703491211,
      "logps/ref_rejected": -101.21935272216797,
      "logps/rejected": -124.5699462890625,
      "loss": 1.0952,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.5435177087783813,
      "rewards/margins": 0.5134809613227844,
      "rewards/rejected": -1.0569987297058105,
      "step": 291
    },
    {
      "epoch": 0.4414210128495843,
      "epsilon_dpo/beta": 0.045158155262470245,
      "epsilon_dpo/beta_margin_grad_mean": -0.40868300199508667,
      "epsilon_dpo/beta_margin_grad_std": 0.17304831743240356,
      "epsilon_dpo/beta_margin_mean": 0.41493427753448486,
      "epsilon_dpo/beta_margin_std": 0.8037520051002502,
      "epsilon_dpo/loss_margin_mean": 9.300040245056152,
      "grad_norm": 16.256256103515625,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.04536629468202591,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.4414181450867465e-07,
      "logits/chosen": -0.3706292510032654,
      "logits/rejected": -0.3315098285675049,
      "logps/chosen": -100.09800720214844,
      "logps/ref_chosen": -87.9404296875,
      "logps/ref_rejected": -103.1168212890625,
      "logps/rejected": -124.57444763183594,
      "loss": 1.1609,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5523610711097717,
      "rewards/margins": 0.4149342477321625,
      "rewards/rejected": -0.9672952890396118,
      "step": 292
    },
    {
      "epoch": 0.4429327286470144,
      "epsilon_dpo/beta": 0.04493334889411926,
      "epsilon_dpo/beta_margin_grad_mean": -0.3718346357345581,
      "epsilon_dpo/beta_margin_grad_std": 0.19872498512268066,
      "epsilon_dpo/beta_margin_mean": 0.6251580119132996,
      "epsilon_dpo/beta_margin_std": 0.9845532178878784,
      "epsilon_dpo/loss_margin_mean": 14.047744750976562,
      "grad_norm": 14.59776782989502,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.0451546311378479,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.4291561391508185e-07,
      "logits/chosen": -0.380871057510376,
      "logits/rejected": -0.24333512783050537,
      "logps/chosen": -88.27153015136719,
      "logps/ref_chosen": -77.02020263671875,
      "logps/ref_rejected": -111.15068054199219,
      "logps/rejected": -136.44973754882812,
      "loss": 1.0647,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5092015266418457,
      "rewards/margins": 0.6251579523086548,
      "rewards/rejected": -1.1343594789505005,
      "step": 293
    },
    {
      "epoch": 0.4444444444444444,
      "epsilon_dpo/beta": 0.044737886637449265,
      "epsilon_dpo/beta_margin_grad_mean": -0.41345012187957764,
      "epsilon_dpo/beta_margin_grad_std": 0.16614548861980438,
      "epsilon_dpo/beta_margin_mean": 0.39459964632987976,
      "epsilon_dpo/beta_margin_std": 0.7768252491950989,
      "epsilon_dpo/loss_margin_mean": 8.932480812072754,
      "grad_norm": 14.383130073547363,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.04492998123168945,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.4168681427203153e-07,
      "logits/chosen": -0.4731680750846863,
      "logits/rejected": -0.32575279474258423,
      "logps/chosen": -89.42033386230469,
      "logps/ref_chosen": -76.86295318603516,
      "logps/ref_rejected": -91.43938446044922,
      "logps/rejected": -112.92924499511719,
      "loss": 1.1666,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.5631007552146912,
      "rewards/margins": 0.3945996165275574,
      "rewards/rejected": -0.9577003717422485,
      "step": 294
    },
    {
      "epoch": 0.4459561602418745,
      "epsilon_dpo/beta": 0.044529028236866,
      "epsilon_dpo/beta_margin_grad_mean": -0.4238334894180298,
      "epsilon_dpo/beta_margin_grad_std": 0.1660003811120987,
      "epsilon_dpo/beta_margin_mean": 0.3427713215351105,
      "epsilon_dpo/beta_margin_std": 0.7789937257766724,
      "epsilon_dpo/loss_margin_mean": 7.807656288146973,
      "grad_norm": 19.5250186920166,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.04473426938056946,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.4045544995169125e-07,
      "logits/chosen": -0.4011607766151428,
      "logits/rejected": -0.5367324352264404,
      "logps/chosen": -83.89892578125,
      "logps/ref_chosen": -70.6540298461914,
      "logps/ref_rejected": -107.9378433227539,
      "logps/rejected": -128.99038696289062,
      "loss": 1.2102,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5908478498458862,
      "rewards/margins": 0.3427712917327881,
      "rewards/rejected": -0.9336191415786743,
      "step": 295
    },
    {
      "epoch": 0.4474678760393046,
      "epsilon_dpo/beta": 0.04436302185058594,
      "epsilon_dpo/beta_margin_grad_mean": -0.38691622018814087,
      "epsilon_dpo/beta_margin_grad_std": 0.17068080604076385,
      "epsilon_dpo/beta_margin_mean": 0.5330710411071777,
      "epsilon_dpo/beta_margin_std": 0.8052643537521362,
      "epsilon_dpo/loss_margin_mean": 12.143484115600586,
      "grad_norm": 13.374835968017578,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.04452555626630783,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.392215553979679e-07,
      "logits/chosen": -0.18664756417274475,
      "logits/rejected": -0.5290898084640503,
      "logps/chosen": -105.62068176269531,
      "logps/ref_chosen": -92.49901580810547,
      "logps/ref_rejected": -109.3675537109375,
      "logps/rejected": -134.63270568847656,
      "loss": 1.0668,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5853240489959717,
      "rewards/margins": 0.5330710411071777,
      "rewards/rejected": -1.1183950901031494,
      "step": 296
    },
    {
      "epoch": 0.4489795918367347,
      "epsilon_dpo/beta": 0.04411409795284271,
      "epsilon_dpo/beta_margin_grad_mean": -0.37734997272491455,
      "epsilon_dpo/beta_margin_grad_std": 0.15403449535369873,
      "epsilon_dpo/beta_margin_mean": 0.5672126412391663,
      "epsilon_dpo/beta_margin_std": 0.7445960640907288,
      "epsilon_dpo/loss_margin_mean": 12.947671890258789,
      "grad_norm": 13.148897171020508,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.04435920715332031,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 3.3798516512554485e-07,
      "logits/chosen": -0.3896023631095886,
      "logits/rejected": -0.3738701343536377,
      "logps/chosen": -91.46431732177734,
      "logps/ref_chosen": -78.97463989257812,
      "logps/ref_rejected": -96.7846908569336,
      "logps/rejected": -122.2220458984375,
      "loss": 1.0188,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.5526552796363831,
      "rewards/margins": 0.5672126412391663,
      "rewards/rejected": -1.1198678016662598,
      "step": 297
    },
    {
      "epoch": 0.4504913076341648,
      "epsilon_dpo/beta": 0.043908704072237015,
      "epsilon_dpo/beta_margin_grad_mean": -0.4079945385456085,
      "epsilon_dpo/beta_margin_grad_std": 0.1935083121061325,
      "epsilon_dpo/beta_margin_mean": 0.4454768896102905,
      "epsilon_dpo/beta_margin_std": 0.9295777678489685,
      "epsilon_dpo/loss_margin_mean": 10.276945114135742,
      "grad_norm": 20.633535385131836,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.04411108419299126,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.367463137189156e-07,
      "logits/chosen": -0.49798786640167236,
      "logits/rejected": -0.21161851286888123,
      "logps/chosen": -90.81427001953125,
      "logps/ref_chosen": -78.96235656738281,
      "logps/ref_rejected": -94.07938385009766,
      "logps/rejected": -116.20824432373047,
      "loss": 1.1815,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5214691162109375,
      "rewards/margins": 0.4454768896102905,
      "rewards/rejected": -0.966946005821228,
      "step": 298
    },
    {
      "epoch": 0.4520030234315949,
      "epsilon_dpo/beta": 0.043745007365942,
      "epsilon_dpo/beta_margin_grad_mean": -0.4169948995113373,
      "epsilon_dpo/beta_margin_grad_std": 0.1756003201007843,
      "epsilon_dpo/beta_margin_mean": 0.4042234420776367,
      "epsilon_dpo/beta_margin_std": 0.862417459487915,
      "epsilon_dpo/loss_margin_mean": 9.364120483398438,
      "grad_norm": 14.022722244262695,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.04390527680516243,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.355050358314172e-07,
      "logits/chosen": -0.2293621301651001,
      "logits/rejected": -0.3756565749645233,
      "logps/chosen": -94.15196228027344,
      "logps/ref_chosen": -81.70744323730469,
      "logps/ref_rejected": -91.36782836914062,
      "logps/rejected": -113.17647552490234,
      "loss": 1.1844,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.5472281575202942,
      "rewards/margins": 0.4042234718799591,
      "rewards/rejected": -0.9514516592025757,
      "step": 299
    },
    {
      "epoch": 0.45351473922902497,
      "epsilon_dpo/beta": 0.043595246970653534,
      "epsilon_dpo/beta_margin_grad_mean": -0.4183122515678406,
      "epsilon_dpo/beta_margin_grad_std": 0.19845078885555267,
      "epsilon_dpo/beta_margin_mean": 0.3808678686618805,
      "epsilon_dpo/beta_margin_std": 0.9299583435058594,
      "epsilon_dpo/loss_margin_mean": 8.885825157165527,
      "grad_norm": 13.588310241699219,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.04374124854803085,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.3426136618426043e-07,
      "logits/chosen": -0.32883864641189575,
      "logits/rejected": -0.4738979935646057,
      "logps/chosen": -95.8285140991211,
      "logps/ref_chosen": -84.14907836914062,
      "logps/ref_rejected": -103.7045669555664,
      "logps/rejected": -124.26982116699219,
      "loss": 1.2376,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5121716260910034,
      "rewards/margins": 0.3808678984642029,
      "rewards/rejected": -0.8930395245552063,
      "step": 300
    },
    {
      "epoch": 0.45351473922902497,
      "eval_epsilon_dpo/beta": 0.04342613369226456,
      "eval_epsilon_dpo/beta_margin_grad_mean": -0.40209266543388367,
      "eval_epsilon_dpo/beta_margin_grad_std": 0.17587903141975403,
      "eval_epsilon_dpo/beta_margin_mean": 0.46374091506004333,
      "eval_epsilon_dpo/beta_margin_std": 0.8474838137626648,
      "eval_epsilon_dpo/loss_margin_mean": 10.808184623718262,
      "eval_kl/n_epsilon_steps": 0.3050176203250885,
      "eval_kl/p_epsilon_steps": 0.6941021084785461,
      "eval_logits/chosen": -0.3509601354598999,
      "eval_logits/rejected": -0.42714956402778625,
      "eval_logps/chosen": -97.03380584716797,
      "eval_logps/ref_chosen": -87.42715454101562,
      "eval_logps/ref_rejected": -104.23548889160156,
      "eval_logps/rejected": -124.65033721923828,
      "eval_loss": 0.5699101686477661,
      "eval_rewards/accuracies": 0.7086267471313477,
      "eval_rewards/chosen": -0.41977250576019287,
      "eval_rewards/margins": 0.46374091506004333,
      "eval_rewards/rejected": -0.8835135102272034,
      "eval_runtime": 47.568,
      "eval_samples_per_second": 48.415,
      "eval_steps_per_second": 1.514,
      "step": 300
    },
    {
      "epoch": 0.455026455026455,
      "epsilon_dpo/beta": 0.043500397354364395,
      "epsilon_dpo/beta_margin_grad_mean": -0.4132575988769531,
      "epsilon_dpo/beta_margin_grad_std": 0.17967145144939423,
      "epsilon_dpo/beta_margin_mean": 0.42292046546936035,
      "epsilon_dpo/beta_margin_std": 0.8556511402130127,
      "epsilon_dpo/loss_margin_mean": 9.865974426269531,
      "grad_norm": 12.259976387023926,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.04359140247106552,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 3.3301533956555885e-07,
      "logits/chosen": -0.6459434628486633,
      "logits/rejected": -0.5689473748207092,
      "logps/chosen": -84.82774353027344,
      "logps/ref_chosen": -75.17066955566406,
      "logps/ref_rejected": -91.63156127929688,
      "logps/rejected": -111.15460968017578,
      "loss": 1.17,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.42266562581062317,
      "rewards/margins": 0.42292046546936035,
      "rewards/rejected": -0.8455861210823059,
      "step": 301
    },
    {
      "epoch": 0.4565381708238851,
      "epsilon_dpo/beta": 0.043323881924152374,
      "epsilon_dpo/beta_margin_grad_mean": -0.4404016435146332,
      "epsilon_dpo/beta_margin_grad_std": 0.1699916273355484,
      "epsilon_dpo/beta_margin_mean": 0.26643675565719604,
      "epsilon_dpo/beta_margin_std": 0.8003353476524353,
      "epsilon_dpo/loss_margin_mean": 6.27188777923584,
      "grad_norm": 15.321131706237793,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.043496254831552505,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.317669908293554e-07,
      "logits/chosen": -0.4655022919178009,
      "logits/rejected": -0.3340109884738922,
      "logps/chosen": -102.78459930419922,
      "logps/ref_chosen": -91.02297973632812,
      "logps/ref_rejected": -114.12149810791016,
      "logps/rejected": -132.15499877929688,
      "loss": 1.2832,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5123536586761475,
      "rewards/margins": 0.26643672585487366,
      "rewards/rejected": -0.7787904143333435,
      "step": 302
    },
    {
      "epoch": 0.4580498866213152,
      "epsilon_dpo/beta": 0.04314858838915825,
      "epsilon_dpo/beta_margin_grad_mean": -0.37916189432144165,
      "epsilon_dpo/beta_margin_grad_std": 0.19821646809577942,
      "epsilon_dpo/beta_margin_mean": 0.5926302671432495,
      "epsilon_dpo/beta_margin_std": 0.9654866456985474,
      "epsilon_dpo/loss_margin_mean": 13.885607719421387,
      "grad_norm": 12.133411407470703,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.04332026466727257,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.3051635489464793e-07,
      "logits/chosen": -0.25000911951065063,
      "logits/rejected": -0.08165319263935089,
      "logps/chosen": -98.95149993896484,
      "logps/ref_chosen": -89.47880554199219,
      "logps/ref_rejected": -114.87516784667969,
      "logps/rejected": -138.2334747314453,
      "loss": 1.0818,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.41259628534317017,
      "rewards/margins": 0.5926302671432495,
      "rewards/rejected": -1.0052266120910645,
      "step": 303
    },
    {
      "epoch": 0.4595616024187453,
      "epsilon_dpo/beta": 0.042906589806079865,
      "epsilon_dpo/beta_margin_grad_mean": -0.39030832052230835,
      "epsilon_dpo/beta_margin_grad_std": 0.14020994305610657,
      "epsilon_dpo/beta_margin_mean": 0.5041561722755432,
      "epsilon_dpo/beta_margin_std": 0.6786003708839417,
      "epsilon_dpo/loss_margin_mean": 11.824028015136719,
      "grad_norm": 11.328267097473145,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.043144989758729935,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 3.292634667444117e-07,
      "logits/chosen": -0.48182913661003113,
      "logits/rejected": -0.5360888242721558,
      "logps/chosen": -89.85015869140625,
      "logps/ref_chosen": -82.58224487304688,
      "logps/ref_rejected": -101.22938537597656,
      "logps/rejected": -120.32133483886719,
      "loss": 1.045,
      "rewards/accuracies": 0.84375,
      "rewards/chosen": -0.3128387928009033,
      "rewards/margins": 0.5041561722755432,
      "rewards/rejected": -0.8169949054718018,
      "step": 304
    },
    {
      "epoch": 0.46107331821617537,
      "epsilon_dpo/beta": 0.0427604503929615,
      "epsilon_dpo/beta_margin_grad_mean": -0.3991040587425232,
      "epsilon_dpo/beta_margin_grad_std": 0.19379828870296478,
      "epsilon_dpo/beta_margin_mean": 0.48293551802635193,
      "epsilon_dpo/beta_margin_std": 0.9312344193458557,
      "epsilon_dpo/loss_margin_mean": 11.445974349975586,
      "grad_norm": 12.086163520812988,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.042903654277324677,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.280083614246217e-07,
      "logits/chosen": -0.22488896548748016,
      "logits/rejected": -0.0361773818731308,
      "logps/chosen": -107.65516662597656,
      "logps/ref_chosen": -97.63542175292969,
      "logps/ref_rejected": -90.60706329345703,
      "logps/rejected": -112.07278442382812,
      "loss": 1.1529,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.43161648511886597,
      "rewards/margins": 0.48293548822402954,
      "rewards/rejected": -0.9145519733428955,
      "step": 305
    },
    {
      "epoch": 0.46258503401360546,
      "epsilon_dpo/beta": 0.0426006019115448,
      "epsilon_dpo/beta_margin_grad_mean": -0.39660578966140747,
      "epsilon_dpo/beta_margin_grad_std": 0.1713494211435318,
      "epsilon_dpo/beta_margin_mean": 0.49307981133461,
      "epsilon_dpo/beta_margin_std": 0.8368701934814453,
      "epsilon_dpo/loss_margin_mean": 11.6995210647583,
      "grad_norm": 12.074493408203125,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.04275668039917946,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.267510740432719e-07,
      "logits/chosen": -0.208884596824646,
      "logits/rejected": -0.4091324210166931,
      "logps/chosen": -82.65934753417969,
      "logps/ref_chosen": -76.59986877441406,
      "logps/ref_rejected": -108.35289764404297,
      "logps/rejected": -126.11189270019531,
      "loss": 1.1049,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.2597709894180298,
      "rewards/margins": 0.4930797517299652,
      "rewards/rejected": -0.7528507709503174,
      "step": 306
    },
    {
      "epoch": 0.46409674981103555,
      "epsilon_dpo/beta": 0.042601197957992554,
      "epsilon_dpo/beta_margin_grad_mean": -0.44173377752304077,
      "epsilon_dpo/beta_margin_grad_std": 0.1646379977464676,
      "epsilon_dpo/beta_margin_mean": 0.2777702510356903,
      "epsilon_dpo/beta_margin_std": 0.7612587213516235,
      "epsilon_dpo/loss_margin_mean": 6.658515930175781,
      "grad_norm": 13.114051818847656,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.04259693995118141,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 3.2549163976939285e-07,
      "logits/chosen": -0.3018755316734314,
      "logits/rejected": -0.5736774802207947,
      "logps/chosen": -84.22529602050781,
      "logps/ref_chosen": -78.99342346191406,
      "logps/ref_rejected": -91.41653442382812,
      "logps/rejected": -103.30691528320312,
      "loss": 1.2598,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.22581946849822998,
      "rewards/margins": 0.2777702510356903,
      "rewards/rejected": -0.5035896897315979,
      "step": 307
    },
    {
      "epoch": 0.4656084656084656,
      "epsilon_dpo/beta": 0.04244144633412361,
      "epsilon_dpo/beta_margin_grad_mean": -0.40081164240837097,
      "epsilon_dpo/beta_margin_grad_std": 0.17875270545482635,
      "epsilon_dpo/beta_margin_mean": 0.46681687235832214,
      "epsilon_dpo/beta_margin_std": 0.8536893725395203,
      "epsilon_dpo/loss_margin_mean": 11.135279655456543,
      "grad_norm": 15.095867156982422,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.04259693995118141,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.2423009383206874e-07,
      "logits/chosen": -0.7949353456497192,
      "logits/rejected": -0.33702436089515686,
      "logps/chosen": -100.32615661621094,
      "logps/ref_chosen": -94.20703887939453,
      "logps/ref_rejected": -103.01887512207031,
      "logps/rejected": -120.27326965332031,
      "loss": 1.1351,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.2623611092567444,
      "rewards/margins": 0.46681690216064453,
      "rewards/rejected": -0.7291780114173889,
      "step": 308
    },
    {
      "epoch": 0.4671201814058957,
      "epsilon_dpo/beta": 0.04219004511833191,
      "epsilon_dpo/beta_margin_grad_mean": -0.4012344479560852,
      "epsilon_dpo/beta_margin_grad_std": 0.1441354751586914,
      "epsilon_dpo/beta_margin_mean": 0.44948306679725647,
      "epsilon_dpo/beta_margin_std": 0.6808764934539795,
      "epsilon_dpo/loss_margin_mean": 10.729732513427734,
      "grad_norm": 12.998229026794434,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.04243779927492142,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 3.229664715194511e-07,
      "logits/chosen": -0.44665130972862244,
      "logits/rejected": -0.2758372128009796,
      "logps/chosen": -95.9297103881836,
      "logps/ref_chosen": -87.41177368164062,
      "logps/ref_rejected": -105.32408142089844,
      "logps/rejected": -124.57174682617188,
      "loss": 1.09,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.36134669184684753,
      "rewards/margins": 0.44948309659957886,
      "rewards/rejected": -0.810829758644104,
      "step": 309
    },
    {
      "epoch": 0.46863189720332576,
      "epsilon_dpo/beta": 0.04217834398150444,
      "epsilon_dpo/beta_margin_grad_mean": -0.4731263518333435,
      "epsilon_dpo/beta_margin_grad_std": 0.174700528383255,
      "epsilon_dpo/beta_margin_mean": 0.12875214219093323,
      "epsilon_dpo/beta_margin_std": 0.8044756054878235,
      "epsilon_dpo/loss_margin_mean": 3.1934990882873535,
      "grad_norm": 15.983020782470703,
      "kl/avg_steps": 0.03125,
      "kl/beta": 0.04218731075525284,
      "kl/n_epsilon_steps": 0.484375,
      "kl/p_epsilon_steps": 0.515625,
      "learning_rate": 3.2170080817777257e-07,
      "logits/chosen": -0.7636397480964661,
      "logits/rejected": -0.5896657705307007,
      "logps/chosen": -111.97227478027344,
      "logps/ref_chosen": -101.82364654541016,
      "logps/ref_rejected": -105.92247009277344,
      "logps/rejected": -119.26460266113281,
      "loss": 1.4115,
      "rewards/accuracies": 0.578125,
      "rewards/chosen": -0.4309285879135132,
      "rewards/margins": 0.12875217199325562,
      "rewards/rejected": -0.5596807599067688,
      "step": 310
    },
    {
      "epoch": 0.47014361300075586,
      "epsilon_dpo/beta": 0.04204654321074486,
      "epsilon_dpo/beta_margin_grad_mean": -0.4101211428642273,
      "epsilon_dpo/beta_margin_grad_std": 0.1667424887418747,
      "epsilon_dpo/beta_margin_mean": 0.4359593689441681,
      "epsilon_dpo/beta_margin_std": 0.8494295477867126,
      "epsilon_dpo/loss_margin_mean": 10.489989280700684,
      "grad_norm": 11.257895469665527,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.04217413440346718,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 3.204331392103574e-07,
      "logits/chosen": -0.3479464650154114,
      "logits/rejected": -0.6605005264282227,
      "logps/chosen": -84.55355834960938,
      "logps/ref_chosen": -81.57022857666016,
      "logps/ref_rejected": -118.00057983398438,
      "logps/rejected": -131.47389221191406,
      "loss": 1.149,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.12689390778541565,
      "rewards/margins": 0.4359593391418457,
      "rewards/rejected": -0.562853217124939,
      "step": 311
    },
    {
      "epoch": 0.47165532879818595,
      "epsilon_dpo/beta": 0.0418432243168354,
      "epsilon_dpo/beta_margin_grad_mean": -0.3847061097621918,
      "epsilon_dpo/beta_margin_grad_std": 0.16446185111999512,
      "epsilon_dpo/beta_margin_mean": 0.5488578081130981,
      "epsilon_dpo/beta_margin_std": 0.7945172786712646,
      "epsilon_dpo/loss_margin_mean": 13.221216201782227,
      "grad_norm": 10.767632484436035,
      "kl/avg_steps": 0.484375,
      "kl/beta": 0.042042750865221024,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.1916350007663176e-07,
      "logits/chosen": -0.32211631536483765,
      "logits/rejected": -0.1509503722190857,
      "logps/chosen": -79.85913848876953,
      "logps/ref_chosen": -73.72250366210938,
      "logps/ref_rejected": -101.14654541015625,
      "logps/rejected": -120.50439453125,
      "loss": 1.0487,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.2573304772377014,
      "rewards/margins": 0.5488578081130981,
      "rewards/rejected": -0.8061882257461548,
      "step": 312
    },
    {
      "epoch": 0.47316704459561604,
      "epsilon_dpo/beta": 0.04171350598335266,
      "epsilon_dpo/beta_margin_grad_mean": -0.43476757407188416,
      "epsilon_dpo/beta_margin_grad_std": 0.1839824914932251,
      "epsilon_dpo/beta_margin_mean": 0.3212515711784363,
      "epsilon_dpo/beta_margin_std": 0.8919322490692139,
      "epsilon_dpo/loss_margin_mean": 7.839280605316162,
      "grad_norm": 12.423453330993652,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.0418400876224041,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 3.178919262911314e-07,
      "logits/chosen": -0.2028963714838028,
      "logits/rejected": -0.39557722210884094,
      "logps/chosen": -86.79026794433594,
      "logps/ref_chosen": -83.71331787109375,
      "logps/ref_rejected": -93.05607604980469,
      "logps/rejected": -103.9723129272461,
      "loss": 1.2671,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.13130736351013184,
      "rewards/margins": 0.3212515711784363,
      "rewards/rejected": -0.4525589346885681,
      "step": 313
    },
    {
      "epoch": 0.47467876039304613,
      "epsilon_dpo/beta": 0.04155748710036278,
      "epsilon_dpo/beta_margin_grad_mean": -0.3984832167625427,
      "epsilon_dpo/beta_margin_grad_std": 0.16552382707595825,
      "epsilon_dpo/beta_margin_mean": 0.4729049801826477,
      "epsilon_dpo/beta_margin_std": 0.7664237022399902,
      "epsilon_dpo/loss_margin_mean": 11.503206253051758,
      "grad_norm": 14.001582145690918,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.041709743440151215,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.166184534225087e-07,
      "logits/chosen": -0.36525222659111023,
      "logits/rejected": -0.4043053090572357,
      "logps/chosen": -102.99305725097656,
      "logps/ref_chosen": -98.91353607177734,
      "logps/ref_rejected": -88.04048919677734,
      "logps/rejected": -103.62322235107422,
      "loss": 1.1009,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.17131654918193817,
      "rewards/margins": 0.4729049801826477,
      "rewards/rejected": -0.6442215442657471,
      "step": 314
    },
    {
      "epoch": 0.47619047619047616,
      "epsilon_dpo/beta": 0.04138924181461334,
      "epsilon_dpo/beta_margin_grad_mean": -0.4095999598503113,
      "epsilon_dpo/beta_margin_grad_std": 0.15082554519176483,
      "epsilon_dpo/beta_margin_mean": 0.41423144936561584,
      "epsilon_dpo/beta_margin_std": 0.6999982595443726,
      "epsilon_dpo/loss_margin_mean": 10.115947723388672,
      "grad_norm": 12.541328430175781,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.04155391454696655,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.1534311709253723e-07,
      "logits/chosen": -0.2650623023509979,
      "logits/rejected": -0.30792877078056335,
      "logps/chosen": -97.30009460449219,
      "logps/ref_chosen": -92.76774597167969,
      "logps/ref_rejected": -102.42605590820312,
      "logps/rejected": -117.07435607910156,
      "loss": 1.1248,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.18921594321727753,
      "rewards/margins": 0.41423147916793823,
      "rewards/rejected": -0.603447437286377,
      "step": 315
    },
    {
      "epoch": 0.47770219198790626,
      "epsilon_dpo/beta": 0.041234713047742844,
      "epsilon_dpo/beta_margin_grad_mean": -0.39376702904701233,
      "epsilon_dpo/beta_margin_grad_std": 0.16725587844848633,
      "epsilon_dpo/beta_margin_mean": 0.4972352981567383,
      "epsilon_dpo/beta_margin_std": 0.7851472496986389,
      "epsilon_dpo/loss_margin_mean": 12.186201095581055,
      "grad_norm": 11.967978477478027,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.04138578474521637,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.1406595297511564e-07,
      "logits/chosen": -0.17552141845226288,
      "logits/rejected": -0.2363368570804596,
      "logps/chosen": -85.0501708984375,
      "logps/ref_chosen": -80.11531066894531,
      "logps/ref_rejected": -124.51179504394531,
      "logps/rejected": -141.6328582763672,
      "loss": 1.0872,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.20370078086853027,
      "rewards/margins": 0.49723532795906067,
      "rewards/rejected": -0.7009360790252686,
      "step": 316
    },
    {
      "epoch": 0.47921390778533635,
      "epsilon_dpo/beta": 0.04100334644317627,
      "epsilon_dpo/beta_margin_grad_mean": -0.3886288106441498,
      "epsilon_dpo/beta_margin_grad_std": 0.171974316239357,
      "epsilon_dpo/beta_margin_mean": 0.5132362246513367,
      "epsilon_dpo/beta_margin_std": 0.8470146059989929,
      "epsilon_dpo/loss_margin_mean": 12.628094673156738,
      "grad_norm": 16.494524002075195,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.041231170296669006,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 3.1278699679526975e-07,
      "logits/chosen": -0.3971477448940277,
      "logits/rejected": -0.32429075241088867,
      "logps/chosen": -83.87468719482422,
      "logps/ref_chosen": -81.12582397460938,
      "logps/ref_rejected": -98.85325622558594,
      "logps/rejected": -114.23020935058594,
      "loss": 1.0944,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.1150696724653244,
      "rewards/margins": 0.5132362842559814,
      "rewards/rejected": -0.6283059120178223,
      "step": 317
    },
    {
      "epoch": 0.48072562358276644,
      "epsilon_dpo/beta": 0.04087650030851364,
      "epsilon_dpo/beta_margin_grad_mean": -0.42562544345855713,
      "epsilon_dpo/beta_margin_grad_std": 0.18235144019126892,
      "epsilon_dpo/beta_margin_mean": 0.36548641324043274,
      "epsilon_dpo/beta_margin_std": 0.8594874143600464,
      "epsilon_dpo/loss_margin_mean": 9.091790199279785,
      "grad_norm": 15.041504859924316,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.04100054129958153,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 3.1150628432815336e-07,
      "logits/chosen": -0.4328976273536682,
      "logits/rejected": -0.25933602452278137,
      "logps/chosen": -80.97486877441406,
      "logps/ref_chosen": -75.95895385742188,
      "logps/ref_rejected": -105.97601318359375,
      "logps/rejected": -120.0837173461914,
      "loss": 1.2196,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.20701169967651367,
      "rewards/margins": 0.3654864430427551,
      "rewards/rejected": -0.572498083114624,
      "step": 318
    },
    {
      "epoch": 0.48223733938019653,
      "epsilon_dpo/beta": 0.04071083664894104,
      "epsilon_dpo/beta_margin_grad_mean": -0.3958146274089813,
      "epsilon_dpo/beta_margin_grad_std": 0.1657380759716034,
      "epsilon_dpo/beta_margin_mean": 0.48401468992233276,
      "epsilon_dpo/beta_margin_std": 0.7800770401954651,
      "epsilon_dpo/loss_margin_mean": 12.014669418334961,
      "grad_norm": 11.02835464477539,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.040872812271118164,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.1022385139804707e-07,
      "logits/chosen": -0.685593843460083,
      "logits/rejected": -0.36105144023895264,
      "logps/chosen": -93.86186218261719,
      "logps/ref_chosen": -88.54928588867188,
      "logps/ref_rejected": -102.58941650390625,
      "logps/rejected": -119.91665649414062,
      "loss": 1.0955,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.21764391660690308,
      "rewards/margins": 0.48401468992233276,
      "rewards/rejected": -0.7016586661338806,
      "step": 319
    },
    {
      "epoch": 0.4837490551776266,
      "epsilon_dpo/beta": 0.040609732270240784,
      "epsilon_dpo/beta_margin_grad_mean": -0.4467881917953491,
      "epsilon_dpo/beta_margin_grad_std": 0.17798320949077606,
      "epsilon_dpo/beta_margin_mean": 0.24270537495613098,
      "epsilon_dpo/beta_margin_std": 0.8327056169509888,
      "epsilon_dpo/loss_margin_mean": 6.119729518890381,
      "grad_norm": 11.728205680847168,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.04070743918418884,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 3.0893973387735683e-07,
      "logits/chosen": -0.24449878931045532,
      "logits/rejected": -0.4534444808959961,
      "logps/chosen": -76.99515533447266,
      "logps/ref_chosen": -72.40655517578125,
      "logps/ref_rejected": -90.68508911132812,
      "logps/rejected": -101.39341735839844,
      "loss": 1.3166,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.1887323260307312,
      "rewards/margins": 0.24270540475845337,
      "rewards/rejected": -0.43143773078918457,
      "step": 320
    },
    {
      "epoch": 0.4852607709750567,
      "epsilon_dpo/beta": 0.040432319045066833,
      "epsilon_dpo/beta_margin_grad_mean": -0.39743614196777344,
      "epsilon_dpo/beta_margin_grad_std": 0.1755134016275406,
      "epsilon_dpo/beta_margin_mean": 0.47886669635772705,
      "epsilon_dpo/beta_margin_std": 0.8305484652519226,
      "epsilon_dpo/loss_margin_mean": 11.975590705871582,
      "grad_norm": 14.635098457336426,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.040605925023555756,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.0765396768561004e-07,
      "logits/chosen": -0.5184362530708313,
      "logits/rejected": -0.4613918662071228,
      "logps/chosen": -74.00281524658203,
      "logps/ref_chosen": -71.84503936767578,
      "logps/ref_rejected": -78.1065673828125,
      "logps/rejected": -92.23993682861328,
      "loss": 1.1179,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.09022434055805206,
      "rewards/margins": 0.47886669635772705,
      "rewards/rejected": -0.5690910220146179,
      "step": 321
    },
    {
      "epoch": 0.48677248677248675,
      "epsilon_dpo/beta": 0.04016774892807007,
      "epsilon_dpo/beta_margin_grad_mean": -0.36837536096572876,
      "epsilon_dpo/beta_margin_grad_std": 0.15891148149967194,
      "epsilon_dpo/beta_margin_mean": 0.6071159243583679,
      "epsilon_dpo/beta_margin_std": 0.7657486796379089,
      "epsilon_dpo/loss_margin_mean": 15.201668739318848,
      "grad_norm": 11.990974426269531,
      "kl/avg_steps": 0.65625,
      "kl/beta": 0.040429048240184784,
      "kl/n_epsilon_steps": 0.171875,
      "kl/p_epsilon_steps": 0.828125,
      "learning_rate": 3.063665887884511e-07,
      "logits/chosen": -0.18207885324954987,
      "logits/rejected": -0.3786749541759491,
      "logps/chosen": -76.58918762207031,
      "logps/ref_chosen": -70.58155822753906,
      "logps/ref_rejected": -99.65203094482422,
      "logps/rejected": -120.861328125,
      "loss": 0.9974,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": -0.2421734780073166,
      "rewards/margins": 0.6071159243583679,
      "rewards/rejected": -0.8492894172668457,
      "step": 322
    },
    {
      "epoch": 0.48828420256991684,
      "epsilon_dpo/beta": 0.04000628739595413,
      "epsilon_dpo/beta_margin_grad_mean": -0.4120200574398041,
      "epsilon_dpo/beta_margin_grad_std": 0.1918712854385376,
      "epsilon_dpo/beta_margin_mean": 0.4309968650341034,
      "epsilon_dpo/beta_margin_std": 0.9564192295074463,
      "epsilon_dpo/loss_margin_mean": 10.92121410369873,
      "grad_norm": 12.162070274353027,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.04016546159982681,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.0507763319663517e-07,
      "logits/chosen": -0.4847533106803894,
      "logits/rejected": -0.21049503982067108,
      "logps/chosen": -100.65248107910156,
      "logps/ref_chosen": -93.55892944335938,
      "logps/ref_rejected": -122.09970092773438,
      "logps/rejected": -140.11447143554688,
      "loss": 1.1997,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.2870478630065918,
      "rewards/margins": 0.4309968948364258,
      "rewards/rejected": -0.7180447578430176,
      "step": 323
    },
    {
      "epoch": 0.4897959183673469,
      "epsilon_dpo/beta": 0.03980691358447075,
      "epsilon_dpo/beta_margin_grad_mean": -0.3884578347206116,
      "epsilon_dpo/beta_margin_grad_std": 0.14908921718597412,
      "epsilon_dpo/beta_margin_mean": 0.512287437915802,
      "epsilon_dpo/beta_margin_std": 0.7012259364128113,
      "epsilon_dpo/loss_margin_mean": 12.961596488952637,
      "grad_norm": 11.008560180664062,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.04000294953584671,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.0378713696502097e-07,
      "logits/chosen": -0.4199288785457611,
      "logits/rejected": -0.7384412288665771,
      "logps/chosen": -85.6984634399414,
      "logps/ref_chosen": -81.87194061279297,
      "logps/ref_rejected": -91.50868225097656,
      "logps/rejected": -108.29679870605469,
      "loss": 1.0477,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.15418805181980133,
      "rewards/margins": 0.512287437915802,
      "rewards/rejected": -0.6664755344390869,
      "step": 324
    },
    {
      "epoch": 0.491307634164777,
      "epsilon_dpo/beta": 0.03965863212943077,
      "epsilon_dpo/beta_margin_grad_mean": -0.3986729383468628,
      "epsilon_dpo/beta_margin_grad_std": 0.1633588671684265,
      "epsilon_dpo/beta_margin_mean": 0.4666576087474823,
      "epsilon_dpo/beta_margin_std": 0.7744911909103394,
      "epsilon_dpo/loss_margin_mean": 11.899657249450684,
      "grad_norm": 11.990095138549805,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.03980392962694168,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.0249513619156206e-07,
      "logits/chosen": -0.3202875256538391,
      "logits/rejected": -0.4885994791984558,
      "logps/chosen": -97.64002990722656,
      "logps/ref_chosen": -86.96577453613281,
      "logps/ref_rejected": -104.21844482421875,
      "logps/rejected": -126.79237365722656,
      "loss": 1.107,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.42622262239456177,
      "rewards/margins": 0.4666576385498047,
      "rewards/rejected": -0.8928803205490112,
      "step": 325
    },
    {
      "epoch": 0.4928193499622071,
      "epsilon_dpo/beta": 0.03953525424003601,
      "epsilon_dpo/beta_margin_grad_mean": -0.42973268032073975,
      "epsilon_dpo/beta_margin_grad_std": 0.16378960013389587,
      "epsilon_dpo/beta_margin_mean": 0.3205667734146118,
      "epsilon_dpo/beta_margin_std": 0.7373978495597839,
      "epsilon_dpo/loss_margin_mean": 8.246063232421875,
      "grad_norm": 12.42955493927002,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.03965522348880768,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 3.012016670162977e-07,
      "logits/chosen": -0.5864256620407104,
      "logits/rejected": -0.6668286323547363,
      "logps/chosen": -110.80265808105469,
      "logps/ref_chosen": -98.51603698730469,
      "logps/ref_rejected": -97.31979370117188,
      "logps/rejected": -117.85248565673828,
      "loss": 1.2173,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.4882323145866394,
      "rewards/margins": 0.3205668032169342,
      "rewards/rejected": -0.8087990880012512,
      "step": 326
    },
    {
      "epoch": 0.4943310657596372,
      "epsilon_dpo/beta": 0.03943679854273796,
      "epsilon_dpo/beta_margin_grad_mean": -0.42828133702278137,
      "epsilon_dpo/beta_margin_grad_std": 0.17430712282657623,
      "epsilon_dpo/beta_margin_mean": 0.3396408259868622,
      "epsilon_dpo/beta_margin_std": 0.842943549156189,
      "epsilon_dpo/loss_margin_mean": 8.755025863647461,
      "grad_norm": 13.681025505065918,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.039531685411930084,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.99906765620341e-07,
      "logits/chosen": -0.374507337808609,
      "logits/rejected": -0.5013106465339661,
      "logps/chosen": -107.53089904785156,
      "logps/ref_chosen": -95.87124633789062,
      "logps/ref_rejected": -108.29940795898438,
      "logps/rejected": -128.71409606933594,
      "loss": 1.2332,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.4616592824459076,
      "rewards/margins": 0.33964085578918457,
      "rewards/rejected": -0.8013001084327698,
      "step": 327
    },
    {
      "epoch": 0.4958427815570673,
      "epsilon_dpo/beta": 0.039276834577322006,
      "epsilon_dpo/beta_margin_grad_mean": -0.4211041033267975,
      "epsilon_dpo/beta_margin_grad_std": 0.16422966122627258,
      "epsilon_dpo/beta_margin_mean": 0.3656752407550812,
      "epsilon_dpo/beta_margin_std": 0.7797321677207947,
      "epsilon_dpo/loss_margin_mean": 9.426528930664062,
      "grad_norm": 11.109094619750977,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.03943310305476189,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.9861046822486766e-07,
      "logits/chosen": -0.2030862420797348,
      "logits/rejected": -0.10589778423309326,
      "logps/chosen": -95.96703338623047,
      "logps/ref_chosen": -85.7418212890625,
      "logps/ref_rejected": -104.25873565673828,
      "logps/rejected": -123.91046905517578,
      "loss": 1.1903,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.4038175642490387,
      "rewards/margins": 0.3656752407550812,
      "rewards/rejected": -0.7694927453994751,
      "step": 328
    },
    {
      "epoch": 0.4973544973544973,
      "epsilon_dpo/beta": 0.03911792114377022,
      "epsilon_dpo/beta_margin_grad_mean": -0.4037947654724121,
      "epsilon_dpo/beta_margin_grad_std": 0.15787221491336823,
      "epsilon_dpo/beta_margin_mean": 0.4557969570159912,
      "epsilon_dpo/beta_margin_std": 0.7563498616218567,
      "epsilon_dpo/loss_margin_mean": 11.770658493041992,
      "grad_norm": 11.235872268676758,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.03927355632185936,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.9731281109010253e-07,
      "logits/chosen": -0.2828035354614258,
      "logits/rejected": -0.26564446091651917,
      "logps/chosen": -95.25900268554688,
      "logps/ref_chosen": -85.7830810546875,
      "logps/ref_rejected": -109.35794830322266,
      "logps/rejected": -130.6045379638672,
      "loss": 1.1076,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.37267088890075684,
      "rewards/margins": 0.4557969570159912,
      "rewards/rejected": -0.828467845916748,
      "step": 329
    },
    {
      "epoch": 0.4988662131519274,
      "epsilon_dpo/beta": 0.03892296925187111,
      "epsilon_dpo/beta_margin_grad_mean": -0.3836939036846161,
      "epsilon_dpo/beta_margin_grad_std": 0.18278218805789948,
      "epsilon_dpo/beta_margin_mean": 0.5425154566764832,
      "epsilon_dpo/beta_margin_std": 0.8813403248786926,
      "epsilon_dpo/loss_margin_mean": 14.080089569091797,
      "grad_norm": 10.5839262008667,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.03911465033888817,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 2.9601383051430505e-07,
      "logits/chosen": -0.24584242701530457,
      "logits/rejected": -0.20333293080329895,
      "logps/chosen": -82.42384338378906,
      "logps/ref_chosen": -75.27607727050781,
      "logps/ref_rejected": -102.55793762207031,
      "logps/rejected": -123.7857894897461,
      "loss": 1.0872,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.2804255187511444,
      "rewards/margins": 0.5425155162811279,
      "rewards/rejected": -0.82294100522995,
      "step": 330
    },
    {
      "epoch": 0.5003779289493575,
      "epsilon_dpo/beta": 0.03874148800969124,
      "epsilon_dpo/beta_margin_grad_mean": -0.38367658853530884,
      "epsilon_dpo/beta_margin_grad_std": 0.17551447451114655,
      "epsilon_dpo/beta_margin_mean": 0.5445398092269897,
      "epsilon_dpo/beta_margin_std": 0.8645876049995422,
      "epsilon_dpo/loss_margin_mean": 14.192091941833496,
      "grad_norm": 11.384100914001465,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.03892005234956741,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.947135628327544e-07,
      "logits/chosen": -0.27640146017074585,
      "logits/rejected": -0.12456096708774567,
      "logps/chosen": -90.9360122680664,
      "logps/ref_chosen": -80.22660064697266,
      "logps/ref_rejected": -90.82720947265625,
      "logps/rejected": -115.72871398925781,
      "loss": 1.0769,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.41756218671798706,
      "rewards/margins": 0.5445398092269897,
      "rewards/rejected": -0.9621019959449768,
      "step": 331
    },
    {
      "epoch": 0.5018896447467877,
      "epsilon_dpo/beta": 0.038560736924409866,
      "epsilon_dpo/beta_margin_grad_mean": -0.3933686316013336,
      "epsilon_dpo/beta_margin_grad_std": 0.1542639285326004,
      "epsilon_dpo/beta_margin_mean": 0.5055604577064514,
      "epsilon_dpo/beta_margin_std": 0.7535479068756104,
      "epsilon_dpo/loss_margin_mean": 13.21468734741211,
      "grad_norm": 12.87071704864502,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.038738466799259186,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.934120444167326e-07,
      "logits/chosen": -0.4173445701599121,
      "logits/rejected": -0.5104435086250305,
      "logps/chosen": -100.83187103271484,
      "logps/ref_chosen": -90.54778289794922,
      "logps/ref_rejected": -107.08577728271484,
      "logps/rejected": -130.58456420898438,
      "loss": 1.0664,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.3975829482078552,
      "rewards/margins": 0.5055604577064514,
      "rewards/rejected": -0.9031434059143066,
      "step": 332
    },
    {
      "epoch": 0.5034013605442177,
      "epsilon_dpo/beta": 0.03841697797179222,
      "epsilon_dpo/beta_margin_grad_mean": -0.3773667812347412,
      "epsilon_dpo/beta_margin_grad_std": 0.14968030154705048,
      "epsilon_dpo/beta_margin_mean": 0.5757911801338196,
      "epsilon_dpo/beta_margin_std": 0.7351418137550354,
      "epsilon_dpo/loss_margin_mean": 15.106307983398438,
      "grad_norm": 11.473505020141602,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.03855772688984871,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.921093116725076e-07,
      "logits/chosen": -0.5246701240539551,
      "logits/rejected": -0.4246484637260437,
      "logps/chosen": -94.96739196777344,
      "logps/ref_chosen": -83.32445526123047,
      "logps/ref_rejected": -112.07743835449219,
      "logps/rejected": -138.82667541503906,
      "loss": 1.0074,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.44865351915359497,
      "rewards/margins": 0.5757912397384644,
      "rewards/rejected": -1.0244446992874146,
      "step": 333
    },
    {
      "epoch": 0.5049130763416477,
      "epsilon_dpo/beta": 0.038321349769830704,
      "epsilon_dpo/beta_margin_grad_mean": -0.42791518568992615,
      "epsilon_dpo/beta_margin_grad_std": 0.2027246654033661,
      "epsilon_dpo/beta_margin_mean": 0.3443240225315094,
      "epsilon_dpo/beta_margin_std": 0.9910604953765869,
      "epsilon_dpo/loss_margin_mean": 9.166866302490234,
      "grad_norm": 14.25706672668457,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.038413673639297485,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 2.9080540104031484e-07,
      "logits/chosen": -0.13559825718402863,
      "logits/rejected": -0.3539199233055115,
      "logps/chosen": -104.35836029052734,
      "logps/ref_chosen": -91.25233459472656,
      "logps/ref_rejected": -114.59999084472656,
      "logps/rejected": -136.8728790283203,
      "loss": 1.2894,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5054439902305603,
      "rewards/margins": 0.3443240523338318,
      "rewards/rejected": -0.8497680425643921,
      "step": 334
    },
    {
      "epoch": 0.5064247921390779,
      "epsilon_dpo/beta": 0.03822590783238411,
      "epsilon_dpo/beta_margin_grad_mean": -0.41665026545524597,
      "epsilon_dpo/beta_margin_grad_std": 0.20269030332565308,
      "epsilon_dpo/beta_margin_mean": 0.4251091480255127,
      "epsilon_dpo/beta_margin_std": 0.9988790154457092,
      "epsilon_dpo/loss_margin_mean": 11.305398941040039,
      "grad_norm": 12.197574615478516,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.038317881524562836,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.895003489933375e-07,
      "logits/chosen": -0.1341536045074463,
      "logits/rejected": -0.1751415729522705,
      "logps/chosen": -99.05448913574219,
      "logps/ref_chosen": -83.96897888183594,
      "logps/ref_rejected": -99.78108215332031,
      "logps/rejected": -126.1719970703125,
      "loss": 1.2237,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.5797243118286133,
      "rewards/margins": 0.4251091480255127,
      "rewards/rejected": -1.004833459854126,
      "step": 335
    },
    {
      "epoch": 0.5079365079365079,
      "epsilon_dpo/beta": 0.03805890679359436,
      "epsilon_dpo/beta_margin_grad_mean": -0.3994874060153961,
      "epsilon_dpo/beta_margin_grad_std": 0.16580943763256073,
      "epsilon_dpo/beta_margin_mean": 0.4782327711582184,
      "epsilon_dpo/beta_margin_std": 0.8129090070724487,
      "epsilon_dpo/loss_margin_mean": 12.692656517028809,
      "grad_norm": 13.025675773620605,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.03822232410311699,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.8819419203668675e-07,
      "logits/chosen": -0.1973276436328888,
      "logits/rejected": -0.1543230414390564,
      "logps/chosen": -113.52323913574219,
      "logps/ref_chosen": -95.66194152832031,
      "logps/ref_rejected": -114.09184265136719,
      "logps/rejected": -144.64581298828125,
      "loss": 1.1081,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6822552680969238,
      "rewards/margins": 0.47823280096054077,
      "rewards/rejected": -1.1604881286621094,
      "step": 336
    },
    {
      "epoch": 0.509448223733938,
      "epsilon_dpo/beta": 0.037964485585689545,
      "epsilon_dpo/beta_margin_grad_mean": -0.4189932942390442,
      "epsilon_dpo/beta_margin_grad_std": 0.1795109510421753,
      "epsilon_dpo/beta_margin_mean": 0.38377320766448975,
      "epsilon_dpo/beta_margin_std": 0.8511192202568054,
      "epsilon_dpo/loss_margin_mean": 10.27182388305664,
      "grad_norm": 10.546476364135742,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.03805582970380783,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.8688696670638053e-07,
      "logits/chosen": -0.20124396681785583,
      "logits/rejected": -0.48997047543525696,
      "logps/chosen": -117.29708862304688,
      "logps/ref_chosen": -99.90296173095703,
      "logps/ref_rejected": -113.39192199707031,
      "logps/rejected": -141.05787658691406,
      "loss": 1.2014,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.6641528606414795,
      "rewards/margins": 0.38377323746681213,
      "rewards/rejected": -1.0479261875152588,
      "step": 337
    },
    {
      "epoch": 0.5109599395313681,
      "epsilon_dpo/beta": 0.03785794600844383,
      "epsilon_dpo/beta_margin_grad_mean": -0.42979949712753296,
      "epsilon_dpo/beta_margin_grad_std": 0.1584760844707489,
      "epsilon_dpo/beta_margin_mean": 0.3304803669452667,
      "epsilon_dpo/beta_margin_std": 0.7403267025947571,
      "epsilon_dpo/loss_margin_mean": 8.861513137817383,
      "grad_norm": 11.558711051940918,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.037960927933454514,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 2.8557870956832133e-07,
      "logits/chosen": -0.06663598120212555,
      "logits/rejected": -0.06503565609455109,
      "logps/chosen": -105.7027587890625,
      "logps/ref_chosen": -86.85198974609375,
      "logps/ref_rejected": -96.50064086914062,
      "logps/rejected": -124.21292114257812,
      "loss": 1.2068,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.7169533967971802,
      "rewards/margins": 0.33048033714294434,
      "rewards/rejected": -1.0474337339401245,
      "step": 338
    },
    {
      "epoch": 0.5124716553287982,
      "epsilon_dpo/beta": 0.03770444914698601,
      "epsilon_dpo/beta_margin_grad_mean": -0.3875303566455841,
      "epsilon_dpo/beta_margin_grad_std": 0.1659288853406906,
      "epsilon_dpo/beta_margin_mean": 0.515190601348877,
      "epsilon_dpo/beta_margin_std": 0.7667047381401062,
      "epsilon_dpo/loss_margin_mean": 13.80580997467041,
      "grad_norm": 15.705618858337402,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.03785446286201477,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.842694572172736e-07,
      "logits/chosen": -0.544682502746582,
      "logits/rejected": -0.5630224943161011,
      "logps/chosen": -78.35090637207031,
      "logps/ref_chosen": -66.23301696777344,
      "logps/ref_rejected": -91.11048126220703,
      "logps/rejected": -117.03418731689453,
      "loss": 1.0689,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.45997846126556396,
      "rewards/margins": 0.515190601348877,
      "rewards/rejected": -0.9751690626144409,
      "step": 339
    },
    {
      "epoch": 0.5139833711262283,
      "epsilon_dpo/beta": 0.0375872403383255,
      "epsilon_dpo/beta_margin_grad_mean": -0.3984226584434509,
      "epsilon_dpo/beta_margin_grad_std": 0.19148968160152435,
      "epsilon_dpo/beta_margin_mean": 0.49994638562202454,
      "epsilon_dpo/beta_margin_std": 0.9249793291091919,
      "epsilon_dpo/loss_margin_mean": 13.473993301391602,
      "grad_norm": 10.675450325012207,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.03770130127668381,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 2.8295924627584004e-07,
      "logits/chosen": -0.1954106092453003,
      "logits/rejected": -0.060511067509651184,
      "logps/chosen": -92.63058471679688,
      "logps/ref_chosen": -76.95579528808594,
      "logps/ref_rejected": -86.38760375976562,
      "logps/rejected": -115.53638458251953,
      "loss": 1.1359,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.5927058458328247,
      "rewards/margins": 0.49994635581970215,
      "rewards/rejected": -1.0926522016525269,
      "step": 340
    },
    {
      "epoch": 0.5154950869236583,
      "epsilon_dpo/beta": 0.037458401173353195,
      "epsilon_dpo/beta_margin_grad_mean": -0.3949894607067108,
      "epsilon_dpo/beta_margin_grad_std": 0.19184111058712006,
      "epsilon_dpo/beta_margin_mean": 0.5000091195106506,
      "epsilon_dpo/beta_margin_std": 0.9141637682914734,
      "epsilon_dpo/loss_margin_mean": 13.519756317138672,
      "grad_norm": 11.776097297668457,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.0375838503241539,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.816481133934373e-07,
      "logits/chosen": -0.3284544348716736,
      "logits/rejected": -0.22923079133033752,
      "logps/chosen": -100.67996215820312,
      "logps/ref_chosen": -84.42140197753906,
      "logps/ref_rejected": -101.77664184570312,
      "logps/rejected": -131.55496215820312,
      "loss": 1.1337,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6120609641075134,
      "rewards/margins": 0.5000091195106506,
      "rewards/rejected": -1.112070083618164,
      "step": 341
    },
    {
      "epoch": 0.5170068027210885,
      "epsilon_dpo/beta": 0.037294965237379074,
      "epsilon_dpo/beta_margin_grad_mean": -0.39244845509529114,
      "epsilon_dpo/beta_margin_grad_std": 0.18085214495658875,
      "epsilon_dpo/beta_margin_mean": 0.5229305028915405,
      "epsilon_dpo/beta_margin_std": 0.8699291944503784,
      "epsilon_dpo/loss_margin_mean": 14.16259479522705,
      "grad_norm": 10.74425220489502,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.03745510056614876,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.8033609524527046e-07,
      "logits/chosen": -0.24709981679916382,
      "logits/rejected": -0.25939422845840454,
      "logps/chosen": -89.46382141113281,
      "logps/ref_chosen": -75.53753662109375,
      "logps/ref_rejected": -91.3760986328125,
      "logps/rejected": -119.46498107910156,
      "loss": 1.0964,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5207824110984802,
      "rewards/margins": 0.5229305028915405,
      "rewards/rejected": -1.043712854385376,
      "step": 342
    },
    {
      "epoch": 0.5185185185185185,
      "epsilon_dpo/beta": 0.03719655051827431,
      "epsilon_dpo/beta_margin_grad_mean": -0.44036367535591125,
      "epsilon_dpo/beta_margin_grad_std": 0.17963874340057373,
      "epsilon_dpo/beta_margin_mean": 0.27663713693618774,
      "epsilon_dpo/beta_margin_std": 0.8233576416969299,
      "epsilon_dpo/loss_margin_mean": 7.593426704406738,
      "grad_norm": 11.922304153442383,
      "kl/avg_steps": 0.265625,
      "kl/beta": 0.03729194775223732,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.7902322853130753e-07,
      "logits/chosen": -0.46268126368522644,
      "logits/rejected": -0.49861472845077515,
      "logps/chosen": -111.53775787353516,
      "logps/ref_chosen": -96.93223571777344,
      "logps/ref_rejected": -104.06831359863281,
      "logps/rejected": -126.26726531982422,
      "loss": 1.2851,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.5451029539108276,
      "rewards/margins": 0.27663713693618774,
      "rewards/rejected": -0.8217400312423706,
      "step": 343
    },
    {
      "epoch": 0.5200302343159486,
      "epsilon_dpo/beta": 0.037034135311841965,
      "epsilon_dpo/beta_margin_grad_mean": -0.39166146516799927,
      "epsilon_dpo/beta_margin_grad_std": 0.17405946552753448,
      "epsilon_dpo/beta_margin_mean": 0.5026716589927673,
      "epsilon_dpo/beta_margin_std": 0.8177103996276855,
      "epsilon_dpo/loss_margin_mean": 13.722208023071289,
      "grad_norm": 11.160215377807617,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.03719315305352211,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.7770954997525274e-07,
      "logits/chosen": -0.29269033670425415,
      "logits/rejected": -0.29436129331588745,
      "logps/chosen": -97.3475112915039,
      "logps/ref_chosen": -78.63787841796875,
      "logps/ref_rejected": -107.91048431396484,
      "logps/rejected": -140.3423309326172,
      "loss": 1.0953,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6958789229393005,
      "rewards/margins": 0.5026716589927673,
      "rewards/rejected": -1.1985507011413574,
      "step": 344
    },
    {
      "epoch": 0.5215419501133787,
      "epsilon_dpo/beta": 0.03689596429467201,
      "epsilon_dpo/beta_margin_grad_mean": -0.4142055809497833,
      "epsilon_dpo/beta_margin_grad_std": 0.18251675367355347,
      "epsilon_dpo/beta_margin_mean": 0.4029293954372406,
      "epsilon_dpo/beta_margin_std": 0.871208667755127,
      "epsilon_dpo/loss_margin_mean": 11.078716278076172,
      "grad_norm": 15.428401947021484,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.037031140178442,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.7639509632351927e-07,
      "logits/chosen": -0.12234549969434738,
      "logits/rejected": -0.16197794675827026,
      "logps/chosen": -91.65048217773438,
      "logps/ref_chosen": -79.6897201538086,
      "logps/ref_rejected": -100.54981994628906,
      "logps/rejected": -123.58929443359375,
      "loss": 1.1932,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.44506821036338806,
      "rewards/margins": 0.4029293954372406,
      "rewards/rejected": -0.8479976654052734,
      "step": 345
    },
    {
      "epoch": 0.5230536659108088,
      "epsilon_dpo/beta": 0.036735061556100845,
      "epsilon_dpo/beta_margin_grad_mean": -0.3954569101333618,
      "epsilon_dpo/beta_margin_grad_std": 0.17873188853263855,
      "epsilon_dpo/beta_margin_mean": 0.5000748038291931,
      "epsilon_dpo/beta_margin_std": 0.8613612055778503,
      "epsilon_dpo/loss_margin_mean": 13.76108169555664,
      "grad_norm": 13.216407775878906,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.03689279407262802,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.7507990434420123e-07,
      "logits/chosen": -0.3659716546535492,
      "logits/rejected": -0.22499585151672363,
      "logps/chosen": -87.64271545410156,
      "logps/ref_chosen": -75.92257690429688,
      "logps/ref_rejected": -116.39082336425781,
      "logps/rejected": -141.87203979492188,
      "loss": 1.1105,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.43288546800613403,
      "rewards/margins": 0.5000748038291931,
      "rewards/rejected": -0.9329602718353271,
      "step": 346
    },
    {
      "epoch": 0.5245653817082389,
      "epsilon_dpo/beta": 0.036632440984249115,
      "epsilon_dpo/beta_margin_grad_mean": -0.4249770939350128,
      "epsilon_dpo/beta_margin_grad_std": 0.1754349023103714,
      "epsilon_dpo/beta_margin_mean": 0.36538615822792053,
      "epsilon_dpo/beta_margin_std": 0.8376450538635254,
      "epsilon_dpo/loss_margin_mean": 10.127693176269531,
      "grad_norm": 11.684128761291504,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.03673208877444267,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 2.737640108260456e-07,
      "logits/chosen": -0.3123420178890228,
      "logits/rejected": -0.3318549692630768,
      "logps/chosen": -96.49603271484375,
      "logps/ref_chosen": -78.1712875366211,
      "logps/ref_rejected": -98.7479248046875,
      "logps/rejected": -127.20036315917969,
      "loss": 1.21,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.674199104309082,
      "rewards/margins": 0.36538615822792053,
      "rewards/rejected": -1.0395852327346802,
      "step": 347
    },
    {
      "epoch": 0.5260770975056689,
      "epsilon_dpo/beta": 0.03654114902019501,
      "epsilon_dpo/beta_margin_grad_mean": -0.4058833718299866,
      "epsilon_dpo/beta_margin_grad_std": 0.17744751274585724,
      "epsilon_dpo/beta_margin_mean": 0.4648371636867523,
      "epsilon_dpo/beta_margin_std": 0.8594658374786377,
      "epsilon_dpo/loss_margin_mean": 12.89057445526123,
      "grad_norm": 14.16436767578125,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.03662906959652901,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.724474525774229e-07,
      "logits/chosen": -0.21772196888923645,
      "logits/rejected": -0.3911592364311218,
      "logps/chosen": -92.62994384765625,
      "logps/ref_chosen": -75.35821533203125,
      "logps/ref_rejected": -90.65567016601562,
      "logps/rejected": -120.8179702758789,
      "loss": 1.136,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.6333014965057373,
      "rewards/margins": 0.4648371636867523,
      "rewards/rejected": -1.098138689994812,
      "step": 348
    },
    {
      "epoch": 0.527588813303099,
      "epsilon_dpo/beta": 0.03637009114027023,
      "epsilon_dpo/beta_margin_grad_mean": -0.40059003233909607,
      "epsilon_dpo/beta_margin_grad_std": 0.18574047088623047,
      "epsilon_dpo/beta_margin_mean": 0.4811083674430847,
      "epsilon_dpo/beta_margin_std": 0.910507321357727,
      "epsilon_dpo/loss_margin_mean": 13.37955093383789,
      "grad_norm": 11.531458854675293,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.03653772547841072,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.711302664252973e-07,
      "logits/chosen": -0.5206562280654907,
      "logits/rejected": -0.34496116638183594,
      "logps/chosen": -90.16403198242188,
      "logps/ref_chosen": -75.35797119140625,
      "logps/ref_rejected": -109.66375732421875,
      "logps/rejected": -137.849365234375,
      "loss": 1.1427,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.5402654409408569,
      "rewards/margins": 0.4811083674430847,
      "rewards/rejected": -1.0213738679885864,
      "step": 349
    },
    {
      "epoch": 0.5291005291005291,
      "epsilon_dpo/beta": 0.036143574863672256,
      "epsilon_dpo/beta_margin_grad_mean": -0.3563748002052307,
      "epsilon_dpo/beta_margin_grad_std": 0.14815759658813477,
      "epsilon_dpo/beta_margin_mean": 0.6667227149009705,
      "epsilon_dpo/beta_margin_std": 0.7182585597038269,
      "epsilon_dpo/loss_margin_mean": 18.544713973999023,
      "grad_norm": 14.325149536132812,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.036367256194353104,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.698124892141971e-07,
      "logits/chosen": -0.4628453254699707,
      "logits/rejected": -0.3299052119255066,
      "logps/chosen": -99.627197265625,
      "logps/ref_chosen": -84.2036361694336,
      "logps/ref_rejected": -115.74954223632812,
      "logps/rejected": -149.71783447265625,
      "loss": 0.9385,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.55904221534729,
      "rewards/margins": 0.6667227149009705,
      "rewards/rejected": -1.2257649898529053,
      "step": 350
    },
    {
      "epoch": 0.5306122448979592,
      "epsilon_dpo/beta": 0.03597555682063103,
      "epsilon_dpo/beta_margin_grad_mean": -0.4049746096134186,
      "epsilon_dpo/beta_margin_grad_std": 0.18007493019104004,
      "epsilon_dpo/beta_margin_mean": 0.4236089587211609,
      "epsilon_dpo/beta_margin_std": 0.8556088805198669,
      "epsilon_dpo/loss_margin_mean": 11.925423622131348,
      "grad_norm": 12.217904090881348,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.036141373217105865,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.6849415780518357e-07,
      "logits/chosen": -0.2202155888080597,
      "logits/rejected": -0.2810223698616028,
      "logps/chosen": -92.83936309814453,
      "logps/ref_chosen": -75.33352661132812,
      "logps/ref_rejected": -103.31353759765625,
      "logps/rejected": -132.74481201171875,
      "loss": 1.1721,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6318588852882385,
      "rewards/margins": 0.4236089289188385,
      "rewards/rejected": -1.0554677248001099,
      "step": 351
    },
    {
      "epoch": 0.5321239606953893,
      "epsilon_dpo/beta": 0.03585267439484596,
      "epsilon_dpo/beta_margin_grad_mean": -0.40181681513786316,
      "epsilon_dpo/beta_margin_grad_std": 0.1694328784942627,
      "epsilon_dpo/beta_margin_mean": 0.4756009876728058,
      "epsilon_dpo/beta_margin_std": 0.8291955590248108,
      "epsilon_dpo/loss_margin_mean": 13.421477317810059,
      "grad_norm": 10.798321723937988,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.03597274795174599,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.6717530907482024e-07,
      "logits/chosen": -0.3640022873878479,
      "logits/rejected": -0.23501214385032654,
      "logps/chosen": -105.84851837158203,
      "logps/ref_chosen": -87.23213195800781,
      "logps/ref_rejected": -112.02339172363281,
      "logps/rejected": -144.06124877929688,
      "loss": 1.1153,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.670296311378479,
      "rewards/margins": 0.4756010174751282,
      "rewards/rejected": -1.145897388458252,
      "step": 352
    },
    {
      "epoch": 0.5336356764928194,
      "epsilon_dpo/beta": 0.03561781346797943,
      "epsilon_dpo/beta_margin_grad_mean": -0.38533923029899597,
      "epsilon_dpo/beta_margin_grad_std": 0.18581627309322357,
      "epsilon_dpo/beta_margin_mean": 0.5274343490600586,
      "epsilon_dpo/beta_margin_std": 0.9047785401344299,
      "epsilon_dpo/loss_margin_mean": 14.941997528076172,
      "grad_norm": 11.600533485412598,
      "kl/avg_steps": 0.65625,
      "kl/beta": 0.03584951534867287,
      "kl/n_epsilon_steps": 0.171875,
      "kl/p_epsilon_steps": 0.828125,
      "learning_rate": 2.658559799141411e-07,
      "logits/chosen": -0.31952571868896484,
      "logits/rejected": -0.42992115020751953,
      "logps/chosen": -104.45259857177734,
      "logps/ref_chosen": -90.74217224121094,
      "logps/ref_rejected": -93.2311019897461,
      "logps/rejected": -121.88351440429688,
      "loss": 1.1077,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.49013230204582214,
      "rewards/margins": 0.5274343490600586,
      "rewards/rejected": -1.0175666809082031,
      "step": 353
    },
    {
      "epoch": 0.5351473922902494,
      "epsilon_dpo/beta": 0.035452380776405334,
      "epsilon_dpo/beta_margin_grad_mean": -0.3861326277256012,
      "epsilon_dpo/beta_margin_grad_std": 0.18367260694503784,
      "epsilon_dpo/beta_margin_mean": 0.5246230959892273,
      "epsilon_dpo/beta_margin_std": 0.8698256015777588,
      "epsilon_dpo/loss_margin_mean": 14.963334083557129,
      "grad_norm": 11.076436042785645,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.03561578691005707,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.6453620722761895e-07,
      "logits/chosen": -0.503691554069519,
      "logits/rejected": -0.20610500872135162,
      "logps/chosen": -73.68644714355469,
      "logps/ref_chosen": -59.317867279052734,
      "logps/ref_rejected": -106.89750671386719,
      "logps/rejected": -136.2294158935547,
      "loss": 1.0983,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.511963963508606,
      "rewards/margins": 0.5246231555938721,
      "rewards/rejected": -1.036587119102478,
      "step": 354
    },
    {
      "epoch": 0.5366591080876795,
      "epsilon_dpo/beta": 0.0352315790951252,
      "epsilon_dpo/beta_margin_grad_mean": -0.3844098150730133,
      "epsilon_dpo/beta_margin_grad_std": 0.17795078456401825,
      "epsilon_dpo/beta_margin_mean": 0.5513162612915039,
      "epsilon_dpo/beta_margin_std": 0.868008017539978,
      "epsilon_dpo/loss_margin_mean": 15.773537635803223,
      "grad_norm": 11.321294784545898,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.035449616611003876,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.632160279321328e-07,
      "logits/chosen": -0.2635769546031952,
      "logits/rejected": -0.6916155815124512,
      "logps/chosen": -80.50833892822266,
      "logps/ref_chosen": -66.28169250488281,
      "logps/ref_rejected": -107.13027954101562,
      "logps/rejected": -137.13046264648438,
      "loss": 1.0733,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.5025330185890198,
      "rewards/margins": 0.5513162612915039,
      "rewards/rejected": -1.053849220275879,
      "step": 355
    },
    {
      "epoch": 0.5381708238851096,
      "epsilon_dpo/beta": 0.03508982062339783,
      "epsilon_dpo/beta_margin_grad_mean": -0.3791634738445282,
      "epsilon_dpo/beta_margin_grad_std": 0.1978328675031662,
      "epsilon_dpo/beta_margin_mean": 0.5767776966094971,
      "epsilon_dpo/beta_margin_std": 0.9672296047210693,
      "epsilon_dpo/loss_margin_mean": 16.622722625732422,
      "grad_norm": 14.604504585266113,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.035229433327913284,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.618954789559356e-07,
      "logits/chosen": -0.3518662452697754,
      "logits/rejected": -0.28898200392723083,
      "logps/chosen": -76.50581359863281,
      "logps/ref_chosen": -66.96154022216797,
      "logps/ref_rejected": -96.09982299804688,
      "logps/rejected": -122.26683044433594,
      "loss": 1.0952,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.33804744482040405,
      "rewards/margins": 0.5767776966094971,
      "rewards/rejected": -0.9148250818252563,
      "step": 356
    },
    {
      "epoch": 0.5396825396825397,
      "epsilon_dpo/beta": 0.03495881333947182,
      "epsilon_dpo/beta_margin_grad_mean": -0.404715359210968,
      "epsilon_dpo/beta_margin_grad_std": 0.18092228472232819,
      "epsilon_dpo/beta_margin_mean": 0.45309287309646606,
      "epsilon_dpo/beta_margin_std": 0.8744781017303467,
      "epsilon_dpo/loss_margin_mean": 13.121901512145996,
      "grad_norm": 12.346394538879395,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.035086892545223236,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.6057459723762076e-07,
      "logits/chosen": -0.34115922451019287,
      "logits/rejected": -0.4912487864494324,
      "logps/chosen": -104.1282958984375,
      "logps/ref_chosen": -88.78244018554688,
      "logps/ref_rejected": -93.55414581298828,
      "logps/rejected": -122.02190399169922,
      "loss": 1.1528,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.5402135252952576,
      "rewards/margins": 0.45309287309646606,
      "rewards/rejected": -0.9933063983917236,
      "step": 357
    },
    {
      "epoch": 0.5411942554799698,
      "epsilon_dpo/beta": 0.03479543328285217,
      "epsilon_dpo/beta_margin_grad_mean": -0.3826284110546112,
      "epsilon_dpo/beta_margin_grad_std": 0.17101824283599854,
      "epsilon_dpo/beta_margin_mean": 0.5514477491378784,
      "epsilon_dpo/beta_margin_std": 0.8156856298446655,
      "epsilon_dpo/loss_margin_mean": 15.996456146240234,
      "grad_norm": 12.181327819824219,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.03495581075549126,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.5925341972508954e-07,
      "logits/chosen": -0.5567905306816101,
      "logits/rejected": -0.4287104606628418,
      "logps/chosen": -105.34233093261719,
      "logps/ref_chosen": -92.37928771972656,
      "logps/ref_rejected": -90.57598876953125,
      "logps/rejected": -119.53549194335938,
      "loss": 1.0559,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.4534444808959961,
      "rewards/margins": 0.5514477491378784,
      "rewards/rejected": -1.004892349243164,
      "step": 358
    },
    {
      "epoch": 0.5427059712773998,
      "epsilon_dpo/beta": 0.03469833359122276,
      "epsilon_dpo/beta_margin_grad_mean": -0.4318692982196808,
      "epsilon_dpo/beta_margin_grad_std": 0.18369096517562866,
      "epsilon_dpo/beta_margin_mean": 0.32085007429122925,
      "epsilon_dpo/beta_margin_std": 0.8615565896034241,
      "epsilon_dpo/loss_margin_mean": 9.426478385925293,
      "grad_norm": 12.058820724487305,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.034792717546224594,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 2.579319833745169e-07,
      "logits/chosen": -0.6475515365600586,
      "logits/rejected": -0.38479065895080566,
      "logps/chosen": -100.14118194580078,
      "logps/ref_chosen": -85.90824127197266,
      "logps/ref_rejected": -104.04725646972656,
      "logps/rejected": -127.70668029785156,
      "loss": 1.2595,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.49748337268829346,
      "rewards/margins": 0.32085007429122925,
      "rewards/rejected": -0.8183333873748779,
      "step": 359
    },
    {
      "epoch": 0.54421768707483,
      "epsilon_dpo/beta": 0.03460101783275604,
      "epsilon_dpo/beta_margin_grad_mean": -0.4063347578048706,
      "epsilon_dpo/beta_margin_grad_std": 0.17292876541614532,
      "epsilon_dpo/beta_margin_mean": 0.4609655439853668,
      "epsilon_dpo/beta_margin_std": 0.8432448506355286,
      "epsilon_dpo/loss_margin_mean": 13.486247062683105,
      "grad_norm": 10.786174774169922,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.03469513729214668,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 2.5661032514931834e-07,
      "logits/chosen": -0.3451042175292969,
      "logits/rejected": -0.5038284063339233,
      "logps/chosen": -98.89404296875,
      "logps/ref_chosen": -84.124267578125,
      "logps/ref_rejected": -118.12923431396484,
      "logps/rejected": -146.38525390625,
      "loss": 1.1324,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.5138739943504333,
      "rewards/margins": 0.46096551418304443,
      "rewards/rejected": -0.9748395085334778,
      "step": 360
    },
    {
      "epoch": 0.54572940287226,
      "epsilon_dpo/beta": 0.03442828729748726,
      "epsilon_dpo/beta_margin_grad_mean": -0.3710825443267822,
      "epsilon_dpo/beta_margin_grad_std": 0.16406892240047455,
      "epsilon_dpo/beta_margin_mean": 0.6091728806495667,
      "epsilon_dpo/beta_margin_std": 0.8011631369590759,
      "epsilon_dpo/loss_margin_mean": 17.82909393310547,
      "grad_norm": 9.879510879516602,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.03459783270955086,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 2.552884820191154e-07,
      "logits/chosen": -0.25418639183044434,
      "logits/rejected": -0.4245424270629883,
      "logps/chosen": -92.85491180419922,
      "logps/ref_chosen": -79.35969543457031,
      "logps/ref_rejected": -100.64299011230469,
      "logps/rejected": -131.96731567382812,
      "loss": 1.0057,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.4663471281528473,
      "rewards/margins": 0.6091729402542114,
      "rewards/rejected": -1.0755200386047363,
      "step": 361
    },
    {
      "epoch": 0.54724111866969,
      "epsilon_dpo/beta": 0.034240804612636566,
      "epsilon_dpo/beta_margin_grad_mean": -0.41057583689689636,
      "epsilon_dpo/beta_margin_grad_std": 0.1774662286043167,
      "epsilon_dpo/beta_margin_mean": 0.424104779958725,
      "epsilon_dpo/beta_margin_std": 0.8874197006225586,
      "epsilon_dpo/loss_margin_mean": 12.533075332641602,
      "grad_norm": 13.939665794372559,
      "kl/avg_steps": 0.546875,
      "kl/beta": 0.034425701946020126,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 2.53966490958702e-07,
      "logits/chosen": -0.49522465467453003,
      "logits/rejected": -0.8078896999359131,
      "logps/chosen": -104.72351837158203,
      "logps/ref_chosen": -87.18423461914062,
      "logps/ref_rejected": -131.90994262695312,
      "logps/rejected": -161.9822998046875,
      "loss": 1.1749,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6019046306610107,
      "rewards/margins": 0.42410480976104736,
      "rewards/rejected": -1.026009440422058,
      "step": 362
    },
    {
      "epoch": 0.5487528344671202,
      "epsilon_dpo/beta": 0.03398507088422775,
      "epsilon_dpo/beta_margin_grad_mean": -0.3623369038105011,
      "epsilon_dpo/beta_margin_grad_std": 0.15423668920993805,
      "epsilon_dpo/beta_margin_mean": 0.6261187195777893,
      "epsilon_dpo/beta_margin_std": 0.7368482947349548,
      "epsilon_dpo/loss_margin_mean": 18.516027450561523,
      "grad_norm": 12.303284645080566,
      "kl/avg_steps": 0.75,
      "kl/beta": 0.034238461405038834,
      "kl/n_epsilon_steps": 0.125,
      "kl/p_epsilon_steps": 0.875,
      "learning_rate": 2.526443889470099e-07,
      "logits/chosen": -0.2638314664363861,
      "logits/rejected": -0.3185631334781647,
      "logps/chosen": -84.76541900634766,
      "logps/ref_chosen": -69.48241424560547,
      "logps/ref_rejected": -115.17513275146484,
      "logps/rejected": -148.9741668701172,
      "loss": 0.9752,
      "rewards/accuracies": 0.875,
      "rewards/chosen": -0.5202293395996094,
      "rewards/margins": 0.6261187791824341,
      "rewards/rejected": -1.146347999572754,
      "step": 363
    },
    {
      "epoch": 0.5502645502645502,
      "epsilon_dpo/beta": 0.0338701531291008,
      "epsilon_dpo/beta_margin_grad_mean": -0.37633389234542847,
      "epsilon_dpo/beta_margin_grad_std": 0.18905964493751526,
      "epsilon_dpo/beta_margin_mean": 0.6139385104179382,
      "epsilon_dpo/beta_margin_std": 0.9477400183677673,
      "epsilon_dpo/loss_margin_mean": 18.3122501373291,
      "grad_norm": 13.421125411987305,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.033983584493398666,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.513222129660744e-07,
      "logits/chosen": -0.2861930727958679,
      "logits/rejected": -0.30931782722473145,
      "logps/chosen": -93.13960266113281,
      "logps/ref_chosen": -81.23245239257812,
      "logps/ref_rejected": -109.52008056640625,
      "logps/rejected": -139.73948669433594,
      "loss": 1.0549,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.4042242765426636,
      "rewards/margins": 0.613938570022583,
      "rewards/rejected": -1.018162727355957,
      "step": 364
    },
    {
      "epoch": 0.5517762660619804,
      "epsilon_dpo/beta": 0.0336800292134285,
      "epsilon_dpo/beta_margin_grad_mean": -0.3741872012615204,
      "epsilon_dpo/beta_margin_grad_std": 0.15171965956687927,
      "epsilon_dpo/beta_margin_mean": 0.5734090209007263,
      "epsilon_dpo/beta_margin_std": 0.7196288704872131,
      "epsilon_dpo/loss_margin_mean": 17.134653091430664,
      "grad_norm": 12.334189414978027,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.03386716544628143,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 2.5e-07,
      "logits/chosen": -0.44367483258247375,
      "logits/rejected": -0.374923974275589,
      "logps/chosen": -95.52092742919922,
      "logps/ref_chosen": -88.40348815917969,
      "logps/ref_rejected": -93.00691986083984,
      "logps/rejected": -117.25901794433594,
      "loss": 1.0079,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.24103927612304688,
      "rewards/margins": 0.5734090209007263,
      "rewards/rejected": -0.8144482970237732,
      "step": 365
    },
    {
      "epoch": 0.5532879818594104,
      "epsilon_dpo/beta": 0.033586371690034866,
      "epsilon_dpo/beta_margin_grad_mean": -0.40513846278190613,
      "epsilon_dpo/beta_margin_grad_std": 0.1917259395122528,
      "epsilon_dpo/beta_margin_mean": 0.47303882241249084,
      "epsilon_dpo/beta_margin_std": 0.9454165101051331,
      "epsilon_dpo/loss_margin_mean": 14.272467613220215,
      "grad_norm": 10.503172874450684,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.033677730709314346,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 2.486777870339255e-07,
      "logits/chosen": -0.5803329944610596,
      "logits/rejected": -0.5538395643234253,
      "logps/chosen": -87.64436340332031,
      "logps/ref_chosen": -77.08778381347656,
      "logps/ref_rejected": -88.84114074707031,
      "logps/rejected": -113.67018127441406,
      "loss": 1.1624,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.3563389480113983,
      "rewards/margins": 0.47303879261016846,
      "rewards/rejected": -0.8293777704238892,
      "step": 366
    },
    {
      "epoch": 0.5547996976568406,
      "epsilon_dpo/beta": 0.03344488888978958,
      "epsilon_dpo/beta_margin_grad_mean": -0.41342422366142273,
      "epsilon_dpo/beta_margin_grad_std": 0.16751711070537567,
      "epsilon_dpo/beta_margin_mean": 0.3815315067768097,
      "epsilon_dpo/beta_margin_std": 0.7867935299873352,
      "epsilon_dpo/loss_margin_mean": 11.564186096191406,
      "grad_norm": 10.828190803527832,
      "kl/avg_steps": 0.421875,
      "kl/beta": 0.033583275973796844,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.4735561105299014e-07,
      "logits/chosen": -0.613606333732605,
      "logits/rejected": -0.5410276651382446,
      "logps/chosen": -92.40072631835938,
      "logps/ref_chosen": -77.62783813476562,
      "logps/ref_rejected": -104.18344116210938,
      "logps/rejected": -130.52052307128906,
      "loss": 1.1813,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.49686622619628906,
      "rewards/margins": 0.3815315365791321,
      "rewards/rejected": -0.8783978223800659,
      "step": 367
    },
    {
      "epoch": 0.5563114134542706,
      "epsilon_dpo/beta": 0.03333056718111038,
      "epsilon_dpo/beta_margin_grad_mean": -0.4020618498325348,
      "epsilon_dpo/beta_margin_grad_std": 0.18135952949523926,
      "epsilon_dpo/beta_margin_mean": 0.4608325660228729,
      "epsilon_dpo/beta_margin_std": 0.8643712401390076,
      "epsilon_dpo/loss_margin_mean": 14.009523391723633,
      "grad_norm": 10.81441593170166,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.033442191779613495,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.46033509041298e-07,
      "logits/chosen": -0.43107640743255615,
      "logits/rejected": -0.27693885564804077,
      "logps/chosen": -113.70527648925781,
      "logps/ref_chosen": -96.74423217773438,
      "logps/ref_rejected": -98.10311889648438,
      "logps/rejected": -129.0736846923828,
      "loss": 1.1441,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.5687465667724609,
      "rewards/margins": 0.4608325660228729,
      "rewards/rejected": -1.0295791625976562,
      "step": 368
    },
    {
      "epoch": 0.5578231292517006,
      "epsilon_dpo/beta": 0.033211126923561096,
      "epsilon_dpo/beta_margin_grad_mean": -0.4330996870994568,
      "epsilon_dpo/beta_margin_grad_std": 0.16043256223201752,
      "epsilon_dpo/beta_margin_mean": 0.30416300892829895,
      "epsilon_dpo/beta_margin_std": 0.7585687637329102,
      "epsilon_dpo/loss_margin_mean": 9.304594039916992,
      "grad_norm": 12.7125244140625,
      "kl/avg_steps": 0.359375,
      "kl/beta": 0.033327627927064896,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.447115179808846e-07,
      "logits/chosen": -0.7121263742446899,
      "logits/rejected": -0.6011459827423096,
      "logps/chosen": -91.87727355957031,
      "logps/ref_chosen": -77.4908676147461,
      "logps/ref_rejected": -105.63996887207031,
      "logps/rejected": -129.33096313476562,
      "loss": 1.2349,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.47804659605026245,
      "rewards/margins": 0.30416297912597656,
      "rewards/rejected": -0.782209575176239,
      "step": 369
    },
    {
      "epoch": 0.5593348450491308,
      "epsilon_dpo/beta": 0.033076684921979904,
      "epsilon_dpo/beta_margin_grad_mean": -0.36861076951026917,
      "epsilon_dpo/beta_margin_grad_std": 0.1980779469013214,
      "epsilon_dpo/beta_margin_mean": 0.6260709166526794,
      "epsilon_dpo/beta_margin_std": 0.9714806079864502,
      "epsilon_dpo/loss_margin_mean": 19.130352020263672,
      "grad_norm": 15.036335945129395,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.03320828452706337,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.4338967485068164e-07,
      "logits/chosen": -0.37563467025756836,
      "logits/rejected": -0.16301874816417694,
      "logps/chosen": -86.26065063476562,
      "logps/ref_chosen": -73.92730712890625,
      "logps/ref_rejected": -99.21125030517578,
      "logps/rejected": -130.67494201660156,
      "loss": 1.0611,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.412090003490448,
      "rewards/margins": 0.6260708570480347,
      "rewards/rejected": -1.0381609201431274,
      "step": 370
    },
    {
      "epoch": 0.5608465608465608,
      "epsilon_dpo/beta": 0.03293251991271973,
      "epsilon_dpo/beta_margin_grad_mean": -0.3986269235610962,
      "epsilon_dpo/beta_margin_grad_std": 0.1933879405260086,
      "epsilon_dpo/beta_margin_mean": 0.48927295207977295,
      "epsilon_dpo/beta_margin_std": 0.940234899520874,
      "epsilon_dpo/loss_margin_mean": 15.045866012573242,
      "grad_norm": 16.158891677856445,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.033073924481868744,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.420680166254831e-07,
      "logits/chosen": -0.33735889196395874,
      "logits/rejected": -0.2781010568141937,
      "logps/chosen": -84.40055847167969,
      "logps/ref_chosen": -70.58553314208984,
      "logps/ref_rejected": -84.94666290283203,
      "logps/rejected": -113.80755615234375,
      "loss": 1.1498,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.4571009874343872,
      "rewards/margins": 0.48927295207977295,
      "rewards/rejected": -0.9463739395141602,
      "step": 371
    },
    {
      "epoch": 0.562358276643991,
      "epsilon_dpo/beta": 0.03283023461699486,
      "epsilon_dpo/beta_margin_grad_mean": -0.4104512333869934,
      "epsilon_dpo/beta_margin_grad_std": 0.19789732992649078,
      "epsilon_dpo/beta_margin_mean": 0.42812153697013855,
      "epsilon_dpo/beta_margin_std": 0.9525101184844971,
      "epsilon_dpo/loss_margin_mean": 13.248870849609375,
      "grad_norm": 11.11093807220459,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.032929856330156326,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 2.4074658027491044e-07,
      "logits/chosen": -0.41160130500793457,
      "logits/rejected": -0.47164422273635864,
      "logps/chosen": -86.42536163330078,
      "logps/ref_chosen": -74.52593994140625,
      "logps/ref_rejected": -100.20262145996094,
      "logps/rejected": -125.35091400146484,
      "loss": 1.2052,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.39410853385925293,
      "rewards/margins": 0.42812150716781616,
      "rewards/rejected": -0.8222300410270691,
      "step": 372
    },
    {
      "epoch": 0.563869992441421,
      "epsilon_dpo/beta": 0.03270743787288666,
      "epsilon_dpo/beta_margin_grad_mean": -0.4025079607963562,
      "epsilon_dpo/beta_margin_grad_std": 0.18594452738761902,
      "epsilon_dpo/beta_margin_mean": 0.4715001881122589,
      "epsilon_dpo/beta_margin_std": 0.8927410244941711,
      "epsilon_dpo/loss_margin_mean": 14.600152969360352,
      "grad_norm": 12.162555694580078,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.03282726928591728,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.394254027623792e-07,
      "logits/chosen": -0.4808266758918762,
      "logits/rejected": -0.5573515892028809,
      "logps/chosen": -105.85910034179688,
      "logps/ref_chosen": -87.2992935180664,
      "logps/ref_rejected": -109.5474853515625,
      "logps/rejected": -142.7074432373047,
      "loss": 1.1456,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6099110245704651,
      "rewards/margins": 0.4715002179145813,
      "rewards/rejected": -1.0814111232757568,
      "step": 373
    },
    {
      "epoch": 0.5653817082388511,
      "epsilon_dpo/beta": 0.03252391517162323,
      "epsilon_dpo/beta_margin_grad_mean": -0.3559419512748718,
      "epsilon_dpo/beta_margin_grad_std": 0.18340633809566498,
      "epsilon_dpo/beta_margin_mean": 0.6854166984558105,
      "epsilon_dpo/beta_margin_std": 0.9182686805725098,
      "epsilon_dpo/loss_margin_mean": 21.238224029541016,
      "grad_norm": 10.829121589660645,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.03270462900400162,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 2.381045210440644e-07,
      "logits/chosen": -0.33610478043556213,
      "logits/rejected": -0.38159722089767456,
      "logps/chosen": -118.37493896484375,
      "logps/ref_chosen": -103.18836212158203,
      "logps/ref_rejected": -101.68389892578125,
      "logps/rejected": -138.10870361328125,
      "loss": 0.9944,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.49699217081069946,
      "rewards/margins": 0.6854166984558105,
      "rewards/rejected": -1.1824089288711548,
      "step": 374
    },
    {
      "epoch": 0.5668934240362812,
      "epsilon_dpo/beta": 0.03239784389734268,
      "epsilon_dpo/beta_margin_grad_mean": -0.39941704273223877,
      "epsilon_dpo/beta_margin_grad_std": 0.1892874836921692,
      "epsilon_dpo/beta_margin_mean": 0.4690830409526825,
      "epsilon_dpo/beta_margin_std": 0.890160858631134,
      "epsilon_dpo/loss_margin_mean": 14.67497730255127,
      "grad_norm": 12.20340633392334,
      "kl/avg_steps": 0.390625,
      "kl/beta": 0.03252169117331505,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.3678397206786715e-07,
      "logits/chosen": -0.3479893207550049,
      "logits/rejected": -0.14140911400318146,
      "logps/chosen": -90.7242660522461,
      "logps/ref_chosen": -80.96290588378906,
      "logps/ref_rejected": -105.99746704101562,
      "logps/rejected": -130.43380737304688,
      "loss": 1.1496,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.31886184215545654,
      "rewards/margins": 0.46908310055732727,
      "rewards/rejected": -0.7879449129104614,
      "step": 375
    },
    {
      "epoch": 0.5684051398337112,
      "epsilon_dpo/beta": 0.0322667732834816,
      "epsilon_dpo/beta_margin_grad_mean": -0.3862917721271515,
      "epsilon_dpo/beta_margin_grad_std": 0.19730228185653687,
      "epsilon_dpo/beta_margin_mean": 0.5730307698249817,
      "epsilon_dpo/beta_margin_std": 0.984718918800354,
      "epsilon_dpo/loss_margin_mean": 17.959718704223633,
      "grad_norm": 10.152912139892578,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.032395150512456894,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.3546379277238103e-07,
      "logits/chosen": -0.24255940318107605,
      "logits/rejected": -0.46518683433532715,
      "logps/chosen": -86.96746826171875,
      "logps/ref_chosen": -73.2122573852539,
      "logps/ref_rejected": -96.98638916015625,
      "logps/rejected": -128.70132446289062,
      "loss": 1.1006,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.44633597135543823,
      "rewards/margins": 0.5730308294296265,
      "rewards/rejected": -1.01936674118042,
      "step": 376
    },
    {
      "epoch": 0.5699168556311414,
      "epsilon_dpo/beta": 0.032126132398843765,
      "epsilon_dpo/beta_margin_grad_mean": -0.4142596423625946,
      "epsilon_dpo/beta_margin_grad_std": 0.16649965941905975,
      "epsilon_dpo/beta_margin_mean": 0.4061944782733917,
      "epsilon_dpo/beta_margin_std": 0.7875421047210693,
      "epsilon_dpo/loss_margin_mean": 12.792858123779297,
      "grad_norm": 11.573308944702148,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.03226407617330551,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.3414402008585886e-07,
      "logits/chosen": -0.2805550694465637,
      "logits/rejected": -0.31607991456985474,
      "logps/chosen": -85.04446411132812,
      "logps/ref_chosen": -68.43621063232422,
      "logps/ref_rejected": -80.13630676269531,
      "logps/rejected": -109.53741455078125,
      "loss": 1.1594,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5347778797149658,
      "rewards/margins": 0.4061944782733917,
      "rewards/rejected": -0.9409723281860352,
      "step": 377
    },
    {
      "epoch": 0.5714285714285714,
      "epsilon_dpo/beta": 0.03196611627936363,
      "epsilon_dpo/beta_margin_grad_mean": -0.41235488653182983,
      "epsilon_dpo/beta_margin_grad_std": 0.17046606540679932,
      "epsilon_dpo/beta_margin_mean": 0.4097154140472412,
      "epsilon_dpo/beta_margin_std": 0.8452909588813782,
      "epsilon_dpo/loss_margin_mean": 12.954933166503906,
      "grad_norm": 11.13086986541748,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.03212353587150574,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 2.3282469092517977e-07,
      "logits/chosen": -0.3559729754924774,
      "logits/rejected": -0.3680075407028198,
      "logps/chosen": -99.54299926757812,
      "logps/ref_chosen": -83.83049011230469,
      "logps/ref_rejected": -101.34911346435547,
      "logps/rejected": -130.0165557861328,
      "loss": 1.1737,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5037399530410767,
      "rewards/margins": 0.4097154140472412,
      "rewards/rejected": -0.9134553074836731,
      "step": 378
    },
    {
      "epoch": 0.5729402872260015,
      "epsilon_dpo/beta": 0.031827058643102646,
      "epsilon_dpo/beta_margin_grad_mean": -0.3888133764266968,
      "epsilon_dpo/beta_margin_grad_std": 0.18203434348106384,
      "epsilon_dpo/beta_margin_mean": 0.5286673307418823,
      "epsilon_dpo/beta_margin_std": 0.8799561262130737,
      "epsilon_dpo/loss_margin_mean": 16.793392181396484,
      "grad_norm": 11.439459800720215,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.03196371719241142,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.3150584219481643e-07,
      "logits/chosen": -0.6511602997779846,
      "logits/rejected": -0.6409453749656677,
      "logps/chosen": -112.08035278320312,
      "logps/ref_chosen": -99.05479431152344,
      "logps/ref_rejected": -126.27124786376953,
      "logps/rejected": -156.09019470214844,
      "loss": 1.0957,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4176122844219208,
      "rewards/margins": 0.5286673903465271,
      "rewards/rejected": -0.9462796449661255,
      "step": 379
    },
    {
      "epoch": 0.5744520030234316,
      "epsilon_dpo/beta": 0.03164863586425781,
      "epsilon_dpo/beta_margin_grad_mean": -0.36535948514938354,
      "epsilon_dpo/beta_margin_grad_std": 0.18868181109428406,
      "epsilon_dpo/beta_margin_mean": 0.6302640438079834,
      "epsilon_dpo/beta_margin_std": 0.9209726452827454,
      "epsilon_dpo/loss_margin_mean": 20.09124183654785,
      "grad_norm": 9.986943244934082,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.03182448446750641,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 2.3018751078580283e-07,
      "logits/chosen": -0.31451958417892456,
      "logits/rejected": -0.2750563621520996,
      "logps/chosen": -93.44203186035156,
      "logps/ref_chosen": -83.15869140625,
      "logps/ref_rejected": -100.08312225341797,
      "logps/rejected": -130.45770263671875,
      "loss": 1.0381,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.32757604122161865,
      "rewards/margins": 0.6302640438079834,
      "rewards/rejected": -0.957840085029602,
      "step": 380
    },
    {
      "epoch": 0.5759637188208617,
      "epsilon_dpo/beta": 0.031570516526699066,
      "epsilon_dpo/beta_margin_grad_mean": -0.4260694086551666,
      "epsilon_dpo/beta_margin_grad_std": 0.17625002562999725,
      "epsilon_dpo/beta_margin_mean": 0.3538646697998047,
      "epsilon_dpo/beta_margin_std": 0.8401235342025757,
      "epsilon_dpo/loss_margin_mean": 11.38886833190918,
      "grad_norm": 10.793598175048828,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.031646475195884705,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.288697335747027e-07,
      "logits/chosen": -0.29458528757095337,
      "logits/rejected": -0.4564410448074341,
      "logps/chosen": -97.15643310546875,
      "logps/ref_chosen": -81.04007720947266,
      "logps/ref_rejected": -89.97328186035156,
      "logps/rejected": -117.47850036621094,
      "loss": 1.2213,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.5116348266601562,
      "rewards/margins": 0.3538646996021271,
      "rewards/rejected": -0.8654994964599609,
      "step": 381
    },
    {
      "epoch": 0.5774754346182918,
      "epsilon_dpo/beta": 0.03141285851597786,
      "epsilon_dpo/beta_margin_grad_mean": -0.40493854880332947,
      "epsilon_dpo/beta_margin_grad_std": 0.1608555167913437,
      "epsilon_dpo/beta_margin_mean": 0.4543093740940094,
      "epsilon_dpo/beta_margin_std": 0.7822091579437256,
      "epsilon_dpo/loss_margin_mean": 14.60008430480957,
      "grad_norm": 13.035445213317871,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.03156755492091179,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 2.2755254742257706e-07,
      "logits/chosen": -0.2647457718849182,
      "logits/rejected": -0.5761418342590332,
      "logps/chosen": -99.72867584228516,
      "logps/ref_chosen": -83.58617401123047,
      "logps/ref_rejected": -105.45875549316406,
      "logps/rejected": -136.2013397216797,
      "loss": 1.1163,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5082204341888428,
      "rewards/margins": 0.4543094038963318,
      "rewards/rejected": -0.9625298380851746,
      "step": 382
    },
    {
      "epoch": 0.5789871504157218,
      "epsilon_dpo/beta": 0.031354743987321854,
      "epsilon_dpo/beta_margin_grad_mean": -0.4083390533924103,
      "epsilon_dpo/beta_margin_grad_std": 0.18964853882789612,
      "epsilon_dpo/beta_margin_mean": 0.45374611020088196,
      "epsilon_dpo/beta_margin_std": 0.9225190281867981,
      "epsilon_dpo/loss_margin_mean": 14.688213348388672,
      "grad_norm": 12.959400177001953,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.03141050413250923,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 2.2623598917395436e-07,
      "logits/chosen": -0.4968550503253937,
      "logits/rejected": -0.37411147356033325,
      "logps/chosen": -116.3802719116211,
      "logps/ref_chosen": -102.12741088867188,
      "logps/ref_rejected": -100.01849365234375,
      "logps/rejected": -128.95956420898438,
      "loss": 1.1697,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.45026350021362305,
      "rewards/margins": 0.45374611020088196,
      "rewards/rejected": -0.9040095806121826,
      "step": 383
    },
    {
      "epoch": 0.5804988662131519,
      "epsilon_dpo/beta": 0.03119313158094883,
      "epsilon_dpo/beta_margin_grad_mean": -0.3977428674697876,
      "epsilon_dpo/beta_margin_grad_std": 0.17535927891731262,
      "epsilon_dpo/beta_margin_mean": 0.48126453161239624,
      "epsilon_dpo/beta_margin_std": 0.8371120691299438,
      "epsilon_dpo/loss_margin_mean": 15.583664894104004,
      "grad_norm": 11.489479064941406,
      "kl/avg_steps": 0.515625,
      "kl/beta": 0.0313517190515995,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 2.2492009565579875e-07,
      "logits/chosen": -0.0987042710185051,
      "logits/rejected": -0.31251096725463867,
      "logps/chosen": -102.03778076171875,
      "logps/ref_chosen": -86.96634674072266,
      "logps/ref_rejected": -101.53166198730469,
      "logps/rejected": -132.18675231933594,
      "loss": 1.1173,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.47207605838775635,
      "rewards/margins": 0.48126456141471863,
      "rewards/rejected": -0.9533406496047974,
      "step": 384
    },
    {
      "epoch": 0.582010582010582,
      "epsilon_dpo/beta": 0.031077032908797264,
      "epsilon_dpo/beta_margin_grad_mean": -0.3920232951641083,
      "epsilon_dpo/beta_margin_grad_std": 0.1606670767068863,
      "epsilon_dpo/beta_margin_mean": 0.5069488286972046,
      "epsilon_dpo/beta_margin_std": 0.7647234797477722,
      "epsilon_dpo/loss_margin_mean": 16.471561431884766,
      "grad_norm": 14.488018989562988,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.031190890818834305,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.2360490367648084e-07,
      "logits/chosen": -0.45997345447540283,
      "logits/rejected": -0.5446884632110596,
      "logps/chosen": -106.83857727050781,
      "logps/ref_chosen": -91.81571960449219,
      "logps/ref_rejected": -110.62493896484375,
      "logps/rejected": -142.11935424804688,
      "loss": 1.0717,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.46806877851486206,
      "rewards/margins": 0.5069488286972046,
      "rewards/rejected": -0.9750176668167114,
      "step": 385
    },
    {
      "epoch": 0.5835222978080121,
      "epsilon_dpo/beta": 0.03096092864871025,
      "epsilon_dpo/beta_margin_grad_mean": -0.42493200302124023,
      "epsilon_dpo/beta_margin_grad_std": 0.17903663218021393,
      "epsilon_dpo/beta_margin_mean": 0.34691476821899414,
      "epsilon_dpo/beta_margin_std": 0.8625679016113281,
      "epsilon_dpo/loss_margin_mean": 11.385979652404785,
      "grad_norm": 12.905364990234375,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.031074361875653267,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.2229045002474724e-07,
      "logits/chosen": -0.26814377307891846,
      "logits/rejected": -0.4823966324329376,
      "logps/chosen": -116.19183349609375,
      "logps/ref_chosen": -99.04656982421875,
      "logps/ref_rejected": -122.45169067382812,
      "logps/rejected": -150.98294067382812,
      "loss": 1.2355,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5320286750793457,
      "rewards/margins": 0.34691473841667175,
      "rewards/rejected": -0.8789434432983398,
      "step": 386
    },
    {
      "epoch": 0.5850340136054422,
      "epsilon_dpo/beta": 0.030787205323576927,
      "epsilon_dpo/beta_margin_grad_mean": -0.3811575174331665,
      "epsilon_dpo/beta_margin_grad_std": 0.17074567079544067,
      "epsilon_dpo/beta_margin_mean": 0.555602490901947,
      "epsilon_dpo/beta_margin_std": 0.809950590133667,
      "epsilon_dpo/loss_margin_mean": 18.20049285888672,
      "grad_norm": 10.500995635986328,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.030958266928792,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 2.209767714686924e-07,
      "logits/chosen": -0.4561399519443512,
      "logits/rejected": -0.2552209794521332,
      "logps/chosen": -83.26592254638672,
      "logps/ref_chosen": -71.10548400878906,
      "logps/ref_rejected": -108.16309356689453,
      "logps/rejected": -138.52401733398438,
      "loss": 1.0512,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.3770449757575989,
      "rewards/margins": 0.555602490901947,
      "rewards/rejected": -0.9326474666595459,
      "step": 387
    },
    {
      "epoch": 0.5865457294028723,
      "epsilon_dpo/beta": 0.030687108635902405,
      "epsilon_dpo/beta_margin_grad_mean": -0.4308612644672394,
      "epsilon_dpo/beta_margin_grad_std": 0.1636020541191101,
      "epsilon_dpo/beta_margin_mean": 0.31852632761001587,
      "epsilon_dpo/beta_margin_std": 0.7362657785415649,
      "epsilon_dpo/loss_margin_mean": 10.55184555053711,
      "grad_norm": 14.64468765258789,
      "kl/avg_steps": 0.328125,
      "kl/beta": 0.030785102397203445,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 2.1966390475472954e-07,
      "logits/chosen": -0.3645251393318176,
      "logits/rejected": -0.34023189544677734,
      "logps/chosen": -116.99144744873047,
      "logps/ref_chosen": -101.09429931640625,
      "logps/ref_rejected": -107.06742858886719,
      "logps/rejected": -133.51641845703125,
      "loss": 1.2187,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.4895590543746948,
      "rewards/margins": 0.31852632761001587,
      "rewards/rejected": -0.8080854415893555,
      "step": 388
    },
    {
      "epoch": 0.5880574452003023,
      "epsilon_dpo/beta": 0.03055322915315628,
      "epsilon_dpo/beta_margin_grad_mean": -0.3908854126930237,
      "epsilon_dpo/beta_margin_grad_std": 0.1603168547153473,
      "epsilon_dpo/beta_margin_mean": 0.5040192604064941,
      "epsilon_dpo/beta_margin_std": 0.7411776781082153,
      "epsilon_dpo/loss_margin_mean": 16.651473999023438,
      "grad_norm": 10.520373344421387,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.030684418976306915,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.1835188660656265e-07,
      "logits/chosen": -0.2951532006263733,
      "logits/rejected": -0.3086988031864166,
      "logps/chosen": -92.45687103271484,
      "logps/ref_chosen": -81.87734985351562,
      "logps/ref_rejected": -99.03115844726562,
      "logps/rejected": -126.26214599609375,
      "loss": 1.0683,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.32412028312683105,
      "rewards/margins": 0.5040192604064941,
      "rewards/rejected": -0.8281395435333252,
      "step": 389
    },
    {
      "epoch": 0.5895691609977324,
      "epsilon_dpo/beta": 0.03042014129459858,
      "epsilon_dpo/beta_margin_grad_mean": -0.3851366937160492,
      "epsilon_dpo/beta_margin_grad_std": 0.1705460548400879,
      "epsilon_dpo/beta_margin_mean": 0.5503158569335938,
      "epsilon_dpo/beta_margin_std": 0.8301871418952942,
      "epsilon_dpo/loss_margin_mean": 18.252700805664062,
      "grad_norm": 9.353094100952148,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.0305507592856884,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.170407537241599e-07,
      "logits/chosen": -0.27739864587783813,
      "logits/rejected": -0.28185367584228516,
      "logps/chosen": -70.11703491210938,
      "logps/ref_chosen": -64.2594985961914,
      "logps/ref_rejected": -85.45353698730469,
      "logps/rejected": -109.56376647949219,
      "loss": 1.0595,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.17991623282432556,
      "rewards/margins": 0.5503158569335938,
      "rewards/rejected": -0.7302320599555969,
      "step": 390
    },
    {
      "epoch": 0.5910808767951625,
      "epsilon_dpo/beta": 0.03024960495531559,
      "epsilon_dpo/beta_margin_grad_mean": -0.36598044633865356,
      "epsilon_dpo/beta_margin_grad_std": 0.18533697724342346,
      "epsilon_dpo/beta_margin_mean": 0.6447353363037109,
      "epsilon_dpo/beta_margin_std": 0.9046013951301575,
      "epsilon_dpo/loss_margin_mean": 21.484485626220703,
      "grad_norm": 9.929374694824219,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.030417680740356445,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 2.1573054278272636e-07,
      "logits/chosen": -0.5862469673156738,
      "logits/rejected": -0.41867655515670776,
      "logps/chosen": -95.09674072265625,
      "logps/ref_chosen": -84.55359649658203,
      "logps/ref_rejected": -111.49552154541016,
      "logps/rejected": -143.52316284179688,
      "loss": 1.0192,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.3208223581314087,
      "rewards/margins": 0.6447353363037109,
      "rewards/rejected": -0.9655576944351196,
      "step": 391
    },
    {
      "epoch": 0.5925925925925926,
      "epsilon_dpo/beta": 0.030118217691779137,
      "epsilon_dpo/beta_margin_grad_mean": -0.3631499409675598,
      "epsilon_dpo/beta_margin_grad_std": 0.1734553724527359,
      "epsilon_dpo/beta_margin_mean": 0.6588443517684937,
      "epsilon_dpo/beta_margin_std": 0.8642693161964417,
      "epsilon_dpo/loss_margin_mean": 22.05181121826172,
      "grad_norm": 9.381688117980957,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.030247539281845093,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.1442129043167873e-07,
      "logits/chosen": -0.8898314833641052,
      "logits/rejected": -0.3909822702407837,
      "logps/chosen": -101.3697509765625,
      "logps/ref_chosen": -95.38569641113281,
      "logps/ref_rejected": -124.9638671875,
      "logps/rejected": -152.99972534179688,
      "loss": 0.9915,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.18112841248512268,
      "rewards/margins": 0.6588443517684937,
      "rewards/rejected": -0.8399727940559387,
      "step": 392
    },
    {
      "epoch": 0.5941043083900227,
      "epsilon_dpo/beta": 0.029968200251460075,
      "epsilon_dpo/beta_margin_grad_mean": -0.3912041187286377,
      "epsilon_dpo/beta_margin_grad_std": 0.15663762390613556,
      "epsilon_dpo/beta_margin_mean": 0.5000752806663513,
      "epsilon_dpo/beta_margin_std": 0.7438558340072632,
      "epsilon_dpo/loss_margin_mean": 16.825716018676758,
      "grad_norm": 10.260003089904785,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.030115783214569092,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 2.131130332936195e-07,
      "logits/chosen": -0.31805336475372314,
      "logits/rejected": -0.3601935803890228,
      "logps/chosen": -87.36988067626953,
      "logps/ref_chosen": -76.34320068359375,
      "logps/ref_rejected": -94.04782104492188,
      "logps/rejected": -121.90022277832031,
      "loss": 1.0705,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.33345675468444824,
      "rewards/margins": 0.5000752806663513,
      "rewards/rejected": -0.8335320949554443,
      "step": 393
    },
    {
      "epoch": 0.5956160241874527,
      "epsilon_dpo/beta": 0.029828468337655067,
      "epsilon_dpo/beta_margin_grad_mean": -0.3802933990955353,
      "epsilon_dpo/beta_margin_grad_std": 0.1510239988565445,
      "epsilon_dpo/beta_margin_mean": 0.5410258173942566,
      "epsilon_dpo/beta_margin_std": 0.7005043029785156,
      "epsilon_dpo/loss_margin_mean": 18.2916202545166,
      "grad_norm": 11.447973251342773,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.029965952038764954,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.1180580796331323e-07,
      "logits/chosen": -0.21706591546535492,
      "logits/rejected": -0.3572559356689453,
      "logps/chosen": -91.35517883300781,
      "logps/ref_chosen": -82.9498291015625,
      "logps/ref_rejected": -92.46846771240234,
      "logps/rejected": -119.16543579101562,
      "loss": 1.0272,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.25249218940734863,
      "rewards/margins": 0.5410258173942566,
      "rewards/rejected": -0.79351806640625,
      "step": 394
    },
    {
      "epoch": 0.5971277399848829,
      "epsilon_dpo/beta": 0.029754459857940674,
      "epsilon_dpo/beta_margin_grad_mean": -0.41406112909317017,
      "epsilon_dpo/beta_margin_grad_std": 0.16929131746292114,
      "epsilon_dpo/beta_margin_mean": 0.40022772550582886,
      "epsilon_dpo/beta_margin_std": 0.7990128397941589,
      "epsilon_dpo/loss_margin_mean": 13.644739151000977,
      "grad_norm": 10.52011489868164,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.029826141893863678,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 2.104996510066625e-07,
      "logits/chosen": -0.3717024028301239,
      "logits/rejected": -0.5349926352500916,
      "logps/chosen": -79.53875732421875,
      "logps/ref_chosen": -70.12308502197266,
      "logps/ref_rejected": -102.75277709960938,
      "logps/rejected": -125.81317138671875,
      "loss": 1.1687,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.2816503345966339,
      "rewards/margins": 0.40022772550582886,
      "rewards/rejected": -0.6818780899047852,
      "step": 395
    },
    {
      "epoch": 0.5986394557823129,
      "epsilon_dpo/beta": 0.02956877090036869,
      "epsilon_dpo/beta_margin_grad_mean": -0.3932345509529114,
      "epsilon_dpo/beta_margin_grad_std": 0.14215026795864105,
      "epsilon_dpo/beta_margin_mean": 0.47859686613082886,
      "epsilon_dpo/beta_margin_std": 0.6916207671165466,
      "epsilon_dpo/loss_margin_mean": 16.287395477294922,
      "grad_norm": 12.903496742248535,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.029751762747764587,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.0919459895968517e-07,
      "logits/chosen": -0.2872779965400696,
      "logits/rejected": -0.3059869408607483,
      "logps/chosen": -81.68611907958984,
      "logps/ref_chosen": -71.49127197265625,
      "logps/ref_rejected": -107.02473449707031,
      "logps/rejected": -133.50697326660156,
      "loss": 1.0691,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": -0.30214953422546387,
      "rewards/margins": 0.47859686613082886,
      "rewards/rejected": -0.780746340751648,
      "step": 396
    },
    {
      "epoch": 0.600151171579743,
      "epsilon_dpo/beta": 0.029431317001581192,
      "epsilon_dpo/beta_margin_grad_mean": -0.4295797646045685,
      "epsilon_dpo/beta_margin_grad_std": 0.17762644588947296,
      "epsilon_dpo/beta_margin_mean": 0.308048814535141,
      "epsilon_dpo/beta_margin_std": 0.8488595485687256,
      "epsilon_dpo/loss_margin_mean": 10.641646385192871,
      "grad_norm": 11.531167984008789,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.029566969722509384,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.078906883274924e-07,
      "logits/chosen": -0.5348297357559204,
      "logits/rejected": -0.1661069393157959,
      "logps/chosen": -107.25617980957031,
      "logps/ref_chosen": -95.6727523803711,
      "logps/ref_rejected": -111.85726928710938,
      "logps/rejected": -134.08233642578125,
      "loss": 1.2647,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.34256458282470703,
      "rewards/margins": 0.3080487847328186,
      "rewards/rejected": -0.6506133675575256,
      "step": 397
    },
    {
      "epoch": 0.6016628873771731,
      "epsilon_dpo/beta": 0.02931239642202854,
      "epsilon_dpo/beta_margin_grad_mean": -0.3774406909942627,
      "epsilon_dpo/beta_margin_grad_std": 0.17114564776420593,
      "epsilon_dpo/beta_margin_mean": 0.585285484790802,
      "epsilon_dpo/beta_margin_std": 0.8242340683937073,
      "epsilon_dpo/loss_margin_mean": 20.152509689331055,
      "grad_norm": 8.442160606384277,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.02942902036011219,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 2.065879555832674e-07,
      "logits/chosen": -0.40959131717681885,
      "logits/rejected": -0.6267037391662598,
      "logps/chosen": -79.64991760253906,
      "logps/ref_chosen": -70.61448669433594,
      "logps/ref_rejected": -98.2459716796875,
      "logps/rejected": -127.43391418457031,
      "loss": 1.0326,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.26629623770713806,
      "rewards/margins": 0.5852854251861572,
      "rewards/rejected": -0.8515816926956177,
      "step": 398
    },
    {
      "epoch": 0.6031746031746031,
      "epsilon_dpo/beta": 0.029175475239753723,
      "epsilon_dpo/beta_margin_grad_mean": -0.38290148973464966,
      "epsilon_dpo/beta_margin_grad_std": 0.1641579419374466,
      "epsilon_dpo/beta_margin_mean": 0.5577808618545532,
      "epsilon_dpo/beta_margin_std": 0.8090330958366394,
      "epsilon_dpo/loss_margin_mean": 19.284090042114258,
      "grad_norm": 15.290122985839844,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.029309948906302452,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.052864371672457e-07,
      "logits/chosen": -0.12391947209835052,
      "logits/rejected": -0.24706780910491943,
      "logps/chosen": -112.02005004882812,
      "logps/ref_chosen": -93.99934387207031,
      "logps/ref_rejected": -146.26950073242188,
      "logps/rejected": -183.5742950439453,
      "loss": 1.0452,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5295411348342896,
      "rewards/margins": 0.5577808618545532,
      "rewards/rejected": -1.0873219966888428,
      "step": 399
    },
    {
      "epoch": 0.6046863189720333,
      "epsilon_dpo/beta": 0.029103174805641174,
      "epsilon_dpo/beta_margin_grad_mean": -0.434611976146698,
      "epsilon_dpo/beta_margin_grad_std": 0.17042642831802368,
      "epsilon_dpo/beta_margin_mean": 0.3079947233200073,
      "epsilon_dpo/beta_margin_std": 0.7888738512992859,
      "epsilon_dpo/loss_margin_mean": 10.77357292175293,
      "grad_norm": 12.158735275268555,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.029173199087381363,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.0398616948569493e-07,
      "logits/chosen": -0.2648681700229645,
      "logits/rejected": -0.5493128299713135,
      "logps/chosen": -118.06231689453125,
      "logps/ref_chosen": -97.5876235961914,
      "logps/ref_rejected": -117.63888549804688,
      "logps/rejected": -148.88714599609375,
      "loss": 1.2438,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.5979068279266357,
      "rewards/margins": 0.3079947233200073,
      "rewards/rejected": -0.9059015512466431,
      "step": 400
    },
    {
      "epoch": 0.6046863189720333,
      "eval_epsilon_dpo/beta": 0.028982818126678467,
      "eval_epsilon_dpo/beta_margin_grad_mean": -0.4003281891345978,
      "eval_epsilon_dpo/beta_margin_grad_std": 0.17058531939983368,
      "eval_epsilon_dpo/beta_margin_mean": 0.47119957208633423,
      "eval_epsilon_dpo/beta_margin_std": 0.8252949118614197,
      "eval_epsilon_dpo/loss_margin_mean": 16.441312789916992,
      "eval_kl/n_epsilon_steps": 0.2922535240650177,
      "eval_kl/p_epsilon_steps": 0.7064260840415955,
      "eval_logits/chosen": -0.35346463322639465,
      "eval_logits/rejected": -0.45723411440849304,
      "eval_logps/chosen": -101.37396240234375,
      "eval_logps/ref_chosen": -87.42715454101562,
      "eval_logps/ref_rejected": -104.23548889160156,
      "eval_logps/rejected": -134.62362670898438,
      "eval_loss": 0.5624240636825562,
      "eval_rewards/accuracies": 0.7178696990013123,
      "eval_rewards/chosen": -0.4064703583717346,
      "eval_rewards/margins": 0.47119957208633423,
      "eval_rewards/rejected": -0.8776699304580688,
      "eval_runtime": 47.6139,
      "eval_samples_per_second": 48.368,
      "eval_steps_per_second": 1.512,
      "step": 400
    },
    {
      "epoch": 0.6061980347694633,
      "epsilon_dpo/beta": 0.02892146445810795,
      "epsilon_dpo/beta_margin_grad_mean": -0.3574426770210266,
      "epsilon_dpo/beta_margin_grad_std": 0.16776859760284424,
      "epsilon_dpo/beta_margin_mean": 0.6677769422531128,
      "epsilon_dpo/beta_margin_std": 0.8188538551330566,
      "epsilon_dpo/loss_margin_mean": 23.23797035217285,
      "grad_norm": 8.932498931884766,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.02910044975578785,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.0268718890989752e-07,
      "logits/chosen": -0.14188948273658752,
      "logits/rejected": -0.18290500342845917,
      "logps/chosen": -81.09709167480469,
      "logps/ref_chosen": -72.36085510253906,
      "logps/ref_rejected": -105.40269470214844,
      "logps/rejected": -137.37689208984375,
      "loss": 0.9714,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": -0.25411131978034973,
      "rewards/margins": 0.6677769422531128,
      "rewards/rejected": -0.9218882918357849,
      "step": 401
    },
    {
      "epoch": 0.6077097505668935,
      "epsilon_dpo/beta": 0.02879149280488491,
      "epsilon_dpo/beta_margin_grad_mean": -0.3803299069404602,
      "epsilon_dpo/beta_margin_grad_std": 0.1708374321460724,
      "epsilon_dpo/beta_margin_mean": 0.5698226690292358,
      "epsilon_dpo/beta_margin_std": 0.8285603523254395,
      "epsilon_dpo/loss_margin_mean": 19.973407745361328,
      "grad_norm": 18.650056838989258,
      "kl/avg_steps": 0.453125,
      "kl/beta": 0.028919700533151627,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.013895317751323e-07,
      "logits/chosen": -0.46771174669265747,
      "logits/rejected": -0.2445870041847229,
      "logps/chosen": -94.94572448730469,
      "logps/ref_chosen": -83.60546112060547,
      "logps/ref_rejected": -94.18315124511719,
      "logps/rejected": -125.496826171875,
      "loss": 1.0448,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.3300950527191162,
      "rewards/margins": 0.5698226690292358,
      "rewards/rejected": -0.899917721748352,
      "step": 402
    },
    {
      "epoch": 0.6092214663643235,
      "epsilon_dpo/beta": 0.02863917127251625,
      "epsilon_dpo/beta_margin_grad_mean": -0.376331090927124,
      "epsilon_dpo/beta_margin_grad_std": 0.18496714532375336,
      "epsilon_dpo/beta_margin_mean": 0.6017536520957947,
      "epsilon_dpo/beta_margin_std": 0.9268775582313538,
      "epsilon_dpo/loss_margin_mean": 21.197481155395508,
      "grad_norm": 13.501971244812012,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.028789250180125237,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 2.0009323437965898e-07,
      "logits/chosen": -0.38072872161865234,
      "logits/rejected": -0.4114571809768677,
      "logps/chosen": -87.80209350585938,
      "logps/ref_chosen": -72.8893814086914,
      "logps/ref_rejected": -107.12028503417969,
      "logps/rejected": -143.23048400878906,
      "loss": 1.0558,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.4293259084224701,
      "rewards/margins": 0.6017535924911499,
      "rewards/rejected": -1.0310795307159424,
      "step": 403
    },
    {
      "epoch": 0.6107331821617535,
      "epsilon_dpo/beta": 0.028505729511380196,
      "epsilon_dpo/beta_margin_grad_mean": -0.3736587464809418,
      "epsilon_dpo/beta_margin_grad_std": 0.17950692772865295,
      "epsilon_dpo/beta_margin_mean": 0.6139946579933167,
      "epsilon_dpo/beta_margin_std": 0.8749144077301025,
      "epsilon_dpo/loss_margin_mean": 21.728633880615234,
      "grad_norm": 9.500164985656738,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.0286371149122715,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.9879833298370237e-07,
      "logits/chosen": -0.43219926953315735,
      "logits/rejected": -0.43257758021354675,
      "logps/chosen": -99.16268920898438,
      "logps/ref_chosen": -86.45829772949219,
      "logps/ref_rejected": -120.54048156738281,
      "logps/rejected": -154.9735107421875,
      "loss": 1.0292,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.3642900586128235,
      "rewards/margins": 0.6139946579933167,
      "rewards/rejected": -0.9782847166061401,
      "step": 404
    },
    {
      "epoch": 0.6122448979591837,
      "epsilon_dpo/beta": 0.028408365324139595,
      "epsilon_dpo/beta_margin_grad_mean": -0.41101425886154175,
      "epsilon_dpo/beta_margin_grad_std": 0.16673527657985687,
      "epsilon_dpo/beta_margin_mean": 0.42524203658103943,
      "epsilon_dpo/beta_margin_std": 0.8040966987609863,
      "epsilon_dpo/loss_margin_mean": 15.15623664855957,
      "grad_norm": 9.812227249145508,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.028503505513072014,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 1.975048638084379e-07,
      "logits/chosen": -0.25932326912879944,
      "logits/rejected": -0.3149704337120056,
      "logps/chosen": -89.40546417236328,
      "logps/ref_chosen": -73.7075424194336,
      "logps/ref_rejected": -89.86683654785156,
      "logps/rejected": -120.72099304199219,
      "loss": 1.148,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.4482584297657013,
      "rewards/margins": 0.42524200677871704,
      "rewards/rejected": -0.8735004663467407,
      "step": 405
    },
    {
      "epoch": 0.6137566137566137,
      "epsilon_dpo/beta": 0.0282400231808424,
      "epsilon_dpo/beta_margin_grad_mean": -0.3645992875099182,
      "epsilon_dpo/beta_margin_grad_std": 0.18260234594345093,
      "epsilon_dpo/beta_margin_mean": 0.6403554677963257,
      "epsilon_dpo/beta_margin_std": 0.8946476578712463,
      "epsilon_dpo/loss_margin_mean": 22.858041763305664,
      "grad_norm": 9.639616012573242,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.02840586006641388,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 1.9621286303497914e-07,
      "logits/chosen": -0.5095548629760742,
      "logits/rejected": -0.504365086555481,
      "logps/chosen": -81.74095153808594,
      "logps/ref_chosen": -69.46755981445312,
      "logps/ref_rejected": -119.13258361816406,
      "logps/rejected": -154.26402282714844,
      "loss": 1.0191,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.34815317392349243,
      "rewards/margins": 0.6403554677963257,
      "rewards/rejected": -0.9885086417198181,
      "step": 406
    },
    {
      "epoch": 0.6152683295540439,
      "epsilon_dpo/beta": 0.028135115280747414,
      "epsilon_dpo/beta_margin_grad_mean": -0.4090394377708435,
      "epsilon_dpo/beta_margin_grad_std": 0.17076851427555084,
      "epsilon_dpo/beta_margin_mean": 0.4270060062408447,
      "epsilon_dpo/beta_margin_std": 0.7996051907539368,
      "epsilon_dpo/loss_margin_mean": 15.369938850402832,
      "grad_norm": 11.385334014892578,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.028238195925951004,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.9492236680336483e-07,
      "logits/chosen": -0.28435638546943665,
      "logits/rejected": -0.6896387338638306,
      "logps/chosen": -140.2555389404297,
      "logps/ref_chosen": -113.93284606933594,
      "logps/ref_rejected": -133.01231384277344,
      "logps/rejected": -174.7049560546875,
      "loss": 1.1481,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.7427991032600403,
      "rewards/margins": 0.4270060062408447,
      "rewards/rejected": -1.1698050498962402,
      "step": 407
    },
    {
      "epoch": 0.6167800453514739,
      "epsilon_dpo/beta": 0.028003625571727753,
      "epsilon_dpo/beta_margin_grad_mean": -0.36860692501068115,
      "epsilon_dpo/beta_margin_grad_std": 0.16879484057426453,
      "epsilon_dpo/beta_margin_mean": 0.6323613524436951,
      "epsilon_dpo/beta_margin_std": 0.8337081074714661,
      "epsilon_dpo/loss_margin_mean": 22.75626564025879,
      "grad_norm": 9.038987159729004,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.028132697567343712,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.9363341121154895e-07,
      "logits/chosen": -0.42244648933410645,
      "logits/rejected": -0.5274006128311157,
      "logps/chosen": -93.25194549560547,
      "logps/ref_chosen": -77.43348693847656,
      "logps/ref_rejected": -100.76187133789062,
      "logps/rejected": -139.33657836914062,
      "loss": 0.9995,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4446289837360382,
      "rewards/margins": 0.6323613524436951,
      "rewards/rejected": -1.0769903659820557,
      "step": 408
    },
    {
      "epoch": 0.618291761148904,
      "epsilon_dpo/beta": 0.027951732277870178,
      "epsilon_dpo/beta_margin_grad_mean": -0.4339336156845093,
      "epsilon_dpo/beta_margin_grad_std": 0.18402144312858582,
      "epsilon_dpo/beta_margin_mean": 0.3334841728210449,
      "epsilon_dpo/beta_margin_std": 0.8810906410217285,
      "epsilon_dpo/loss_margin_mean": 12.15471076965332,
      "grad_norm": 10.198472023010254,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.028001440688967705,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 1.9234603231438994e-07,
      "logits/chosen": -0.12168080359697342,
      "logits/rejected": -0.14667344093322754,
      "logps/chosen": -110.5352554321289,
      "logps/ref_chosen": -86.81268310546875,
      "logps/ref_rejected": -84.29527282714844,
      "logps/rejected": -120.17254638671875,
      "loss": 1.2535,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.6660339832305908,
      "rewards/margins": 0.3334841728210449,
      "rewards/rejected": -0.9995181560516357,
      "step": 409
    },
    {
      "epoch": 0.6198034769463341,
      "epsilon_dpo/beta": 0.027829542756080627,
      "epsilon_dpo/beta_margin_grad_mean": -0.37669700384140015,
      "epsilon_dpo/beta_margin_grad_std": 0.16883623600006104,
      "epsilon_dpo/beta_margin_mean": 0.5865784883499146,
      "epsilon_dpo/beta_margin_std": 0.8136049509048462,
      "epsilon_dpo/loss_margin_mean": 21.269351959228516,
      "grad_norm": 10.556927680969238,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.02794903703033924,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.9106026612264315e-07,
      "logits/chosen": -0.5301246643066406,
      "logits/rejected": -0.44228461384773254,
      "logps/chosen": -105.40447998046875,
      "logps/ref_chosen": -85.89543151855469,
      "logps/ref_rejected": -96.34490203857422,
      "logps/rejected": -137.123291015625,
      "loss": 1.0278,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5454895496368408,
      "rewards/margins": 0.5865784883499146,
      "rewards/rejected": -1.1320680379867554,
      "step": 410
    },
    {
      "epoch": 0.6213151927437641,
      "epsilon_dpo/beta": 0.027725713327527046,
      "epsilon_dpo/beta_margin_grad_mean": -0.3819405436515808,
      "epsilon_dpo/beta_margin_grad_std": 0.1781851202249527,
      "epsilon_dpo/beta_margin_mean": 0.5704288482666016,
      "epsilon_dpo/beta_margin_std": 0.8644275069236755,
      "epsilon_dpo/loss_margin_mean": 20.78996467590332,
      "grad_norm": 17.217391967773438,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.027827292680740356,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.8977614860195296e-07,
      "logits/chosen": -0.2461598962545395,
      "logits/rejected": -0.2894115447998047,
      "logps/chosen": -96.726318359375,
      "logps/ref_chosen": -75.55298614501953,
      "logps/ref_rejected": -100.882568359375,
      "logps/rejected": -142.84585571289062,
      "loss": 1.0574,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5909495949745178,
      "rewards/margins": 0.5704289078712463,
      "rewards/rejected": -1.1613783836364746,
      "step": 411
    },
    {
      "epoch": 0.6228269085411943,
      "epsilon_dpo/beta": 0.027596134692430496,
      "epsilon_dpo/beta_margin_grad_mean": -0.3919138014316559,
      "epsilon_dpo/beta_margin_grad_std": 0.1753304898738861,
      "epsilon_dpo/beta_margin_mean": 0.5203702449798584,
      "epsilon_dpo/beta_margin_std": 0.8557584881782532,
      "epsilon_dpo/loss_margin_mean": 19.04148292541504,
      "grad_norm": 11.123086929321289,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.02772332914173603,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.8849371567184662e-07,
      "logits/chosen": -0.10348869115114212,
      "logits/rejected": -0.18310517072677612,
      "logps/chosen": -103.1690673828125,
      "logps/ref_chosen": -78.12739562988281,
      "logps/ref_rejected": -97.01484680175781,
      "logps/rejected": -141.09799194335938,
      "loss": 1.0914,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6940072774887085,
      "rewards/margins": 0.5203702449798584,
      "rewards/rejected": -1.214377522468567,
      "step": 412
    },
    {
      "epoch": 0.6243386243386243,
      "epsilon_dpo/beta": 0.027523396536707878,
      "epsilon_dpo/beta_margin_grad_mean": -0.4108419120311737,
      "epsilon_dpo/beta_margin_grad_std": 0.1892181932926178,
      "epsilon_dpo/beta_margin_mean": 0.42955899238586426,
      "epsilon_dpo/beta_margin_std": 0.9341761469841003,
      "epsilon_dpo/loss_margin_mean": 15.84644603729248,
      "grad_norm": 14.241096496582031,
      "kl/avg_steps": 0.265625,
      "kl/beta": 0.027593983337283134,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 1.872130032047302e-07,
      "logits/chosen": -0.3025868535041809,
      "logits/rejected": -0.5660539865493774,
      "logps/chosen": -118.65052795410156,
      "logps/ref_chosen": -91.40047454833984,
      "logps/ref_rejected": -99.7619857788086,
      "logps/rejected": -142.8584747314453,
      "loss": 1.1932,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.7531765699386597,
      "rewards/margins": 0.42955899238586426,
      "rewards/rejected": -1.182735562324524,
      "step": 413
    },
    {
      "epoch": 0.6258503401360545,
      "epsilon_dpo/beta": 0.02738601341843605,
      "epsilon_dpo/beta_margin_grad_mean": -0.3774614930152893,
      "epsilon_dpo/beta_margin_grad_std": 0.1806628406047821,
      "epsilon_dpo/beta_margin_mean": 0.6020256876945496,
      "epsilon_dpo/beta_margin_std": 0.9087315201759338,
      "epsilon_dpo/loss_margin_mean": 22.174726486206055,
      "grad_norm": 10.976476669311523,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.027520880103111267,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.8593404702488436e-07,
      "logits/chosen": -0.06367478519678116,
      "logits/rejected": -0.1593630015850067,
      "logps/chosen": -100.70257568359375,
      "logps/ref_chosen": -75.29246520996094,
      "logps/ref_rejected": -100.37710571289062,
      "logps/rejected": -147.96194458007812,
      "loss": 1.0476,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6971333622932434,
      "rewards/margins": 0.6020256280899048,
      "rewards/rejected": -1.299159049987793,
      "step": 414
    },
    {
      "epoch": 0.6273620559334845,
      "epsilon_dpo/beta": 0.02725832350552082,
      "epsilon_dpo/beta_margin_grad_mean": -0.39092764258384705,
      "epsilon_dpo/beta_margin_grad_std": 0.1978784054517746,
      "epsilon_dpo/beta_margin_mean": 0.5205000042915344,
      "epsilon_dpo/beta_margin_std": 0.9865575432777405,
      "epsilon_dpo/loss_margin_mean": 19.31855583190918,
      "grad_norm": 10.629053115844727,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.027383960783481598,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.846568829074628e-07,
      "logits/chosen": -0.2597837448120117,
      "logits/rejected": -0.2640593647956848,
      "logps/chosen": -102.90240478515625,
      "logps/ref_chosen": -77.00994873046875,
      "logps/ref_rejected": -80.96036529541016,
      "logps/rejected": -126.17137908935547,
      "loss": 1.1433,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7089916467666626,
      "rewards/margins": 0.5205000638961792,
      "rewards/rejected": -1.2294917106628418,
      "step": 415
    },
    {
      "epoch": 0.6288737717309146,
      "epsilon_dpo/beta": 0.02716521918773651,
      "epsilon_dpo/beta_margin_grad_mean": -0.4312796890735626,
      "epsilon_dpo/beta_margin_grad_std": 0.18393133580684662,
      "epsilon_dpo/beta_margin_mean": 0.33430424332618713,
      "epsilon_dpo/beta_margin_std": 0.8818397521972656,
      "epsilon_dpo/loss_margin_mean": 12.520179748535156,
      "grad_norm": 11.289984703063965,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.0272561963647604,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 1.8338154657749128e-07,
      "logits/chosen": -0.38363879919052124,
      "logits/rejected": -0.6178092956542969,
      "logps/chosen": -110.30681610107422,
      "logps/ref_chosen": -79.6038589477539,
      "logps/ref_rejected": -93.3397216796875,
      "logps/rejected": -136.5628662109375,
      "loss": 1.2532,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8370293378829956,
      "rewards/margins": 0.3343042731285095,
      "rewards/rejected": -1.1713335514068604,
      "step": 416
    },
    {
      "epoch": 0.6303854875283447,
      "epsilon_dpo/beta": 0.027021225541830063,
      "epsilon_dpo/beta_margin_grad_mean": -0.36599236726760864,
      "epsilon_dpo/beta_margin_grad_std": 0.18626999855041504,
      "epsilon_dpo/beta_margin_mean": 0.647750735282898,
      "epsilon_dpo/beta_margin_std": 0.9404109716415405,
      "epsilon_dpo/loss_margin_mean": 24.17540740966797,
      "grad_norm": 10.080581665039062,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.02716282568871975,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.8210807370886849e-07,
      "logits/chosen": -0.2293497771024704,
      "logits/rejected": -0.2854086458683014,
      "logps/chosen": -105.18743133544922,
      "logps/ref_chosen": -79.04769897460938,
      "logps/ref_rejected": -101.35102844238281,
      "logps/rejected": -151.66616821289062,
      "loss": 1.0277,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.7088662385940552,
      "rewards/margins": 0.647750735282898,
      "rewards/rejected": -1.3566169738769531,
      "step": 417
    },
    {
      "epoch": 0.6318972033257747,
      "epsilon_dpo/beta": 0.02697976492345333,
      "epsilon_dpo/beta_margin_grad_mean": -0.44641682505607605,
      "epsilon_dpo/beta_margin_grad_std": 0.19468529522418976,
      "epsilon_dpo/beta_margin_mean": 0.26206421852111816,
      "epsilon_dpo/beta_margin_std": 0.9137290120124817,
      "epsilon_dpo/loss_margin_mean": 9.974386215209961,
      "grad_norm": 13.358625411987305,
      "kl/avg_steps": 0.15625,
      "kl/beta": 0.027019284665584564,
      "kl/n_epsilon_steps": 0.421875,
      "kl/p_epsilon_steps": 0.578125,
      "learning_rate": 1.8083649992336825e-07,
      "logits/chosen": -0.29838278889656067,
      "logits/rejected": -0.4096614718437195,
      "logps/chosen": -131.08416748046875,
      "logps/ref_chosen": -96.25605773925781,
      "logps/ref_rejected": -97.95796203613281,
      "logps/rejected": -142.76046752929688,
      "loss": 1.3312,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.9437500834465027,
      "rewards/margins": 0.2620641887187958,
      "rewards/rejected": -1.205814242362976,
      "step": 418
    },
    {
      "epoch": 0.6334089191232048,
      "epsilon_dpo/beta": 0.026811206713318825,
      "epsilon_dpo/beta_margin_grad_mean": -0.3356010615825653,
      "epsilon_dpo/beta_margin_grad_std": 0.17441126704216003,
      "epsilon_dpo/beta_margin_mean": 0.7969227433204651,
      "epsilon_dpo/beta_margin_std": 0.8698088526725769,
      "epsilon_dpo/loss_margin_mean": 29.899383544921875,
      "grad_norm": 10.102495193481445,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.026977133005857468,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 1.7956686078964255e-07,
      "logits/chosen": -0.16823619604110718,
      "logits/rejected": -0.3001464605331421,
      "logps/chosen": -100.81460571289062,
      "logps/ref_chosen": -81.72348022460938,
      "logps/ref_rejected": -99.48829650878906,
      "logps/rejected": -148.4788055419922,
      "loss": 0.9006,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.51449054479599,
      "rewards/margins": 0.7969227433204651,
      "rewards/rejected": -1.311413288116455,
      "step": 419
    },
    {
      "epoch": 0.6349206349206349,
      "epsilon_dpo/beta": 0.026736844331026077,
      "epsilon_dpo/beta_margin_grad_mean": -0.41811904311180115,
      "epsilon_dpo/beta_margin_grad_std": 0.2055063396692276,
      "epsilon_dpo/beta_margin_mean": 0.407371461391449,
      "epsilon_dpo/beta_margin_std": 1.029210090637207,
      "epsilon_dpo/loss_margin_mean": 15.509881973266602,
      "grad_norm": 12.732593536376953,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.02680957317352295,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 1.782991918222275e-07,
      "logits/chosen": -0.293887197971344,
      "logits/rejected": -0.454693078994751,
      "logps/chosen": -109.46027374267578,
      "logps/ref_chosen": -78.56198120117188,
      "logps/ref_rejected": -85.0513916015625,
      "logps/rejected": -131.45956420898438,
      "loss": 1.25,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8302141427993774,
      "rewards/margins": 0.4073714315891266,
      "rewards/rejected": -1.2375855445861816,
      "step": 420
    },
    {
      "epoch": 0.636432350718065,
      "epsilon_dpo/beta": 0.026653503999114037,
      "epsilon_dpo/beta_margin_grad_mean": -0.4126037061214447,
      "epsilon_dpo/beta_margin_grad_std": 0.21195949614048004,
      "epsilon_dpo/beta_margin_mean": 0.41447150707244873,
      "epsilon_dpo/beta_margin_std": 1.0345741510391235,
      "epsilon_dpo/loss_margin_mean": 15.841025352478027,
      "grad_norm": 11.281766891479492,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.026734383776783943,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 1.7703352848054887e-07,
      "logits/chosen": -0.1268298327922821,
      "logits/rejected": -0.5857471823692322,
      "logps/chosen": -110.41204071044922,
      "logps/ref_chosen": -82.40538024902344,
      "logps/ref_rejected": -100.89399719238281,
      "logps/rejected": -144.74168395996094,
      "loss": 1.2508,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.7509020566940308,
      "rewards/margins": 0.41447150707244873,
      "rewards/rejected": -1.1653735637664795,
      "step": 421
    },
    {
      "epoch": 0.6379440665154951,
      "epsilon_dpo/beta": 0.026520494371652603,
      "epsilon_dpo/beta_margin_grad_mean": -0.3952094614505768,
      "epsilon_dpo/beta_margin_grad_std": 0.1741863489151001,
      "epsilon_dpo/beta_margin_mean": 0.5068070888519287,
      "epsilon_dpo/beta_margin_std": 0.8488343954086304,
      "epsilon_dpo/loss_margin_mean": 19.29708480834961,
      "grad_norm": 11.663961410522461,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.02665109746158123,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.7576990616793137e-07,
      "logits/chosen": -0.09396468847990036,
      "logits/rejected": -0.29736435413360596,
      "logps/chosen": -119.38473510742188,
      "logps/ref_chosen": -93.38703918457031,
      "logps/ref_rejected": -103.51509094238281,
      "logps/rejected": -148.80987548828125,
      "loss": 1.0988,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6915041208267212,
      "rewards/margins": 0.5068070888519287,
      "rewards/rejected": -1.19831120967865,
      "step": 422
    },
    {
      "epoch": 0.6394557823129252,
      "epsilon_dpo/beta": 0.02638855203986168,
      "epsilon_dpo/beta_margin_grad_mean": -0.3787572979927063,
      "epsilon_dpo/beta_margin_grad_std": 0.169308602809906,
      "epsilon_dpo/beta_margin_mean": 0.5847669243812561,
      "epsilon_dpo/beta_margin_std": 0.862210750579834,
      "epsilon_dpo/loss_margin_mean": 22.34361457824707,
      "grad_norm": 9.964740753173828,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.026518505066633224,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.745083602306071e-07,
      "logits/chosen": -0.1422388255596161,
      "logits/rejected": -0.5984101295471191,
      "logps/chosen": -104.98013305664062,
      "logps/ref_chosen": -78.10965728759766,
      "logps/ref_rejected": -99.3348388671875,
      "logps/rejected": -148.54891967773438,
      "loss": 1.0417,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.7118363976478577,
      "rewards/margins": 0.5847669839859009,
      "rewards/rejected": -1.2966034412384033,
      "step": 423
    },
    {
      "epoch": 0.6409674981103552,
      "epsilon_dpo/beta": 0.0262325257062912,
      "epsilon_dpo/beta_margin_grad_mean": -0.3739382028579712,
      "epsilon_dpo/beta_margin_grad_std": 0.17924833297729492,
      "epsilon_dpo/beta_margin_mean": 0.5975351929664612,
      "epsilon_dpo/beta_margin_std": 0.8674590587615967,
      "epsilon_dpo/loss_margin_mean": 22.970008850097656,
      "grad_norm": 10.012207984924316,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.026386572048068047,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 1.7324892595672804e-07,
      "logits/chosen": -0.6137542724609375,
      "logits/rejected": -0.353971928358078,
      "logps/chosen": -101.5662841796875,
      "logps/ref_chosen": -73.11288452148438,
      "logps/ref_rejected": -98.70126342773438,
      "logps/rejected": -150.12466430664062,
      "loss": 1.0399,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.7486594915390015,
      "rewards/margins": 0.5975351333618164,
      "rewards/rejected": -1.3461946249008179,
      "step": 424
    },
    {
      "epoch": 0.6424792139077853,
      "epsilon_dpo/beta": 0.026085887104272842,
      "epsilon_dpo/beta_margin_grad_mean": -0.3714931607246399,
      "epsilon_dpo/beta_margin_grad_std": 0.17195554077625275,
      "epsilon_dpo/beta_margin_mean": 0.6200801730155945,
      "epsilon_dpo/beta_margin_std": 0.8400947451591492,
      "epsilon_dpo/loss_margin_mean": 23.949697494506836,
      "grad_norm": 10.907929420471191,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.0262308269739151,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 1.7199163857537824e-07,
      "logits/chosen": -0.07301705330610275,
      "logits/rejected": -0.19498585164546967,
      "logps/chosen": -109.0750732421875,
      "logps/ref_chosen": -83.90113830566406,
      "logps/ref_rejected": -93.6903076171875,
      "logps/rejected": -142.81394958496094,
      "loss": 1.0115,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.6586664915084839,
      "rewards/margins": 0.6200801730155945,
      "rewards/rejected": -1.2787466049194336,
      "step": 425
    },
    {
      "epoch": 0.6439909297052154,
      "epsilon_dpo/beta": 0.026054104790091515,
      "epsilon_dpo/beta_margin_grad_mean": -0.45818114280700684,
      "epsilon_dpo/beta_margin_grad_std": 0.2074817568063736,
      "epsilon_dpo/beta_margin_mean": 0.21456298232078552,
      "epsilon_dpo/beta_margin_std": 1.0599706172943115,
      "epsilon_dpo/loss_margin_mean": 8.526104927062988,
      "grad_norm": 16.841733932495117,
      "kl/avg_steps": 0.125,
      "kl/beta": 0.02608410455286503,
      "kl/n_epsilon_steps": 0.4375,
      "kl/p_epsilon_steps": 0.5625,
      "learning_rate": 1.7073653325558828e-07,
      "logits/chosen": -0.15785779058933258,
      "logits/rejected": -0.3029636740684509,
      "logps/chosen": -130.49227905273438,
      "logps/ref_chosen": -95.90599060058594,
      "logps/ref_rejected": -96.2318115234375,
      "logps/rejected": -139.34420776367188,
      "loss": 1.427,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.9045907258987427,
      "rewards/margins": 0.2145630121231079,
      "rewards/rejected": -1.1191537380218506,
      "step": 426
    },
    {
      "epoch": 0.6455026455026455,
      "epsilon_dpo/beta": 0.02594830095767975,
      "epsilon_dpo/beta_margin_grad_mean": -0.39626121520996094,
      "epsilon_dpo/beta_margin_grad_std": 0.18655137717723846,
      "epsilon_dpo/beta_margin_mean": 0.5007147789001465,
      "epsilon_dpo/beta_margin_std": 0.9225812554359436,
      "epsilon_dpo/loss_margin_mean": 19.527267456054688,
      "grad_norm": 9.823708534240723,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.026051539927721024,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 1.6948364510535218e-07,
      "logits/chosen": -0.28479713201522827,
      "logits/rejected": -0.48131686449050903,
      "logps/chosen": -112.27393341064453,
      "logps/ref_chosen": -81.92965698242188,
      "logps/ref_rejected": -105.96514892578125,
      "logps/rejected": -155.83670043945312,
      "loss": 1.1312,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.7907149791717529,
      "rewards/margins": 0.5007147789001465,
      "rewards/rejected": -1.2914297580718994,
      "step": 427
    },
    {
      "epoch": 0.6470143613000756,
      "epsilon_dpo/beta": 0.025835201144218445,
      "epsilon_dpo/beta_margin_grad_mean": -0.38718709349632263,
      "epsilon_dpo/beta_margin_grad_std": 0.1809690296649933,
      "epsilon_dpo/beta_margin_mean": 0.5327649116516113,
      "epsilon_dpo/beta_margin_std": 0.869602382183075,
      "epsilon_dpo/loss_margin_mean": 20.841089248657227,
      "grad_norm": 12.027029991149902,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.025946132838726044,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.6823300917064458e-07,
      "logits/chosen": -0.12337511777877808,
      "logits/rejected": -0.30909281969070435,
      "logps/chosen": -126.36972045898438,
      "logps/ref_chosen": -95.26530456542969,
      "logps/ref_rejected": -106.78121948242188,
      "logps/rejected": -158.72671508789062,
      "loss": 1.0898,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.8065395355224609,
      "rewards/margins": 0.5327649116516113,
      "rewards/rejected": -1.3393044471740723,
      "step": 428
    },
    {
      "epoch": 0.6485260770975056,
      "epsilon_dpo/beta": 0.025714591145515442,
      "epsilon_dpo/beta_margin_grad_mean": -0.38070225715637207,
      "epsilon_dpo/beta_margin_grad_std": 0.20825932919979095,
      "epsilon_dpo/beta_margin_mean": 0.5661614537239075,
      "epsilon_dpo/beta_margin_std": 1.029008150100708,
      "epsilon_dpo/loss_margin_mean": 22.288246154785156,
      "grad_norm": 10.108731269836426,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.025833113119006157,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.669846604344412e-07,
      "logits/chosen": -0.21388304233551025,
      "logits/rejected": -0.1304803490638733,
      "logps/chosen": -116.25877380371094,
      "logps/ref_chosen": -84.36827850341797,
      "logps/ref_rejected": -79.90157318115234,
      "logps/rejected": -134.080322265625,
      "loss": 1.1294,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.8244999051094055,
      "rewards/margins": 0.5661614537239075,
      "rewards/rejected": -1.390661358833313,
      "step": 429
    },
    {
      "epoch": 0.6500377928949358,
      "epsilon_dpo/beta": 0.025586578994989395,
      "epsilon_dpo/beta_margin_grad_mean": -0.3593744933605194,
      "epsilon_dpo/beta_margin_grad_std": 0.17643196880817413,
      "epsilon_dpo/beta_margin_mean": 0.6664764285087585,
      "epsilon_dpo/beta_margin_std": 0.8701526522636414,
      "epsilon_dpo/loss_margin_mean": 26.25492286682129,
      "grad_norm": 9.700910568237305,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.02571258507668972,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.6573863381573954e-07,
      "logits/chosen": -0.302243709564209,
      "logits/rejected": -0.07065822929143906,
      "logps/chosen": -108.92573547363281,
      "logps/ref_chosen": -82.68994140625,
      "logps/ref_rejected": -89.63719177246094,
      "logps/rejected": -142.12791442871094,
      "loss": 0.9903,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.6739843487739563,
      "rewards/margins": 0.6664764881134033,
      "rewards/rejected": -1.3404607772827148,
      "step": 430
    },
    {
      "epoch": 0.6515495086923658,
      "epsilon_dpo/beta": 0.02546728029847145,
      "epsilon_dpo/beta_margin_grad_mean": -0.38510796427726746,
      "epsilon_dpo/beta_margin_grad_std": 0.1891801953315735,
      "epsilon_dpo/beta_margin_mean": 0.5427370667457581,
      "epsilon_dpo/beta_margin_std": 0.9203450083732605,
      "epsilon_dpo/loss_margin_mean": 21.542346954345703,
      "grad_norm": 10.811869621276855,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.025584662333130836,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.6449496416858282e-07,
      "logits/chosen": -0.36825886368751526,
      "logits/rejected": -0.3587014079093933,
      "logps/chosen": -94.90249633789062,
      "logps/ref_chosen": -71.41302490234375,
      "logps/ref_rejected": -94.84451293945312,
      "logps/rejected": -139.8763427734375,
      "loss": 1.1017,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6012818813323975,
      "rewards/margins": 0.5427370071411133,
      "rewards/rejected": -1.1440188884735107,
      "step": 431
    },
    {
      "epoch": 0.6530612244897959,
      "epsilon_dpo/beta": 0.02537233754992485,
      "epsilon_dpo/beta_margin_grad_mean": -0.39833539724349976,
      "epsilon_dpo/beta_margin_grad_std": 0.20775966346263885,
      "epsilon_dpo/beta_margin_mean": 0.5033900141716003,
      "epsilon_dpo/beta_margin_std": 1.0366266965866089,
      "epsilon_dpo/loss_margin_mean": 20.118471145629883,
      "grad_norm": 12.222615242004395,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.02546529471874237,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.632536862810844e-07,
      "logits/chosen": -0.6639053821563721,
      "logits/rejected": -0.31408679485321045,
      "logps/chosen": -117.53076171875,
      "logps/ref_chosen": -87.24708557128906,
      "logps/ref_rejected": -109.61161804199219,
      "logps/rejected": -160.01376342773438,
      "loss": 1.1779,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.7728521823883057,
      "rewards/margins": 0.5033900141716003,
      "rewards/rejected": -1.2762421369552612,
      "step": 432
    },
    {
      "epoch": 0.654572940287226,
      "epsilon_dpo/beta": 0.025245830416679382,
      "epsilon_dpo/beta_margin_grad_mean": -0.35882607102394104,
      "epsilon_dpo/beta_margin_grad_std": 0.21039634943008423,
      "epsilon_dpo/beta_margin_mean": 0.7088367342948914,
      "epsilon_dpo/beta_margin_std": 1.0585843324661255,
      "epsilon_dpo/loss_margin_mean": 28.348417282104492,
      "grad_norm": 11.30290699005127,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.0253701563924551,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.6201483487445515e-07,
      "logits/chosen": 0.08845943212509155,
      "logits/rejected": -0.13681207597255707,
      "logps/chosen": -109.51736450195312,
      "logps/ref_chosen": -84.12983703613281,
      "logps/ref_rejected": -88.07145690917969,
      "logps/rejected": -141.80740356445312,
      "loss": 1.0359,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6448712348937988,
      "rewards/margins": 0.7088366746902466,
      "rewards/rejected": -1.3537079095840454,
      "step": 433
    },
    {
      "epoch": 0.656084656084656,
      "epsilon_dpo/beta": 0.025136008858680725,
      "epsilon_dpo/beta_margin_grad_mean": -0.3597789704799652,
      "epsilon_dpo/beta_margin_grad_std": 0.1990552842617035,
      "epsilon_dpo/beta_margin_mean": 0.6943358182907104,
      "epsilon_dpo/beta_margin_std": 0.9835460782051086,
      "epsilon_dpo/loss_margin_mean": 27.884300231933594,
      "grad_norm": 9.337661743164062,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.025243936106562614,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.6077844460203204e-07,
      "logits/chosen": -0.3891775906085968,
      "logits/rejected": -0.4619581699371338,
      "logps/chosen": -87.47248077392578,
      "logps/ref_chosen": -67.95369720458984,
      "logps/ref_rejected": -95.22476196289062,
      "logps/rejected": -142.6278533935547,
      "loss": 1.0156,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.49363699555397034,
      "rewards/margins": 0.6943358778953552,
      "rewards/rejected": -1.1879727840423584,
      "step": 434
    },
    {
      "epoch": 0.6575963718820862,
      "epsilon_dpo/beta": 0.025065790861845016,
      "epsilon_dpo/beta_margin_grad_mean": -0.3998577892780304,
      "epsilon_dpo/beta_margin_grad_std": 0.19081562757492065,
      "epsilon_dpo/beta_margin_mean": 0.48510000109672546,
      "epsilon_dpo/beta_margin_std": 0.910292387008667,
      "epsilon_dpo/loss_margin_mean": 19.630142211914062,
      "grad_norm": 12.830538749694824,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.025133974850177765,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 1.5954455004830878e-07,
      "logits/chosen": -0.03043718822300434,
      "logits/rejected": 0.05432761833071709,
      "logps/chosen": -109.69837951660156,
      "logps/ref_chosen": -79.04033660888672,
      "logps/ref_rejected": -95.648193359375,
      "logps/rejected": -145.93637084960938,
      "loss": 1.1425,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.7722083926200867,
      "rewards/margins": 0.48510000109672546,
      "rewards/rejected": -1.2573083639144897,
      "step": 435
    },
    {
      "epoch": 0.6591080876795162,
      "epsilon_dpo/beta": 0.02495632693171501,
      "epsilon_dpo/beta_margin_grad_mean": -0.39338260889053345,
      "epsilon_dpo/beta_margin_grad_std": 0.19476363062858582,
      "epsilon_dpo/beta_margin_mean": 0.512516975402832,
      "epsilon_dpo/beta_margin_std": 0.9612401127815247,
      "epsilon_dpo/loss_margin_mean": 20.7864990234375,
      "grad_norm": 11.167028427124023,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.02506348490715027,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.5831318572796847e-07,
      "logits/chosen": -0.384287029504776,
      "logits/rejected": -0.22708836197853088,
      "logps/chosen": -105.12644958496094,
      "logps/ref_chosen": -79.86819458007812,
      "logps/ref_rejected": -94.28229522705078,
      "logps/rejected": -140.32705688476562,
      "loss": 1.1392,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.632888674736023,
      "rewards/margins": 0.512516975402832,
      "rewards/rejected": -1.145405650138855,
      "step": 436
    },
    {
      "epoch": 0.6606198034769464,
      "epsilon_dpo/beta": 0.024886613711714745,
      "epsilon_dpo/beta_margin_grad_mean": -0.40573927760124207,
      "epsilon_dpo/beta_margin_grad_std": 0.202871635556221,
      "epsilon_dpo/beta_margin_mean": 0.4717595875263214,
      "epsilon_dpo/beta_margin_std": 1.0110931396484375,
      "epsilon_dpo/loss_margin_mean": 19.250438690185547,
      "grad_norm": 17.93876075744629,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.024954309687018394,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 1.5708438608491815e-07,
      "logits/chosen": -0.05259464681148529,
      "logits/rejected": -0.2141057848930359,
      "logps/chosen": -112.15205383300781,
      "logps/ref_chosen": -78.60440826416016,
      "logps/ref_rejected": -120.1206283569336,
      "logps/rejected": -172.91871643066406,
      "loss": 1.1904,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.8363392949104309,
      "rewards/margins": 0.471759557723999,
      "rewards/rejected": -1.3080989122390747,
      "step": 437
    },
    {
      "epoch": 0.6621315192743764,
      "epsilon_dpo/beta": 0.024754598736763,
      "epsilon_dpo/beta_margin_grad_mean": -0.3753092885017395,
      "epsilon_dpo/beta_margin_grad_std": 0.18493610620498657,
      "epsilon_dpo/beta_margin_mean": 0.6025840640068054,
      "epsilon_dpo/beta_margin_std": 0.9036944508552551,
      "epsilon_dpo/loss_margin_mean": 24.554054260253906,
      "grad_norm": 9.134082794189453,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.02488432265818119,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.558581854913253e-07,
      "logits/chosen": -0.3997645974159241,
      "logits/rejected": -0.19586080312728882,
      "logps/chosen": -85.06373596191406,
      "logps/ref_chosen": -61.14134216308594,
      "logps/ref_rejected": -89.01496887207031,
      "logps/rejected": -137.49142456054688,
      "loss": 1.0492,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5958795547485352,
      "rewards/margins": 0.6025840640068054,
      "rewards/rejected": -1.1984636783599854,
      "step": 438
    },
    {
      "epoch": 0.6636432350718064,
      "epsilon_dpo/beta": 0.02466246485710144,
      "epsilon_dpo/beta_margin_grad_mean": -0.38297438621520996,
      "epsilon_dpo/beta_margin_grad_std": 0.1793513000011444,
      "epsilon_dpo/beta_margin_mean": 0.5795184373855591,
      "epsilon_dpo/beta_margin_std": 0.8874919414520264,
      "epsilon_dpo/loss_margin_mean": 23.736835479736328,
      "grad_norm": 9.677699089050293,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.024752821773290634,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.5463461824665658e-07,
      "logits/chosen": -0.28539150953292847,
      "logits/rejected": -0.40400251746177673,
      "logps/chosen": -133.99826049804688,
      "logps/ref_chosen": -105.3409652709961,
      "logps/ref_rejected": -118.54618072509766,
      "logps/rejected": -170.9403076171875,
      "loss": 1.0571,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.7099634408950806,
      "rewards/margins": 0.5795184373855591,
      "rewards/rejected": -1.2894818782806396,
      "step": 439
    },
    {
      "epoch": 0.6651549508692366,
      "epsilon_dpo/beta": 0.02450096420943737,
      "epsilon_dpo/beta_margin_grad_mean": -0.3457765281200409,
      "epsilon_dpo/beta_margin_grad_std": 0.17008066177368164,
      "epsilon_dpo/beta_margin_mean": 0.7509934902191162,
      "epsilon_dpo/beta_margin_std": 0.8576542139053345,
      "epsilon_dpo/loss_margin_mean": 30.81147003173828,
      "grad_norm": 11.456929206848145,
      "kl/avg_steps": 0.65625,
      "kl/beta": 0.024660347029566765,
      "kl/n_epsilon_steps": 0.171875,
      "kl/p_epsilon_steps": 0.828125,
      "learning_rate": 1.534137185767178e-07,
      "logits/chosen": -0.3186664581298828,
      "logits/rejected": -0.3347044289112091,
      "logps/chosen": -84.53070068359375,
      "logps/ref_chosen": -66.07489776611328,
      "logps/ref_rejected": -93.529296875,
      "logps/rejected": -142.79656982421875,
      "loss": 0.9247,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.4537859559059143,
      "rewards/margins": 0.7509934902191162,
      "rewards/rejected": -1.2047793865203857,
      "step": 440
    },
    {
      "epoch": 0.6666666666666666,
      "epsilon_dpo/beta": 0.02435653656721115,
      "epsilon_dpo/beta_margin_grad_mean": -0.36488860845565796,
      "epsilon_dpo/beta_margin_grad_std": 0.17126530408859253,
      "epsilon_dpo/beta_margin_mean": 0.6469001173973083,
      "epsilon_dpo/beta_margin_std": 0.8343473076820374,
      "epsilon_dpo/loss_margin_mean": 26.742786407470703,
      "grad_norm": 10.66884994506836,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.024499567225575447,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 1.521955206326976e-07,
      "logits/chosen": -0.12676770985126495,
      "logits/rejected": -0.3456115126609802,
      "logps/chosen": -100.19915771484375,
      "logps/ref_chosen": -80.23723602294922,
      "logps/ref_rejected": -110.27183532714844,
      "logps/rejected": -156.97653198242188,
      "loss": 0.9907,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.48816603422164917,
      "rewards/margins": 0.6469000577926636,
      "rewards/rejected": -1.1350661516189575,
      "step": 441
    },
    {
      "epoch": 0.6681783824640968,
      "epsilon_dpo/beta": 0.024269822984933853,
      "epsilon_dpo/beta_margin_grad_mean": -0.3921697437763214,
      "epsilon_dpo/beta_margin_grad_std": 0.18120039999485016,
      "epsilon_dpo/beta_margin_mean": 0.5139843821525574,
      "epsilon_dpo/beta_margin_std": 0.8636588454246521,
      "epsilon_dpo/loss_margin_mean": 21.431564331054688,
      "grad_norm": 9.974164962768555,
      "kl/avg_steps": 0.359375,
      "kl/beta": 0.0243549607694149,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 1.5098005849021078e-07,
      "logits/chosen": -0.4980059862136841,
      "logits/rejected": -0.2725337743759155,
      "logps/chosen": -117.87562561035156,
      "logps/ref_chosen": -88.29069519042969,
      "logps/ref_rejected": -115.02373504638672,
      "logps/rejected": -166.04022216796875,
      "loss": 1.1021,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.7207363247871399,
      "rewards/margins": 0.5139843821525574,
      "rewards/rejected": -1.2347207069396973,
      "step": 442
    },
    {
      "epoch": 0.6696900982615268,
      "epsilon_dpo/beta": 0.024133656173944473,
      "epsilon_dpo/beta_margin_grad_mean": -0.36061957478523254,
      "epsilon_dpo/beta_margin_grad_std": 0.19436120986938477,
      "epsilon_dpo/beta_margin_mean": 0.6958425045013428,
      "epsilon_dpo/beta_margin_std": 1.0084832906723022,
      "epsilon_dpo/loss_margin_mean": 29.060047149658203,
      "grad_norm": 9.462386131286621,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.02426774799823761,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 1.4976736614834662e-07,
      "logits/chosen": -0.018897585570812225,
      "logits/rejected": -0.1660281866788864,
      "logps/chosen": -105.89450073242188,
      "logps/ref_chosen": -79.12504577636719,
      "logps/ref_rejected": -106.16142272949219,
      "logps/rejected": -161.99093627929688,
      "loss": 1.0179,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6496339440345764,
      "rewards/margins": 0.6958425045013428,
      "rewards/rejected": -1.3454763889312744,
      "step": 443
    },
    {
      "epoch": 0.671201814058957,
      "epsilon_dpo/beta": 0.02405519038438797,
      "epsilon_dpo/beta_margin_grad_mean": -0.4312235116958618,
      "epsilon_dpo/beta_margin_grad_std": 0.1837351769208908,
      "epsilon_dpo/beta_margin_mean": 0.33834654092788696,
      "epsilon_dpo/beta_margin_std": 0.907697856426239,
      "epsilon_dpo/loss_margin_mean": 14.312912940979004,
      "grad_norm": 14.25627326965332,
      "kl/avg_steps": 0.328125,
      "kl/beta": 0.024132005870342255,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 1.4855747752871654e-07,
      "logits/chosen": -0.4729708135128021,
      "logits/rejected": -0.5134774446487427,
      "logps/chosen": -112.18655395507812,
      "logps/ref_chosen": -80.70797729492188,
      "logps/ref_rejected": -115.05545806884766,
      "logps/rejected": -160.84695434570312,
      "loss": 1.2564,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.7601273059844971,
      "rewards/margins": 0.33834654092788696,
      "rewards/rejected": -1.0984739065170288,
      "step": 444
    },
    {
      "epoch": 0.672713529856387,
      "epsilon_dpo/beta": 0.023935209959745407,
      "epsilon_dpo/beta_margin_grad_mean": -0.3562336564064026,
      "epsilon_dpo/beta_margin_grad_std": 0.17927402257919312,
      "epsilon_dpo/beta_margin_mean": 0.6905251741409302,
      "epsilon_dpo/beta_margin_std": 0.8721351027488708,
      "epsilon_dpo/loss_margin_mean": 29.084095001220703,
      "grad_norm": 13.974209785461426,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.02405308187007904,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.473504264745062e-07,
      "logits/chosen": -0.3294023871421814,
      "logits/rejected": -0.16536982357501984,
      "logps/chosen": -113.95536041259766,
      "logps/ref_chosen": -84.2096939086914,
      "logps/ref_rejected": -96.03083038330078,
      "logps/rejected": -154.860595703125,
      "loss": 0.9751,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.7159168720245361,
      "rewards/margins": 0.6905251741409302,
      "rewards/rejected": -1.4064419269561768,
      "step": 445
    },
    {
      "epoch": 0.674225245653817,
      "epsilon_dpo/beta": 0.02380117028951645,
      "epsilon_dpo/beta_margin_grad_mean": -0.3451477587223053,
      "epsilon_dpo/beta_margin_grad_std": 0.15824930369853973,
      "epsilon_dpo/beta_margin_mean": 0.7407541275024414,
      "epsilon_dpo/beta_margin_std": 0.7942438721656799,
      "epsilon_dpo/loss_margin_mean": 31.29979133605957,
      "grad_norm": 9.890649795532227,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.02393341436982155,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 1.461462467495284e-07,
      "logits/chosen": -0.07980051636695862,
      "logits/rejected": -0.21256430447101593,
      "logps/chosen": -90.020263671875,
      "logps/ref_chosen": -67.17405700683594,
      "logps/ref_rejected": -93.05891418457031,
      "logps/rejected": -147.2049102783203,
      "loss": 0.9097,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.5450651049613953,
      "rewards/margins": 0.7407541871070862,
      "rewards/rejected": -1.2858192920684814,
      "step": 446
    },
    {
      "epoch": 0.6757369614512472,
      "epsilon_dpo/beta": 0.023675473406910896,
      "epsilon_dpo/beta_margin_grad_mean": -0.37413400411605835,
      "epsilon_dpo/beta_margin_grad_std": 0.1963834911584854,
      "epsilon_dpo/beta_margin_mean": 0.6092168688774109,
      "epsilon_dpo/beta_margin_std": 0.9781890511512756,
      "epsilon_dpo/loss_margin_mean": 25.984163284301758,
      "grad_norm": 11.772629737854004,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.02379954233765602,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.4494497203727843e-07,
      "logits/chosen": -0.23869886994361877,
      "logits/rejected": -0.692590057849884,
      "logps/chosen": -97.84403991699219,
      "logps/ref_chosen": -76.00971984863281,
      "logps/ref_rejected": -110.74910736083984,
      "logps/rejected": -158.56759643554688,
      "loss": 1.0732,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5195660591125488,
      "rewards/margins": 0.6092169284820557,
      "rewards/rejected": -1.1287829875946045,
      "step": 447
    },
    {
      "epoch": 0.6772486772486772,
      "epsilon_dpo/beta": 0.023565160110592842,
      "epsilon_dpo/beta_margin_grad_mean": -0.3909008204936981,
      "epsilon_dpo/beta_margin_grad_std": 0.18127767741680145,
      "epsilon_dpo/beta_margin_mean": 0.5344966650009155,
      "epsilon_dpo/beta_margin_std": 0.8911596536636353,
      "epsilon_dpo/loss_margin_mean": 22.905698776245117,
      "grad_norm": 10.394296646118164,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.023673774674534798,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.4374663593999256e-07,
      "logits/chosen": -0.38351887464523315,
      "logits/rejected": -0.38271769881248474,
      "logps/chosen": -107.19539642333984,
      "logps/ref_chosen": -78.36839294433594,
      "logps/ref_rejected": -104.13056945800781,
      "logps/rejected": -155.86328125,
      "loss": 1.0935,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6800907850265503,
      "rewards/margins": 0.5344966650009155,
      "rewards/rejected": -1.2145874500274658,
      "step": 448
    },
    {
      "epoch": 0.6787603930461074,
      "epsilon_dpo/beta": 0.023492034524679184,
      "epsilon_dpo/beta_margin_grad_mean": -0.4497864544391632,
      "epsilon_dpo/beta_margin_grad_std": 0.19710615277290344,
      "epsilon_dpo/beta_margin_mean": 0.2375793606042862,
      "epsilon_dpo/beta_margin_std": 0.9235098958015442,
      "epsilon_dpo/loss_margin_mean": 10.408994674682617,
      "grad_norm": 24.015352249145508,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.02356332167983055,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 1.4255127197770707e-07,
      "logits/chosen": -0.5230458974838257,
      "logits/rejected": -0.2880995273590088,
      "logps/chosen": -136.14390563964844,
      "logps/ref_chosen": -101.54728698730469,
      "logps/ref_rejected": -109.60847473144531,
      "logps/rejected": -154.6140899658203,
      "loss": 1.3573,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.814102053642273,
      "rewards/margins": 0.2375793755054474,
      "rewards/rejected": -1.051681399345398,
      "step": 449
    },
    {
      "epoch": 0.6802721088435374,
      "epsilon_dpo/beta": 0.023389486595988274,
      "epsilon_dpo/beta_margin_grad_mean": -0.40028244256973267,
      "epsilon_dpo/beta_margin_grad_std": 0.19258946180343628,
      "epsilon_dpo/beta_margin_mean": 0.4728687107563019,
      "epsilon_dpo/beta_margin_std": 0.9459613561630249,
      "epsilon_dpo/loss_margin_mean": 20.483104705810547,
      "grad_norm": 12.487074851989746,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.02348991669714451,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.4135891358732205e-07,
      "logits/chosen": -0.49118733406066895,
      "logits/rejected": -0.6007115840911865,
      "logps/chosen": -92.49281311035156,
      "logps/ref_chosen": -66.52513122558594,
      "logps/ref_rejected": -100.74227905273438,
      "logps/rejected": -147.19308471679688,
      "loss": 1.1646,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6101650595664978,
      "rewards/margins": 0.4728686809539795,
      "rewards/rejected": -1.083033800125122,
      "step": 450
    },
    {
      "epoch": 0.6817838246409675,
      "epsilon_dpo/beta": 0.023302221670746803,
      "epsilon_dpo/beta_margin_grad_mean": -0.40755635499954224,
      "epsilon_dpo/beta_margin_grad_std": 0.18733078241348267,
      "epsilon_dpo/beta_margin_mean": 0.44163015484809875,
      "epsilon_dpo/beta_margin_std": 0.8944984674453735,
      "epsilon_dpo/loss_margin_mean": 19.218761444091797,
      "grad_norm": 9.85672378540039,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.023387596011161804,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.4016959412166437e-07,
      "logits/chosen": -0.24648326635360718,
      "logits/rejected": -0.27126193046569824,
      "logps/chosen": -112.3831787109375,
      "logps/ref_chosen": -86.15756225585938,
      "logps/ref_rejected": -100.31735229492188,
      "logps/rejected": -145.76173400878906,
      "loss": 1.1708,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6132830381393433,
      "rewards/margins": 0.44163012504577637,
      "rewards/rejected": -1.0549132823944092,
      "step": 451
    },
    {
      "epoch": 0.6832955404383976,
      "epsilon_dpo/beta": 0.023193318396806717,
      "epsilon_dpo/beta_margin_grad_mean": -0.3813394606113434,
      "epsilon_dpo/beta_margin_grad_std": 0.19726483523845673,
      "epsilon_dpo/beta_margin_mean": 0.5745446085929871,
      "epsilon_dpo/beta_margin_std": 0.9635257720947266,
      "epsilon_dpo/loss_margin_mean": 25.042280197143555,
      "grad_norm": 10.472002029418945,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.02330021932721138,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.3898334684855645e-07,
      "logits/chosen": -0.33625802397727966,
      "logits/rejected": -0.4327930212020874,
      "logps/chosen": -102.27005004882812,
      "logps/ref_chosen": -75.09101867675781,
      "logps/ref_rejected": -104.30365753173828,
      "logps/rejected": -156.52496337890625,
      "loss": 1.0946,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.634591281414032,
      "rewards/margins": 0.5745445489883423,
      "rewards/rejected": -1.2091357707977295,
      "step": 452
    },
    {
      "epoch": 0.6848072562358276,
      "epsilon_dpo/beta": 0.02309594303369522,
      "epsilon_dpo/beta_margin_grad_mean": -0.40049630403518677,
      "epsilon_dpo/beta_margin_grad_std": 0.2005462944507599,
      "epsilon_dpo/beta_margin_mean": 0.48215773701667786,
      "epsilon_dpo/beta_margin_std": 0.9639610052108765,
      "epsilon_dpo/loss_margin_mean": 21.156658172607422,
      "grad_norm": 13.386679649353027,
      "kl/avg_steps": 0.421875,
      "kl/beta": 0.023191509768366814,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 1.3780020494988445e-07,
      "logits/chosen": -0.18490701913833618,
      "logits/rejected": -0.4014432430267334,
      "logps/chosen": -113.76175689697266,
      "logps/ref_chosen": -85.76741027832031,
      "logps/ref_rejected": -97.41229248046875,
      "logps/rejected": -146.56329345703125,
      "loss": 1.1674,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6490232348442078,
      "rewards/margins": 0.48215776681900024,
      "rewards/rejected": -1.131181001663208,
      "step": 453
    },
    {
      "epoch": 0.6863189720332578,
      "epsilon_dpo/beta": 0.022973690181970596,
      "epsilon_dpo/beta_margin_grad_mean": -0.36466944217681885,
      "epsilon_dpo/beta_margin_grad_std": 0.17821235954761505,
      "epsilon_dpo/beta_margin_mean": 0.6326720714569092,
      "epsilon_dpo/beta_margin_std": 0.8994981050491333,
      "epsilon_dpo/loss_margin_mean": 27.766921997070312,
      "grad_norm": 10.431668281555176,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.02309408038854599,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.366202015206706e-07,
      "logits/chosen": -0.28234240412712097,
      "logits/rejected": -0.38511741161346436,
      "logps/chosen": -92.9898452758789,
      "logps/ref_chosen": -73.73841857910156,
      "logps/ref_rejected": -89.2447280883789,
      "logps/rejected": -136.26307678222656,
      "loss": 1.0238,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.44317740201950073,
      "rewards/margins": 0.6326720714569092,
      "rewards/rejected": -1.0758495330810547,
      "step": 454
    },
    {
      "epoch": 0.6878306878306878,
      "epsilon_dpo/beta": 0.02285228855907917,
      "epsilon_dpo/beta_margin_grad_mean": -0.36160367727279663,
      "epsilon_dpo/beta_margin_grad_std": 0.1697792261838913,
      "epsilon_dpo/beta_margin_mean": 0.6611186861991882,
      "epsilon_dpo/beta_margin_std": 0.8520635962486267,
      "epsilon_dpo/loss_margin_mean": 29.138427734375,
      "grad_norm": 10.137971878051758,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.022972041741013527,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.354433695681474e-07,
      "logits/chosen": -0.44800513982772827,
      "logits/rejected": -0.49094438552856445,
      "logps/chosen": -125.23015594482422,
      "logps/ref_chosen": -97.0819320678711,
      "logps/ref_rejected": -106.07518768310547,
      "logps/rejected": -163.36184692382812,
      "loss": 0.9844,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6450259685516357,
      "rewards/margins": 0.661118745803833,
      "rewards/rejected": -1.3061447143554688,
      "step": 455
    },
    {
      "epoch": 0.6893424036281179,
      "epsilon_dpo/beta": 0.022717243060469627,
      "epsilon_dpo/beta_margin_grad_mean": -0.38223299384117126,
      "epsilon_dpo/beta_margin_grad_std": 0.17249758541584015,
      "epsilon_dpo/beta_margin_mean": 0.5534289479255676,
      "epsilon_dpo/beta_margin_std": 0.8530722856521606,
      "epsilon_dpo/loss_margin_mean": 24.55746841430664,
      "grad_norm": 10.840579986572266,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.022850647568702698,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 1.3426974201083439e-07,
      "logits/chosen": -0.3131183385848999,
      "logits/rejected": -0.05182047188282013,
      "logps/chosen": -105.27415466308594,
      "logps/ref_chosen": -78.9945297241211,
      "logps/ref_rejected": -106.11909484863281,
      "logps/rejected": -156.9561767578125,
      "loss": 1.0652,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.5996953248977661,
      "rewards/margins": 0.5534289479255676,
      "rewards/rejected": -1.1531243324279785,
      "step": 456
    },
    {
      "epoch": 0.690854119425548,
      "epsilon_dpo/beta": 0.022618653252720833,
      "epsilon_dpo/beta_margin_grad_mean": -0.4127366840839386,
      "epsilon_dpo/beta_margin_grad_std": 0.159461110830307,
      "epsilon_dpo/beta_margin_mean": 0.40310725569725037,
      "epsilon_dpo/beta_margin_std": 0.7608078718185425,
      "epsilon_dpo/loss_margin_mean": 18.023801803588867,
      "grad_norm": 12.661500930786133,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.022715773433446884,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.3309935167761717e-07,
      "logits/chosen": -0.10714979469776154,
      "logits/rejected": -0.5415033102035522,
      "logps/chosen": -98.71037292480469,
      "logps/ref_chosen": -66.06265258789062,
      "logps/ref_rejected": -87.40638732910156,
      "logps/rejected": -138.07791137695312,
      "loss": 1.1523,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7403136491775513,
      "rewards/margins": 0.40310725569725037,
      "rewards/rejected": -1.143420934677124,
      "step": 457
    },
    {
      "epoch": 0.6923658352229781,
      "epsilon_dpo/beta": 0.022498922422528267,
      "epsilon_dpo/beta_margin_grad_mean": -0.36385974287986755,
      "epsilon_dpo/beta_margin_grad_std": 0.16366532444953918,
      "epsilon_dpo/beta_margin_mean": 0.6369403004646301,
      "epsilon_dpo/beta_margin_std": 0.7969620823860168,
      "epsilon_dpo/loss_margin_mean": 28.51059341430664,
      "grad_norm": 11.44316291809082,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.022616824135184288,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.3193223130682936e-07,
      "logits/chosen": -0.06593091785907745,
      "logits/rejected": -0.5136197805404663,
      "logps/chosen": -93.28556823730469,
      "logps/ref_chosen": -72.29696655273438,
      "logps/ref_rejected": -112.85076904296875,
      "logps/rejected": -162.34996032714844,
      "loss": 0.9855,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.47399500012397766,
      "rewards/margins": 0.6369403004646301,
      "rewards/rejected": -1.1109352111816406,
      "step": 458
    },
    {
      "epoch": 0.6938775510204082,
      "epsilon_dpo/beta": 0.022362414747476578,
      "epsilon_dpo/beta_margin_grad_mean": -0.3720557391643524,
      "epsilon_dpo/beta_margin_grad_std": 0.15929129719734192,
      "epsilon_dpo/beta_margin_mean": 0.6010655164718628,
      "epsilon_dpo/beta_margin_std": 0.7777773141860962,
      "epsilon_dpo/loss_margin_mean": 27.049121856689453,
      "grad_norm": 11.10675048828125,
      "kl/avg_steps": 0.609375,
      "kl/beta": 0.022497307509183884,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 1.3076841354533658e-07,
      "logits/chosen": -0.25048884749412537,
      "logits/rejected": -0.319161981344223,
      "logps/chosen": -113.13943481445312,
      "logps/ref_chosen": -90.7015609741211,
      "logps/ref_rejected": -112.55254364013672,
      "logps/rejected": -162.03955078125,
      "loss": 1.0037,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.5026448965072632,
      "rewards/margins": 0.6010655164718628,
      "rewards/rejected": -1.103710412979126,
      "step": 459
    },
    {
      "epoch": 0.6953892668178382,
      "epsilon_dpo/beta": 0.022265443578362465,
      "epsilon_dpo/beta_margin_grad_mean": -0.3674984574317932,
      "epsilon_dpo/beta_margin_grad_std": 0.17106232047080994,
      "epsilon_dpo/beta_margin_mean": 0.6420856714248657,
      "epsilon_dpo/beta_margin_std": 0.8475755453109741,
      "epsilon_dpo/loss_margin_mean": 29.068513870239258,
      "grad_norm": 9.10698413848877,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.022361045703291893,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.2960793094762345e-07,
      "logits/chosen": -0.18218980729579926,
      "logits/rejected": -0.37891292572021484,
      "logps/chosen": -115.58694458007812,
      "logps/ref_chosen": -85.61819458007812,
      "logps/ref_rejected": -122.83721923828125,
      "logps/rejected": -181.87448120117188,
      "loss": 0.9972,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6702232360839844,
      "rewards/margins": 0.6420857310295105,
      "rewards/rejected": -1.3123090267181396,
      "step": 460
    },
    {
      "epoch": 0.6969009826152683,
      "epsilon_dpo/beta": 0.022154536098241806,
      "epsilon_dpo/beta_margin_grad_mean": -0.3609759509563446,
      "epsilon_dpo/beta_margin_grad_std": 0.16306711733341217,
      "epsilon_dpo/beta_margin_mean": 0.6506234407424927,
      "epsilon_dpo/beta_margin_std": 0.7753090858459473,
      "epsilon_dpo/loss_margin_mean": 29.590946197509766,
      "grad_norm": 10.064002990722656,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.02226364053785801,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.2845081597488286e-07,
      "logits/chosen": -0.2814715504646301,
      "logits/rejected": -0.2578463554382324,
      "logps/chosen": -92.06724548339844,
      "logps/ref_chosen": -68.47746276855469,
      "logps/ref_rejected": -101.3572998046875,
      "logps/rejected": -154.5380401611328,
      "loss": 0.97,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5240465402603149,
      "rewards/margins": 0.6506234407424927,
      "rewards/rejected": -1.1746699810028076,
      "step": 461
    },
    {
      "epoch": 0.6984126984126984,
      "epsilon_dpo/beta": 0.022016622126102448,
      "epsilon_dpo/beta_margin_grad_mean": -0.364952951669693,
      "epsilon_dpo/beta_margin_grad_std": 0.17581763863563538,
      "epsilon_dpo/beta_margin_mean": 0.6401377320289612,
      "epsilon_dpo/beta_margin_std": 0.8566480875015259,
      "epsilon_dpo/loss_margin_mean": 29.284711837768555,
      "grad_norm": 11.0882568359375,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.02215287648141384,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 1.27297100994108e-07,
      "logits/chosen": 0.07481776177883148,
      "logits/rejected": -0.13942140340805054,
      "logps/chosen": -109.2104263305664,
      "logps/ref_chosen": -82.09750366210938,
      "logps/ref_rejected": -96.87604522705078,
      "logps/rejected": -153.273681640625,
      "loss": 1.0044,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.5979229807853699,
      "rewards/margins": 0.6401377320289612,
      "rewards/rejected": -1.238060712814331,
      "step": 462
    },
    {
      "epoch": 0.6999244142101285,
      "epsilon_dpo/beta": 0.02194179780781269,
      "epsilon_dpo/beta_margin_grad_mean": -0.40211692452430725,
      "epsilon_dpo/beta_margin_grad_std": 0.17251704633235931,
      "epsilon_dpo/beta_margin_mean": 0.47483089566230774,
      "epsilon_dpo/beta_margin_std": 0.8205597400665283,
      "epsilon_dpo/loss_margin_mean": 21.893447875976562,
      "grad_norm": 10.746212005615234,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.022015281021595,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 1.2614681827718695e-07,
      "logits/chosen": -0.27809298038482666,
      "logits/rejected": -0.3278348445892334,
      "logps/chosen": -124.42726135253906,
      "logps/ref_chosen": -94.08966064453125,
      "logps/ref_rejected": -94.4760971069336,
      "logps/rejected": -146.7071533203125,
      "loss": 1.1157,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.6673296689987183,
      "rewards/margins": 0.4748309552669525,
      "rewards/rejected": -1.1421605348587036,
      "step": 463
    },
    {
      "epoch": 0.7014361300075586,
      "epsilon_dpo/beta": 0.021859774366021156,
      "epsilon_dpo/beta_margin_grad_mean": -0.38281190395355225,
      "epsilon_dpo/beta_margin_grad_std": 0.2125449925661087,
      "epsilon_dpo/beta_margin_mean": 0.5871289372444153,
      "epsilon_dpo/beta_margin_std": 1.0753085613250732,
      "epsilon_dpo/loss_margin_mean": 27.19884490966797,
      "grad_norm": 10.855381965637207,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.02193986251950264,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.2500000000000005e-07,
      "logits/chosen": -0.21308638155460358,
      "logits/rejected": -0.13446223735809326,
      "logps/chosen": -114.7242202758789,
      "logps/ref_chosen": -81.69437408447266,
      "logps/ref_rejected": -90.02086639404297,
      "logps/rejected": -150.2495574951172,
      "loss": 1.1303,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.7249494791030884,
      "rewards/margins": 0.5871289372444153,
      "rewards/rejected": -1.3120784759521484,
      "step": 464
    },
    {
      "epoch": 0.7029478458049887,
      "epsilon_dpo/beta": 0.021750781685113907,
      "epsilon_dpo/beta_margin_grad_mean": -0.3837484121322632,
      "epsilon_dpo/beta_margin_grad_std": 0.19969280064105988,
      "epsilon_dpo/beta_margin_mean": 0.587910532951355,
      "epsilon_dpo/beta_margin_std": 0.9850619435310364,
      "epsilon_dpo/loss_margin_mean": 27.304344177246094,
      "grad_norm": 10.320209503173828,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.02185789681971073,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.238566782415197e-07,
      "logits/chosen": -0.0753406435251236,
      "logits/rejected": -0.21963469684123993,
      "logps/chosen": -116.1302719116211,
      "logps/ref_chosen": -81.77964782714844,
      "logps/ref_rejected": -109.90087890625,
      "logps/rejected": -171.55584716796875,
      "loss": 1.0913,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.7500728368759155,
      "rewards/margins": 0.587910532951355,
      "rewards/rejected": -1.3379833698272705,
      "step": 465
    },
    {
      "epoch": 0.7044595616024187,
      "epsilon_dpo/beta": 0.021703744307160378,
      "epsilon_dpo/beta_margin_grad_mean": -0.4434705674648285,
      "epsilon_dpo/beta_margin_grad_std": 0.17293018102645874,
      "epsilon_dpo/beta_margin_mean": 0.2722676396369934,
      "epsilon_dpo/beta_margin_std": 0.8159366250038147,
      "epsilon_dpo/loss_margin_mean": 12.814265251159668,
      "grad_norm": 19.282155990600586,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.02174915000796318,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 1.2271688498291334e-07,
      "logits/chosen": -0.382304847240448,
      "logits/rejected": -0.16424314677715302,
      "logps/chosen": -136.80435180664062,
      "logps/ref_chosen": -99.00543212890625,
      "logps/ref_rejected": -97.36400604248047,
      "logps/rejected": -147.97720336914062,
      "loss": 1.283,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.8247905373573303,
      "rewards/margins": 0.272267609834671,
      "rewards/rejected": -1.0970580577850342,
      "step": 466
    },
    {
      "epoch": 0.7059712773998488,
      "epsilon_dpo/beta": 0.02159532904624939,
      "epsilon_dpo/beta_margin_grad_mean": -0.3907034397125244,
      "epsilon_dpo/beta_margin_grad_std": 0.1725900024175644,
      "epsilon_dpo/beta_margin_mean": 0.5272072553634644,
      "epsilon_dpo/beta_margin_std": 0.8612512350082397,
      "epsilon_dpo/loss_margin_mean": 24.633930206298828,
      "grad_norm": 11.24698543548584,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.02170167863368988,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.2158065210664848e-07,
      "logits/chosen": -0.05803515762090683,
      "logits/rejected": -0.4103708267211914,
      "logps/chosen": -98.90908813476562,
      "logps/ref_chosen": -68.85592651367188,
      "logps/ref_rejected": -114.98614501953125,
      "logps/rejected": -169.67323303222656,
      "loss": 1.0855,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6503223180770874,
      "rewards/margins": 0.5272072553634644,
      "rewards/rejected": -1.1775295734405518,
      "step": 467
    },
    {
      "epoch": 0.7074829931972789,
      "epsilon_dpo/beta": 0.021474391222000122,
      "epsilon_dpo/beta_margin_grad_mean": -0.3783760070800781,
      "epsilon_dpo/beta_margin_grad_std": 0.19576546549797058,
      "epsilon_dpo/beta_margin_mean": 0.5842656493186951,
      "epsilon_dpo/beta_margin_std": 0.9758591651916504,
      "epsilon_dpo/loss_margin_mean": 27.47597312927246,
      "grad_norm": 9.588836669921875,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.02159370854496956,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 1.204480113956011e-07,
      "logits/chosen": -0.25211167335510254,
      "logits/rejected": -0.08560752868652344,
      "logps/chosen": -119.99714660644531,
      "logps/ref_chosen": -86.93231201171875,
      "logps/ref_rejected": -97.35225677490234,
      "logps/rejected": -157.89306640625,
      "loss": 1.0908,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.7126418352127075,
      "rewards/margins": 0.5842655897140503,
      "rewards/rejected": -1.2969074249267578,
      "step": 468
    },
    {
      "epoch": 0.708994708994709,
      "epsilon_dpo/beta": 0.02141467295587063,
      "epsilon_dpo/beta_margin_grad_mean": -0.3991638720035553,
      "epsilon_dpo/beta_margin_grad_std": 0.1622338443994522,
      "epsilon_dpo/beta_margin_mean": 0.4792685806751251,
      "epsilon_dpo/beta_margin_std": 0.7716690301895142,
      "epsilon_dpo/loss_margin_mean": 22.63409996032715,
      "grad_norm": 10.76986312866211,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.021472923457622528,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 1.1931899453216697e-07,
      "logits/chosen": -0.527290940284729,
      "logits/rejected": -0.24194341897964478,
      "logps/chosen": -113.86764526367188,
      "logps/ref_chosen": -81.70521545410156,
      "logps/ref_rejected": -95.8597640991211,
      "logps/rejected": -150.65628051757812,
      "loss": 1.0949,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.6916900873184204,
      "rewards/margins": 0.47926855087280273,
      "rewards/rejected": -1.1709586381912231,
      "step": 469
    },
    {
      "epoch": 0.7105064247921391,
      "epsilon_dpo/beta": 0.021307768300175667,
      "epsilon_dpo/beta_margin_grad_mean": -0.3925294578075409,
      "epsilon_dpo/beta_margin_grad_std": 0.17281277477741241,
      "epsilon_dpo/beta_margin_mean": 0.5017194747924805,
      "epsilon_dpo/beta_margin_std": 0.8305301070213318,
      "epsilon_dpo/loss_margin_mean": 23.785242080688477,
      "grad_norm": 13.406644821166992,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.021412700414657593,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.1819363309737438e-07,
      "logits/chosen": 0.09025856852531433,
      "logits/rejected": -0.2041073441505432,
      "logps/chosen": -101.21160888671875,
      "logps/ref_chosen": -69.24808502197266,
      "logps/ref_rejected": -91.63481140136719,
      "logps/rejected": -147.38357543945312,
      "loss": 1.0982,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6838077306747437,
      "rewards/margins": 0.5017194151878357,
      "rewards/rejected": -1.1855272054672241,
      "step": 470
    },
    {
      "epoch": 0.7120181405895691,
      "epsilon_dpo/beta": 0.021201759576797485,
      "epsilon_dpo/beta_margin_grad_mean": -0.36601752042770386,
      "epsilon_dpo/beta_margin_grad_std": 0.18967947363853455,
      "epsilon_dpo/beta_margin_mean": 0.6323056817054749,
      "epsilon_dpo/beta_margin_std": 0.9497935175895691,
      "epsilon_dpo/loss_margin_mean": 30.101776123046875,
      "grad_norm": 10.747753143310547,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.021306170150637627,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.1707195857000215e-07,
      "logits/chosen": -0.4186813235282898,
      "logits/rejected": -0.3955724835395813,
      "logps/chosen": -109.00920867919922,
      "logps/ref_chosen": -80.25093078613281,
      "logps/ref_rejected": -102.29495239257812,
      "logps/rejected": -161.15499877929688,
      "loss": 1.0449,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6124268174171448,
      "rewards/margins": 0.6323057413101196,
      "rewards/rejected": -1.2447326183319092,
      "step": 471
    },
    {
      "epoch": 0.7135298563869993,
      "epsilon_dpo/beta": 0.021109528839588165,
      "epsilon_dpo/beta_margin_grad_mean": -0.3851298987865448,
      "epsilon_dpo/beta_margin_grad_std": 0.19840273261070251,
      "epsilon_dpo/beta_margin_mean": 0.5623181462287903,
      "epsilon_dpo/beta_margin_std": 1.0013166666030884,
      "epsilon_dpo/loss_margin_mean": 26.94999885559082,
      "grad_norm": 10.546019554138184,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.021200168877840042,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.1595400232569768e-07,
      "logits/chosen": -0.40806734561920166,
      "logits/rejected": -0.21839666366577148,
      "logps/chosen": -111.0872802734375,
      "logps/ref_chosen": -82.03518676757812,
      "logps/ref_rejected": -102.50106811523438,
      "logps/rejected": -158.503173828125,
      "loss": 1.1144,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.6167609095573425,
      "rewards/margins": 0.5623180866241455,
      "rewards/rejected": -1.1790790557861328,
      "step": 472
    },
    {
      "epoch": 0.7150415721844293,
      "epsilon_dpo/beta": 0.02099778689444065,
      "epsilon_dpo/beta_margin_grad_mean": -0.3725453317165375,
      "epsilon_dpo/beta_margin_grad_std": 0.19428926706314087,
      "epsilon_dpo/beta_margin_mean": 0.6166351437568665,
      "epsilon_dpo/beta_margin_std": 0.9741420745849609,
      "epsilon_dpo/loss_margin_mean": 29.64267921447754,
      "grad_norm": 9.246530532836914,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.02110782265663147,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.1483979563610069e-07,
      "logits/chosen": -0.1701044887304306,
      "logits/rejected": -0.309928834438324,
      "logps/chosen": -100.15250396728516,
      "logps/ref_chosen": -75.42779541015625,
      "logps/ref_rejected": -120.35392761230469,
      "logps/rejected": -174.7213134765625,
      "loss": 1.0653,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5218599438667297,
      "rewards/margins": 0.6166351437568665,
      "rewards/rejected": -1.1384950876235962,
      "step": 473
    },
    {
      "epoch": 0.7165532879818595,
      "epsilon_dpo/beta": 0.020906511694192886,
      "epsilon_dpo/beta_margin_grad_mean": -0.4070540964603424,
      "epsilon_dpo/beta_margin_grad_std": 0.18986035883426666,
      "epsilon_dpo/beta_margin_mean": 0.44220811128616333,
      "epsilon_dpo/beta_margin_std": 0.9186369776725769,
      "epsilon_dpo/loss_margin_mean": 21.432273864746094,
      "grad_norm": 13.415764808654785,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.020996280014514923,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.1372936966796709e-07,
      "logits/chosen": -0.3077600598335266,
      "logits/rejected": -0.49201875925064087,
      "logps/chosen": -107.1239013671875,
      "logps/ref_chosen": -70.26007080078125,
      "logps/ref_rejected": -95.32382202148438,
      "logps/rejected": -153.61993408203125,
      "loss": 1.1793,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7730291485786438,
      "rewards/margins": 0.44220811128616333,
      "rewards/rejected": -1.2152371406555176,
      "step": 474
    },
    {
      "epoch": 0.7180650037792895,
      "epsilon_dpo/beta": 0.02077624201774597,
      "epsilon_dpo/beta_margin_grad_mean": -0.3475864827632904,
      "epsilon_dpo/beta_margin_grad_std": 0.15696246922016144,
      "epsilon_dpo/beta_margin_mean": 0.7207489013671875,
      "epsilon_dpo/beta_margin_std": 0.7813100814819336,
      "epsilon_dpo/loss_margin_mean": 34.88127899169922,
      "grad_norm": 10.983023643493652,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.020904820412397385,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 1.126227554822985e-07,
      "logits/chosen": -0.3970502018928528,
      "logits/rejected": -0.2638678252696991,
      "logps/chosen": -116.77337646484375,
      "logps/ref_chosen": -82.82109069824219,
      "logps/ref_rejected": -113.2535400390625,
      "logps/rejected": -182.08709716796875,
      "loss": 0.9198,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": -0.7075306177139282,
      "rewards/margins": 0.7207489013671875,
      "rewards/rejected": -1.4282796382904053,
      "step": 475
    },
    {
      "epoch": 0.7195767195767195,
      "epsilon_dpo/beta": 0.020699141547083855,
      "epsilon_dpo/beta_margin_grad_mean": -0.4025791883468628,
      "epsilon_dpo/beta_margin_grad_std": 0.1724308729171753,
      "epsilon_dpo/beta_margin_mean": 0.45882412791252136,
      "epsilon_dpo/beta_margin_std": 0.8210156559944153,
      "epsilon_dpo/loss_margin_mean": 22.426555633544922,
      "grad_norm": 10.899810791015625,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.020774977281689644,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.1151998403347243e-07,
      "logits/chosen": -0.178768128156662,
      "logits/rejected": -0.3625830411911011,
      "logps/chosen": -143.37841796875,
      "logps/ref_chosen": -102.48684692382812,
      "logps/ref_rejected": -102.62210083007812,
      "logps/rejected": -165.94021606445312,
      "loss": 1.1291,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.8488463163375854,
      "rewards/margins": 0.45882412791252136,
      "rewards/rejected": -1.3076703548431396,
      "step": 476
    },
    {
      "epoch": 0.7210884353741497,
      "epsilon_dpo/beta": 0.0206476841121912,
      "epsilon_dpo/beta_margin_grad_mean": -0.41202253103256226,
      "epsilon_dpo/beta_margin_grad_std": 0.20400650799274445,
      "epsilon_dpo/beta_margin_mean": 0.42743879556655884,
      "epsilon_dpo/beta_margin_std": 0.9944959878921509,
      "epsilon_dpo/loss_margin_mean": 21.062976837158203,
      "grad_norm": 14.26219654083252,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.02069736272096634,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 1.1042108616837692e-07,
      "logits/chosen": -0.4083220660686493,
      "logits/rejected": -0.20560237765312195,
      "logps/chosen": -125.48551940917969,
      "logps/ref_chosen": -82.20433044433594,
      "logps/ref_rejected": -106.65583801269531,
      "logps/rejected": -171.0,
      "loss": 1.2225,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8975530862808228,
      "rewards/margins": 0.42743879556655884,
      "rewards/rejected": -1.3249918222427368,
      "step": 477
    },
    {
      "epoch": 0.7226001511715797,
      "epsilon_dpo/beta": 0.02059619314968586,
      "epsilon_dpo/beta_margin_grad_mean": -0.42403674125671387,
      "epsilon_dpo/beta_margin_grad_std": 0.21408548951148987,
      "epsilon_dpo/beta_margin_mean": 0.3858782947063446,
      "epsilon_dpo/beta_margin_std": 1.0612176656723022,
      "epsilon_dpo/loss_margin_mean": 19.113130569458008,
      "grad_norm": 11.53897476196289,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.02064574882388115,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 1.0932609262554746e-07,
      "logits/chosen": -0.12969213724136353,
      "logits/rejected": -0.06466308236122131,
      "logps/chosen": -120.49067687988281,
      "logps/ref_chosen": -82.88982391357422,
      "logps/ref_rejected": -76.89048767089844,
      "logps/rejected": -133.60446166992188,
      "loss": 1.2838,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.7774056196212769,
      "rewards/margins": 0.3858782947063446,
      "rewards/rejected": -1.1632839441299438,
      "step": 478
    },
    {
      "epoch": 0.7241118669690099,
      "epsilon_dpo/beta": 0.02049977518618107,
      "epsilon_dpo/beta_margin_grad_mean": -0.40637218952178955,
      "epsilon_dpo/beta_margin_grad_std": 0.19583311676979065,
      "epsilon_dpo/beta_margin_mean": 0.43407750129699707,
      "epsilon_dpo/beta_margin_std": 0.9609951972961426,
      "epsilon_dpo/loss_margin_mean": 21.468713760375977,
      "grad_norm": 10.94175910949707,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.020594261586666107,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.0823503403430734e-07,
      "logits/chosen": 0.0649048462510109,
      "logits/rejected": -0.108011893928051,
      "logps/chosen": -119.17659759521484,
      "logps/ref_chosen": -82.81025695800781,
      "logps/ref_rejected": -93.0564956665039,
      "logps/rejected": -150.89154052734375,
      "loss": 1.2025,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.7476523518562317,
      "rewards/margins": 0.43407750129699707,
      "rewards/rejected": -1.181729793548584,
      "step": 479
    },
    {
      "epoch": 0.7256235827664399,
      "epsilon_dpo/beta": 0.020410537719726562,
      "epsilon_dpo/beta_margin_grad_mean": -0.38558438420295715,
      "epsilon_dpo/beta_margin_grad_std": 0.1957198679447174,
      "epsilon_dpo/beta_margin_mean": 0.5398164987564087,
      "epsilon_dpo/beta_margin_std": 0.9710685014724731,
      "epsilon_dpo/loss_margin_mean": 26.758392333984375,
      "grad_norm": 14.283851623535156,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.020498177036643028,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.0714794091391072e-07,
      "logits/chosen": -0.2883099913597107,
      "logits/rejected": 0.03229174017906189,
      "logps/chosen": -122.21525573730469,
      "logps/ref_chosen": -86.16099548339844,
      "logps/ref_rejected": -89.46211242675781,
      "logps/rejected": -152.27476501464844,
      "loss": 1.1232,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.7387363314628601,
      "rewards/margins": 0.5398164987564087,
      "rewards/rejected": -1.278552770614624,
      "step": 480
    },
    {
      "epoch": 0.72713529856387,
      "epsilon_dpo/beta": 0.020308874547481537,
      "epsilon_dpo/beta_margin_grad_mean": -0.38809558749198914,
      "epsilon_dpo/beta_margin_grad_std": 0.18458719551563263,
      "epsilon_dpo/beta_margin_mean": 0.5430381894111633,
      "epsilon_dpo/beta_margin_std": 0.9227668046951294,
      "epsilon_dpo/loss_margin_mean": 27.00404930114746,
      "grad_norm": 9.960450172424316,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.020408889278769493,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.0606484367268906e-07,
      "logits/chosen": -0.2094596028327942,
      "logits/rejected": -0.17691798508167267,
      "logps/chosen": -128.1065216064453,
      "logps/ref_chosen": -91.20343017578125,
      "logps/ref_rejected": -95.93585968017578,
      "logps/rejected": -159.84300231933594,
      "loss": 1.098,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7517346143722534,
      "rewards/margins": 0.5430382490158081,
      "rewards/rejected": -1.2947728633880615,
      "step": 481
    },
    {
      "epoch": 0.7286470143613001,
      "epsilon_dpo/beta": 0.02025860734283924,
      "epsilon_dpo/beta_margin_grad_mean": -0.4095396399497986,
      "epsilon_dpo/beta_margin_grad_std": 0.21444030106067657,
      "epsilon_dpo/beta_margin_mean": 0.4701303243637085,
      "epsilon_dpo/beta_margin_std": 1.0936204195022583,
      "epsilon_dpo/loss_margin_mean": 23.593067169189453,
      "grad_norm": 14.232986450195312,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.020307350903749466,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 1.0498577260720048e-07,
      "logits/chosen": 0.031998008489608765,
      "logits/rejected": -0.07167947292327881,
      "logps/chosen": -129.4146728515625,
      "logps/ref_chosen": -84.04693603515625,
      "logps/ref_rejected": -138.25186157226562,
      "logps/rejected": -207.21267700195312,
      "loss": 1.2254,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.9225196242332458,
      "rewards/margins": 0.4701303243637085,
      "rewards/rejected": -1.3926498889923096,
      "step": 482
    },
    {
      "epoch": 0.7301587301587301,
      "epsilon_dpo/beta": 0.020132116973400116,
      "epsilon_dpo/beta_margin_grad_mean": -0.3548187017440796,
      "epsilon_dpo/beta_margin_grad_std": 0.17624545097351074,
      "epsilon_dpo/beta_margin_mean": 0.688190758228302,
      "epsilon_dpo/beta_margin_std": 0.8882216215133667,
      "epsilon_dpo/loss_margin_mean": 34.3974609375,
      "grad_norm": 10.093025207519531,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.020256709307432175,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 1.0391075790138232e-07,
      "logits/chosen": -0.36327123641967773,
      "logits/rejected": -0.43129682540893555,
      "logps/chosen": -111.67202758789062,
      "logps/ref_chosen": -77.54084777832031,
      "logps/ref_rejected": -106.26319122314453,
      "logps/rejected": -174.7918243408203,
      "loss": 0.9805,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.6897982358932495,
      "rewards/margins": 0.6881906986236572,
      "rewards/rejected": -1.3779890537261963,
      "step": 483
    },
    {
      "epoch": 0.7316704459561603,
      "epsilon_dpo/beta": 0.02001648023724556,
      "epsilon_dpo/beta_margin_grad_mean": -0.39846718311309814,
      "epsilon_dpo/beta_margin_grad_std": 0.15919949114322662,
      "epsilon_dpo/beta_margin_mean": 0.46173810958862305,
      "epsilon_dpo/beta_margin_std": 0.7505614161491394,
      "epsilon_dpo/loss_margin_mean": 23.269933700561523,
      "grad_norm": 12.176405906677246,
      "kl/avg_steps": 0.578125,
      "kl/beta": 0.020130891352891922,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 1.0283982962570681e-07,
      "logits/chosen": -0.49070245027542114,
      "logits/rejected": -0.2863999307155609,
      "logps/chosen": -113.4175033569336,
      "logps/ref_chosen": -79.68392181396484,
      "logps/ref_rejected": -97.50027465820312,
      "logps/rejected": -154.5037841796875,
      "loss": 1.1034,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.6778692603111267,
      "rewards/margins": 0.46173810958862305,
      "rewards/rejected": -1.1396074295043945,
      "step": 484
    },
    {
      "epoch": 0.7331821617535903,
      "epsilon_dpo/beta": 0.019917093217372894,
      "epsilon_dpo/beta_margin_grad_mean": -0.3898495137691498,
      "epsilon_dpo/beta_margin_grad_std": 0.170601487159729,
      "epsilon_dpo/beta_margin_mean": 0.5218725800514221,
      "epsilon_dpo/beta_margin_std": 0.8386232256889343,
      "epsilon_dpo/loss_margin_mean": 26.436914443969727,
      "grad_norm": 10.812078475952148,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.020015178248286247,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.0177301773633992e-07,
      "logits/chosen": -0.07163044810295105,
      "logits/rejected": -0.2883603274822235,
      "logps/chosen": -121.81181335449219,
      "logps/ref_chosen": -83.94019317626953,
      "logps/ref_rejected": -98.3631362915039,
      "logps/rejected": -162.67166137695312,
      "loss": 1.0839,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.7567333579063416,
      "rewards/margins": 0.5218726396560669,
      "rewards/rejected": -1.2786059379577637,
      "step": 485
    },
    {
      "epoch": 0.7346938775510204,
      "epsilon_dpo/beta": 0.019861575216054916,
      "epsilon_dpo/beta_margin_grad_mean": -0.4068491756916046,
      "epsilon_dpo/beta_margin_grad_std": 0.2024412304162979,
      "epsilon_dpo/beta_margin_mean": 0.47068801522254944,
      "epsilon_dpo/beta_margin_std": 1.022111177444458,
      "epsilon_dpo/loss_margin_mean": 24.056474685668945,
      "grad_norm": 9.771635055541992,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.01991560123860836,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 1.007103520743035e-07,
      "logits/chosen": 0.19127486646175385,
      "logits/rejected": -0.22449460625648499,
      "logps/chosen": -122.89358520507812,
      "logps/ref_chosen": -78.64771270751953,
      "logps/ref_rejected": -116.00846862792969,
      "logps/rejected": -184.31082153320312,
      "loss": 1.1946,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8815645575523376,
      "rewards/margins": 0.47068801522254944,
      "rewards/rejected": -1.3522526025772095,
      "step": 486
    },
    {
      "epoch": 0.7362055933484505,
      "epsilon_dpo/beta": 0.019787250086665154,
      "epsilon_dpo/beta_margin_grad_mean": -0.40424415469169617,
      "epsilon_dpo/beta_margin_grad_std": 0.17017613351345062,
      "epsilon_dpo/beta_margin_mean": 0.4518623650074005,
      "epsilon_dpo/beta_margin_std": 0.8180162906646729,
      "epsilon_dpo/loss_margin_mean": 23.10869789123535,
      "grad_norm": 9.558370590209961,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.01985974609851837,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 9.965186236464046e-08,
      "logits/chosen": -0.13608111441135406,
      "logits/rejected": -0.3792567551136017,
      "logps/chosen": -131.393798828125,
      "logps/ref_chosen": -85.29930114746094,
      "logps/ref_rejected": -108.99090576171875,
      "logps/rejected": -178.19412231445312,
      "loss": 1.1325,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.9149771928787231,
      "rewards/margins": 0.4518623650074005,
      "rewards/rejected": -1.3668395280838013,
      "step": 487
    },
    {
      "epoch": 0.7377173091458806,
      "epsilon_dpo/beta": 0.01972569152712822,
      "epsilon_dpo/beta_margin_grad_mean": -0.3908286988735199,
      "epsilon_dpo/beta_margin_grad_std": 0.18784664571285248,
      "epsilon_dpo/beta_margin_mean": 0.5446649789810181,
      "epsilon_dpo/beta_margin_std": 0.9466550946235657,
      "epsilon_dpo/loss_margin_mean": 27.929880142211914,
      "grad_norm": 11.954972267150879,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.019785549491643906,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 9.859757821558337e-08,
      "logits/chosen": -0.36887866258621216,
      "logits/rejected": -0.5798311233520508,
      "logps/chosen": -125.43516540527344,
      "logps/ref_chosen": -87.57951354980469,
      "logps/ref_rejected": -112.03506469726562,
      "logps/rejected": -177.8206024169922,
      "loss": 1.1054,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.7492825984954834,
      "rewards/margins": 0.5446649789810181,
      "rewards/rejected": -1.2939475774765015,
      "step": 488
    },
    {
      "epoch": 0.7392290249433107,
      "epsilon_dpo/beta": 0.01966424286365509,
      "epsilon_dpo/beta_margin_grad_mean": -0.4462127089500427,
      "epsilon_dpo/beta_margin_grad_std": 0.1676499992609024,
      "epsilon_dpo/beta_margin_mean": 0.25041741132736206,
      "epsilon_dpo/beta_margin_std": 0.7841642498970032,
      "epsilon_dpo/loss_margin_mean": 12.99974536895752,
      "grad_norm": 10.487590789794922,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.019723912701010704,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 9.754752911772615e-08,
      "logits/chosen": -0.23175035417079926,
      "logits/rejected": -0.553443968296051,
      "logps/chosen": -133.7505645751953,
      "logps/ref_chosen": -91.55049896240234,
      "logps/ref_rejected": -108.33108520507812,
      "logps/rejected": -163.53089904785156,
      "loss": 1.2916,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8317919373512268,
      "rewards/margins": 0.2504173517227173,
      "rewards/rejected": -1.0822093486785889,
      "step": 489
    },
    {
      "epoch": 0.7407407407407407,
      "epsilon_dpo/beta": 0.01957840286195278,
      "epsilon_dpo/beta_margin_grad_mean": -0.3897029459476471,
      "epsilon_dpo/beta_margin_grad_std": 0.19855265319347382,
      "epsilon_dpo/beta_margin_mean": 0.5350128412246704,
      "epsilon_dpo/beta_margin_std": 0.9706516265869141,
      "epsilon_dpo/loss_margin_mean": 27.660192489624023,
      "grad_norm": 9.340205192565918,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.019662467762827873,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 9.650174444319956e-08,
      "logits/chosen": -0.056915752589702606,
      "logits/rejected": -0.16137561202049255,
      "logps/chosen": -122.04594421386719,
      "logps/ref_chosen": -83.35502624511719,
      "logps/ref_rejected": -97.42626953125,
      "logps/rejected": -163.77737426757812,
      "loss": 1.1271,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.7606915235519409,
      "rewards/margins": 0.5350128412246704,
      "rewards/rejected": -1.2957043647766113,
      "step": 490
    },
    {
      "epoch": 0.7422524565381708,
      "epsilon_dpo/beta": 0.019499236717820168,
      "epsilon_dpo/beta_margin_grad_mean": -0.399957537651062,
      "epsilon_dpo/beta_margin_grad_std": 0.17069052159786224,
      "epsilon_dpo/beta_margin_mean": 0.4838140904903412,
      "epsilon_dpo/beta_margin_std": 0.8225111961364746,
      "epsilon_dpo/loss_margin_mean": 25.07085609436035,
      "grad_norm": 10.31070327758789,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.019576817750930786,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 9.546025344484868e-08,
      "logits/chosen": 0.1169687956571579,
      "logits/rejected": -0.1125451996922493,
      "logps/chosen": -119.66255187988281,
      "logps/ref_chosen": -78.405517578125,
      "logps/ref_rejected": -98.35725402832031,
      "logps/rejected": -164.68515014648438,
      "loss": 1.1083,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.8066112399101257,
      "rewards/margins": 0.4838140606880188,
      "rewards/rejected": -1.2904253005981445,
      "step": 491
    },
    {
      "epoch": 0.7437641723356009,
      "epsilon_dpo/beta": 0.019438622519373894,
      "epsilon_dpo/beta_margin_grad_mean": -0.42248454689979553,
      "epsilon_dpo/beta_margin_grad_std": 0.1789523810148239,
      "epsilon_dpo/beta_margin_mean": 0.3718709647655487,
      "epsilon_dpo/beta_margin_std": 0.8428933024406433,
      "epsilon_dpo/loss_margin_mean": 19.4351749420166,
      "grad_norm": 9.827570915222168,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.01949760876595974,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 9.442308525541589e-08,
      "logits/chosen": -0.11652935296297073,
      "logits/rejected": -0.3884758949279785,
      "logps/chosen": -143.1727294921875,
      "logps/ref_chosen": -91.67491149902344,
      "logps/ref_rejected": -114.05909729003906,
      "logps/rejected": -184.99209594726562,
      "loss": 1.2084,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -1.0039048194885254,
      "rewards/margins": 0.3718709349632263,
      "rewards/rejected": -1.375775694847107,
      "step": 492
    },
    {
      "epoch": 0.745275888133031,
      "epsilon_dpo/beta": 0.019341617822647095,
      "epsilon_dpo/beta_margin_grad_mean": -0.3907162845134735,
      "epsilon_dpo/beta_margin_grad_std": 0.18976947665214539,
      "epsilon_dpo/beta_margin_mean": 0.5252335667610168,
      "epsilon_dpo/beta_margin_std": 0.924012303352356,
      "epsilon_dpo/loss_margin_mean": 27.459110260009766,
      "grad_norm": 10.20104694366455,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.019436869770288467,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 9.339026888672468e-08,
      "logits/chosen": 0.04558924585580826,
      "logits/rejected": -0.18673455715179443,
      "logps/chosen": -119.65789794921875,
      "logps/ref_chosen": -80.52119445800781,
      "logps/ref_rejected": -105.2249755859375,
      "logps/rejected": -171.82077026367188,
      "loss": 1.1155,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.7599451541900635,
      "rewards/margins": 0.5252335071563721,
      "rewards/rejected": -1.2851786613464355,
      "step": 493
    },
    {
      "epoch": 0.7467876039304611,
      "epsilon_dpo/beta": 0.019257480278611183,
      "epsilon_dpo/beta_margin_grad_mean": -0.4015882611274719,
      "epsilon_dpo/beta_margin_grad_std": 0.18990257382392883,
      "epsilon_dpo/beta_margin_mean": 0.477631151676178,
      "epsilon_dpo/beta_margin_std": 0.9436517953872681,
      "epsilon_dpo/loss_margin_mean": 25.109621047973633,
      "grad_norm": 9.032830238342285,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.019340168684720993,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 9.236183322886945e-08,
      "logits/chosen": -0.42797666788101196,
      "logits/rejected": -0.1930256187915802,
      "logps/chosen": -141.51026916503906,
      "logps/ref_chosen": -99.7325439453125,
      "logps/ref_rejected": -124.01393127441406,
      "logps/rejected": -190.90127563476562,
      "loss": 1.1578,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.8071942329406738,
      "rewards/margins": 0.47763118147850037,
      "rewards/rejected": -1.284825325012207,
      "step": 494
    },
    {
      "epoch": 0.7482993197278912,
      "epsilon_dpo/beta": 0.019185630604624748,
      "epsilon_dpo/beta_margin_grad_mean": -0.4156641960144043,
      "epsilon_dpo/beta_margin_grad_std": 0.1824708878993988,
      "epsilon_dpo/beta_margin_mean": 0.3921421468257904,
      "epsilon_dpo/beta_margin_std": 0.8667027950286865,
      "epsilon_dpo/loss_margin_mean": 20.752164840698242,
      "grad_norm": 16.976558685302734,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.019255923107266426,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 9.133780704940594e-08,
      "logits/chosen": 0.06622087955474854,
      "logits/rejected": -0.3368031978607178,
      "logps/chosen": -115.75942993164062,
      "logps/ref_chosen": -73.57146453857422,
      "logps/ref_rejected": -95.85881042480469,
      "logps/rejected": -158.7989501953125,
      "loss": 1.2008,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.810094952583313,
      "rewards/margins": 0.392142117023468,
      "rewards/rejected": -1.2022371292114258,
      "step": 495
    },
    {
      "epoch": 0.7498110355253212,
      "epsilon_dpo/beta": 0.019101964309811592,
      "epsilon_dpo/beta_margin_grad_mean": -0.3908800482749939,
      "epsilon_dpo/beta_margin_grad_std": 0.17948482930660248,
      "epsilon_dpo/beta_margin_mean": 0.5241735577583313,
      "epsilon_dpo/beta_margin_std": 0.8593854308128357,
      "epsilon_dpo/loss_margin_mean": 27.73284149169922,
      "grad_norm": 10.025064468383789,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.019183984026312828,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 9.031821899254797e-08,
      "logits/chosen": -0.03599054738879204,
      "logits/rejected": -0.2996342182159424,
      "logps/chosen": -119.87254333496094,
      "logps/ref_chosen": -77.81381225585938,
      "logps/ref_rejected": -127.28571319580078,
      "logps/rejected": -197.07728576660156,
      "loss": 1.0917,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.8057615756988525,
      "rewards/margins": 0.5241735577583313,
      "rewards/rejected": -1.329935073852539,
      "step": 496
    },
    {
      "epoch": 0.7513227513227513,
      "epsilon_dpo/beta": 0.01898891106247902,
      "epsilon_dpo/beta_margin_grad_mean": -0.37980324029922485,
      "epsilon_dpo/beta_margin_grad_std": 0.1680688112974167,
      "epsilon_dpo/beta_margin_mean": 0.5702401399612427,
      "epsilon_dpo/beta_margin_std": 0.8251023292541504,
      "epsilon_dpo/loss_margin_mean": 30.249765396118164,
      "grad_norm": 10.120041847229004,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.01910042017698288,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 8.930309757836516e-08,
      "logits/chosen": -0.0725216418504715,
      "logits/rejected": -0.25850170850753784,
      "logps/chosen": -143.61422729492188,
      "logps/ref_chosen": -96.18318176269531,
      "logps/ref_rejected": -113.01858520507812,
      "logps/rejected": -190.69940185546875,
      "loss": 1.0426,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.9025899767875671,
      "rewards/margins": 0.5702401399612427,
      "rewards/rejected": -1.472830057144165,
      "step": 497
    },
    {
      "epoch": 0.7528344671201814,
      "epsilon_dpo/beta": 0.018912434577941895,
      "epsilon_dpo/beta_margin_grad_mean": -0.3832828998565674,
      "epsilon_dpo/beta_margin_grad_std": 0.16757036745548248,
      "epsilon_dpo/beta_margin_mean": 0.5597019195556641,
      "epsilon_dpo/beta_margin_std": 0.8137642741203308,
      "epsilon_dpo/loss_margin_mean": 29.86823272705078,
      "grad_norm": 9.708072662353516,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.018987679854035378,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 8.829247120198563e-08,
      "logits/chosen": -0.2942112386226654,
      "logits/rejected": -0.17943748831748962,
      "logps/chosen": -126.88603210449219,
      "logps/ref_chosen": -87.08412170410156,
      "logps/ref_rejected": -97.77659606933594,
      "logps/rejected": -167.44674682617188,
      "loss": 1.0466,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7544086575508118,
      "rewards/margins": 0.5597019195556641,
      "rewards/rejected": -1.314110517501831,
      "step": 498
    },
    {
      "epoch": 0.7543461829176115,
      "epsilon_dpo/beta": 0.018847733736038208,
      "epsilon_dpo/beta_margin_grad_mean": -0.4056819677352905,
      "epsilon_dpo/beta_margin_grad_std": 0.2031472772359848,
      "epsilon_dpo/beta_margin_mean": 0.44904473423957825,
      "epsilon_dpo/beta_margin_std": 0.961423933506012,
      "epsilon_dpo/loss_margin_mean": 24.202713012695312,
      "grad_norm": 10.174310684204102,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.018910855054855347,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 8.728636813280163e-08,
      "logits/chosen": -0.3366241455078125,
      "logits/rejected": -0.31855422258377075,
      "logps/chosen": -132.77659606933594,
      "logps/ref_chosen": -84.11980438232422,
      "logps/ref_rejected": -111.52287292480469,
      "logps/rejected": -184.3823699951172,
      "loss": 1.1947,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.920242428779602,
      "rewards/margins": 0.44904473423957825,
      "rewards/rejected": -1.3692872524261475,
      "step": 499
    },
    {
      "epoch": 0.7558578987150416,
      "epsilon_dpo/beta": 0.01876549795269966,
      "epsilon_dpo/beta_margin_grad_mean": -0.3985728323459625,
      "epsilon_dpo/beta_margin_grad_std": 0.17415080964565277,
      "epsilon_dpo/beta_margin_mean": 0.48147937655448914,
      "epsilon_dpo/beta_margin_std": 0.8304465413093567,
      "epsilon_dpo/loss_margin_mean": 25.937110900878906,
      "grad_norm": 10.142269134521484,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.018846072256565094,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 8.628481651367875e-08,
      "logits/chosen": -0.3174499571323395,
      "logits/rejected": -0.3522814214229584,
      "logps/chosen": -139.42095947265625,
      "logps/ref_chosen": -93.23562622070312,
      "logps/ref_rejected": -96.9388427734375,
      "logps/rejected": -169.061279296875,
      "loss": 1.1147,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.8695238828659058,
      "rewards/margins": 0.48147934675216675,
      "rewards/rejected": -1.3510031700134277,
      "step": 500
    },
    {
      "epoch": 0.7558578987150416,
      "eval_epsilon_dpo/beta": 0.018684910610318184,
      "eval_epsilon_dpo/beta_margin_grad_mean": -0.39601635932922363,
      "eval_epsilon_dpo/beta_margin_grad_std": 0.1754881739616394,
      "eval_epsilon_dpo/beta_margin_mean": 0.4919276833534241,
      "eval_epsilon_dpo/beta_margin_std": 0.8456751108169556,
      "eval_epsilon_dpo/loss_margin_mean": 26.623165130615234,
      "eval_kl/n_epsilon_steps": 0.28433099389076233,
      "eval_kl/p_epsilon_steps": 0.7156690359115601,
      "eval_logits/chosen": -0.19369195401668549,
      "eval_logits/rejected": -0.3317781388759613,
      "eval_logps/chosen": -130.41204833984375,
      "eval_logps/ref_chosen": -87.42715454101562,
      "eval_logps/ref_rejected": -104.23548889160156,
      "eval_logps/rejected": -173.8435516357422,
      "eval_loss": 0.5588386058807373,
      "eval_rewards/accuracies": 0.7205105423927307,
      "eval_rewards/chosen": -0.8058603405952454,
      "eval_rewards/margins": 0.4919276833534241,
      "eval_rewards/rejected": -1.2977880239486694,
      "eval_runtime": 47.5848,
      "eval_samples_per_second": 48.398,
      "eval_steps_per_second": 1.513,
      "step": 500
    },
    {
      "epoch": 0.7573696145124716,
      "epsilon_dpo/beta": 0.018677890300750732,
      "epsilon_dpo/beta_margin_grad_mean": -0.4091666638851166,
      "epsilon_dpo/beta_margin_grad_std": 0.14577420055866241,
      "epsilon_dpo/beta_margin_mean": 0.4170517921447754,
      "epsilon_dpo/beta_margin_std": 0.6771373748779297,
      "epsilon_dpo/loss_margin_mean": 22.541135787963867,
      "grad_norm": 8.958548545837402,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.018763979896903038,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 8.528784436016878e-08,
      "logits/chosen": -0.31239911913871765,
      "logits/rejected": -0.3793390989303589,
      "logps/chosen": -135.99786376953125,
      "logps/ref_chosen": -88.00139617919922,
      "logps/ref_rejected": -91.72904968261719,
      "logps/rejected": -162.26666259765625,
      "loss": 1.1154,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.8980482220649719,
      "rewards/margins": 0.4170517921447754,
      "rewards/rejected": -1.3150999546051025,
      "step": 501
    },
    {
      "epoch": 0.7588813303099018,
      "epsilon_dpo/beta": 0.01859947293996811,
      "epsilon_dpo/beta_margin_grad_mean": -0.4219035804271698,
      "epsilon_dpo/beta_margin_grad_std": 0.15145091712474823,
      "epsilon_dpo/beta_margin_mean": 0.35392406582832336,
      "epsilon_dpo/beta_margin_std": 0.7095279097557068,
      "epsilon_dpo/loss_margin_mean": 19.268327713012695,
      "grad_norm": 11.017980575561523,
      "kl/avg_steps": 0.421875,
      "kl/beta": 0.018676433712244034,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 8.4295479559726e-08,
      "logits/chosen": -0.4218502938747406,
      "logits/rejected": -0.41790133714675903,
      "logps/chosen": -133.26429748535156,
      "logps/ref_chosen": -89.93152618408203,
      "logps/ref_rejected": -114.26683044433594,
      "logps/rejected": -176.867919921875,
      "loss": 1.1771,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.8082338571548462,
      "rewards/margins": 0.353924036026001,
      "rewards/rejected": -1.1621580123901367,
      "step": 502
    },
    {
      "epoch": 0.7603930461073318,
      "epsilon_dpo/beta": 0.018501022830605507,
      "epsilon_dpo/beta_margin_grad_mean": -0.38354945182800293,
      "epsilon_dpo/beta_margin_grad_std": 0.17253164947032928,
      "epsilon_dpo/beta_margin_mean": 0.5339672565460205,
      "epsilon_dpo/beta_margin_std": 0.8142776489257812,
      "epsilon_dpo/loss_margin_mean": 29.133380889892578,
      "grad_norm": 9.433905601501465,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.018597973510622978,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 8.330774987092712e-08,
      "logits/chosen": -0.2990953326225281,
      "logits/rejected": 0.10869180411100388,
      "logps/chosen": -115.01774597167969,
      "logps/ref_chosen": -72.67813110351562,
      "logps/ref_rejected": -80.22515106201172,
      "logps/rejected": -151.69813537597656,
      "loss": 1.0705,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.786790132522583,
      "rewards/margins": 0.5339672565460205,
      "rewards/rejected": -1.3207573890686035,
      "step": 503
    },
    {
      "epoch": 0.7619047619047619,
      "epsilon_dpo/beta": 0.018385907635092735,
      "epsilon_dpo/beta_margin_grad_mean": -0.3638831079006195,
      "epsilon_dpo/beta_margin_grad_std": 0.15275923907756805,
      "epsilon_dpo/beta_margin_mean": 0.63335782289505,
      "epsilon_dpo/beta_margin_std": 0.7385804057121277,
      "epsilon_dpo/loss_margin_mean": 34.64598846435547,
      "grad_norm": 9.538182258605957,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.018499692901968956,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 8.232468292269479e-08,
      "logits/chosen": -0.1938086897134781,
      "logits/rejected": -0.06743744015693665,
      "logps/chosen": -132.69613647460938,
      "logps/ref_chosen": -92.39486694335938,
      "logps/ref_rejected": -106.31135559082031,
      "logps/rejected": -181.25860595703125,
      "loss": 0.9686,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.7422640919685364,
      "rewards/margins": 0.63335782289505,
      "rewards/rejected": -1.3756219148635864,
      "step": 504
    },
    {
      "epoch": 0.763416477702192,
      "epsilon_dpo/beta": 0.018323423340916634,
      "epsilon_dpo/beta_margin_grad_mean": -0.42362314462661743,
      "epsilon_dpo/beta_margin_grad_std": 0.19126078486442566,
      "epsilon_dpo/beta_margin_mean": 0.3635737895965576,
      "epsilon_dpo/beta_margin_std": 0.9163693189620972,
      "epsilon_dpo/loss_margin_mean": 20.19578742980957,
      "grad_norm": 12.260894775390625,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.018384788185358047,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 8.134630621352483e-08,
      "logits/chosen": -0.25550952553749084,
      "logits/rejected": -0.604439377784729,
      "logps/chosen": -130.39122009277344,
      "logps/ref_chosen": -84.77385711669922,
      "logps/ref_rejected": -103.21849060058594,
      "logps/rejected": -169.03164672851562,
      "loss": 1.2431,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8400043249130249,
      "rewards/margins": 0.3635737895965576,
      "rewards/rejected": -1.2035781145095825,
      "step": 505
    },
    {
      "epoch": 0.764928193499622,
      "epsilon_dpo/beta": 0.0182606503367424,
      "epsilon_dpo/beta_margin_grad_mean": -0.4268198311328888,
      "epsilon_dpo/beta_margin_grad_std": 0.19903631508350372,
      "epsilon_dpo/beta_margin_mean": 0.33908024430274963,
      "epsilon_dpo/beta_margin_std": 0.9598965048789978,
      "epsilon_dpo/loss_margin_mean": 18.941492080688477,
      "grad_norm": 13.514118194580078,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.0183218065649271,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 8.037264711071698e-08,
      "logits/chosen": -0.545729398727417,
      "logits/rejected": -0.5596314072608948,
      "logps/chosen": -143.47811889648438,
      "logps/ref_chosen": -95.78478240966797,
      "logps/ref_rejected": -108.5623779296875,
      "logps/rejected": -175.19720458984375,
      "loss": 1.2821,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.8736474514007568,
      "rewards/margins": 0.33908021450042725,
      "rewards/rejected": -1.2127277851104736,
      "step": 506
    },
    {
      "epoch": 0.7664399092970522,
      "epsilon_dpo/beta": 0.018180975690484047,
      "epsilon_dpo/beta_margin_grad_mean": -0.40145301818847656,
      "epsilon_dpo/beta_margin_grad_std": 0.17579539120197296,
      "epsilon_dpo/beta_margin_mean": 0.47464361786842346,
      "epsilon_dpo/beta_margin_std": 0.8506050109863281,
      "epsilon_dpo/loss_margin_mean": 26.402099609375,
      "grad_norm": 11.599726676940918,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.018259041011333466,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 7.940373284960933e-08,
      "logits/chosen": -0.22546380758285522,
      "logits/rejected": -0.4367539882659912,
      "logps/chosen": -143.24029541015625,
      "logps/ref_chosen": -94.66133880615234,
      "logps/ref_rejected": -119.80830383300781,
      "logps/rejected": -194.78936767578125,
      "loss": 1.1256,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.8854374885559082,
      "rewards/margins": 0.47464364767074585,
      "rewards/rejected": -1.3600811958312988,
      "step": 507
    },
    {
      "epoch": 0.7679516250944822,
      "epsilon_dpo/beta": 0.01809609867632389,
      "epsilon_dpo/beta_margin_grad_mean": -0.39532774686813354,
      "epsilon_dpo/beta_margin_grad_std": 0.17735745012760162,
      "epsilon_dpo/beta_margin_mean": 0.5011078119277954,
      "epsilon_dpo/beta_margin_std": 0.8593124151229858,
      "epsilon_dpo/loss_margin_mean": 27.98598861694336,
      "grad_norm": 9.803098678588867,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.018179506063461304,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 7.843959053281663e-08,
      "logits/chosen": -0.011693110689520836,
      "logits/rejected": -0.5169368982315063,
      "logps/chosen": -128.99636840820312,
      "logps/ref_chosen": -83.423583984375,
      "logps/ref_rejected": -127.66143798828125,
      "logps/rejected": -201.22021484375,
      "loss": 1.1084,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.8275296688079834,
      "rewards/margins": 0.5011078119277954,
      "rewards/rejected": -1.3286374807357788,
      "step": 508
    },
    {
      "epoch": 0.7694633408919124,
      "epsilon_dpo/beta": 0.01802298054099083,
      "epsilon_dpo/beta_margin_grad_mean": -0.3995635509490967,
      "epsilon_dpo/beta_margin_grad_std": 0.1715720146894455,
      "epsilon_dpo/beta_margin_mean": 0.4618225693702698,
      "epsilon_dpo/beta_margin_std": 0.780823290348053,
      "epsilon_dpo/loss_margin_mean": 25.937952041625977,
      "grad_norm": 11.328001022338867,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.018094688653945923,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 7.748024712947204e-08,
      "logits/chosen": -0.03571470454335213,
      "logits/rejected": -0.03329726681113243,
      "logps/chosen": -132.56443786621094,
      "logps/ref_chosen": -89.12423706054688,
      "logps/ref_rejected": -98.25785064697266,
      "logps/rejected": -167.63600158691406,
      "loss": 1.1163,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.7848267555236816,
      "rewards/margins": 0.4618225693702698,
      "rewards/rejected": -1.2466492652893066,
      "step": 509
    },
    {
      "epoch": 0.7709750566893424,
      "epsilon_dpo/beta": 0.017944425344467163,
      "epsilon_dpo/beta_margin_grad_mean": -0.37997183203697205,
      "epsilon_dpo/beta_margin_grad_std": 0.18886786699295044,
      "epsilon_dpo/beta_margin_mean": 0.5730319619178772,
      "epsilon_dpo/beta_margin_std": 0.9132055044174194,
      "epsilon_dpo/loss_margin_mean": 32.2667350769043,
      "grad_norm": 11.396764755249023,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.01802147552371025,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 7.652572947447272e-08,
      "logits/chosen": -0.37277430295944214,
      "logits/rejected": -0.651774525642395,
      "logps/chosen": -112.097412109375,
      "logps/ref_chosen": -71.99481201171875,
      "logps/ref_rejected": -115.5155029296875,
      "logps/rejected": -187.88485717773438,
      "loss": 1.0765,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7218503952026367,
      "rewards/margins": 0.573032021522522,
      "rewards/rejected": -1.2948825359344482,
      "step": 510
    },
    {
      "epoch": 0.7724867724867724,
      "epsilon_dpo/beta": 0.0178606528788805,
      "epsilon_dpo/beta_margin_grad_mean": -0.3695341646671295,
      "epsilon_dpo/beta_margin_grad_std": 0.1857190579175949,
      "epsilon_dpo/beta_margin_mean": 0.6305907368659973,
      "epsilon_dpo/beta_margin_std": 0.9014586806297302,
      "epsilon_dpo/loss_margin_mean": 35.62781524658203,
      "grad_norm": 10.467903137207031,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.017942974343895912,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 7.557606426772961e-08,
      "logits/chosen": -0.44700726866722107,
      "logits/rejected": -0.35447224974632263,
      "logps/chosen": -122.42686462402344,
      "logps/ref_chosen": -79.94530487060547,
      "logps/ref_rejected": -112.25093078613281,
      "logps/rejected": -190.36029052734375,
      "loss": 1.0288,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.7609899044036865,
      "rewards/margins": 0.6305907964706421,
      "rewards/rejected": -1.3915807008743286,
      "step": 511
    },
    {
      "epoch": 0.7739984882842026,
      "epsilon_dpo/beta": 0.01781081035733223,
      "epsilon_dpo/beta_margin_grad_mean": -0.4195144772529602,
      "epsilon_dpo/beta_margin_grad_std": 0.1699885129928589,
      "epsilon_dpo/beta_margin_mean": 0.3944665491580963,
      "epsilon_dpo/beta_margin_std": 0.8180778622627258,
      "epsilon_dpo/loss_margin_mean": 22.461170196533203,
      "grad_norm": 10.9349946975708,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.017859259620308876,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 7.463127807341966e-08,
      "logits/chosen": -0.30398204922676086,
      "logits/rejected": -0.6556848287582397,
      "logps/chosen": -131.9627227783203,
      "logps/ref_chosen": -86.12109375,
      "logps/ref_rejected": -89.51139068603516,
      "logps/rejected": -157.814208984375,
      "loss": 1.1781,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8200229406356812,
      "rewards/margins": 0.3944665491580963,
      "rewards/rejected": -1.214489459991455,
      "step": 512
    },
    {
      "epoch": 0.7755102040816326,
      "epsilon_dpo/beta": 0.01772189699113369,
      "epsilon_dpo/beta_margin_grad_mean": -0.39932936429977417,
      "epsilon_dpo/beta_margin_grad_std": 0.1577375829219818,
      "epsilon_dpo/beta_margin_mean": 0.47727683186531067,
      "epsilon_dpo/beta_margin_std": 0.7684574723243713,
      "epsilon_dpo/loss_margin_mean": 27.176280975341797,
      "grad_norm": 8.833486557006836,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.017809171229600906,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 7.369139731924401e-08,
      "logits/chosen": -0.011374262161552906,
      "logits/rejected": -0.04961850494146347,
      "logps/chosen": -115.10137939453125,
      "logps/ref_chosen": -74.36705017089844,
      "logps/ref_rejected": -93.00857543945312,
      "logps/rejected": -160.919189453125,
      "loss": 1.0937,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.7242238521575928,
      "rewards/margins": 0.4772768020629883,
      "rewards/rejected": -1.201500654220581,
      "step": 513
    },
    {
      "epoch": 0.7770219198790628,
      "epsilon_dpo/beta": 0.01762819103896618,
      "epsilon_dpo/beta_margin_grad_mean": -0.37859275937080383,
      "epsilon_dpo/beta_margin_grad_std": 0.16942396759986877,
      "epsilon_dpo/beta_margin_mean": 0.5625271201133728,
      "epsilon_dpo/beta_margin_std": 0.8023771047592163,
      "epsilon_dpo/loss_margin_mean": 32.183902740478516,
      "grad_norm": 11.468728065490723,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.017720568925142288,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 7.275644829568747e-08,
      "logits/chosen": -0.3658566474914551,
      "logits/rejected": -0.33812397718429565,
      "logps/chosen": -136.5714874267578,
      "logps/ref_chosen": -91.89642333984375,
      "logps/ref_rejected": -109.63691711425781,
      "logps/rejected": -186.49588012695312,
      "loss": 1.044,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.790207028388977,
      "rewards/margins": 0.562527060508728,
      "rewards/rejected": -1.352734088897705,
      "step": 514
    },
    {
      "epoch": 0.7785336356764928,
      "epsilon_dpo/beta": 0.017551563680171967,
      "epsilon_dpo/beta_margin_grad_mean": -0.40302208065986633,
      "epsilon_dpo/beta_margin_grad_std": 0.17570269107818604,
      "epsilon_dpo/beta_margin_mean": 0.45750582218170166,
      "epsilon_dpo/beta_margin_std": 0.8222211599349976,
      "epsilon_dpo/loss_margin_mean": 26.378746032714844,
      "grad_norm": 10.036144256591797,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.01762692630290985,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 7.182645715528435e-08,
      "logits/chosen": -0.2181454449892044,
      "logits/rejected": -0.17116093635559082,
      "logps/chosen": -128.15139770507812,
      "logps/ref_chosen": -80.23466491699219,
      "logps/ref_rejected": -113.73807525634766,
      "logps/rejected": -188.03355407714844,
      "loss": 1.1322,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.8435226678848267,
      "rewards/margins": 0.45750582218170166,
      "rewards/rejected": -1.3010284900665283,
      "step": 515
    },
    {
      "epoch": 0.780045351473923,
      "epsilon_dpo/beta": 0.017458654940128326,
      "epsilon_dpo/beta_margin_grad_mean": -0.4042704999446869,
      "epsilon_dpo/beta_margin_grad_std": 0.16673848032951355,
      "epsilon_dpo/beta_margin_mean": 0.4266871213912964,
      "epsilon_dpo/beta_margin_std": 0.7753763794898987,
      "epsilon_dpo/loss_margin_mean": 24.717382431030273,
      "grad_norm": 10.534723281860352,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.017550144344568253,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 7.090144991188568e-08,
      "logits/chosen": -0.0516197606921196,
      "logits/rejected": -0.16222411394119263,
      "logps/chosen": -121.34104919433594,
      "logps/ref_chosen": -82.21078491210938,
      "logps/ref_rejected": -100.97323608398438,
      "logps/rejected": -164.82089233398438,
      "loss": 1.1413,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6861926317214966,
      "rewards/margins": 0.426687091588974,
      "rewards/rejected": -1.112879753112793,
      "step": 516
    },
    {
      "epoch": 0.781557067271353,
      "epsilon_dpo/beta": 0.017393674701452255,
      "epsilon_dpo/beta_margin_grad_mean": -0.43510809540748596,
      "epsilon_dpo/beta_margin_grad_std": 0.1779409646987915,
      "epsilon_dpo/beta_margin_mean": 0.32335782051086426,
      "epsilon_dpo/beta_margin_std": 0.8659263253211975,
      "epsilon_dpo/loss_margin_mean": 18.899690628051758,
      "grad_norm": 10.160877227783203,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.017457401379942894,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 6.998145243993284e-08,
      "logits/chosen": -0.191817507147789,
      "logits/rejected": -0.15607379376888275,
      "logps/chosen": -131.53024291992188,
      "logps/ref_chosen": -83.18726348876953,
      "logps/ref_rejected": -87.62814331054688,
      "logps/rejected": -154.87081909179688,
      "loss": 1.2549,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.8429970741271973,
      "rewards/margins": 0.32335782051086426,
      "rewards/rejected": -1.1663548946380615,
      "step": 517
    },
    {
      "epoch": 0.783068783068783,
      "epsilon_dpo/beta": 0.017355870455503464,
      "epsilon_dpo/beta_margin_grad_mean": -0.41893237829208374,
      "epsilon_dpo/beta_margin_grad_std": 0.17085179686546326,
      "epsilon_dpo/beta_margin_mean": 0.38197755813598633,
      "epsilon_dpo/beta_margin_std": 0.8127449750900269,
      "epsilon_dpo/loss_margin_mean": 22.35114097595215,
      "grad_norm": 9.441346168518066,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.017392180860042572,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 6.906649047373245e-08,
      "logits/chosen": -0.2893047034740448,
      "logits/rejected": -0.5340129733085632,
      "logps/chosen": -120.58061218261719,
      "logps/ref_chosen": -82.70620727539062,
      "logps/ref_rejected": -103.10424041748047,
      "logps/rejected": -163.32977294921875,
      "loss": 1.188,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.6589775085449219,
      "rewards/margins": 0.38197755813598633,
      "rewards/rejected": -1.0409550666809082,
      "step": 518
    },
    {
      "epoch": 0.7845804988662132,
      "epsilon_dpo/beta": 0.017307139933109283,
      "epsilon_dpo/beta_margin_grad_mean": -0.4324077367782593,
      "epsilon_dpo/beta_margin_grad_std": 0.21184632182121277,
      "epsilon_dpo/beta_margin_mean": 0.31561097502708435,
      "epsilon_dpo/beta_margin_std": 1.0064094066619873,
      "epsilon_dpo/loss_margin_mean": 18.674875259399414,
      "grad_norm": 9.743617057800293,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.017354218289256096,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 6.815658960673781e-08,
      "logits/chosen": -0.023153584450483322,
      "logits/rejected": -0.30077916383743286,
      "logps/chosen": -133.44569396972656,
      "logps/ref_chosen": -83.7763442993164,
      "logps/ref_rejected": -104.38874816894531,
      "logps/rejected": -172.73297119140625,
      "loss": 1.3247,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.8625408411026001,
      "rewards/margins": 0.31561097502708435,
      "rewards/rejected": -1.1781518459320068,
      "step": 519
    },
    {
      "epoch": 0.7860922146636432,
      "epsilon_dpo/beta": 0.017226146534085274,
      "epsilon_dpo/beta_margin_grad_mean": -0.4106348156929016,
      "epsilon_dpo/beta_margin_grad_std": 0.14547856152057648,
      "epsilon_dpo/beta_margin_mean": 0.40368935465812683,
      "epsilon_dpo/beta_margin_std": 0.6645292043685913,
      "epsilon_dpo/loss_margin_mean": 23.6751651763916,
      "grad_norm": 10.935470581054688,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.017305545508861542,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 6.725177529083209e-08,
      "logits/chosen": -0.020371049642562866,
      "logits/rejected": -0.517693281173706,
      "logps/chosen": -128.94769287109375,
      "logps/ref_chosen": -87.44279479980469,
      "logps/ref_rejected": -101.50048828125,
      "logps/rejected": -166.68057250976562,
      "loss": 1.1235,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7172441482543945,
      "rewards/margins": 0.40368932485580444,
      "rewards/rejected": -1.1209335327148438,
      "step": 520
    },
    {
      "epoch": 0.7876039304610734,
      "epsilon_dpo/beta": 0.01715654507279396,
      "epsilon_dpo/beta_margin_grad_mean": -0.390647292137146,
      "epsilon_dpo/beta_margin_grad_std": 0.17379479110240936,
      "epsilon_dpo/beta_margin_mean": 0.5044295191764832,
      "epsilon_dpo/beta_margin_std": 0.830818235874176,
      "epsilon_dpo/loss_margin_mean": 29.717973709106445,
      "grad_norm": 8.840238571166992,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.017224805429577827,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 6.63520728356167e-08,
      "logits/chosen": 0.1258752942085266,
      "logits/rejected": -0.2747128903865814,
      "logps/chosen": -136.19918823242188,
      "logps/ref_chosen": -93.29463195800781,
      "logps/ref_rejected": -115.53233337402344,
      "logps/rejected": -188.15487670898438,
      "loss": 1.0978,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.7379418611526489,
      "rewards/margins": 0.5044295191764832,
      "rewards/rejected": -1.2423713207244873,
      "step": 521
    },
    {
      "epoch": 0.7891156462585034,
      "epsilon_dpo/beta": 0.017097851261496544,
      "epsilon_dpo/beta_margin_grad_mean": -0.42847952246665955,
      "epsilon_dpo/beta_margin_grad_std": 0.19407261908054352,
      "epsilon_dpo/beta_margin_mean": 0.3344910442829132,
      "epsilon_dpo/beta_margin_std": 0.9310140609741211,
      "epsilon_dpo/loss_margin_mean": 19.93511962890625,
      "grad_norm": 12.875117301940918,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.017155112698674202,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 6.545750740770336e-08,
      "logits/chosen": -0.20354172587394714,
      "logits/rejected": -0.12197837978601456,
      "logps/chosen": -129.2339324951172,
      "logps/ref_chosen": -84.8248291015625,
      "logps/ref_rejected": -90.75812530517578,
      "logps/rejected": -155.10235595703125,
      "loss": 1.2743,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.7611554861068726,
      "rewards/margins": 0.3344910144805908,
      "rewards/rejected": -1.0956465005874634,
      "step": 522
    },
    {
      "epoch": 0.7906273620559335,
      "epsilon_dpo/beta": 0.01701790653169155,
      "epsilon_dpo/beta_margin_grad_mean": -0.4060684144496918,
      "epsilon_dpo/beta_margin_grad_std": 0.1653435081243515,
      "epsilon_dpo/beta_margin_mean": 0.42627379298210144,
      "epsilon_dpo/beta_margin_std": 0.7653840780258179,
      "epsilon_dpo/loss_margin_mean": 25.342716217041016,
      "grad_norm": 10.446045875549316,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.017096344381570816,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 6.456810403001012e-08,
      "logits/chosen": -0.18978756666183472,
      "logits/rejected": -0.33216989040374756,
      "logps/chosen": -134.27200317382812,
      "logps/ref_chosen": -87.69473266601562,
      "logps/ref_rejected": -127.14915466308594,
      "logps/rejected": -199.0691375732422,
      "loss": 1.138,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7956066727638245,
      "rewards/margins": 0.42627379298210144,
      "rewards/rejected": -1.2218804359436035,
      "step": 523
    },
    {
      "epoch": 0.7921390778533636,
      "epsilon_dpo/beta": 0.01693318784236908,
      "epsilon_dpo/beta_margin_grad_mean": -0.41478970646858215,
      "epsilon_dpo/beta_margin_grad_std": 0.16659562289714813,
      "epsilon_dpo/beta_margin_mean": 0.37822225689888,
      "epsilon_dpo/beta_margin_std": 0.748524010181427,
      "epsilon_dpo/loss_margin_mean": 22.62693977355957,
      "grad_norm": 9.329117774963379,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.01701657846570015,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 6.368388758106134e-08,
      "logits/chosen": -0.4357072114944458,
      "logits/rejected": -0.3554335832595825,
      "logps/chosen": -138.2708740234375,
      "logps/ref_chosen": -101.48445129394531,
      "logps/ref_rejected": -116.5414047241211,
      "logps/rejected": -175.95477294921875,
      "loss": 1.1738,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6247434616088867,
      "rewards/margins": 0.37822225689888,
      "rewards/rejected": -1.0029656887054443,
      "step": 524
    },
    {
      "epoch": 0.7936507936507936,
      "epsilon_dpo/beta": 0.016896570101380348,
      "epsilon_dpo/beta_margin_grad_mean": -0.43039360642433167,
      "epsilon_dpo/beta_margin_grad_std": 0.15428684651851654,
      "epsilon_dpo/beta_margin_mean": 0.3196568191051483,
      "epsilon_dpo/beta_margin_std": 0.7115899324417114,
      "epsilon_dpo/loss_margin_mean": 19.21969223022461,
      "grad_norm": 8.687216758728027,
      "kl/avg_steps": 0.21875,
      "kl/beta": 0.016931919381022453,
      "kl/n_epsilon_steps": 0.390625,
      "kl/p_epsilon_steps": 0.609375,
      "learning_rate": 6.280488279429185e-08,
      "logits/chosen": -0.5150451064109802,
      "logits/rejected": -0.6030400395393372,
      "logps/chosen": -145.2410888671875,
      "logps/ref_chosen": -100.77320861816406,
      "logps/ref_rejected": -106.02122497558594,
      "logps/rejected": -169.70880126953125,
      "loss": 1.2076,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.7533972859382629,
      "rewards/margins": 0.3196568191051483,
      "rewards/rejected": -1.0730540752410889,
      "step": 525
    },
    {
      "epoch": 0.7951625094482238,
      "epsilon_dpo/beta": 0.016849128529429436,
      "epsilon_dpo/beta_margin_grad_mean": -0.4284980595111847,
      "epsilon_dpo/beta_margin_grad_std": 0.15590006113052368,
      "epsilon_dpo/beta_margin_mean": 0.3317577838897705,
      "epsilon_dpo/beta_margin_std": 0.725226879119873,
      "epsilon_dpo/loss_margin_mean": 19.985027313232422,
      "grad_norm": 8.52363109588623,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.016894960775971413,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 6.193111425735515e-08,
      "logits/chosen": -0.17107675969600677,
      "logits/rejected": -0.36276209354400635,
      "logps/chosen": -129.8316192626953,
      "logps/ref_chosen": -83.02246856689453,
      "logps/ref_rejected": -108.54130554199219,
      "logps/rejected": -175.33547973632812,
      "loss": 1.2012,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.7910605669021606,
      "rewards/margins": 0.3317577838897705,
      "rewards/rejected": -1.1228183507919312,
      "step": 526
    },
    {
      "epoch": 0.7966742252456538,
      "epsilon_dpo/beta": 0.016817670315504074,
      "epsilon_dpo/beta_margin_grad_mean": -0.43790262937545776,
      "epsilon_dpo/beta_margin_grad_std": 0.1645323485136032,
      "epsilon_dpo/beta_margin_mean": 0.28252390027046204,
      "epsilon_dpo/beta_margin_std": 0.7518173456192017,
      "epsilon_dpo/loss_margin_mean": 17.13136863708496,
      "grad_norm": 9.994187355041504,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.016847576946020126,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 6.106260641143546e-08,
      "logits/chosen": -0.1380855292081833,
      "logits/rejected": -0.5256826877593994,
      "logps/chosen": -134.24891662597656,
      "logps/ref_chosen": -82.10969543457031,
      "logps/ref_rejected": -118.33562469482422,
      "logps/rejected": -187.60621643066406,
      "loss": 1.2541,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.8798421621322632,
      "rewards/margins": 0.28252387046813965,
      "rewards/rejected": -1.1623661518096924,
      "step": 527
    },
    {
      "epoch": 0.7981859410430839,
      "epsilon_dpo/beta": 0.01674940623342991,
      "epsilon_dpo/beta_margin_grad_mean": -0.414455384016037,
      "epsilon_dpo/beta_margin_grad_std": 0.17547161877155304,
      "epsilon_dpo/beta_margin_mean": 0.38497909903526306,
      "epsilon_dpo/beta_margin_std": 0.8312879800796509,
      "epsilon_dpo/loss_margin_mean": 23.30469512939453,
      "grad_norm": 8.494281768798828,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.016816047951579094,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 6.019938355056422e-08,
      "logits/chosen": 0.061112236231565475,
      "logits/rejected": -0.18409264087677002,
      "logps/chosen": -120.67262268066406,
      "logps/ref_chosen": -80.03038787841797,
      "logps/ref_rejected": -102.83058166503906,
      "logps/rejected": -166.7775115966797,
      "loss": 1.1941,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6828817129135132,
      "rewards/margins": 0.38497909903526306,
      "rewards/rejected": -1.0678608417510986,
      "step": 528
    },
    {
      "epoch": 0.799697656840514,
      "epsilon_dpo/beta": 0.016650233417749405,
      "epsilon_dpo/beta_margin_grad_mean": -0.34349969029426575,
      "epsilon_dpo/beta_margin_grad_std": 0.16111795604228973,
      "epsilon_dpo/beta_margin_mean": 0.7405751943588257,
      "epsilon_dpo/beta_margin_std": 0.7790158987045288,
      "epsilon_dpo/loss_margin_mean": 44.73019790649414,
      "grad_norm": 8.767058372497559,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.016748009249567986,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 5.934146982094049e-08,
      "logits/chosen": 0.014895215630531311,
      "logits/rejected": -0.07006177306175232,
      "logps/chosen": -113.50651550292969,
      "logps/ref_chosen": -78.1475830078125,
      "logps/ref_rejected": -98.78194427490234,
      "logps/rejected": -178.87107849121094,
      "loss": 0.9084,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.5910122990608215,
      "rewards/margins": 0.7405751943588257,
      "rewards/rejected": -1.331587553024292,
      "step": 529
    },
    {
      "epoch": 0.8012093726379441,
      "epsilon_dpo/beta": 0.016588380560278893,
      "epsilon_dpo/beta_margin_grad_mean": -0.3937232792377472,
      "epsilon_dpo/beta_margin_grad_std": 0.1646030843257904,
      "epsilon_dpo/beta_margin_mean": 0.49445220828056335,
      "epsilon_dpo/beta_margin_std": 0.761608362197876,
      "epsilon_dpo/loss_margin_mean": 30.12356948852539,
      "grad_norm": 9.491631507873535,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.016649154946208,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 5.848888922025552e-08,
      "logits/chosen": -0.3080604076385498,
      "logits/rejected": -0.39112964272499084,
      "logps/chosen": -124.18809509277344,
      "logps/ref_chosen": -84.73670196533203,
      "logps/ref_rejected": -100.79402160644531,
      "logps/rejected": -170.36898803710938,
      "loss": 1.0825,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.6560468673706055,
      "rewards/margins": 0.49445223808288574,
      "rewards/rejected": -1.1504991054534912,
      "step": 530
    },
    {
      "epoch": 0.8027210884353742,
      "epsilon_dpo/beta": 0.01652122102677822,
      "epsilon_dpo/beta_margin_grad_mean": -0.41418811678886414,
      "epsilon_dpo/beta_margin_grad_std": 0.14102323353290558,
      "epsilon_dpo/beta_margin_mean": 0.38079819083213806,
      "epsilon_dpo/beta_margin_std": 0.6443190574645996,
      "epsilon_dpo/loss_margin_mean": 23.29291343688965,
      "grad_norm": 12.684146881103516,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.01658695377409458,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 5.7641665597021435e-08,
      "logits/chosen": -0.11376606673002243,
      "logits/rejected": -0.1980658918619156,
      "logps/chosen": -117.44862365722656,
      "logps/ref_chosen": -75.68649291992188,
      "logps/ref_rejected": -102.33787536621094,
      "logps/rejected": -167.39292907714844,
      "loss": 1.1365,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6924214959144592,
      "rewards/margins": 0.38079819083213806,
      "rewards/rejected": -1.0732197761535645,
      "step": 531
    },
    {
      "epoch": 0.8042328042328042,
      "epsilon_dpo/beta": 0.016438886523246765,
      "epsilon_dpo/beta_margin_grad_mean": -0.40522220730781555,
      "epsilon_dpo/beta_margin_grad_std": 0.16460780799388885,
      "epsilon_dpo/beta_margin_mean": 0.4439644515514374,
      "epsilon_dpo/beta_margin_std": 0.7691618204116821,
      "epsilon_dpo/loss_margin_mean": 27.288150787353516,
      "grad_norm": 8.678123474121094,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.01651984080672264,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 5.679982264990424e-08,
      "logits/chosen": -0.21569831669330597,
      "logits/rejected": -0.23878905177116394,
      "logps/chosen": -130.5865478515625,
      "logps/ref_chosen": -83.01626586914062,
      "logps/ref_rejected": -103.29268646240234,
      "logps/rejected": -178.151123046875,
      "loss": 1.1238,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.7830289602279663,
      "rewards/margins": 0.44396448135375977,
      "rewards/rejected": -1.226993441581726,
      "step": 532
    },
    {
      "epoch": 0.8057445200302343,
      "epsilon_dpo/beta": 0.016387924551963806,
      "epsilon_dpo/beta_margin_grad_mean": -0.4147703945636749,
      "epsilon_dpo/beta_margin_grad_std": 0.18282870948314667,
      "epsilon_dpo/beta_margin_mean": 0.40231624245643616,
      "epsilon_dpo/beta_margin_std": 0.8481965065002441,
      "epsilon_dpo/loss_margin_mean": 24.933364868164062,
      "grad_norm": 10.612200736999512,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.016437653452157974,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 5.596338392706076e-08,
      "logits/chosen": 0.08782052993774414,
      "logits/rejected": -0.4669248163700104,
      "logps/chosen": -113.88616943359375,
      "logps/ref_chosen": -78.34988403320312,
      "logps/ref_rejected": -96.21743774414062,
      "logps/rejected": -156.6870880126953,
      "loss": 1.1873,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.5851448774337769,
      "rewards/margins": 0.40231621265411377,
      "rewards/rejected": -0.9874610900878906,
      "step": 533
    },
    {
      "epoch": 0.8072562358276644,
      "epsilon_dpo/beta": 0.016306143254041672,
      "epsilon_dpo/beta_margin_grad_mean": -0.39130768179893494,
      "epsilon_dpo/beta_margin_grad_std": 0.16498181223869324,
      "epsilon_dpo/beta_margin_mean": 0.4981834590435028,
      "epsilon_dpo/beta_margin_std": 0.7556285858154297,
      "epsilon_dpo/loss_margin_mean": 30.85486602783203,
      "grad_norm": 10.962309837341309,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.016386445611715317,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 5.513237282548033e-08,
      "logits/chosen": -0.48020732402801514,
      "logits/rejected": -0.33289915323257446,
      "logps/chosen": -116.89681243896484,
      "logps/ref_chosen": -84.2362060546875,
      "logps/ref_rejected": -98.3175048828125,
      "logps/rejected": -161.83297729492188,
      "loss": 1.0786,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5349410176277161,
      "rewards/margins": 0.4981834292411804,
      "rewards/rejected": -1.0331244468688965,
      "step": 534
    },
    {
      "epoch": 0.8087679516250945,
      "epsilon_dpo/beta": 0.01623521000146866,
      "epsilon_dpo/beta_margin_grad_mean": -0.4018467962741852,
      "epsilon_dpo/beta_margin_grad_std": 0.1728149652481079,
      "epsilon_dpo/beta_margin_mean": 0.4536625146865845,
      "epsilon_dpo/beta_margin_std": 0.8184735178947449,
      "epsilon_dpo/loss_margin_mean": 28.273239135742188,
      "grad_norm": 7.956579685211182,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.016304921358823776,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 5.430681259032957e-08,
      "logits/chosen": -0.20313288271427155,
      "logits/rejected": -0.416176974773407,
      "logps/chosen": -125.93107604980469,
      "logps/ref_chosen": -84.17073822021484,
      "logps/ref_rejected": -104.83849334716797,
      "logps/rejected": -174.8720703125,
      "loss": 1.1333,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.6805884838104248,
      "rewards/margins": 0.4536625146865845,
      "rewards/rejected": -1.1342509984970093,
      "step": 535
    },
    {
      "epoch": 0.8102796674225246,
      "epsilon_dpo/beta": 0.016134047880768776,
      "epsilon_dpo/beta_margin_grad_mean": -0.37715545296669006,
      "epsilon_dpo/beta_margin_grad_std": 0.1385839879512787,
      "epsilon_dpo/beta_margin_mean": 0.5649297833442688,
      "epsilon_dpo/beta_margin_std": 0.6600634455680847,
      "epsilon_dpo/loss_margin_mean": 35.17201614379883,
      "grad_norm": 7.054563999176025,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.01623389683663845,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 5.3486726314303175e-08,
      "logits/chosen": -0.08141159266233444,
      "logits/rejected": -0.3341004550457001,
      "logps/chosen": -105.59613037109375,
      "logps/ref_chosen": -70.18797302246094,
      "logps/ref_rejected": -97.0616455078125,
      "logps/rejected": -167.64183044433594,
      "loss": 0.9949,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5719255805015564,
      "rewards/margins": 0.5649298429489136,
      "rewards/rejected": -1.1368553638458252,
      "step": 536
    },
    {
      "epoch": 0.8117913832199547,
      "epsilon_dpo/beta": 0.016084259375929832,
      "epsilon_dpo/beta_margin_grad_mean": -0.4204951226711273,
      "epsilon_dpo/beta_margin_grad_std": 0.15833215415477753,
      "epsilon_dpo/beta_margin_mean": 0.35688626766204834,
      "epsilon_dpo/beta_margin_std": 0.7070753574371338,
      "epsilon_dpo/loss_margin_mean": 22.500165939331055,
      "grad_norm": 10.414916038513184,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.01613306626677513,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 5.267213693697695e-08,
      "logits/chosen": -0.13419167697429657,
      "logits/rejected": -0.4870593547821045,
      "logps/chosen": -133.16604614257812,
      "logps/ref_chosen": -80.709716796875,
      "logps/ref_rejected": -116.72985076904297,
      "logps/rejected": -191.68634033203125,
      "loss": 1.1774,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.8454914093017578,
      "rewards/margins": 0.35688626766204834,
      "rewards/rejected": -1.2023777961730957,
      "step": 537
    },
    {
      "epoch": 0.8133030990173847,
      "epsilon_dpo/beta": 0.016014045104384422,
      "epsilon_dpo/beta_margin_grad_mean": -0.4055761396884918,
      "epsilon_dpo/beta_margin_grad_std": 0.16074000298976898,
      "epsilon_dpo/beta_margin_mean": 0.437906414270401,
      "epsilon_dpo/beta_margin_std": 0.7546932697296143,
      "epsilon_dpo/loss_margin_mean": 27.643733978271484,
      "grad_norm": 8.242533683776855,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.016082806512713432,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 5.1863067244167144e-08,
      "logits/chosen": -0.13110458850860596,
      "logits/rejected": -0.23731385171413422,
      "logps/chosen": -143.7471466064453,
      "logps/ref_chosen": -94.1431884765625,
      "logps/ref_rejected": -103.38029479980469,
      "logps/rejected": -180.62799072265625,
      "loss": 1.1233,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.7969592809677124,
      "rewards/margins": 0.4379063844680786,
      "rewards/rejected": -1.234865665435791,
      "step": 538
    },
    {
      "epoch": 0.8148148148148148,
      "epsilon_dpo/beta": 0.015919266268610954,
      "epsilon_dpo/beta_margin_grad_mean": -0.4092956781387329,
      "epsilon_dpo/beta_margin_grad_std": 0.13377299904823303,
      "epsilon_dpo/beta_margin_mean": 0.40943920612335205,
      "epsilon_dpo/beta_margin_std": 0.6206133961677551,
      "epsilon_dpo/loss_margin_mean": 25.90907859802246,
      "grad_norm": 9.922149658203125,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.016012750566005707,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 5.105953986729195e-08,
      "logits/chosen": 0.24353787302970886,
      "logits/rejected": -0.5500741004943848,
      "logps/chosen": -129.62351989746094,
      "logps/ref_chosen": -84.61248779296875,
      "logps/ref_rejected": -107.14148712158203,
      "logps/rejected": -178.06158447265625,
      "loss": 1.1052,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.7179849147796631,
      "rewards/margins": 0.40943920612335205,
      "rewards/rejected": -1.1274241209030151,
      "step": 539
    },
    {
      "epoch": 0.8163265306122449,
      "epsilon_dpo/beta": 0.015845203772187233,
      "epsilon_dpo/beta_margin_grad_mean": -0.3944285809993744,
      "epsilon_dpo/beta_margin_grad_std": 0.1632717400789261,
      "epsilon_dpo/beta_margin_mean": 0.48868271708488464,
      "epsilon_dpo/beta_margin_std": 0.7638451457023621,
      "epsilon_dpo/loss_margin_mean": 31.13498306274414,
      "grad_norm": 9.602672576904297,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.01591823622584343,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 5.026157728273966e-08,
      "logits/chosen": -0.14873777329921722,
      "logits/rejected": -0.4375326335430145,
      "logps/chosen": -132.97463989257812,
      "logps/ref_chosen": -89.30429077148438,
      "logps/ref_rejected": -122.46199035644531,
      "logps/rejected": -197.267333984375,
      "loss": 1.0872,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6940083503723145,
      "rewards/margins": 0.48868274688720703,
      "rewards/rejected": -1.1826910972595215,
      "step": 540
    },
    {
      "epoch": 0.817838246409675,
      "epsilon_dpo/beta": 0.015781180933117867,
      "epsilon_dpo/beta_margin_grad_mean": -0.38085731863975525,
      "epsilon_dpo/beta_margin_grad_std": 0.16132763028144836,
      "epsilon_dpo/beta_margin_mean": 0.5413305163383484,
      "epsilon_dpo/beta_margin_std": 0.7297938466072083,
      "epsilon_dpo/loss_margin_mean": 34.643150329589844,
      "grad_norm": 12.654080390930176,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.015843968838453293,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 4.9469201811239035e-08,
      "logits/chosen": -0.13464142382144928,
      "logits/rejected": 0.12418127059936523,
      "logps/chosen": -125.31890869140625,
      "logps/ref_chosen": -87.92848205566406,
      "logps/ref_rejected": -81.73931884765625,
      "logps/rejected": -153.7729034423828,
      "loss": 1.0383,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5931801199913025,
      "rewards/margins": 0.5413305163383484,
      "rewards/rejected": -1.1345106363296509,
      "step": 541
    },
    {
      "epoch": 0.8193499622071051,
      "epsilon_dpo/beta": 0.01569266989827156,
      "epsilon_dpo/beta_margin_grad_mean": -0.37570175528526306,
      "epsilon_dpo/beta_margin_grad_std": 0.16721844673156738,
      "epsilon_dpo/beta_margin_mean": 0.5741678476333618,
      "epsilon_dpo/beta_margin_std": 0.7826369404792786,
      "epsilon_dpo/loss_margin_mean": 36.878963470458984,
      "grad_norm": 9.048364639282227,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.01577986218035221,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 4.868243561723534e-08,
      "logits/chosen": -0.06194941699504852,
      "logits/rejected": -0.1134922131896019,
      "logps/chosen": -103.31153869628906,
      "logps/ref_chosen": -71.24203491210938,
      "logps/ref_rejected": -88.06472778320312,
      "logps/rejected": -157.01319885253906,
      "loss": 1.0295,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5051934123039246,
      "rewards/margins": 0.5741678476333618,
      "rewards/rejected": -1.0793613195419312,
      "step": 542
    },
    {
      "epoch": 0.8208616780045351,
      "epsilon_dpo/beta": 0.015595084056258202,
      "epsilon_dpo/beta_margin_grad_mean": -0.3929816782474518,
      "epsilon_dpo/beta_margin_grad_std": 0.16900266706943512,
      "epsilon_dpo/beta_margin_mean": 0.49183303117752075,
      "epsilon_dpo/beta_margin_std": 0.8085128664970398,
      "epsilon_dpo/loss_margin_mean": 31.800132751464844,
      "grad_norm": 7.8205671310424805,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.015691597014665604,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 4.790130070827028e-08,
      "logits/chosen": -0.1205248311161995,
      "logits/rejected": -0.30794036388397217,
      "logps/chosen": -114.30192565917969,
      "logps/ref_chosen": -73.00680541992188,
      "logps/ref_rejected": -99.84335327148438,
      "logps/rejected": -172.9385986328125,
      "loss": 1.0993,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.6449280977249146,
      "rewards/margins": 0.49183300137519836,
      "rewards/rejected": -1.1367610692977905,
      "step": 543
    },
    {
      "epoch": 0.8223733938019653,
      "epsilon_dpo/beta": 0.015512841753661633,
      "epsilon_dpo/beta_margin_grad_mean": -0.3808680474758148,
      "epsilon_dpo/beta_margin_grad_std": 0.17236299812793732,
      "epsilon_dpo/beta_margin_mean": 0.5604837536811829,
      "epsilon_dpo/beta_margin_std": 0.8075725436210632,
      "epsilon_dpo/loss_margin_mean": 36.44947052001953,
      "grad_norm": 11.447324752807617,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.015594134107232094,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 4.7125818934366454e-08,
      "logits/chosen": -0.10196401923894882,
      "logits/rejected": -0.42052143812179565,
      "logps/chosen": -129.619140625,
      "logps/ref_chosen": -87.3775863647461,
      "logps/ref_rejected": -112.98512268066406,
      "logps/rejected": -191.6761474609375,
      "loss": 1.0477,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6571992635726929,
      "rewards/margins": 0.5604837536811829,
      "rewards/rejected": -1.2176830768585205,
      "step": 544
    },
    {
      "epoch": 0.8238851095993953,
      "epsilon_dpo/beta": 0.015464799478650093,
      "epsilon_dpo/beta_margin_grad_mean": -0.41932106018066406,
      "epsilon_dpo/beta_margin_grad_std": 0.16810117661952972,
      "epsilon_dpo/beta_margin_mean": 0.37158551812171936,
      "epsilon_dpo/beta_margin_std": 0.780260443687439,
      "epsilon_dpo/loss_margin_mean": 24.382139205932617,
      "grad_norm": 7.527310848236084,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.015511727891862392,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.635601198741607e-08,
      "logits/chosen": -0.010136006399989128,
      "logits/rejected": -0.3866921663284302,
      "logps/chosen": -129.83676147460938,
      "logps/ref_chosen": -85.00263977050781,
      "logps/ref_rejected": -105.95549011230469,
      "logps/rejected": -175.1717529296875,
      "loss": 1.1873,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.6954314708709717,
      "rewards/margins": 0.371585488319397,
      "rewards/rejected": -1.0670169591903687,
      "step": 545
    },
    {
      "epoch": 0.8253968253968254,
      "epsilon_dpo/beta": 0.015387626364827156,
      "epsilon_dpo/beta_margin_grad_mean": -0.3998827040195465,
      "epsilon_dpo/beta_margin_grad_std": 0.1473892629146576,
      "epsilon_dpo/beta_margin_mean": 0.45398467779159546,
      "epsilon_dpo/beta_margin_std": 0.682119607925415,
      "epsilon_dpo/loss_margin_mean": 29.757774353027344,
      "grad_norm": 8.235089302062988,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.015463404357433319,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 4.559190140057428e-08,
      "logits/chosen": -0.2149541974067688,
      "logits/rejected": -0.11502528190612793,
      "logps/chosen": -117.99137878417969,
      "logps/ref_chosen": -79.49039459228516,
      "logps/ref_rejected": -87.46519470214844,
      "logps/rejected": -155.7239532470703,
      "loss": 1.0882,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5945260524749756,
      "rewards/margins": 0.45398467779159546,
      "rewards/rejected": -1.0485107898712158,
      "step": 546
    },
    {
      "epoch": 0.8269085411942555,
      "epsilon_dpo/beta": 0.015287027694284916,
      "epsilon_dpo/beta_margin_grad_mean": -0.3833127021789551,
      "epsilon_dpo/beta_margin_grad_std": 0.15344193577766418,
      "epsilon_dpo/beta_margin_mean": 0.5260194540023804,
      "epsilon_dpo/beta_margin_std": 0.7001830339431763,
      "epsilon_dpo/loss_margin_mean": 34.64781951904297,
      "grad_norm": 8.841313362121582,
      "kl/avg_steps": 0.65625,
      "kl/beta": 0.015386472456157207,
      "kl/n_epsilon_steps": 0.171875,
      "kl/p_epsilon_steps": 0.828125,
      "learning_rate": 4.483350854765672e-08,
      "logits/chosen": -0.3627888560295105,
      "logits/rejected": -0.21551699936389923,
      "logps/chosen": -110.71125793457031,
      "logps/ref_chosen": -75.10725402832031,
      "logps/ref_rejected": -97.79402160644531,
      "logps/rejected": -168.0458526611328,
      "loss": 1.0396,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": -0.5463103652000427,
      "rewards/margins": 0.5260194540023804,
      "rewards/rejected": -1.0723297595977783,
      "step": 547
    },
    {
      "epoch": 0.8284202569916855,
      "epsilon_dpo/beta": 0.015235133469104767,
      "epsilon_dpo/beta_margin_grad_mean": -0.4219856262207031,
      "epsilon_dpo/beta_margin_grad_std": 0.16018447279930115,
      "epsilon_dpo/beta_margin_mean": 0.3609781563282013,
      "epsilon_dpo/beta_margin_std": 0.7288519144058228,
      "epsilon_dpo/loss_margin_mean": 24.029176712036133,
      "grad_norm": 9.881436347961426,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.01528615690767765,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.4080854642541826e-08,
      "logits/chosen": 0.08564075827598572,
      "logits/rejected": -0.3461897671222687,
      "logps/chosen": -133.748779296875,
      "logps/ref_chosen": -85.9015121459961,
      "logps/ref_rejected": -108.32438659667969,
      "logps/rejected": -180.20083618164062,
      "loss": 1.1796,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.7319462299346924,
      "rewards/margins": 0.3609781265258789,
      "rewards/rejected": -1.0929243564605713,
      "step": 548
    },
    {
      "epoch": 0.8299319727891157,
      "epsilon_dpo/beta": 0.015168659389019012,
      "epsilon_dpo/beta_margin_grad_mean": -0.42518672347068787,
      "epsilon_dpo/beta_margin_grad_std": 0.17199109494686127,
      "epsilon_dpo/beta_margin_mean": 0.3430107533931732,
      "epsilon_dpo/beta_margin_std": 0.7894731163978577,
      "epsilon_dpo/loss_margin_mean": 22.958250045776367,
      "grad_norm": 16.42542839050293,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.015233790501952171,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.333396073857723e-08,
      "logits/chosen": -0.14731433987617493,
      "logits/rejected": -0.31095337867736816,
      "logps/chosen": -131.48519897460938,
      "logps/ref_chosen": -87.12271118164062,
      "logps/ref_rejected": -121.61825561523438,
      "logps/rejected": -188.93899536132812,
      "loss": 1.2156,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.6761950254440308,
      "rewards/margins": 0.3430107533931732,
      "rewards/rejected": -1.0192058086395264,
      "step": 549
    },
    {
      "epoch": 0.8314436885865457,
      "epsilon_dpo/beta": 0.015102584846317768,
      "epsilon_dpo/beta_margin_grad_mean": -0.4151040017604828,
      "epsilon_dpo/beta_margin_grad_std": 0.1606675386428833,
      "epsilon_dpo/beta_margin_mean": 0.3868383765220642,
      "epsilon_dpo/beta_margin_std": 0.7236196994781494,
      "epsilon_dpo/loss_margin_mean": 25.927637100219727,
      "grad_norm": 9.797503471374512,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.015167432837188244,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.259284772799099e-08,
      "logits/chosen": -0.009366992861032486,
      "logits/rejected": -0.278700590133667,
      "logps/chosen": -124.78021240234375,
      "logps/ref_chosen": -80.63383483886719,
      "logps/ref_rejected": -87.28580474853516,
      "logps/rejected": -157.35983276367188,
      "loss": 1.1575,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6684355735778809,
      "rewards/margins": 0.3868383765220642,
      "rewards/rejected": -1.0552740097045898,
      "step": 550
    },
    {
      "epoch": 0.8329554043839759,
      "epsilon_dpo/beta": 0.01503679994493723,
      "epsilon_dpo/beta_margin_grad_mean": -0.41068235039711,
      "epsilon_dpo/beta_margin_grad_std": 0.13416306674480438,
      "epsilon_dpo/beta_margin_mean": 0.39317819476127625,
      "epsilon_dpo/beta_margin_std": 0.6135088205337524,
      "epsilon_dpo/loss_margin_mean": 26.400938034057617,
      "grad_norm": 8.979781150817871,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.015101364813745022,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 4.1857536341307176e-08,
      "logits/chosen": 0.00559891015291214,
      "logits/rejected": -0.38543951511383057,
      "logps/chosen": -133.9377899169922,
      "logps/ref_chosen": -92.95742797851562,
      "logps/ref_rejected": -112.68172454833984,
      "logps/rejected": -180.06301879882812,
      "loss": 1.1175,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.6161550283432007,
      "rewards/margins": 0.39317816495895386,
      "rewards/rejected": -1.0093331336975098,
      "step": 551
    },
    {
      "epoch": 0.8344671201814059,
      "epsilon_dpo/beta": 0.014957202598452568,
      "epsilon_dpo/beta_margin_grad_mean": -0.4155219495296478,
      "epsilon_dpo/beta_margin_grad_std": 0.1576501727104187,
      "epsilon_dpo/beta_margin_mean": 0.3671215772628784,
      "epsilon_dpo/beta_margin_std": 0.7070900797843933,
      "epsilon_dpo/loss_margin_mean": 24.845149993896484,
      "grad_norm": 10.171217918395996,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.015035583637654781,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 4.112804714676593e-08,
      "logits/chosen": -0.25429147481918335,
      "logits/rejected": -0.29028773307800293,
      "logps/chosen": -131.58714294433594,
      "logps/ref_chosen": -88.42652130126953,
      "logps/ref_rejected": -111.27716064453125,
      "logps/rejected": -179.28292846679688,
      "loss": 1.1692,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6473616361618042,
      "rewards/margins": 0.3671215772628784,
      "rewards/rejected": -1.0144832134246826,
      "step": 552
    },
    {
      "epoch": 0.8359788359788359,
      "epsilon_dpo/beta": 0.014906208962202072,
      "epsilon_dpo/beta_margin_grad_mean": -0.4175128936767578,
      "epsilon_dpo/beta_margin_grad_std": 0.18361634016036987,
      "epsilon_dpo/beta_margin_mean": 0.3894207775592804,
      "epsilon_dpo/beta_margin_std": 0.86497563123703,
      "epsilon_dpo/loss_margin_mean": 26.525848388671875,
      "grad_norm": 9.084397315979004,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.014956129714846611,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 4.0404400549748144e-08,
      "logits/chosen": 0.05000998079776764,
      "logits/rejected": -0.2641226649284363,
      "logps/chosen": -126.92143249511719,
      "logps/ref_chosen": -78.53276824951172,
      "logps/ref_rejected": -111.15892028808594,
      "logps/rejected": -186.07342529296875,
      "loss": 1.2031,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.7230408191680908,
      "rewards/margins": 0.3894208073616028,
      "rewards/rejected": -1.1124615669250488,
      "step": 553
    },
    {
      "epoch": 0.8374905517762661,
      "epsilon_dpo/beta": 0.01482719462364912,
      "epsilon_dpo/beta_margin_grad_mean": -0.39437729120254517,
      "epsilon_dpo/beta_margin_grad_std": 0.15337437391281128,
      "epsilon_dpo/beta_margin_mean": 0.4819382429122925,
      "epsilon_dpo/beta_margin_std": 0.7117973566055298,
      "epsilon_dpo/loss_margin_mean": 32.775718688964844,
      "grad_norm": 8.914608001708984,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.014904893934726715,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 3.968661679220467e-08,
      "logits/chosen": -0.3598790466785431,
      "logits/rejected": -0.19130371510982513,
      "logps/chosen": -129.90811157226562,
      "logps/ref_chosen": -85.90342712402344,
      "logps/ref_rejected": -94.58822631835938,
      "logps/rejected": -171.36862182617188,
      "loss": 1.0757,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6546410918235779,
      "rewards/margins": 0.4819382429122925,
      "rewards/rejected": -1.1365792751312256,
      "step": 554
    },
    {
      "epoch": 0.8390022675736961,
      "epsilon_dpo/beta": 0.014753474853932858,
      "epsilon_dpo/beta_margin_grad_mean": -0.40565571188926697,
      "epsilon_dpo/beta_margin_grad_std": 0.1690124273300171,
      "epsilon_dpo/beta_margin_mean": 0.42326241731643677,
      "epsilon_dpo/beta_margin_std": 0.7654484510421753,
      "epsilon_dpo/loss_margin_mean": 29.026073455810547,
      "grad_norm": 9.91170597076416,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.014826130121946335,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.89747159520904e-08,
      "logits/chosen": -0.008408933877944946,
      "logits/rejected": -0.022642306983470917,
      "logps/chosen": -136.08340454101562,
      "logps/ref_chosen": -87.41081237792969,
      "logps/ref_rejected": -94.39276123046875,
      "logps/rejected": -172.09144592285156,
      "loss": 1.1425,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.7206442952156067,
      "rewards/margins": 0.42326241731643677,
      "rewards/rejected": -1.1439067125320435,
      "step": 555
    },
    {
      "epoch": 0.8405139833711263,
      "epsilon_dpo/beta": 0.014689295552670956,
      "epsilon_dpo/beta_margin_grad_mean": -0.419183611869812,
      "epsilon_dpo/beta_margin_grad_std": 0.1609971672296524,
      "epsilon_dpo/beta_margin_mean": 0.37554728984832764,
      "epsilon_dpo/beta_margin_std": 0.7430645823478699,
      "epsilon_dpo/loss_margin_mean": 25.880294799804688,
      "grad_norm": 7.737394332885742,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.014752368442714214,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.826871794280192e-08,
      "logits/chosen": 0.10626913607120514,
      "logits/rejected": -0.08207418769598007,
      "logps/chosen": -123.96664428710938,
      "logps/ref_chosen": -71.60616302490234,
      "logps/ref_rejected": -90.15902709960938,
      "logps/rejected": -168.39981079101562,
      "loss": 1.1713,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.7713250517845154,
      "rewards/margins": 0.37554728984832764,
      "rewards/rejected": -1.1468722820281982,
      "step": 556
    },
    {
      "epoch": 0.8420256991685563,
      "epsilon_dpo/beta": 0.014616128988564014,
      "epsilon_dpo/beta_margin_grad_mean": -0.39464807510375977,
      "epsilon_dpo/beta_margin_grad_std": 0.14853645861148834,
      "epsilon_dpo/beta_margin_mean": 0.4820627272129059,
      "epsilon_dpo/beta_margin_std": 0.6935344338417053,
      "epsilon_dpo/loss_margin_mean": 33.26234436035156,
      "grad_norm": 8.162524223327637,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.014688108116388321,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.756864251262143e-08,
      "logits/chosen": -0.1140388548374176,
      "logits/rejected": -0.6151837706565857,
      "logps/chosen": -125.14369201660156,
      "logps/ref_chosen": -78.79825592041016,
      "logps/ref_rejected": -105.23713684082031,
      "logps/rejected": -184.84490966796875,
      "loss": 1.069,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6800782680511475,
      "rewards/margins": 0.4820627272129059,
      "rewards/rejected": -1.1621410846710205,
      "step": 557
    },
    {
      "epoch": 0.8435374149659864,
      "epsilon_dpo/beta": 0.014561682008206844,
      "epsilon_dpo/beta_margin_grad_mean": -0.40123188495635986,
      "epsilon_dpo/beta_margin_grad_std": 0.1584625244140625,
      "epsilon_dpo/beta_margin_mean": 0.4439917504787445,
      "epsilon_dpo/beta_margin_std": 0.7239269614219666,
      "epsilon_dpo/loss_margin_mean": 30.838037490844727,
      "grad_norm": 10.327901840209961,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.014615032821893692,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 3.687450924416341e-08,
      "logits/chosen": -0.21190424263477325,
      "logits/rejected": -0.20192435383796692,
      "logps/chosen": -131.4257354736328,
      "logps/ref_chosen": -86.41659545898438,
      "logps/ref_rejected": -115.81890869140625,
      "logps/rejected": -191.6660919189453,
      "loss": 1.1108,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6573868989944458,
      "rewards/margins": 0.4439917504787445,
      "rewards/rejected": -1.1013786792755127,
      "step": 558
    },
    {
      "epoch": 0.8450491307634165,
      "epsilon_dpo/beta": 0.014511831104755402,
      "epsilon_dpo/beta_margin_grad_mean": -0.41161999106407166,
      "epsilon_dpo/beta_margin_grad_std": 0.18268904089927673,
      "epsilon_dpo/beta_margin_mean": 0.41245096921920776,
      "epsilon_dpo/beta_margin_std": 0.8520277142524719,
      "epsilon_dpo/loss_margin_mean": 28.83552360534668,
      "grad_norm": 8.330303192138672,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.014560431241989136,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.6186337553827743e-08,
      "logits/chosen": -0.13291501998901367,
      "logits/rejected": -0.49170762300491333,
      "logps/chosen": -128.95718383789062,
      "logps/ref_chosen": -84.51522064208984,
      "logps/ref_rejected": -105.63461303710938,
      "logps/rejected": -178.912109375,
      "loss": 1.1804,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6472938060760498,
      "rewards/margins": 0.41245096921920776,
      "rewards/rejected": -1.0597448348999023,
      "step": 559
    },
    {
      "epoch": 0.8465608465608465,
      "epsilon_dpo/beta": 0.014443977735936642,
      "epsilon_dpo/beta_margin_grad_mean": -0.38788938522338867,
      "epsilon_dpo/beta_margin_grad_std": 0.1522243767976761,
      "epsilon_dpo/beta_margin_mean": 0.5074370503425598,
      "epsilon_dpo/beta_margin_std": 0.6997931003570557,
      "epsilon_dpo/loss_margin_mean": 35.44127655029297,
      "grad_norm": 7.828211784362793,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.01451055146753788,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.550414669125573e-08,
      "logits/chosen": 0.03868032246828079,
      "logits/rejected": 0.07886442542076111,
      "logps/chosen": -128.6056671142578,
      "logps/ref_chosen": -84.04110717773438,
      "logps/ref_rejected": -98.00209045410156,
      "logps/rejected": -178.0079345703125,
      "loss": 1.053,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.6456395983695984,
      "rewards/margins": 0.5074370503425598,
      "rewards/rejected": -1.1530766487121582,
      "step": 560
    },
    {
      "epoch": 0.8480725623582767,
      "epsilon_dpo/beta": 0.014367558993399143,
      "epsilon_dpo/beta_margin_grad_mean": -0.38891005516052246,
      "epsilon_dpo/beta_margin_grad_std": 0.1498725563287735,
      "epsilon_dpo/beta_margin_mean": 0.5143095850944519,
      "epsilon_dpo/beta_margin_std": 0.7102375626564026,
      "epsilon_dpo/loss_margin_mean": 36.05918884277344,
      "grad_norm": 7.392366409301758,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.014442849904298782,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 3.482795573879241e-08,
      "logits/chosen": -0.23694953322410583,
      "logits/rejected": -0.20725828409194946,
      "logps/chosen": -127.86459350585938,
      "logps/ref_chosen": -85.260009765625,
      "logps/ref_rejected": -99.17951202392578,
      "logps/rejected": -177.84329223632812,
      "loss": 1.0484,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.61405348777771,
      "rewards/margins": 0.5143096446990967,
      "rewards/rejected": -1.128363013267517,
      "step": 561
    },
    {
      "epoch": 0.8495842781557067,
      "epsilon_dpo/beta": 0.01430959440767765,
      "epsilon_dpo/beta_margin_grad_mean": -0.402746319770813,
      "epsilon_dpo/beta_margin_grad_std": 0.14407731592655182,
      "epsilon_dpo/beta_margin_mean": 0.441642165184021,
      "epsilon_dpo/beta_margin_std": 0.6657573580741882,
      "epsilon_dpo/loss_margin_mean": 31.14838218688965,
      "grad_norm": 8.244582176208496,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.014366528019309044,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 3.415778361095226e-08,
      "logits/chosen": -0.13111603260040283,
      "logits/rejected": -0.1937008500099182,
      "logps/chosen": -150.1193389892578,
      "logps/ref_chosen": -101.58757019042969,
      "logps/ref_rejected": -116.84658813476562,
      "logps/rejected": -196.52674865722656,
      "loss": 1.0929,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.6962746977806091,
      "rewards/margins": 0.441642165184021,
      "rewards/rejected": -1.1379168033599854,
      "step": 562
    },
    {
      "epoch": 0.8510959939531368,
      "epsilon_dpo/beta": 0.014247225597500801,
      "epsilon_dpo/beta_margin_grad_mean": -0.39969557523727417,
      "epsilon_dpo/beta_margin_grad_std": 0.17446556687355042,
      "epsilon_dpo/beta_margin_mean": 0.466126948595047,
      "epsilon_dpo/beta_margin_std": 0.8048774003982544,
      "epsilon_dpo/loss_margin_mean": 33.10220718383789,
      "grad_norm": 8.794232368469238,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.014308400452136993,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.349364905389032e-08,
      "logits/chosen": -0.10935737937688828,
      "logits/rejected": -0.1448805034160614,
      "logps/chosen": -109.391845703125,
      "logps/ref_chosen": -69.69686889648438,
      "logps/ref_rejected": -90.02664184570312,
      "logps/rejected": -162.82382202148438,
      "loss": 1.1205,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5677230358123779,
      "rewards/margins": 0.4661269187927246,
      "rewards/rejected": -1.0338499546051025,
      "step": 563
    },
    {
      "epoch": 0.8526077097505669,
      "epsilon_dpo/beta": 0.014180713333189487,
      "epsilon_dpo/beta_margin_grad_mean": -0.4073896110057831,
      "epsilon_dpo/beta_margin_grad_std": 0.1713608354330063,
      "epsilon_dpo/beta_margin_mean": 0.4204823076725006,
      "epsilon_dpo/beta_margin_std": 0.78926020860672,
      "epsilon_dpo/loss_margin_mean": 30.017087936401367,
      "grad_norm": 8.327298164367676,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.014246073551476002,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 3.283557064487785e-08,
      "logits/chosen": -0.06540387868881226,
      "logits/rejected": -0.05503258854150772,
      "logps/chosen": -117.8368911743164,
      "logps/ref_chosen": -79.32344055175781,
      "logps/ref_rejected": -90.12184143066406,
      "logps/rejected": -158.65237426757812,
      "loss": 1.1517,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5486175417900085,
      "rewards/margins": 0.4204823076725006,
      "rewards/rejected": -0.9690998792648315,
      "step": 564
    },
    {
      "epoch": 0.854119425547997,
      "epsilon_dpo/beta": 0.01411898247897625,
      "epsilon_dpo/beta_margin_grad_mean": -0.41273975372314453,
      "epsilon_dpo/beta_margin_grad_std": 0.15865632891654968,
      "epsilon_dpo/beta_margin_mean": 0.39410722255706787,
      "epsilon_dpo/beta_margin_std": 0.7258629202842712,
      "epsilon_dpo/loss_margin_mean": 28.23537254333496,
      "grad_norm": 6.96705436706543,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.014179606921970844,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.218356679178252e-08,
      "logits/chosen": -0.21644821763038635,
      "logits/rejected": -0.06834931671619415,
      "logps/chosen": -133.81097412109375,
      "logps/ref_chosen": -85.25615692138672,
      "logps/ref_rejected": -108.77261352539062,
      "logps/rejected": -185.5627899169922,
      "loss": 1.1514,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.6872435212135315,
      "rewards/margins": 0.39410722255706787,
      "rewards/rejected": -1.0813508033752441,
      "step": 565
    },
    {
      "epoch": 0.8556311413454271,
      "epsilon_dpo/beta": 0.014070717617869377,
      "epsilon_dpo/beta_margin_grad_mean": -0.4168214201927185,
      "epsilon_dpo/beta_margin_grad_std": 0.1676950603723526,
      "epsilon_dpo/beta_margin_mean": 0.3780567944049835,
      "epsilon_dpo/beta_margin_std": 0.7630917429924011,
      "epsilon_dpo/loss_margin_mean": 27.264877319335938,
      "grad_norm": 12.438030242919922,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.014117840677499771,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.1537655732553764e-08,
      "logits/chosen": -0.2886316180229187,
      "logits/rejected": -0.3793463706970215,
      "logps/chosen": -137.4063262939453,
      "logps/ref_chosen": -93.50871276855469,
      "logps/ref_rejected": -94.66215515136719,
      "logps/rejected": -165.82464599609375,
      "loss": 1.1777,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.6199392080307007,
      "rewards/margins": 0.3780567944049835,
      "rewards/rejected": -0.9979960918426514,
      "step": 566
    },
    {
      "epoch": 0.8571428571428571,
      "epsilon_dpo/beta": 0.013991734944283962,
      "epsilon_dpo/beta_margin_grad_mean": -0.3932262063026428,
      "epsilon_dpo/beta_margin_grad_std": 0.14714936912059784,
      "epsilon_dpo/beta_margin_mean": 0.48000258207321167,
      "epsilon_dpo/beta_margin_std": 0.6740255355834961,
      "epsilon_dpo/loss_margin_mean": 34.59055709838867,
      "grad_norm": 8.408370971679688,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.01406947709619999,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 3.089785553471233e-08,
      "logits/chosen": -0.0031261444091796875,
      "logits/rejected": -0.3361750841140747,
      "logps/chosen": -118.34280395507812,
      "logps/ref_chosen": -74.65908813476562,
      "logps/ref_rejected": -104.87959289550781,
      "logps/rejected": -183.15386962890625,
      "loss": 1.0661,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.6130213737487793,
      "rewards/margins": 0.48000258207321167,
      "rewards/rejected": -1.0930240154266357,
      "step": 567
    },
    {
      "epoch": 0.8586545729402872,
      "epsilon_dpo/beta": 0.013922218233346939,
      "epsilon_dpo/beta_margin_grad_mean": -0.3934234082698822,
      "epsilon_dpo/beta_margin_grad_std": 0.1543276011943817,
      "epsilon_dpo/beta_margin_mean": 0.4908979535102844,
      "epsilon_dpo/beta_margin_std": 0.7174976468086243,
      "epsilon_dpo/loss_margin_mean": 35.56275177001953,
      "grad_norm": 7.462749004364014,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.013990779407322407,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.026418409484513e-08,
      "logits/chosen": -0.2286403328180313,
      "logits/rejected": -0.3169150948524475,
      "logps/chosen": -112.17529296875,
      "logps/ref_chosen": -75.05364227294922,
      "logps/ref_rejected": -115.28018188476562,
      "logps/rejected": -187.96458435058594,
      "loss": 1.0702,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5188618302345276,
      "rewards/margins": 0.4908979535102844,
      "rewards/rejected": -1.009759783744812,
      "step": 568
    },
    {
      "epoch": 0.8601662887377173,
      "epsilon_dpo/beta": 0.013870356604456902,
      "epsilon_dpo/beta_margin_grad_mean": -0.4265102446079254,
      "epsilon_dpo/beta_margin_grad_std": 0.15814505517482758,
      "epsilon_dpo/beta_margin_mean": 0.3268107771873474,
      "epsilon_dpo/beta_margin_std": 0.7094203233718872,
      "epsilon_dpo/loss_margin_mean": 23.92220687866211,
      "grad_norm": 9.30414867401123,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.01392117328941822,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.963665913810451e-08,
      "logits/chosen": -0.32446902990341187,
      "logits/rejected": -0.19666746258735657,
      "logps/chosen": -131.11181640625,
      "logps/ref_chosen": -87.77325439453125,
      "logps/ref_rejected": -98.90798950195312,
      "logps/rejected": -166.16876220703125,
      "loss": 1.2032,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6034388542175293,
      "rewards/margins": 0.3268108069896698,
      "rewards/rejected": -0.9302496910095215,
      "step": 569
    },
    {
      "epoch": 0.8616780045351474,
      "epsilon_dpo/beta": 0.013783859089016914,
      "epsilon_dpo/beta_margin_grad_mean": -0.3789195418357849,
      "epsilon_dpo/beta_margin_grad_std": 0.14751386642456055,
      "epsilon_dpo/beta_margin_mean": 0.554857611656189,
      "epsilon_dpo/beta_margin_std": 0.6948018074035645,
      "epsilon_dpo/loss_margin_mean": 40.50212097167969,
      "grad_norm": 7.067422866821289,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.013869163580238819,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.9015298217712453e-08,
      "logits/chosen": -0.28197193145751953,
      "logits/rejected": -0.05522429943084717,
      "logps/chosen": -109.03466796875,
      "logps/ref_chosen": -74.21205139160156,
      "logps/ref_rejected": -107.4862060546875,
      "logps/rejected": -182.81094360351562,
      "loss": 1.0136,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.4814842939376831,
      "rewards/margins": 0.554857611656189,
      "rewards/rejected": -1.036341905593872,
      "step": 570
    },
    {
      "epoch": 0.8631897203325775,
      "epsilon_dpo/beta": 0.013724091462790966,
      "epsilon_dpo/beta_margin_grad_mean": -0.4350747764110565,
      "epsilon_dpo/beta_margin_grad_std": 0.1576703041791916,
      "epsilon_dpo/beta_margin_mean": 0.2830599248409271,
      "epsilon_dpo/beta_margin_std": 0.6972734332084656,
      "epsilon_dpo/loss_margin_mean": 20.96841049194336,
      "grad_norm": 9.307204246520996,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.013783019967377186,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.840011871446962e-08,
      "logits/chosen": -0.12085636705160141,
      "logits/rejected": -0.09424938261508942,
      "logps/chosen": -122.22715759277344,
      "logps/ref_chosen": -77.99315643310547,
      "logps/ref_rejected": -92.53704833984375,
      "logps/rejected": -157.73947143554688,
      "loss": 1.2378,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.6095610857009888,
      "rewards/margins": 0.2830599546432495,
      "rewards/rejected": -0.8926210403442383,
      "step": 571
    },
    {
      "epoch": 0.8647014361300076,
      "epsilon_dpo/beta": 0.013642866164445877,
      "epsilon_dpo/beta_margin_grad_mean": -0.41651660203933716,
      "epsilon_dpo/beta_margin_grad_std": 0.13157060742378235,
      "epsilon_dpo/beta_margin_mean": 0.3664814531803131,
      "epsilon_dpo/beta_margin_std": 0.597045361995697,
      "epsilon_dpo/loss_margin_mean": 27.090225219726562,
      "grad_norm": 8.313310623168945,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.013722982257604599,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 2.7791137836269158e-08,
      "logits/chosen": -0.28194281458854675,
      "logits/rejected": -0.2647858262062073,
      "logps/chosen": -144.88211059570312,
      "logps/ref_chosen": -98.92315673828125,
      "logps/ref_rejected": -86.06552124023438,
      "logps/rejected": -159.1147003173828,
      "loss": 1.1352,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.62903892993927,
      "rewards/margins": 0.3664814233779907,
      "rewards/rejected": -0.9955204129219055,
      "step": 572
    },
    {
      "epoch": 0.8662131519274376,
      "epsilon_dpo/beta": 0.013583657331764698,
      "epsilon_dpo/beta_margin_grad_mean": -0.4110637307167053,
      "epsilon_dpo/beta_margin_grad_std": 0.17743809521198273,
      "epsilon_dpo/beta_margin_mean": 0.42062875628471375,
      "epsilon_dpo/beta_margin_std": 0.8343303799629211,
      "epsilon_dpo/loss_margin_mean": 31.3470401763916,
      "grad_norm": 6.728772163391113,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.013641982339322567,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.718837261761528e-08,
      "logits/chosen": -0.49589985609054565,
      "logits/rejected": -0.6189365386962891,
      "logps/chosen": -141.02272033691406,
      "logps/ref_chosen": -94.72535705566406,
      "logps/ref_rejected": -111.88986206054688,
      "logps/rejected": -189.5342559814453,
      "loss": 1.1658,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6313270330429077,
      "rewards/margins": 0.42062878608703613,
      "rewards/rejected": -1.0519558191299438,
      "step": 573
    },
    {
      "epoch": 0.8677248677248677,
      "epsilon_dpo/beta": 0.013522343710064888,
      "epsilon_dpo/beta_margin_grad_mean": -0.3827129304409027,
      "epsilon_dpo/beta_margin_grad_std": 0.1437026411294937,
      "epsilon_dpo/beta_margin_mean": 0.5322164297103882,
      "epsilon_dpo/beta_margin_std": 0.659810483455658,
      "epsilon_dpo/loss_margin_mean": 39.66382598876953,
      "grad_norm": 7.087203502655029,
      "kl/avg_steps": 0.453125,
      "kl/beta": 0.013582559302449226,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.659183991914696e-08,
      "logits/chosen": -0.19681048393249512,
      "logits/rejected": -0.04592698812484741,
      "logps/chosen": -123.19125366210938,
      "logps/ref_chosen": -80.30760192871094,
      "logps/ref_rejected": -112.79890441894531,
      "logps/rejected": -195.3463897705078,
      "loss": 1.0216,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5813426971435547,
      "rewards/margins": 0.5322164297103882,
      "rewards/rejected": -1.1135591268539429,
      "step": 574
    },
    {
      "epoch": 0.8692365835222978,
      "epsilon_dpo/beta": 0.013476158492267132,
      "epsilon_dpo/beta_margin_grad_mean": -0.42741483449935913,
      "epsilon_dpo/beta_margin_grad_std": 0.15975911915302277,
      "epsilon_dpo/beta_margin_mean": 0.3417285084724426,
      "epsilon_dpo/beta_margin_std": 0.7441447973251343,
      "epsilon_dpo/loss_margin_mean": 25.704574584960938,
      "grad_norm": 6.972466945648193,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.013521290384232998,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.600155642716606e-08,
      "logits/chosen": -0.1062895804643631,
      "logits/rejected": -0.19543448090553284,
      "logps/chosen": -135.07362365722656,
      "logps/ref_chosen": -87.42359161376953,
      "logps/ref_rejected": -117.86439514160156,
      "logps/rejected": -191.218994140625,
      "loss": 1.199,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6450673341751099,
      "rewards/margins": 0.3417285084724426,
      "rewards/rejected": -0.9867959022521973,
      "step": 575
    },
    {
      "epoch": 0.8707482993197279,
      "epsilon_dpo/beta": 0.013392090797424316,
      "epsilon_dpo/beta_margin_grad_mean": -0.38101744651794434,
      "epsilon_dpo/beta_margin_grad_std": 0.1552283763885498,
      "epsilon_dpo/beta_margin_mean": 0.5403695702552795,
      "epsilon_dpo/beta_margin_std": 0.7085763216018677,
      "epsilon_dpo/loss_margin_mean": 40.633262634277344,
      "grad_norm": 8.821754455566406,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.013474970124661922,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.5417538653170754e-08,
      "logits/chosen": -0.09155446290969849,
      "logits/rejected": -0.23096878826618195,
      "logps/chosen": -109.19087219238281,
      "logps/ref_chosen": -73.79335021972656,
      "logps/ref_rejected": -111.63740539550781,
      "logps/rejected": -187.66818237304688,
      "loss": 1.0313,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.475741982460022,
      "rewards/margins": 0.5403695702552795,
      "rewards/rejected": -1.0161116123199463,
      "step": 576
    },
    {
      "epoch": 0.872260015117158,
      "epsilon_dpo/beta": 0.013346577063202858,
      "epsilon_dpo/beta_margin_grad_mean": -0.4340493381023407,
      "epsilon_dpo/beta_margin_grad_std": 0.15020710229873657,
      "epsilon_dpo/beta_margin_mean": 0.29871833324432373,
      "epsilon_dpo/beta_margin_std": 0.6770634055137634,
      "epsilon_dpo/loss_margin_mean": 22.716121673583984,
      "grad_norm": 8.71159553527832,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.013391274958848953,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.4839802933393607e-08,
      "logits/chosen": -0.28973859548568726,
      "logits/rejected": -0.2949914336204529,
      "logps/chosen": -130.9017333984375,
      "logps/ref_chosen": -86.85696411132812,
      "logps/ref_rejected": -90.04165649414062,
      "logps/rejected": -156.8025665283203,
      "loss": 1.2161,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5898460149765015,
      "rewards/margins": 0.29871833324432373,
      "rewards/rejected": -0.8885643482208252,
      "step": 577
    },
    {
      "epoch": 0.873771730914588,
      "epsilon_dpo/beta": 0.01332170981913805,
      "epsilon_dpo/beta_margin_grad_mean": -0.4408455193042755,
      "epsilon_dpo/beta_margin_grad_std": 0.14023886620998383,
      "epsilon_dpo/beta_margin_mean": 0.26660671830177307,
      "epsilon_dpo/beta_margin_std": 0.6218577027320862,
      "epsilon_dpo/loss_margin_mean": 20.344070434570312,
      "grad_norm": 7.295518398284912,
      "kl/avg_steps": 0.1875,
      "kl/beta": 0.013345399871468544,
      "kl/n_epsilon_steps": 0.40625,
      "kl/p_epsilon_steps": 0.59375,
      "learning_rate": 2.4268365428344733e-08,
      "logits/chosen": 0.05835941433906555,
      "logits/rejected": -0.23089148104190826,
      "logps/chosen": -125.45712280273438,
      "logps/ref_chosen": -80.96696472167969,
      "logps/ref_rejected": -90.08758544921875,
      "logps/rejected": -154.92181396484375,
      "loss": 1.228,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.5944623947143555,
      "rewards/margins": 0.26660671830177307,
      "rewards/rejected": -0.8610690832138062,
      "step": 578
    },
    {
      "epoch": 0.8752834467120182,
      "epsilon_dpo/beta": 0.013238495215773582,
      "epsilon_dpo/beta_margin_grad_mean": -0.396721214056015,
      "epsilon_dpo/beta_margin_grad_std": 0.14099986851215363,
      "epsilon_dpo/beta_margin_mean": 0.4538329839706421,
      "epsilon_dpo/beta_margin_std": 0.6490737795829773,
      "epsilon_dpo/loss_margin_mean": 34.52947235107422,
      "grad_norm": 7.298642635345459,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.01332042459398508,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.3703242122359357e-08,
      "logits/chosen": 0.07702606916427612,
      "logits/rejected": 0.09959688782691956,
      "logps/chosen": -129.31687927246094,
      "logps/ref_chosen": -80.56574249267578,
      "logps/ref_rejected": -92.1923828125,
      "logps/rejected": -175.47299194335938,
      "loss": 1.0793,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": -0.6462571620941162,
      "rewards/margins": 0.4538329839706421,
      "rewards/rejected": -1.1000901460647583,
      "step": 579
    },
    {
      "epoch": 0.8767951625094482,
      "epsilon_dpo/beta": 0.013181092217564583,
      "epsilon_dpo/beta_margin_grad_mean": -0.4260570704936981,
      "epsilon_dpo/beta_margin_grad_std": 0.15772272646427155,
      "epsilon_dpo/beta_margin_mean": 0.33158257603645325,
      "epsilon_dpo/beta_margin_std": 0.7125260233879089,
      "epsilon_dpo/loss_margin_mean": 25.510705947875977,
      "grad_norm": 9.164182662963867,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.013237688690423965,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.3144448823151392e-08,
      "logits/chosen": -0.31698185205459595,
      "logits/rejected": -0.5376484990119934,
      "logps/chosen": -122.604248046875,
      "logps/ref_chosen": -81.20346069335938,
      "logps/ref_rejected": -105.1043701171875,
      "logps/rejected": -172.015869140625,
      "loss": 1.1996,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.5469723343849182,
      "rewards/margins": 0.33158260583877563,
      "rewards/rejected": -0.8785548806190491,
      "step": 580
    },
    {
      "epoch": 0.8783068783068783,
      "epsilon_dpo/beta": 0.013111318461596966,
      "epsilon_dpo/beta_margin_grad_mean": -0.40876150131225586,
      "epsilon_dpo/beta_margin_grad_std": 0.1576615869998932,
      "epsilon_dpo/beta_margin_mean": 0.4062279760837555,
      "epsilon_dpo/beta_margin_std": 0.7067951560020447,
      "epsilon_dpo/loss_margin_mean": 31.322311401367188,
      "grad_norm": 8.796152114868164,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.013180025853216648,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 2.259200116137039e-08,
      "logits/chosen": -0.1651584953069687,
      "logits/rejected": -0.24735815823078156,
      "logps/chosen": -129.39312744140625,
      "logps/ref_chosen": -80.71034240722656,
      "logps/ref_rejected": -107.85765838623047,
      "logps/rejected": -187.86276245117188,
      "loss": 1.1368,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.6408957839012146,
      "rewards/margins": 0.4062279462814331,
      "rewards/rejected": -1.0471237897872925,
      "step": 581
    },
    {
      "epoch": 0.8798185941043084,
      "epsilon_dpo/beta": 0.013060449622571468,
      "epsilon_dpo/beta_margin_grad_mean": -0.4191682040691376,
      "epsilon_dpo/beta_margin_grad_std": 0.152615487575531,
      "epsilon_dpo/beta_margin_mean": 0.3628266453742981,
      "epsilon_dpo/beta_margin_std": 0.6817345023155212,
      "epsilon_dpo/loss_margin_mean": 28.12277603149414,
      "grad_norm": 8.426379203796387,
      "kl/avg_steps": 0.390625,
      "kl/beta": 0.01311037689447403,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.204591459016525e-08,
      "logits/chosen": -0.1529655158519745,
      "logits/rejected": -0.05211438238620758,
      "logps/chosen": -135.3836212158203,
      "logps/ref_chosen": -85.75233459472656,
      "logps/ref_rejected": -77.65898895263672,
      "logps/rejected": -155.41305541992188,
      "loss": 1.1641,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.6505056619644165,
      "rewards/margins": 0.3628266453742981,
      "rewards/rejected": -1.0133322477340698,
      "step": 582
    },
    {
      "epoch": 0.8813303099017384,
      "epsilon_dpo/beta": 0.013015775009989738,
      "epsilon_dpo/beta_margin_grad_mean": -0.43119898438453674,
      "epsilon_dpo/beta_margin_grad_std": 0.1673259735107422,
      "epsilon_dpo/beta_margin_mean": 0.31490734219551086,
      "epsilon_dpo/beta_margin_std": 0.7513457536697388,
      "epsilon_dpo/loss_margin_mean": 24.607948303222656,
      "grad_norm": 10.292269706726074,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.013059364631772041,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 2.1506204384751064e-08,
      "logits/chosen": -0.07927985489368439,
      "logits/rejected": -0.35749489068984985,
      "logps/chosen": -120.08602905273438,
      "logps/ref_chosen": -71.96379089355469,
      "logps/ref_rejected": -112.45161437988281,
      "logps/rejected": -185.18179321289062,
      "loss": 1.2273,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.6287330389022827,
      "rewards/margins": 0.3149073123931885,
      "rewards/rejected": -0.9436403512954712,
      "step": 583
    },
    {
      "epoch": 0.8828420256991686,
      "epsilon_dpo/beta": 0.012958982959389687,
      "epsilon_dpo/beta_margin_grad_mean": -0.41085541248321533,
      "epsilon_dpo/beta_margin_grad_std": 0.15736804902553558,
      "epsilon_dpo/beta_margin_mean": 0.40208297967910767,
      "epsilon_dpo/beta_margin_std": 0.7118176817893982,
      "epsilon_dpo/loss_margin_mean": 31.387554168701172,
      "grad_norm": 7.973084926605225,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.013014626689255238,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 2.09728856419826e-08,
      "logits/chosen": 0.2169530838727951,
      "logits/rejected": -0.1978251039981842,
      "logps/chosen": -100.49431610107422,
      "logps/ref_chosen": -65.38093566894531,
      "logps/ref_rejected": -107.81880187988281,
      "logps/rejected": -174.31973266601562,
      "loss": 1.1409,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.4563906788825989,
      "rewards/margins": 0.40208297967910767,
      "rewards/rejected": -0.8584736585617065,
      "step": 584
    },
    {
      "epoch": 0.8843537414965986,
      "epsilon_dpo/beta": 0.012898484244942665,
      "epsilon_dpo/beta_margin_grad_mean": -0.42835086584091187,
      "epsilon_dpo/beta_margin_grad_std": 0.14225082099437714,
      "epsilon_dpo/beta_margin_mean": 0.3104247748851776,
      "epsilon_dpo/beta_margin_std": 0.6384141445159912,
      "epsilon_dpo/loss_margin_mean": 24.375465393066406,
      "grad_norm": 7.945513725280762,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.012957935221493244,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.044597327993153e-08,
      "logits/chosen": -0.2803378701210022,
      "logits/rejected": -0.17878082394599915,
      "logps/chosen": -124.75691223144531,
      "logps/ref_chosen": -80.82450866699219,
      "logps/ref_rejected": -100.2001953125,
      "logps/rejected": -168.508056640625,
      "loss": 1.1949,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5681923627853394,
      "rewards/margins": 0.3104247450828552,
      "rewards/rejected": -0.8786171674728394,
      "step": 585
    },
    {
      "epoch": 0.8858654572940288,
      "epsilon_dpo/beta": 0.01282621268182993,
      "epsilon_dpo/beta_margin_grad_mean": -0.39538994431495667,
      "epsilon_dpo/beta_margin_grad_std": 0.1316211223602295,
      "epsilon_dpo/beta_margin_mean": 0.4575497508049011,
      "epsilon_dpo/beta_margin_std": 0.5963794589042664,
      "epsilon_dpo/loss_margin_mean": 35.92658233642578,
      "grad_norm": 8.383708000183105,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.01289747841656208,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 1.9925482037469187e-08,
      "logits/chosen": 0.13802039623260498,
      "logits/rejected": 0.03304041922092438,
      "logps/chosen": -113.84898376464844,
      "logps/ref_chosen": -74.60169982910156,
      "logps/ref_rejected": -90.7721939086914,
      "logps/rejected": -165.94606018066406,
      "loss": 1.062,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5045956969261169,
      "rewards/margins": 0.4575497508049011,
      "rewards/rejected": -0.9621454477310181,
      "step": 586
    },
    {
      "epoch": 0.8873771730914588,
      "epsilon_dpo/beta": 0.01277050282806158,
      "epsilon_dpo/beta_margin_grad_mean": -0.39474213123321533,
      "epsilon_dpo/beta_margin_grad_std": 0.16267690062522888,
      "epsilon_dpo/beta_margin_mean": 0.47096407413482666,
      "epsilon_dpo/beta_margin_std": 0.749113142490387,
      "epsilon_dpo/loss_margin_mean": 37.267826080322266,
      "grad_norm": 10.214089393615723,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.012825336307287216,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 1.9411426473854687e-08,
      "logits/chosen": -0.44531530141830444,
      "logits/rejected": -0.30305105447769165,
      "logps/chosen": -123.94343566894531,
      "logps/ref_chosen": -84.01087188720703,
      "logps/ref_rejected": -85.56326293945312,
      "logps/rejected": -162.76364135742188,
      "loss": 1.0979,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5123590230941772,
      "rewards/margins": 0.47096407413482666,
      "rewards/rejected": -0.9833230972290039,
      "step": 587
    },
    {
      "epoch": 0.8888888888888888,
      "epsilon_dpo/beta": 0.012702901847660542,
      "epsilon_dpo/beta_margin_grad_mean": -0.4087964594364166,
      "epsilon_dpo/beta_margin_grad_std": 0.14179039001464844,
      "epsilon_dpo/beta_margin_mean": 0.3948504626750946,
      "epsilon_dpo/beta_margin_std": 0.6233694553375244,
      "epsilon_dpo/loss_margin_mean": 31.392719268798828,
      "grad_norm": 9.330618858337402,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.012769469991326332,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.890382096832699e-08,
      "logits/chosen": -0.4431498050689697,
      "logits/rejected": -0.33868607878685,
      "logps/chosen": -128.64813232421875,
      "logps/ref_chosen": -83.79899597167969,
      "logps/ref_rejected": -105.11346435546875,
      "logps/rejected": -181.35531616210938,
      "loss": 1.1214,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.5720565319061279,
      "rewards/margins": 0.3948504328727722,
      "rewards/rejected": -0.9669069647789001,
      "step": 588
    },
    {
      "epoch": 0.890400604686319,
      "epsilon_dpo/beta": 0.012623865157365799,
      "epsilon_dpo/beta_margin_grad_mean": -0.404887318611145,
      "epsilon_dpo/beta_margin_grad_std": 0.13321241736412048,
      "epsilon_dpo/beta_margin_mean": 0.41894814372062683,
      "epsilon_dpo/beta_margin_std": 0.5990987420082092,
      "epsilon_dpo/loss_margin_mean": 33.42301940917969,
      "grad_norm": 9.14968204498291,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.012701990082859993,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 1.840267971970344e-08,
      "logits/chosen": -0.6417205929756165,
      "logits/rejected": -0.3048175871372223,
      "logps/chosen": -121.8003158569336,
      "logps/ref_chosen": -80.36790466308594,
      "logps/ref_rejected": -97.85157775878906,
      "logps/rejected": -172.70700073242188,
      "loss": 1.0935,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.5246583819389343,
      "rewards/margins": 0.41894811391830444,
      "rewards/rejected": -0.9436064958572388,
      "step": 589
    },
    {
      "epoch": 0.891912320483749,
      "epsilon_dpo/beta": 0.012557252310216427,
      "epsilon_dpo/beta_margin_grad_mean": -0.3977271616458893,
      "epsilon_dpo/beta_margin_grad_std": 0.13341881334781647,
      "epsilon_dpo/beta_margin_mean": 0.45191994309425354,
      "epsilon_dpo/beta_margin_std": 0.6039386987686157,
      "epsilon_dpo/loss_margin_mean": 36.24995803833008,
      "grad_norm": 7.618983268737793,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.012623095884919167,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.7908016745981856e-08,
      "logits/chosen": -0.39597877860069275,
      "logits/rejected": -0.14002303779125214,
      "logps/chosen": -138.19757080078125,
      "logps/ref_chosen": -93.33839416503906,
      "logps/ref_rejected": -107.83362579345703,
      "logps/rejected": -188.94276428222656,
      "loss": 1.0684,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5645281076431274,
      "rewards/margins": 0.4519199728965759,
      "rewards/rejected": -1.0164480209350586,
      "step": 590
    },
    {
      "epoch": 0.8934240362811792,
      "epsilon_dpo/beta": 0.012490932829678059,
      "epsilon_dpo/beta_margin_grad_mean": -0.4069448709487915,
      "epsilon_dpo/beta_margin_grad_std": 0.14455479383468628,
      "epsilon_dpo/beta_margin_mean": 0.41144993901252747,
      "epsilon_dpo/beta_margin_std": 0.6489174962043762,
      "epsilon_dpo/loss_margin_mean": 33.24910354614258,
      "grad_norm": 9.191134452819824,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.012556389905512333,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.7419845883949098e-08,
      "logits/chosen": -0.2569578289985657,
      "logits/rejected": -0.37240132689476013,
      "logps/chosen": -116.85206604003906,
      "logps/ref_chosen": -81.7674560546875,
      "logps/ref_rejected": -107.47393798828125,
      "logps/rejected": -175.80764770507812,
      "loss": 1.1141,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4404708743095398,
      "rewards/margins": 0.41144996881484985,
      "rewards/rejected": -0.8519208431243896,
      "step": 591
    },
    {
      "epoch": 0.8949357520786092,
      "epsilon_dpo/beta": 0.012424926273524761,
      "epsilon_dpo/beta_margin_grad_mean": -0.41350051760673523,
      "epsilon_dpo/beta_margin_grad_std": 0.1312229186296463,
      "epsilon_dpo/beta_margin_mean": 0.3863595724105835,
      "epsilon_dpo/beta_margin_std": 0.600444495677948,
      "epsilon_dpo/loss_margin_mean": 31.340322494506836,
      "grad_norm": 7.968548774719238,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.012490036897361279,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.6938180788793556e-08,
      "logits/chosen": -0.11009342223405838,
      "logits/rejected": -0.2954660654067993,
      "logps/chosen": -112.07117462158203,
      "logps/ref_chosen": -72.5201416015625,
      "logps/ref_rejected": -109.68800354003906,
      "logps/rejected": -180.57936096191406,
      "loss": 1.1191,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.49197059869766235,
      "rewards/margins": 0.3863595724105835,
      "rewards/rejected": -0.8783301711082458,
      "step": 592
    },
    {
      "epoch": 0.8964474678760394,
      "epsilon_dpo/beta": 0.012367033399641514,
      "epsilon_dpo/beta_margin_grad_mean": -0.41395166516304016,
      "epsilon_dpo/beta_margin_grad_std": 0.14591091871261597,
      "epsilon_dpo/beta_margin_mean": 0.38128843903541565,
      "epsilon_dpo/beta_margin_std": 0.65277498960495,
      "epsilon_dpo/loss_margin_mean": 31.176050186157227,
      "grad_norm": 6.928558826446533,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.01242403406649828,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.6463034933723336e-08,
      "logits/chosen": 0.08199097216129303,
      "logits/rejected": -0.12141124159097672,
      "logps/chosen": -98.86445617675781,
      "logps/ref_chosen": -62.098533630371094,
      "logps/ref_rejected": -90.62057495117188,
      "logps/rejected": -158.56256103515625,
      "loss": 1.1399,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.4569452404975891,
      "rewards/margins": 0.38128843903541565,
      "rewards/rejected": -0.8382337093353271,
      "step": 593
    },
    {
      "epoch": 0.8979591836734694,
      "epsilon_dpo/beta": 0.01231706328690052,
      "epsilon_dpo/beta_margin_grad_mean": -0.42870694398880005,
      "epsilon_dpo/beta_margin_grad_std": 0.13863804936408997,
      "epsilon_dpo/beta_margin_mean": 0.314374178647995,
      "epsilon_dpo/beta_margin_std": 0.6136248111724854,
      "epsilon_dpo/loss_margin_mean": 25.851320266723633,
      "grad_norm": 8.039793014526367,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.012366068549454212,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 1.5994421609589385e-08,
      "logits/chosen": -0.09637521952390671,
      "logits/rejected": -0.3282632827758789,
      "logps/chosen": -136.2696990966797,
      "logps/ref_chosen": -89.39488220214844,
      "logps/ref_rejected": -96.3311767578125,
      "logps/rejected": -169.05731201171875,
      "loss": 1.1847,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.5789014101028442,
      "rewards/margins": 0.314374178647995,
      "rewards/rejected": -0.8932756185531616,
      "step": 594
    },
    {
      "epoch": 0.8994708994708994,
      "epsilon_dpo/beta": 0.012251830659806728,
      "epsilon_dpo/beta_margin_grad_mean": -0.3980482816696167,
      "epsilon_dpo/beta_margin_grad_std": 0.1583113670349121,
      "epsilon_dpo/beta_margin_mean": 0.4468056857585907,
      "epsilon_dpo/beta_margin_std": 0.7155683636665344,
      "epsilon_dpo/loss_margin_mean": 36.836448669433594,
      "grad_norm": 7.929884910583496,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.01231603417545557,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.553235392451377e-08,
      "logits/chosen": 0.14382167160511017,
      "logits/rejected": -0.2625676095485687,
      "logps/chosen": -119.94728088378906,
      "logps/ref_chosen": -76.57876586914062,
      "logps/ref_rejected": -110.7374267578125,
      "logps/rejected": -190.9423828125,
      "loss": 1.1072,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.5333274602890015,
      "rewards/margins": 0.4468056857585907,
      "rewards/rejected": -0.9801331758499146,
      "step": 595
    },
    {
      "epoch": 0.9009826152683296,
      "epsilon_dpo/beta": 0.01225217618048191,
      "epsilon_dpo/beta_margin_grad_mean": -0.4713704288005829,
      "epsilon_dpo/beta_margin_grad_std": 0.13841482996940613,
      "epsilon_dpo/beta_margin_mean": 0.13456059992313385,
      "epsilon_dpo/beta_margin_std": 0.6147952079772949,
      "epsilon_dpo/loss_margin_mean": 11.352704048156738,
      "grad_norm": 7.926502704620361,
      "kl/avg_steps": 0.0,
      "kl/beta": 0.012250951491296291,
      "kl/n_epsilon_steps": 0.5,
      "kl/p_epsilon_steps": 0.5,
      "learning_rate": 1.507684480352292e-08,
      "logits/chosen": -0.09347207844257355,
      "logits/rejected": -0.08081246167421341,
      "logps/chosen": -143.42486572265625,
      "logps/ref_chosen": -92.4140625,
      "logps/ref_rejected": -84.22576904296875,
      "logps/rejected": -146.5892791748047,
      "loss": 1.3453,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": -0.627160906791687,
      "rewards/margins": 0.13456059992313385,
      "rewards/rejected": -0.7617214918136597,
      "step": 596
    },
    {
      "epoch": 0.9024943310657596,
      "epsilon_dpo/beta": 0.012204297818243504,
      "epsilon_dpo/beta_margin_grad_mean": -0.41553056240081787,
      "epsilon_dpo/beta_margin_grad_std": 0.13693609833717346,
      "epsilon_dpo/beta_margin_mean": 0.37261930108070374,
      "epsilon_dpo/beta_margin_std": 0.6076087951660156,
      "epsilon_dpo/loss_margin_mean": 30.870195388793945,
      "grad_norm": 7.076600551605225,
      "kl/avg_steps": 0.390625,
      "kl/beta": 0.012250951491296291,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.4627906988186111e-08,
      "logits/chosen": -0.1559733748435974,
      "logits/rejected": 0.09400632977485657,
      "logps/chosen": -108.35760498046875,
      "logps/ref_chosen": -71.6175537109375,
      "logps/ref_rejected": -76.79582214355469,
      "logps/rejected": -144.40606689453125,
      "loss": 1.1341,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.44950392842292786,
      "rewards/margins": 0.37261927127838135,
      "rewards/rejected": -0.8221231698989868,
      "step": 597
    },
    {
      "epoch": 0.9040060468631897,
      "epsilon_dpo/beta": 0.012158735655248165,
      "epsilon_dpo/beta_margin_grad_mean": -0.4356701672077179,
      "epsilon_dpo/beta_margin_grad_std": 0.13405224680900574,
      "epsilon_dpo/beta_margin_mean": 0.28293609619140625,
      "epsilon_dpo/beta_margin_std": 0.5979596972465515,
      "epsilon_dpo/loss_margin_mean": 23.588106155395508,
      "grad_norm": 7.478119850158691,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.012203281745314598,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.4185553036259095e-08,
      "logits/chosen": -0.34806156158447266,
      "logits/rejected": -0.42723479866981506,
      "logps/chosen": -133.13076782226562,
      "logps/ref_chosen": -79.51602172851562,
      "logps/ref_rejected": -104.75083923339844,
      "logps/rejected": -181.95370483398438,
      "loss": 1.2068,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.6541522741317749,
      "rewards/margins": 0.28293609619140625,
      "rewards/rejected": -0.9370884299278259,
      "step": 598
    },
    {
      "epoch": 0.9055177626606198,
      "epsilon_dpo/beta": 0.012111391872167587,
      "epsilon_dpo/beta_margin_grad_mean": -0.424478679895401,
      "epsilon_dpo/beta_margin_grad_std": 0.1403839886188507,
      "epsilon_dpo/beta_margin_mean": 0.3317543566226959,
      "epsilon_dpo/beta_margin_std": 0.6224579215049744,
      "epsilon_dpo/loss_margin_mean": 27.74344825744629,
      "grad_norm": 9.905296325683594,
      "kl/avg_steps": 0.390625,
      "kl/beta": 0.012157690711319447,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.3749795321332885e-08,
      "logits/chosen": -0.06255094707012177,
      "logits/rejected": -0.120047926902771,
      "logps/chosen": -126.98910522460938,
      "logps/ref_chosen": -77.9075927734375,
      "logps/ref_rejected": -96.17262268066406,
      "logps/rejected": -172.99758911132812,
      "loss": 1.1725,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5957086682319641,
      "rewards/margins": 0.3317543864250183,
      "rewards/rejected": -0.9274630546569824,
      "step": 599
    },
    {
      "epoch": 0.9070294784580499,
      "epsilon_dpo/beta": 0.01207374595105648,
      "epsilon_dpo/beta_margin_grad_mean": -0.4346500635147095,
      "epsilon_dpo/beta_margin_grad_std": 0.12690189480781555,
      "epsilon_dpo/beta_margin_mean": 0.28666380047798157,
      "epsilon_dpo/beta_margin_std": 0.5536522269248962,
      "epsilon_dpo/loss_margin_mean": 24.055171966552734,
      "grad_norm": 8.634313583374023,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.012110384181141853,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 1.3320646032487393e-08,
      "logits/chosen": -0.2699419856071472,
      "logits/rejected": -0.45036131143569946,
      "logps/chosen": -131.400634765625,
      "logps/ref_chosen": -85.35400390625,
      "logps/ref_rejected": -104.27677154541016,
      "logps/rejected": -174.37857055664062,
      "loss": 1.1925,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.5574197173118591,
      "rewards/margins": 0.28666380047798157,
      "rewards/rejected": -0.8440835475921631,
      "step": 600
    },
    {
      "epoch": 0.9070294784580499,
      "eval_epsilon_dpo/beta": 0.012021305970847607,
      "eval_epsilon_dpo/beta_margin_grad_mean": -0.4186500012874603,
      "eval_epsilon_dpo/beta_margin_grad_std": 0.13757552206516266,
      "eval_epsilon_dpo/beta_margin_mean": 0.35871562361717224,
      "eval_epsilon_dpo/beta_margin_std": 0.6142193675041199,
      "eval_epsilon_dpo/loss_margin_mean": 30.170278549194336,
      "eval_kl/n_epsilon_steps": 0.28169015049934387,
      "eval_kl/p_epsilon_steps": 0.7169894576072693,
      "eval_logits/chosen": -0.20889337360858917,
      "eval_logits/rejected": -0.34235018491744995,
      "eval_logps/chosen": -131.7626953125,
      "eval_logps/ref_chosen": -87.42715454101562,
      "eval_logps/ref_rejected": -104.23548889160156,
      "eval_logps/rejected": -178.7413330078125,
      "eval_loss": 0.5753073692321777,
      "eval_rewards/accuracies": 0.7235915660858154,
      "eval_rewards/chosen": -0.5348218083381653,
      "eval_rewards/margins": 0.35871562361717224,
      "eval_rewards/rejected": -0.8935373425483704,
      "eval_runtime": 47.731,
      "eval_samples_per_second": 48.25,
      "eval_steps_per_second": 1.508,
      "step": 600
    },
    {
      "epoch": 0.90854119425548,
      "epsilon_dpo/beta": 0.012013494968414307,
      "epsilon_dpo/beta_margin_grad_mean": -0.4087972640991211,
      "epsilon_dpo/beta_margin_grad_std": 0.14293596148490906,
      "epsilon_dpo/beta_margin_mean": 0.40643519163131714,
      "epsilon_dpo/beta_margin_std": 0.6445590853691101,
      "epsilon_dpo/loss_margin_mean": 34.13679504394531,
      "grad_norm": 7.5616302490234375,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.012072657234966755,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.2898117173950868e-08,
      "logits/chosen": -0.26879027485847473,
      "logits/rejected": -0.3316187262535095,
      "logps/chosen": -113.0506591796875,
      "logps/ref_chosen": -76.60881042480469,
      "logps/ref_rejected": -104.78749084472656,
      "logps/rejected": -175.3661346435547,
      "loss": 1.1165,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.43959861993789673,
      "rewards/margins": 0.40643519163131714,
      "rewards/rejected": -0.8460338115692139,
      "step": 601
    },
    {
      "epoch": 0.91005291005291,
      "epsilon_dpo/beta": 0.011964989826083183,
      "epsilon_dpo/beta_margin_grad_mean": -0.4060874879360199,
      "epsilon_dpo/beta_margin_grad_std": 0.1276295930147171,
      "epsilon_dpo/beta_margin_mean": 0.41427186131477356,
      "epsilon_dpo/beta_margin_std": 0.5662131309509277,
      "epsilon_dpo/loss_margin_mean": 34.92207717895508,
      "grad_norm": 12.094139099121094,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.012012594379484653,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 1.2482220564763667e-08,
      "logits/chosen": -0.035280682146549225,
      "logits/rejected": -0.3258655071258545,
      "logps/chosen": -114.74050903320312,
      "logps/ref_chosen": -82.1366958618164,
      "logps/ref_rejected": -95.36003875732422,
      "logps/rejected": -162.88592529296875,
      "loss": 1.0888,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.3918892741203308,
      "rewards/margins": 0.41427183151245117,
      "rewards/rejected": -0.8061611652374268,
      "step": 602
    },
    {
      "epoch": 0.9115646258503401,
      "epsilon_dpo/beta": 0.011905360966920853,
      "epsilon_dpo/beta_margin_grad_mean": -0.4135136306285858,
      "epsilon_dpo/beta_margin_grad_std": 0.13839876651763916,
      "epsilon_dpo/beta_margin_mean": 0.3756512701511383,
      "epsilon_dpo/beta_margin_std": 0.612551748752594,
      "epsilon_dpo/loss_margin_mean": 31.871030807495117,
      "grad_norm": 8.700664520263672,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.011963990516960621,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.2072967838448051e-08,
      "logits/chosen": -0.3890395164489746,
      "logits/rejected": -0.2390347272157669,
      "logps/chosen": -116.29725646972656,
      "logps/ref_chosen": -72.34424591064453,
      "logps/ref_rejected": -94.63508605957031,
      "logps/rejected": -170.45913696289062,
      "loss": 1.1334,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5251051783561707,
      "rewards/margins": 0.3756512403488159,
      "rewards/rejected": -0.9007564783096313,
      "step": 603
    },
    {
      "epoch": 0.9130763416477702,
      "epsilon_dpo/beta": 0.011849851347506046,
      "epsilon_dpo/beta_margin_grad_mean": -0.41967111825942993,
      "epsilon_dpo/beta_margin_grad_std": 0.14534930884838104,
      "epsilon_dpo/beta_margin_mean": 0.35719671845436096,
      "epsilon_dpo/beta_margin_std": 0.6550381183624268,
      "epsilon_dpo/loss_margin_mean": 30.46562957763672,
      "grad_norm": 6.926849842071533,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.011904468759894371,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.1670370442682459e-08,
      "logits/chosen": -0.06593503057956696,
      "logits/rejected": -0.2122724950313568,
      "logps/chosen": -130.89698791503906,
      "logps/ref_chosen": -96.72001647949219,
      "logps/ref_rejected": -91.02459716796875,
      "logps/rejected": -155.66720581054688,
      "loss": 1.1601,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4063175618648529,
      "rewards/margins": 0.35719674825668335,
      "rewards/rejected": -0.7635142803192139,
      "step": 604
    },
    {
      "epoch": 0.9145880574452003,
      "epsilon_dpo/beta": 0.011801970191299915,
      "epsilon_dpo/beta_margin_grad_mean": -0.4265349209308624,
      "epsilon_dpo/beta_margin_grad_std": 0.13410498201847076,
      "epsilon_dpo/beta_margin_mean": 0.3190124034881592,
      "epsilon_dpo/beta_margin_std": 0.5920915007591248,
      "epsilon_dpo/loss_margin_mean": 27.37565803527832,
      "grad_norm": 8.321650505065918,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.011848926544189453,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 1.1274439638981532e-08,
      "logits/chosen": -0.27900180220603943,
      "logits/rejected": -0.18813619017601013,
      "logps/chosen": -131.6309356689453,
      "logps/ref_chosen": -84.01524353027344,
      "logps/ref_rejected": -109.6864013671875,
      "logps/rejected": -184.67776489257812,
      "loss": 1.1748,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5635058283805847,
      "rewards/margins": 0.3190124034881592,
      "rewards/rejected": -0.8825182318687439,
      "step": 605
    },
    {
      "epoch": 0.9160997732426304,
      "epsilon_dpo/beta": 0.011743154376745224,
      "epsilon_dpo/beta_margin_grad_mean": -0.4185470938682556,
      "epsilon_dpo/beta_margin_grad_std": 0.11598649621009827,
      "epsilon_dpo/beta_margin_mean": 0.35316967964172363,
      "epsilon_dpo/beta_margin_std": 0.5112137198448181,
      "epsilon_dpo/loss_margin_mean": 30.316177368164062,
      "grad_norm": 7.579667091369629,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.011800984852015972,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.0885186502381016e-08,
      "logits/chosen": -0.10875275731086731,
      "logits/rejected": -0.19334736466407776,
      "logps/chosen": -115.47592163085938,
      "logps/ref_chosen": -75.84971618652344,
      "logps/ref_rejected": -102.4177017211914,
      "logps/rejected": -172.36007690429688,
      "loss": 1.1251,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.46557578444480896,
      "rewards/margins": 0.35316967964172363,
      "rewards/rejected": -0.8187454342842102,
      "step": 606
    },
    {
      "epoch": 0.9176114890400605,
      "epsilon_dpo/beta": 0.011677390895783901,
      "epsilon_dpo/beta_margin_grad_mean": -0.41183313727378845,
      "epsilon_dpo/beta_margin_grad_std": 0.12496085464954376,
      "epsilon_dpo/beta_margin_mean": 0.380126029253006,
      "epsilon_dpo/beta_margin_std": 0.5489794015884399,
      "epsilon_dpo/loss_margin_mean": 32.82114791870117,
      "grad_norm": 8.1083984375,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.011742273345589638,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 1.0502621921127774e-08,
      "logits/chosen": -0.30899912118911743,
      "logits/rejected": -0.4363841414451599,
      "logps/chosen": -133.69430541992188,
      "logps/ref_chosen": -87.947998046875,
      "logps/ref_rejected": -99.3750991821289,
      "logps/rejected": -177.94256591796875,
      "loss": 1.1127,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5348155498504639,
      "rewards/margins": 0.380126029253006,
      "rewards/rejected": -0.9149416089057922,
      "step": 607
    },
    {
      "epoch": 0.9191232048374905,
      "epsilon_dpo/beta": 0.011623021215200424,
      "epsilon_dpo/beta_margin_grad_mean": -0.4458061158657074,
      "epsilon_dpo/beta_margin_grad_std": 0.1285128891468048,
      "epsilon_dpo/beta_margin_mean": 0.2374372035264969,
      "epsilon_dpo/beta_margin_std": 0.5633123517036438,
      "epsilon_dpo/loss_margin_mean": 20.721681594848633,
      "grad_norm": 8.346819877624512,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.011676592752337456,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.0126756596375685e-08,
      "logits/chosen": -0.0955449789762497,
      "logits/rejected": -0.5488607883453369,
      "logps/chosen": -138.93194580078125,
      "logps/ref_chosen": -88.26110076904297,
      "logps/ref_rejected": -116.95655822753906,
      "logps/rejected": -188.34909057617188,
      "loss": 1.2379,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.590623676776886,
      "rewards/margins": 0.2374371886253357,
      "rewards/rejected": -0.8280608654022217,
      "step": 608
    },
    {
      "epoch": 0.9206349206349206,
      "epsilon_dpo/beta": 0.011557895690202713,
      "epsilon_dpo/beta_margin_grad_mean": -0.4109853506088257,
      "epsilon_dpo/beta_margin_grad_std": 0.11524208635091782,
      "epsilon_dpo/beta_margin_mean": 0.3793385326862335,
      "epsilon_dpo/beta_margin_std": 0.5043821930885315,
      "epsilon_dpo/loss_margin_mean": 33.06473159790039,
      "grad_norm": 8.368597030639648,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.011622114107012749,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 9.757601041885694e-09,
      "logits/chosen": -0.08982004970312119,
      "logits/rejected": -0.21673482656478882,
      "logps/chosen": -110.45091247558594,
      "logps/ref_chosen": -72.53919982910156,
      "logps/ref_rejected": -86.69680786132812,
      "logps/rejected": -157.67324829101562,
      "loss": 1.1025,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.4392332434654236,
      "rewards/margins": 0.3793385326862335,
      "rewards/rejected": -0.8185718059539795,
      "step": 609
    },
    {
      "epoch": 0.9221466364323507,
      "epsilon_dpo/beta": 0.011511306278407574,
      "epsilon_dpo/beta_margin_grad_mean": -0.42309725284576416,
      "epsilon_dpo/beta_margin_grad_std": 0.15262646973133087,
      "epsilon_dpo/beta_margin_mean": 0.34350430965423584,
      "epsilon_dpo/beta_margin_std": 0.6798610091209412,
      "epsilon_dpo/loss_margin_mean": 30.249845504760742,
      "grad_norm": 11.12037181854248,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.011557105928659439,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 9.395165583732379e-09,
      "logits/chosen": -0.2149120271205902,
      "logits/rejected": -0.40282920002937317,
      "logps/chosen": -153.38609313964844,
      "logps/ref_chosen": -107.02760314941406,
      "logps/ref_rejected": -110.5920181274414,
      "logps/rejected": -187.20034790039062,
      "loss": 1.1798,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.5368313789367676,
      "rewards/margins": 0.34350430965423584,
      "rewards/rejected": -0.8803356885910034,
      "step": 610
    },
    {
      "epoch": 0.9236583522297808,
      "epsilon_dpo/beta": 0.011471925303339958,
      "epsilon_dpo/beta_margin_grad_mean": -0.4368804097175598,
      "epsilon_dpo/beta_margin_grad_std": 0.11818581819534302,
      "epsilon_dpo/beta_margin_mean": 0.27571043372154236,
      "epsilon_dpo/beta_margin_std": 0.5203703045845032,
      "epsilon_dpo/loss_margin_mean": 24.31031608581543,
      "grad_norm": 8.035173416137695,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.01151034515351057,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 9.03946036001449e-09,
      "logits/chosen": -0.575390636920929,
      "logits/rejected": -0.44690048694610596,
      "logps/chosen": -112.02863311767578,
      "logps/ref_chosen": -70.45919036865234,
      "logps/ref_rejected": -97.04055786132812,
      "logps/rejected": -162.92031860351562,
      "loss": 1.1931,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.4776163399219513,
      "rewards/margins": 0.27571046352386475,
      "rewards/rejected": -0.7533267736434937,
      "step": 611
    },
    {
      "epoch": 0.9251700680272109,
      "epsilon_dpo/beta": 0.011414701119065285,
      "epsilon_dpo/beta_margin_grad_mean": -0.4051789343357086,
      "epsilon_dpo/beta_margin_grad_std": 0.1485532522201538,
      "epsilon_dpo/beta_margin_mean": 0.4205136001110077,
      "epsilon_dpo/beta_margin_std": 0.6696256995201111,
      "epsilon_dpo/loss_margin_mean": 37.19691848754883,
      "grad_norm": 6.956040382385254,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.011470913887023926,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 8.690495320571839e-09,
      "logits/chosen": -0.20279085636138916,
      "logits/rejected": -0.087235227227211,
      "logps/chosen": -128.58370971679688,
      "logps/ref_chosen": -83.53645324707031,
      "logps/ref_rejected": -112.86213684082031,
      "logps/rejected": -195.10629272460938,
      "loss": 1.113,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5160987377166748,
      "rewards/margins": 0.4205136001110077,
      "rewards/rejected": -0.9366123676300049,
      "step": 612
    },
    {
      "epoch": 0.926681783824641,
      "epsilon_dpo/beta": 0.011357910931110382,
      "epsilon_dpo/beta_margin_grad_mean": -0.3938443064689636,
      "epsilon_dpo/beta_margin_grad_std": 0.14389140903949738,
      "epsilon_dpo/beta_margin_mean": 0.4673750102519989,
      "epsilon_dpo/beta_margin_std": 0.6484513878822327,
      "epsilon_dpo/loss_margin_mean": 41.5140266418457,
      "grad_norm": 7.930556774139404,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.011413844302296638,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 8.348280226706722e-09,
      "logits/chosen": -0.13990065455436707,
      "logits/rejected": 0.2776636481285095,
      "logps/chosen": -111.08456420898438,
      "logps/ref_chosen": -77.24560546875,
      "logps/ref_rejected": -84.56202697753906,
      "logps/rejected": -159.91500854492188,
      "loss": 1.0697,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.38656705617904663,
      "rewards/margins": 0.4673749804496765,
      "rewards/rejected": -0.8539420366287231,
      "step": 613
    },
    {
      "epoch": 0.9281934996220711,
      "epsilon_dpo/beta": 0.011297854594886303,
      "epsilon_dpo/beta_margin_grad_mean": -0.41276368498802185,
      "epsilon_dpo/beta_margin_grad_std": 0.12965954840183258,
      "epsilon_dpo/beta_margin_mean": 0.3738853335380554,
      "epsilon_dpo/beta_margin_std": 0.5611046552658081,
      "epsilon_dpo/loss_margin_mean": 33.407039642333984,
      "grad_norm": 8.850772857666016,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.011357058770954609,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 8.012824650910937e-09,
      "logits/chosen": 0.023061083629727364,
      "logits/rejected": 0.16475391387939453,
      "logps/chosen": -129.9523162841797,
      "logps/ref_chosen": -85.42471313476562,
      "logps/ref_rejected": -95.38154602050781,
      "logps/rejected": -173.3162078857422,
      "loss": 1.1217,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.5050544738769531,
      "rewards/margins": 0.3738853335380554,
      "rewards/rejected": -0.8789397478103638,
      "step": 614
    },
    {
      "epoch": 0.9297052154195011,
      "epsilon_dpo/beta": 0.011255805380642414,
      "epsilon_dpo/beta_margin_grad_mean": -0.4168703258037567,
      "epsilon_dpo/beta_margin_grad_std": 0.1420803964138031,
      "epsilon_dpo/beta_margin_mean": 0.3664174973964691,
      "epsilon_dpo/beta_margin_std": 0.6301956176757812,
      "epsilon_dpo/loss_margin_mean": 32.92771530151367,
      "grad_norm": 7.011085510253906,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.011297043412923813,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 7.684137976598088e-09,
      "logits/chosen": -0.35232728719711304,
      "logits/rejected": -0.32909321784973145,
      "logps/chosen": -147.59274291992188,
      "logps/ref_chosen": -98.17005157470703,
      "logps/ref_rejected": -123.09346008300781,
      "logps/rejected": -205.44387817382812,
      "loss": 1.1459,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.5581832528114319,
      "rewards/margins": 0.3664174973964691,
      "rewards/rejected": -0.9246007800102234,
      "step": 615
    },
    {
      "epoch": 0.9312169312169312,
      "epsilon_dpo/beta": 0.011213753372430801,
      "epsilon_dpo/beta_margin_grad_mean": -0.42112424969673157,
      "epsilon_dpo/beta_margin_grad_std": 0.13007456064224243,
      "epsilon_dpo/beta_margin_mean": 0.34851300716400146,
      "epsilon_dpo/beta_margin_std": 0.5745180249214172,
      "epsilon_dpo/loss_margin_mean": 31.411874771118164,
      "grad_norm": 8.707462310791016,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.011254837736487389,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 7.36222939784098e-09,
      "logits/chosen": -0.14188328385353088,
      "logits/rejected": -0.38138338923454285,
      "logps/chosen": -123.18547058105469,
      "logps/ref_chosen": -78.42754364013672,
      "logps/ref_rejected": -102.52302551269531,
      "logps/rejected": -178.6928253173828,
      "loss": 1.145,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.5035969018936157,
      "rewards/margins": 0.34851300716400146,
      "rewards/rejected": -0.852109968662262,
      "step": 616
    },
    {
      "epoch": 0.9327286470143613,
      "epsilon_dpo/beta": 0.011143824085593224,
      "epsilon_dpo/beta_margin_grad_mean": -0.42190423607826233,
      "epsilon_dpo/beta_margin_grad_std": 0.10602878034114838,
      "epsilon_dpo/beta_margin_mean": 0.3308263123035431,
      "epsilon_dpo/beta_margin_std": 0.46238380670547485,
      "epsilon_dpo/loss_margin_mean": 29.89263153076172,
      "grad_norm": 8.340224266052246,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.011212789453566074,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 7.047107919114586e-09,
      "logits/chosen": 0.02730114758014679,
      "logits/rejected": 0.01917155086994171,
      "logps/chosen": -133.4224853515625,
      "logps/ref_chosen": -82.33126831054688,
      "logps/ref_rejected": -103.87673950195312,
      "logps/rejected": -184.86058044433594,
      "loss": 1.1331,
      "rewards/accuracies": 0.84375,
      "rewards/chosen": -0.5706514120101929,
      "rewards/margins": 0.3308263123035431,
      "rewards/rejected": -0.9014778137207031,
      "step": 617
    },
    {
      "epoch": 0.9342403628117913,
      "epsilon_dpo/beta": 0.01110595092177391,
      "epsilon_dpo/beta_margin_grad_mean": -0.42988893389701843,
      "epsilon_dpo/beta_margin_grad_std": 0.13014353811740875,
      "epsilon_dpo/beta_margin_mean": 0.3075876832008362,
      "epsilon_dpo/beta_margin_std": 0.5768880844116211,
      "epsilon_dpo/loss_margin_mean": 28.036251068115234,
      "grad_norm": 9.315959930419922,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.011143145151436329,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 6.738782355044048e-09,
      "logits/chosen": -0.13639292120933533,
      "logits/rejected": -0.4161806106567383,
      "logps/chosen": -121.63025665283203,
      "logps/ref_chosen": -79.3497085571289,
      "logps/ref_rejected": -116.9796142578125,
      "logps/rejected": -187.29641723632812,
      "loss": 1.1801,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.4715430438518524,
      "rewards/margins": 0.3075876832008362,
      "rewards/rejected": -0.779130756855011,
      "step": 618
    },
    {
      "epoch": 0.9357520786092215,
      "epsilon_dpo/beta": 0.011060964316129684,
      "epsilon_dpo/beta_margin_grad_mean": -0.412675142288208,
      "epsilon_dpo/beta_margin_grad_std": 0.1362818479537964,
      "epsilon_dpo/beta_margin_mean": 0.382579505443573,
      "epsilon_dpo/beta_margin_std": 0.6046925783157349,
      "epsilon_dpo/loss_margin_mean": 34.932315826416016,
      "grad_norm": 6.323122501373291,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.01110497210174799,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 6.437261330158206e-09,
      "logits/chosen": -0.009692233055830002,
      "logits/rejected": -0.21985238790512085,
      "logps/chosen": -120.16893005371094,
      "logps/ref_chosen": -75.64820861816406,
      "logps/ref_rejected": -106.817138671875,
      "logps/rejected": -186.27017211914062,
      "loss": 1.1253,
      "rewards/accuracies": 0.734375,
      "rewards/chosen": -0.4942760467529297,
      "rewards/margins": 0.3825794756412506,
      "rewards/rejected": -0.8768554925918579,
      "step": 619
    },
    {
      "epoch": 0.9372637944066515,
      "epsilon_dpo/beta": 0.011012754403054714,
      "epsilon_dpo/beta_margin_grad_mean": -0.4311310350894928,
      "epsilon_dpo/beta_margin_grad_std": 0.12295855581760406,
      "epsilon_dpo/beta_margin_mean": 0.2955702841281891,
      "epsilon_dpo/beta_margin_std": 0.5336819291114807,
      "epsilon_dpo/loss_margin_mean": 27.1448917388916,
      "grad_norm": 7.733489990234375,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.011060040444135666,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 6.142553278648238e-09,
      "logits/chosen": -0.19426310062408447,
      "logits/rejected": -0.00829014927148819,
      "logps/chosen": -125.66661071777344,
      "logps/ref_chosen": -82.22972106933594,
      "logps/ref_rejected": -85.6536865234375,
      "logps/rejected": -156.2354736328125,
      "loss": 1.1801,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.4805551767349243,
      "rewards/margins": 0.2955702841281891,
      "rewards/rejected": -0.776125431060791,
      "step": 620
    },
    {
      "epoch": 0.9387755102040817,
      "epsilon_dpo/beta": 0.010961341671645641,
      "epsilon_dpo/beta_margin_grad_mean": -0.4333516061306,
      "epsilon_dpo/beta_margin_grad_std": 0.11945909261703491,
      "epsilon_dpo/beta_margin_mean": 0.2852727174758911,
      "epsilon_dpo/beta_margin_std": 0.5157901048660278,
      "epsilon_dpo/loss_margin_mean": 26.318817138671875,
      "grad_norm": 7.324526786804199,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.011011863127350807,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 5.854666444131934e-09,
      "logits/chosen": -0.1203552633523941,
      "logits/rejected": -0.34329789876937866,
      "logps/chosen": -121.84413146972656,
      "logps/ref_chosen": -75.2481460571289,
      "logps/ref_rejected": -113.55223083496094,
      "logps/rejected": -186.467041015625,
      "loss": 1.1846,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.5119953155517578,
      "rewards/margins": 0.2852727174758911,
      "rewards/rejected": -0.7972680330276489,
      "step": 621
    },
    {
      "epoch": 0.9402872260015117,
      "epsilon_dpo/beta": 0.010906773619353771,
      "epsilon_dpo/beta_margin_grad_mean": -0.42391106486320496,
      "epsilon_dpo/beta_margin_grad_std": 0.13388563692569733,
      "epsilon_dpo/beta_margin_mean": 0.32644522190093994,
      "epsilon_dpo/beta_margin_std": 0.5833149552345276,
      "epsilon_dpo/loss_margin_mean": 30.275888442993164,
      "grad_norm": 10.996826171875,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.01096048578619957,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 5.573608879422875e-09,
      "logits/chosen": -0.2559853494167328,
      "logits/rejected": -0.31877291202545166,
      "logps/chosen": -130.95132446289062,
      "logps/ref_chosen": -85.68588256835938,
      "logps/ref_rejected": -102.61013793945312,
      "logps/rejected": -178.15147399902344,
      "loss": 1.1669,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.49637672305107117,
      "rewards/margins": 0.32644522190093994,
      "rewards/rejected": -0.8228219747543335,
      "step": 622
    },
    {
      "epoch": 0.9417989417989417,
      "epsilon_dpo/beta": 0.010855920612812042,
      "epsilon_dpo/beta_margin_grad_mean": -0.42936843633651733,
      "epsilon_dpo/beta_margin_grad_std": 0.11718784272670746,
      "epsilon_dpo/beta_margin_mean": 0.3037678897380829,
      "epsilon_dpo/beta_margin_std": 0.5088984370231628,
      "epsilon_dpo/loss_margin_mean": 28.262224197387695,
      "grad_norm": 7.055470943450928,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.010905956849455833,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 5.299388446305342e-09,
      "logits/chosen": -0.12515604496002197,
      "logits/rejected": -0.36976397037506104,
      "logps/chosen": -140.57472229003906,
      "logps/ref_chosen": -88.4764404296875,
      "logps/ref_rejected": -115.94358825683594,
      "logps/rejected": -196.30409240722656,
      "loss": 1.1668,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.5674319863319397,
      "rewards/margins": 0.3037678897380829,
      "rewards/rejected": -0.8711998462677002,
      "step": 623
    },
    {
      "epoch": 0.9433106575963719,
      "epsilon_dpo/beta": 0.010778130032122135,
      "epsilon_dpo/beta_margin_grad_mean": -0.3959210216999054,
      "epsilon_dpo/beta_margin_grad_std": 0.11694164574146271,
      "epsilon_dpo/beta_margin_mean": 0.4492229223251343,
      "epsilon_dpo/beta_margin_std": 0.5222044587135315,
      "epsilon_dpo/loss_margin_mean": 41.864505767822266,
      "grad_norm": 7.092720031738281,
      "kl/avg_steps": 0.71875,
      "kl/beta": 0.010855073109269142,
      "kl/n_epsilon_steps": 0.140625,
      "kl/p_epsilon_steps": 0.859375,
      "learning_rate": 5.03201281531429e-09,
      "logits/chosen": 0.03635944426059723,
      "logits/rejected": -0.2513130307197571,
      "logps/chosen": -106.5604248046875,
      "logps/ref_chosen": -70.26908874511719,
      "logps/ref_rejected": -98.937255859375,
      "logps/rejected": -177.09310913085938,
      "loss": 1.05,
      "rewards/accuracies": 0.84375,
      "rewards/chosen": -0.3919287323951721,
      "rewards/margins": 0.4492229223251343,
      "rewards/rejected": -0.8411516547203064,
      "step": 624
    },
    {
      "epoch": 0.9448223733938019,
      "epsilon_dpo/beta": 0.010751740075647831,
      "epsilon_dpo/beta_margin_grad_mean": -0.4558924436569214,
      "epsilon_dpo/beta_margin_grad_std": 0.12500344216823578,
      "epsilon_dpo/beta_margin_mean": 0.19076864421367645,
      "epsilon_dpo/beta_margin_std": 0.5383822917938232,
      "epsilon_dpo/loss_margin_mean": 18.10923194885254,
      "grad_norm": 9.223965644836426,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.010777609422802925,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 4.7714894655209174e-09,
      "logits/chosen": 0.09123232215642929,
      "logits/rejected": -0.23934797942638397,
      "logps/chosen": -128.52659606933594,
      "logps/ref_chosen": -79.21810913085938,
      "logps/ref_rejected": -113.8106689453125,
      "logps/rejected": -181.2283935546875,
      "loss": 1.274,
      "rewards/accuracies": 0.609375,
      "rewards/chosen": -0.5313090085983276,
      "rewards/margins": 0.19076864421367645,
      "rewards/rejected": -0.7220776677131653,
      "step": 625
    },
    {
      "epoch": 0.9463340891912321,
      "epsilon_dpo/beta": 0.0107182078063488,
      "epsilon_dpo/beta_margin_grad_mean": -0.4198099672794342,
      "epsilon_dpo/beta_margin_grad_std": 0.14759738743305206,
      "epsilon_dpo/beta_margin_mean": 0.35964328050613403,
      "epsilon_dpo/beta_margin_std": 0.6498256921768188,
      "epsilon_dpo/loss_margin_mean": 34.002769470214844,
      "grad_norm": 6.630099296569824,
      "kl/avg_steps": 0.3125,
      "kl/beta": 0.010750732384622097,
      "kl/n_epsilon_steps": 0.34375,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 4.517825684323323e-09,
      "logits/chosen": -0.03878095746040344,
      "logits/rejected": -0.2872091829776764,
      "logps/chosen": -108.78828430175781,
      "logps/ref_chosen": -66.3774642944336,
      "logps/ref_rejected": -118.52725219726562,
      "logps/rejected": -194.94082641601562,
      "loss": 1.1576,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.45677587389945984,
      "rewards/margins": 0.35964328050613403,
      "rewards/rejected": -0.8164191246032715,
      "step": 626
    },
    {
      "epoch": 0.9478458049886621,
      "epsilon_dpo/beta": 0.01065802201628685,
      "epsilon_dpo/beta_margin_grad_mean": -0.40233418345451355,
      "epsilon_dpo/beta_margin_grad_std": 0.12860505282878876,
      "epsilon_dpo/beta_margin_mean": 0.4263233542442322,
      "epsilon_dpo/beta_margin_std": 0.5683040022850037,
      "epsilon_dpo/loss_margin_mean": 40.29022216796875,
      "grad_norm": 6.485427379608154,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.01071724109351635,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 4.271028567242818e-09,
      "logits/chosen": -0.12933257222175598,
      "logits/rejected": -0.4183410406112671,
      "logps/chosen": -127.08903503417969,
      "logps/ref_chosen": -82.48866271972656,
      "logps/ref_rejected": -122.58235168457031,
      "logps/rejected": -207.4729461669922,
      "loss": 1.0801,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.47629666328430176,
      "rewards/margins": 0.4263233542442322,
      "rewards/rejected": -0.9026199579238892,
      "step": 627
    },
    {
      "epoch": 0.9493575207860923,
      "epsilon_dpo/beta": 0.010608398355543613,
      "epsilon_dpo/beta_margin_grad_mean": -0.40769657492637634,
      "epsilon_dpo/beta_margin_grad_std": 0.15037523210048676,
      "epsilon_dpo/beta_margin_mean": 0.4015233516693115,
      "epsilon_dpo/beta_margin_std": 0.6635918021202087,
      "epsilon_dpo/loss_margin_mean": 38.265560150146484,
      "grad_norm": 7.2923197746276855,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.0106572937220335,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.0311050177251895e-09,
      "logits/chosen": -0.4370567202568054,
      "logits/rejected": 0.030434168875217438,
      "logps/chosen": -116.41726684570312,
      "logps/ref_chosen": -75.95014190673828,
      "logps/ref_rejected": -99.55660247802734,
      "logps/rejected": -178.28929138183594,
      "loss": 1.128,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4314713478088379,
      "rewards/margins": 0.4015233516693115,
      "rewards/rejected": -0.8329946994781494,
      "step": 628
    },
    {
      "epoch": 0.9508692365835223,
      "epsilon_dpo/beta": 0.010562218725681305,
      "epsilon_dpo/beta_margin_grad_mean": -0.42028841376304626,
      "epsilon_dpo/beta_margin_grad_std": 0.1128428652882576,
      "epsilon_dpo/beta_margin_mean": 0.3450172245502472,
      "epsilon_dpo/beta_margin_std": 0.4945928454399109,
      "epsilon_dpo/loss_margin_mean": 32.94646453857422,
      "grad_norm": 8.863780975341797,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.01060757040977478,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 3.798061746947995e-09,
      "logits/chosen": -0.1133829802274704,
      "logits/rejected": -0.4618754982948303,
      "logps/chosen": -136.7560272216797,
      "logps/ref_chosen": -96.68063354492188,
      "logps/ref_rejected": -99.79010772705078,
      "logps/rejected": -172.81198120117188,
      "loss": 1.1281,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.4247984290122986,
      "rewards/margins": 0.3450172543525696,
      "rewards/rejected": -0.7698156833648682,
      "step": 629
    },
    {
      "epoch": 0.9523809523809523,
      "epsilon_dpo/beta": 0.01050630770623684,
      "epsilon_dpo/beta_margin_grad_mean": -0.4169251024723053,
      "epsilon_dpo/beta_margin_grad_std": 0.11476609110832214,
      "epsilon_dpo/beta_margin_mean": 0.3586493134498596,
      "epsilon_dpo/beta_margin_std": 0.5017598271369934,
      "epsilon_dpo/loss_margin_mean": 34.3900260925293,
      "grad_norm": 5.832938194274902,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.010561364702880383,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 3.5719052736323806e-09,
      "logits/chosen": -0.20752973854541779,
      "logits/rejected": -0.19195277988910675,
      "logps/chosen": -120.63751220703125,
      "logps/ref_chosen": -80.01641845703125,
      "logps/ref_rejected": -98.34365844726562,
      "logps/rejected": -173.35476684570312,
      "loss": 1.1187,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4271647334098816,
      "rewards/margins": 0.3586493134498596,
      "rewards/rejected": -0.7858140468597412,
      "step": 630
    },
    {
      "epoch": 0.9538926681783825,
      "epsilon_dpo/beta": 0.010454071685671806,
      "epsilon_dpo/beta_margin_grad_mean": -0.40833863615989685,
      "epsilon_dpo/beta_margin_grad_std": 0.13670286536216736,
      "epsilon_dpo/beta_margin_mean": 0.404376357793808,
      "epsilon_dpo/beta_margin_std": 0.6010540723800659,
      "epsilon_dpo/loss_margin_mean": 39.03636169433594,
      "grad_norm": 7.70844030380249,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.010505554266273975,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.352641923861144e-09,
      "logits/chosen": -0.5213108658790588,
      "logits/rejected": -0.5298713445663452,
      "logps/chosen": -124.63873291015625,
      "logps/ref_chosen": -84.44837951660156,
      "logps/ref_rejected": -122.06558227539062,
      "logps/rejected": -201.29229736328125,
      "loss": 1.1068,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.4222160577774048,
      "rewards/margins": 0.40437638759613037,
      "rewards/rejected": -0.8265924453735352,
      "step": 631
    },
    {
      "epoch": 0.9554043839758125,
      "epsilon_dpo/beta": 0.01040206104516983,
      "epsilon_dpo/beta_margin_grad_mean": -0.4239082932472229,
      "epsilon_dpo/beta_margin_grad_std": 0.12038099765777588,
      "epsilon_dpo/beta_margin_mean": 0.3268932104110718,
      "epsilon_dpo/beta_margin_std": 0.5195305347442627,
      "epsilon_dpo/loss_margin_mean": 31.721187591552734,
      "grad_norm": 7.052772521972656,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.01045328751206398,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 3.140277830901428e-09,
      "logits/chosen": -0.21344169974327087,
      "logits/rejected": -0.29157426953315735,
      "logps/chosen": -124.53128814697266,
      "logps/ref_chosen": -81.35279846191406,
      "logps/ref_rejected": -89.2701416015625,
      "logps/rejected": -164.16981506347656,
      "loss": 1.15,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.45100072026252747,
      "rewards/margins": 0.3268932104110718,
      "rewards/rejected": -0.7778939008712769,
      "step": 632
    },
    {
      "epoch": 0.9569160997732427,
      "epsilon_dpo/beta": 0.01034705899655819,
      "epsilon_dpo/beta_margin_grad_mean": -0.40773069858551025,
      "epsilon_dpo/beta_margin_grad_std": 0.12108438462018967,
      "epsilon_dpo/beta_margin_mean": 0.4030759036540985,
      "epsilon_dpo/beta_margin_std": 0.5361250638961792,
      "epsilon_dpo/loss_margin_mean": 39.24986267089844,
      "grad_norm": 7.865797996520996,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.010401281528174877,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 2.9348189350335007e-09,
      "logits/chosen": -0.14327946305274963,
      "logits/rejected": -0.15693995356559753,
      "logps/chosen": -105.8614501953125,
      "logps/ref_chosen": -73.86644744873047,
      "logps/ref_rejected": -92.09278106689453,
      "logps/rejected": -163.337646484375,
      "loss": 1.0902,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.3324025869369507,
      "rewards/margins": 0.4030759036540985,
      "rewards/rejected": -0.7354785203933716,
      "step": 633
    },
    {
      "epoch": 0.9584278155706727,
      "epsilon_dpo/beta": 0.010321483016014099,
      "epsilon_dpo/beta_margin_grad_mean": -0.4732949435710907,
      "epsilon_dpo/beta_margin_grad_std": 0.12383994460105896,
      "epsilon_dpo/beta_margin_mean": 0.11720895022153854,
      "epsilon_dpo/beta_margin_std": 0.532695472240448,
      "epsilon_dpo/loss_margin_mean": 11.713577270507812,
      "grad_norm": 7.904653549194336,
      "kl/avg_steps": 0.25,
      "kl/beta": 0.01034631673246622,
      "kl/n_epsilon_steps": 0.375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 2.736270983384276e-09,
      "logits/chosen": -0.1411525011062622,
      "logits/rejected": -0.4012359380722046,
      "logps/chosen": -134.27059936523438,
      "logps/ref_chosen": -84.89143371582031,
      "logps/ref_rejected": -81.0973892211914,
      "logps/rejected": -142.1901397705078,
      "loss": 1.3407,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": -0.5117350816726685,
      "rewards/margins": 0.11720894277095795,
      "rewards/rejected": -0.6289440393447876,
      "step": 634
    },
    {
      "epoch": 0.9599395313681028,
      "epsilon_dpo/beta": 0.010292517952620983,
      "epsilon_dpo/beta_margin_grad_mean": -0.44807276129722595,
      "epsilon_dpo/beta_margin_grad_std": 0.1326771378517151,
      "epsilon_dpo/beta_margin_mean": 0.22640717029571533,
      "epsilon_dpo/beta_margin_std": 0.5779455304145813,
      "epsilon_dpo/loss_margin_mean": 22.384584426879883,
      "grad_norm": 6.7423930168151855,
      "kl/avg_steps": 0.28125,
      "kl/beta": 0.010320515371859074,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.640625,
      "learning_rate": 2.5446395297668287e-09,
      "logits/chosen": -0.38996532559394836,
      "logits/rejected": -0.4668923318386078,
      "logps/chosen": -147.9334716796875,
      "logps/ref_chosen": -91.09562683105469,
      "logps/ref_rejected": -113.29243469238281,
      "logps/rejected": -192.51486206054688,
      "loss": 1.252,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.5871849656105042,
      "rewards/margins": 0.22640718519687653,
      "rewards/rejected": -0.8135921955108643,
      "step": 635
    },
    {
      "epoch": 0.9614512471655329,
      "epsilon_dpo/beta": 0.010228270664811134,
      "epsilon_dpo/beta_margin_grad_mean": -0.4065358638763428,
      "epsilon_dpo/beta_margin_grad_std": 0.11634913831949234,
      "epsilon_dpo/beta_margin_mean": 0.402926504611969,
      "epsilon_dpo/beta_margin_std": 0.5106287598609924,
      "epsilon_dpo/loss_margin_mean": 39.64336013793945,
      "grad_norm": 6.623480319976807,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.010291569866240025,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 2.359929934524829e-09,
      "logits/chosen": -0.20538179576396942,
      "logits/rejected": -0.4749613404273987,
      "logps/chosen": -111.59320068359375,
      "logps/ref_chosen": -73.47183990478516,
      "logps/ref_rejected": -103.75514221191406,
      "logps/rejected": -181.51986694335938,
      "loss": 1.0845,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -0.3905521333217621,
      "rewards/margins": 0.402926504611969,
      "rewards/rejected": -0.7934786081314087,
      "step": 636
    },
    {
      "epoch": 0.9629629629629629,
      "epsilon_dpo/beta": 0.010177526623010635,
      "epsilon_dpo/beta_margin_grad_mean": -0.4286331534385681,
      "epsilon_dpo/beta_margin_grad_std": 0.12078320980072021,
      "epsilon_dpo/beta_margin_mean": 0.3108992874622345,
      "epsilon_dpo/beta_margin_std": 0.5280004143714905,
      "epsilon_dpo/loss_margin_mean": 30.848106384277344,
      "grad_norm": 5.800489902496338,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.010227647610008717,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 2.1821473643827137e-09,
      "logits/chosen": -0.2502748966217041,
      "logits/rejected": -0.46591269969940186,
      "logps/chosen": -155.5239715576172,
      "logps/ref_chosen": -100.47197723388672,
      "logps/ref_rejected": -114.3409423828125,
      "logps/rejected": -200.24105834960938,
      "loss": 1.1651,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5617954730987549,
      "rewards/margins": 0.3108992874622345,
      "rewards/rejected": -0.8726947903633118,
      "step": 637
    },
    {
      "epoch": 0.9644746787603931,
      "epsilon_dpo/beta": 0.010130072012543678,
      "epsilon_dpo/beta_margin_grad_mean": -0.4317198693752289,
      "epsilon_dpo/beta_margin_grad_std": 0.11745569854974747,
      "epsilon_dpo/beta_margin_mean": 0.2932567894458771,
      "epsilon_dpo/beta_margin_std": 0.5088880658149719,
      "epsilon_dpo/loss_margin_mean": 29.258907318115234,
      "grad_norm": 7.930274486541748,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.01017676293849945,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 2.0112967923011646e-09,
      "logits/chosen": 0.013463929295539856,
      "logits/rejected": -0.2810739278793335,
      "logps/chosen": -139.33770751953125,
      "logps/ref_chosen": -90.44658660888672,
      "logps/ref_rejected": -111.11790466308594,
      "logps/rejected": -189.2679443359375,
      "loss": 1.1759,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.4958798885345459,
      "rewards/margins": 0.2932567894458771,
      "rewards/rejected": -0.7891367077827454,
      "step": 638
    },
    {
      "epoch": 0.9659863945578231,
      "epsilon_dpo/beta": 0.01009230688214302,
      "epsilon_dpo/beta_margin_grad_mean": -0.41252967715263367,
      "epsilon_dpo/beta_margin_grad_std": 0.12411058694124222,
      "epsilon_dpo/beta_margin_mean": 0.38042405247688293,
      "epsilon_dpo/beta_margin_std": 0.540318489074707,
      "epsilon_dpo/loss_margin_mean": 38.056053161621094,
      "grad_norm": 6.623051643371582,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.010129282251000404,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 1.847382997337943e-09,
      "logits/chosen": -0.16237635910511017,
      "logits/rejected": -0.4033578336238861,
      "logps/chosen": -113.76445007324219,
      "logps/ref_chosen": -74.6663818359375,
      "logps/ref_rejected": -102.23210144042969,
      "logps/rejected": -179.38623046875,
      "loss": 1.1104,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.3960699737071991,
      "rewards/margins": 0.3804240822792053,
      "rewards/rejected": -0.776494026184082,
      "step": 639
    },
    {
      "epoch": 0.9674981103552532,
      "epsilon_dpo/beta": 0.010041986592113972,
      "epsilon_dpo/beta_margin_grad_mean": -0.4485768973827362,
      "epsilon_dpo/beta_margin_grad_std": 0.12330426275730133,
      "epsilon_dpo/beta_margin_mean": 0.21434807777404785,
      "epsilon_dpo/beta_margin_std": 0.5284790396690369,
      "epsilon_dpo/loss_margin_mean": 21.6812801361084,
      "grad_norm": 6.14130163192749,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.010091439820826054,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.690410564514244e-09,
      "logits/chosen": -0.05122403800487518,
      "logits/rejected": -0.4697457551956177,
      "logps/chosen": -117.28038024902344,
      "logps/ref_chosen": -73.5439453125,
      "logps/ref_rejected": -96.87593078613281,
      "logps/rejected": -162.29364013671875,
      "loss": 1.2506,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.440321147441864,
      "rewards/margins": 0.21434809267520905,
      "rewards/rejected": -0.6546692252159119,
      "step": 640
    },
    {
      "epoch": 0.9690098261526833,
      "epsilon_dpo/beta": 0.009992025792598724,
      "epsilon_dpo/beta_margin_grad_mean": -0.4373061954975128,
      "epsilon_dpo/beta_margin_grad_std": 0.11379314959049225,
      "epsilon_dpo/beta_margin_mean": 0.26784640550613403,
      "epsilon_dpo/beta_margin_std": 0.4950565993785858,
      "epsilon_dpo/loss_margin_mean": 27.0919189453125,
      "grad_norm": 7.052608966827393,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.010041233152151108,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.5403838846864692e-09,
      "logits/chosen": -0.345502108335495,
      "logits/rejected": -0.09203990548849106,
      "logps/chosen": -152.26617431640625,
      "logps/ref_chosen": -100.62752532958984,
      "logps/ref_rejected": -110.3574447631836,
      "logps/rejected": -189.0880126953125,
      "loss": 1.1944,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.5173914432525635,
      "rewards/margins": 0.26784640550613403,
      "rewards/rejected": -0.7852379083633423,
      "step": 641
    },
    {
      "epoch": 0.9705215419501134,
      "epsilon_dpo/beta": 0.00994543731212616,
      "epsilon_dpo/beta_margin_grad_mean": -0.4270346462726593,
      "epsilon_dpo/beta_margin_grad_std": 0.11673079431056976,
      "epsilon_dpo/beta_margin_mean": 0.30874237418174744,
      "epsilon_dpo/beta_margin_std": 0.5001073479652405,
      "epsilon_dpo/loss_margin_mean": 31.388866424560547,
      "grad_norm": 7.703180313110352,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.009991277009248734,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 1.3973071544233218e-09,
      "logits/chosen": -0.22989991307258606,
      "logits/rejected": -0.21155939996242523,
      "logps/chosen": -143.5499267578125,
      "logps/ref_chosen": -95.99467468261719,
      "logps/ref_rejected": -96.79623413085938,
      "logps/rejected": -175.74032592773438,
      "loss": 1.1611,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.47523415088653564,
      "rewards/margins": 0.3087424039840698,
      "rewards/rejected": -0.7839765548706055,
      "step": 642
    },
    {
      "epoch": 0.9720332577475435,
      "epsilon_dpo/beta": 0.009895927272737026,
      "epsilon_dpo/beta_margin_grad_mean": -0.4284159243106842,
      "epsilon_dpo/beta_margin_grad_std": 0.12825217843055725,
      "epsilon_dpo/beta_margin_mean": 0.3099651634693146,
      "epsilon_dpo/beta_margin_std": 0.5610890984535217,
      "epsilon_dpo/loss_margin_mean": 31.65646743774414,
      "grad_norm": 6.967833995819092,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.009944661520421505,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.261184375888541e-09,
      "logits/chosen": -0.30852293968200684,
      "logits/rejected": -0.6907504796981812,
      "logps/chosen": -134.0585174560547,
      "logps/ref_chosen": -89.49581909179688,
      "logps/ref_rejected": -114.9364013671875,
      "logps/rejected": -191.15557861328125,
      "loss": 1.1745,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.44213464856147766,
      "rewards/margins": 0.3099651336669922,
      "rewards/rejected": -0.7520997524261475,
      "step": 643
    },
    {
      "epoch": 0.9735449735449735,
      "epsilon_dpo/beta": 0.009863686747848988,
      "epsilon_dpo/beta_margin_grad_mean": -0.4447920322418213,
      "epsilon_dpo/beta_margin_grad_std": 0.13148462772369385,
      "epsilon_dpo/beta_margin_mean": 0.2381734848022461,
      "epsilon_dpo/beta_margin_std": 0.5650452375411987,
      "epsilon_dpo/loss_margin_mean": 24.548532485961914,
      "grad_norm": 6.671364784240723,
      "kl/avg_steps": 0.328125,
      "kl/beta": 0.009895185008645058,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.65625,
      "learning_rate": 1.1320193567288527e-09,
      "logits/chosen": 0.001531161367893219,
      "logits/rejected": -0.12236610054969788,
      "logps/chosen": -117.81045532226562,
      "logps/ref_chosen": -69.89152526855469,
      "logps/ref_rejected": -86.02226257324219,
      "logps/rejected": -158.48971557617188,
      "loss": 1.2387,
      "rewards/accuracies": 0.65625,
      "rewards/chosen": -0.47450461983680725,
      "rewards/margins": 0.2381734848022461,
      "rewards/rejected": -0.712678074836731,
      "step": 644
    },
    {
      "epoch": 0.9750566893424036,
      "epsilon_dpo/beta": 0.009811407886445522,
      "epsilon_dpo/beta_margin_grad_mean": -0.417654424905777,
      "epsilon_dpo/beta_margin_grad_std": 0.11769741028547287,
      "epsilon_dpo/beta_margin_mean": 0.35267549753189087,
      "epsilon_dpo/beta_margin_std": 0.5047261714935303,
      "epsilon_dpo/loss_margin_mean": 36.249839782714844,
      "grad_norm": 6.893727779388428,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.009862823411822319,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 1.0098157099674987e-09,
      "logits/chosen": -0.04132319614291191,
      "logits/rejected": 0.27628880739212036,
      "logps/chosen": -126.49514770507812,
      "logps/ref_chosen": -83.9632339477539,
      "logps/ref_rejected": -95.219970703125,
      "logps/rejected": -174.001708984375,
      "loss": 1.1251,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.41895389556884766,
      "rewards/margins": 0.35267549753189087,
      "rewards/rejected": -0.7716293931007385,
      "step": 645
    },
    {
      "epoch": 0.9765684051398337,
      "epsilon_dpo/beta": 0.009748812764883041,
      "epsilon_dpo/beta_margin_grad_mean": -0.4187504053115845,
      "epsilon_dpo/beta_margin_grad_std": 0.11697126179933548,
      "epsilon_dpo/beta_margin_mean": 0.34697264432907104,
      "epsilon_dpo/beta_margin_std": 0.5104357600212097,
      "epsilon_dpo/loss_margin_mean": 35.826202392578125,
      "grad_norm": 6.112748146057129,
      "kl/avg_steps": 0.640625,
      "kl/beta": 0.009810703806579113,
      "kl/n_epsilon_steps": 0.171875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 8.945768539031783e-10,
      "logits/chosen": -0.2783118784427643,
      "logits/rejected": -0.2582191526889801,
      "logps/chosen": -134.0037841796875,
      "logps/ref_chosen": -83.46397399902344,
      "logps/ref_rejected": -108.6168212890625,
      "logps/rejected": -194.9828338623047,
      "loss": 1.1307,
      "rewards/accuracies": 0.796875,
      "rewards/chosen": -0.4940229058265686,
      "rewards/margins": 0.34697264432907104,
      "rewards/rejected": -0.8409955501556396,
      "step": 646
    },
    {
      "epoch": 0.9780801209372638,
      "epsilon_dpo/beta": 0.009676109068095684,
      "epsilon_dpo/beta_margin_grad_mean": -0.3990939259529114,
      "epsilon_dpo/beta_margin_grad_std": 0.10093998163938522,
      "epsilon_dpo/beta_margin_mean": 0.42904818058013916,
      "epsilon_dpo/beta_margin_std": 0.44319963455200195,
      "epsilon_dpo/loss_margin_mean": 44.51857376098633,
      "grad_norm": 9.283509254455566,
      "kl/avg_steps": 0.75,
      "kl/beta": 0.009748253971338272,
      "kl/n_epsilon_steps": 0.125,
      "kl/p_epsilon_steps": 0.875,
      "learning_rate": 7.863060120144316e-10,
      "logits/chosen": -0.45879900455474854,
      "logits/rejected": -0.3617364466190338,
      "logps/chosen": -138.1648406982422,
      "logps/ref_chosen": -88.60772705078125,
      "logps/ref_rejected": -128.72372436523438,
      "logps/rejected": -222.79940795898438,
      "loss": 1.0486,
      "rewards/accuracies": 0.875,
      "rewards/chosen": -0.481179416179657,
      "rewards/margins": 0.42904818058013916,
      "rewards/rejected": -0.9102275967597961,
      "step": 647
    },
    {
      "epoch": 0.9795918367346939,
      "epsilon_dpo/beta": 0.009631294757127762,
      "epsilon_dpo/beta_margin_grad_mean": -0.4343121647834778,
      "epsilon_dpo/beta_margin_grad_std": 0.11647993326187134,
      "epsilon_dpo/beta_margin_mean": 0.27863410115242004,
      "epsilon_dpo/beta_margin_std": 0.49984973669052124,
      "epsilon_dpo/loss_margin_mean": 29.250940322875977,
      "grad_norm": 7.5205888748168945,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.009675686247646809,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 6.850062128694045e-10,
      "logits/chosen": -0.032209426164627075,
      "logits/rejected": -0.2213985025882721,
      "logps/chosen": -143.01754760742188,
      "logps/ref_chosen": -91.4552001953125,
      "logps/ref_rejected": -111.99140167236328,
      "logps/rejected": -192.80470275878906,
      "loss": 1.1867,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.49837782979011536,
      "rewards/margins": 0.27863410115242004,
      "rewards/rejected": -0.7770119309425354,
      "step": 648
    },
    {
      "epoch": 0.981103552532124,
      "epsilon_dpo/beta": 0.009580339305102825,
      "epsilon_dpo/beta_margin_grad_mean": -0.42864057421684265,
      "epsilon_dpo/beta_margin_grad_std": 0.12436851859092712,
      "epsilon_dpo/beta_margin_mean": 0.30313241481781006,
      "epsilon_dpo/beta_margin_std": 0.5281475186347961,
      "epsilon_dpo/loss_margin_mean": 31.997285842895508,
      "grad_norm": 7.824967861175537,
      "kl/avg_steps": 0.53125,
      "kl/beta": 0.009630543179810047,
      "kl/n_epsilon_steps": 0.234375,
      "kl/p_epsilon_steps": 0.765625,
      "learning_rate": 5.906802900412788e-10,
      "logits/chosen": 0.16560648381710052,
      "logits/rejected": -0.0949397012591362,
      "logps/chosen": -118.04146575927734,
      "logps/ref_chosen": -72.111328125,
      "logps/ref_rejected": -96.00141906738281,
      "logps/rejected": -173.9288330078125,
      "loss": 1.1731,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.44223999977111816,
      "rewards/margins": 0.30313241481781006,
      "rewards/rejected": -0.7453724145889282,
      "step": 649
    },
    {
      "epoch": 0.982615268329554,
      "epsilon_dpo/beta": 0.009541687555611134,
      "epsilon_dpo/beta_margin_grad_mean": -0.42905229330062866,
      "epsilon_dpo/beta_margin_grad_std": 0.12155012786388397,
      "epsilon_dpo/beta_margin_mean": 0.30361443758010864,
      "epsilon_dpo/beta_margin_std": 0.5176792740821838,
      "epsilon_dpo/loss_margin_mean": 32.194278717041016,
      "grad_norm": 7.0833353996276855,
      "kl/avg_steps": 0.40625,
      "kl/beta": 0.009579651057720184,
      "kl/n_epsilon_steps": 0.296875,
      "kl/p_epsilon_steps": 0.703125,
      "learning_rate": 5.033308820289184e-10,
      "logits/chosen": 0.3099375069141388,
      "logits/rejected": -0.09590494632720947,
      "logps/chosen": -118.29649353027344,
      "logps/ref_chosen": -77.13442993164062,
      "logps/ref_rejected": -101.20818328857422,
      "logps/rejected": -174.5645294189453,
      "loss": 1.1698,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.3945963978767395,
      "rewards/margins": 0.30361443758010864,
      "rewards/rejected": -0.6982108354568481,
      "step": 650
    },
    {
      "epoch": 0.9841269841269841,
      "epsilon_dpo/beta": 0.009497118182480335,
      "epsilon_dpo/beta_margin_grad_mean": -0.4275834858417511,
      "epsilon_dpo/beta_margin_grad_std": 0.10842897742986679,
      "epsilon_dpo/beta_margin_mean": 0.3093123435974121,
      "epsilon_dpo/beta_margin_std": 0.4653792381286621,
      "epsilon_dpo/loss_margin_mean": 32.86103439331055,
      "grad_norm": 10.239365577697754,
      "kl/avg_steps": 0.46875,
      "kl/beta": 0.009540891274809837,
      "kl/n_epsilon_steps": 0.265625,
      "kl/p_epsilon_steps": 0.734375,
      "learning_rate": 4.2296043218295606e-10,
      "logits/chosen": -0.19762015342712402,
      "logits/rejected": -0.52814120054245,
      "logps/chosen": -121.38260650634766,
      "logps/ref_chosen": -79.18367004394531,
      "logps/ref_rejected": -103.46632385253906,
      "logps/rejected": -178.52630615234375,
      "loss": 1.1523,
      "rewards/accuracies": 0.71875,
      "rewards/chosen": -0.40324467420578003,
      "rewards/margins": 0.3093123435974121,
      "rewards/rejected": -0.7125570178031921,
      "step": 651
    },
    {
      "epoch": 0.9856386999244142,
      "epsilon_dpo/beta": 0.009472084231674671,
      "epsilon_dpo/beta_margin_grad_mean": -0.44292983412742615,
      "epsilon_dpo/beta_margin_grad_std": 0.12006277590990067,
      "epsilon_dpo/beta_margin_mean": 0.24692460894584656,
      "epsilon_dpo/beta_margin_std": 0.5136936902999878,
      "epsilon_dpo/loss_margin_mean": 26.458864212036133,
      "grad_norm": 8.063827514648438,
      "kl/avg_steps": 0.265625,
      "kl/beta": 0.009496376849710941,
      "kl/n_epsilon_steps": 0.359375,
      "kl/p_epsilon_steps": 0.625,
      "learning_rate": 3.4957118863768176e-10,
      "logits/chosen": -0.5991553068161011,
      "logits/rejected": -0.3117263913154602,
      "logps/chosen": -135.54934692382812,
      "logps/ref_chosen": -85.69632720947266,
      "logps/ref_rejected": -109.90357971191406,
      "logps/rejected": -186.2154541015625,
      "loss": 1.2177,
      "rewards/accuracies": 0.640625,
      "rewards/chosen": -0.4741317927837372,
      "rewards/margins": 0.24692459404468536,
      "rewards/rejected": -0.7210564017295837,
      "step": 652
    },
    {
      "epoch": 0.9871504157218443,
      "epsilon_dpo/beta": 0.009436645545065403,
      "epsilon_dpo/beta_margin_grad_mean": -0.42172205448150635,
      "epsilon_dpo/beta_margin_grad_std": 0.1203337237238884,
      "epsilon_dpo/beta_margin_mean": 0.33542388677597046,
      "epsilon_dpo/beta_margin_std": 0.5146583914756775,
      "epsilon_dpo/loss_margin_mean": 35.929100036621094,
      "grad_norm": 7.69780158996582,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.009471219033002853,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.831652042480093e-10,
      "logits/chosen": -0.05788855254650116,
      "logits/rejected": -0.04992194101214409,
      "logps/chosen": -126.68917846679688,
      "logps/ref_chosen": -85.57878112792969,
      "logps/ref_rejected": -99.29658508300781,
      "logps/rejected": -176.33609008789062,
      "loss": 1.142,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.3894695043563843,
      "rewards/margins": 0.33542391657829285,
      "rewards/rejected": -0.7248933911323547,
      "step": 653
    },
    {
      "epoch": 0.9886621315192744,
      "epsilon_dpo/beta": 0.009401390329003334,
      "epsilon_dpo/beta_margin_grad_mean": -0.44078996777534485,
      "epsilon_dpo/beta_margin_grad_std": 0.12072242051362991,
      "epsilon_dpo/beta_margin_mean": 0.2574543356895447,
      "epsilon_dpo/beta_margin_std": 0.5202062129974365,
      "epsilon_dpo/loss_margin_mean": 27.730667114257812,
      "grad_norm": 8.079339981079102,
      "kl/avg_steps": 0.375,
      "kl/beta": 0.009435834363102913,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.6875,
      "learning_rate": 2.2374433653205016e-10,
      "logits/chosen": -0.04486121982336044,
      "logits/rejected": -0.5683233737945557,
      "logps/chosen": -131.8297119140625,
      "logps/ref_chosen": -83.4058837890625,
      "logps/ref_rejected": -116.10549926757812,
      "logps/rejected": -192.260009765625,
      "loss": 1.2098,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.4570308327674866,
      "rewards/margins": 0.2574543356895447,
      "rewards/rejected": -0.7144851684570312,
      "step": 654
    },
    {
      "epoch": 0.9901738473167044,
      "epsilon_dpo/beta": 0.009354515001177788,
      "epsilon_dpo/beta_margin_grad_mean": -0.4245755076408386,
      "epsilon_dpo/beta_margin_grad_std": 0.10435869544744492,
      "epsilon_dpo/beta_margin_mean": 0.3234991431236267,
      "epsilon_dpo/beta_margin_std": 0.45337221026420593,
      "epsilon_dpo/loss_margin_mean": 34.833831787109375,
      "grad_norm": 5.97894811630249,
      "kl/avg_steps": 0.5,
      "kl/beta": 0.009400582872331142,
      "kl/n_epsilon_steps": 0.25,
      "kl/p_epsilon_steps": 0.75,
      "learning_rate": 1.7131024761923852e-10,
      "logits/chosen": -0.18475328385829926,
      "logits/rejected": -0.643438458442688,
      "logps/chosen": -108.57415771484375,
      "logps/ref_chosen": -72.27444458007812,
      "logps/ref_rejected": -107.04484558105469,
      "logps/rejected": -178.17840576171875,
      "loss": 1.1373,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -0.33998844027519226,
      "rewards/margins": 0.3234991431236267,
      "rewards/rejected": -0.6634875535964966,
      "step": 655
    },
    {
      "epoch": 0.9916855631141346,
      "epsilon_dpo/beta": 0.009299205616116524,
      "epsilon_dpo/beta_margin_grad_mean": -0.4235363304615021,
      "epsilon_dpo/beta_margin_grad_std": 0.10802899301052094,
      "epsilon_dpo/beta_margin_mean": 0.32424455881118774,
      "epsilon_dpo/beta_margin_std": 0.4628637135028839,
      "epsilon_dpo/loss_margin_mean": 35.1312255859375,
      "grad_norm": 6.119517803192139,
      "kl/avg_steps": 0.59375,
      "kl/beta": 0.009353813715279102,
      "kl/n_epsilon_steps": 0.203125,
      "kl/p_epsilon_steps": 0.796875,
      "learning_rate": 1.2586440420372934e-10,
      "logits/chosen": -0.306098610162735,
      "logits/rejected": -0.34763431549072266,
      "logps/chosen": -147.2364501953125,
      "logps/ref_chosen": -96.3607177734375,
      "logps/ref_rejected": -111.83285522460938,
      "logps/rejected": -197.8397979736328,
      "loss": 1.1393,
      "rewards/accuracies": 0.78125,
      "rewards/chosen": -0.47428518533706665,
      "rewards/margins": 0.32424455881118774,
      "rewards/rejected": -0.7985297441482544,
      "step": 656
    },
    {
      "epoch": 0.9931972789115646,
      "epsilon_dpo/beta": 0.009247222915291786,
      "epsilon_dpo/beta_margin_grad_mean": -0.40973320603370667,
      "epsilon_dpo/beta_margin_grad_std": 0.11478579789400101,
      "epsilon_dpo/beta_margin_mean": 0.3866519033908844,
      "epsilon_dpo/beta_margin_std": 0.4957241117954254,
      "epsilon_dpo/loss_margin_mean": 42.10652160644531,
      "grad_norm": 8.461531639099121,
      "kl/avg_steps": 0.5625,
      "kl/beta": 0.009298603050410748,
      "kl/n_epsilon_steps": 0.21875,
      "kl/p_epsilon_steps": 0.78125,
      "learning_rate": 8.740807750345913e-11,
      "logits/chosen": -0.10227912664413452,
      "logits/rejected": -0.3732077479362488,
      "logps/chosen": -112.02146911621094,
      "logps/ref_chosen": -71.39437866210938,
      "logps/ref_rejected": -100.89120483398438,
      "logps/rejected": -183.62481689453125,
      "loss": 1.0948,
      "rewards/accuracies": 0.765625,
      "rewards/chosen": -0.37659162282943726,
      "rewards/margins": 0.3866519033908844,
      "rewards/rejected": -0.763243556022644,
      "step": 657
    },
    {
      "epoch": 0.9947089947089947,
      "epsilon_dpo/beta": 0.009207057766616344,
      "epsilon_dpo/beta_margin_grad_mean": -0.4259795546531677,
      "epsilon_dpo/beta_margin_grad_std": 0.12907136976718903,
      "epsilon_dpo/beta_margin_mean": 0.31728169322013855,
      "epsilon_dpo/beta_margin_std": 0.5570127964019775,
      "epsilon_dpo/loss_margin_mean": 34.85756301879883,
      "grad_norm": 7.888942241668701,
      "kl/avg_steps": 0.4375,
      "kl/beta": 0.009246590547263622,
      "kl/n_epsilon_steps": 0.28125,
      "kl/p_epsilon_steps": 0.71875,
      "learning_rate": 5.594234322453539e-11,
      "logits/chosen": -0.19734932482242584,
      "logits/rejected": -0.16385933756828308,
      "logps/chosen": -130.331787109375,
      "logps/ref_chosen": -87.25668334960938,
      "logps/ref_rejected": -103.51670837402344,
      "logps/rejected": -181.44937133789062,
      "loss": 1.1679,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.3980225622653961,
      "rewards/margins": 0.31728169322013855,
      "rewards/rejected": -0.7153042554855347,
      "step": 658
    },
    {
      "epoch": 0.9962207105064248,
      "epsilon_dpo/beta": 0.009175584651529789,
      "epsilon_dpo/beta_margin_grad_mean": -0.4530555009841919,
      "epsilon_dpo/beta_margin_grad_std": 0.11484676599502563,
      "epsilon_dpo/beta_margin_mean": 0.19890394806861877,
      "epsilon_dpo/beta_margin_std": 0.4893724024295807,
      "epsilon_dpo/loss_margin_mean": 22.028242111206055,
      "grad_norm": 5.81763219833374,
      "kl/avg_steps": 0.34375,
      "kl/beta": 0.009206313639879227,
      "kl/n_epsilon_steps": 0.328125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.146808153123293e-11,
      "logits/chosen": -0.03687793016433716,
      "logits/rejected": -0.44385725259780884,
      "logps/chosen": -132.07925415039062,
      "logps/ref_chosen": -80.2817153930664,
      "logps/ref_rejected": -101.70068359375,
      "logps/rejected": -175.52645874023438,
      "loss": 1.255,
      "rewards/accuracies": 0.703125,
      "rewards/chosen": -0.4763162136077881,
      "rewards/margins": 0.19890394806861877,
      "rewards/rejected": -0.6752201318740845,
      "step": 659
    },
    {
      "epoch": 0.9977324263038548,
      "epsilon_dpo/beta": 0.009118344634771347,
      "epsilon_dpo/beta_margin_grad_mean": -0.4164464771747589,
      "epsilon_dpo/beta_margin_grad_std": 0.09780213981866837,
      "epsilon_dpo/beta_margin_mean": 0.35128533840179443,
      "epsilon_dpo/beta_margin_std": 0.4149182140827179,
      "epsilon_dpo/loss_margin_mean": 38.761817932128906,
      "grad_norm": 5.7514753341674805,
      "kl/avg_steps": 0.625,
      "kl/beta": 0.009174775332212448,
      "kl/n_epsilon_steps": 0.1875,
      "kl/p_epsilon_steps": 0.8125,
      "learning_rate": 1.3985977021235829e-11,
      "logits/chosen": -0.07328711450099945,
      "logits/rejected": -0.3468150198459625,
      "logps/chosen": -122.07150268554688,
      "logps/ref_chosen": -76.87039184570312,
      "logps/ref_rejected": -109.3966064453125,
      "logps/rejected": -193.35952758789062,
      "loss": 1.1069,
      "rewards/accuracies": 0.828125,
      "rewards/chosen": -0.4130452573299408,
      "rewards/margins": 0.35128533840179443,
      "rewards/rejected": -0.7643306255340576,
      "step": 660
    },
    {
      "epoch": 0.999244142101285,
      "epsilon_dpo/beta": 0.009085915982723236,
      "epsilon_dpo/beta_margin_grad_mean": -0.4587934911251068,
      "epsilon_dpo/beta_margin_grad_std": 0.11900065094232559,
      "epsilon_dpo/beta_margin_mean": 0.17370952665805817,
      "epsilon_dpo/beta_margin_std": 0.5091543793678284,
      "epsilon_dpo/loss_margin_mean": 19.486536026000977,
      "grad_norm": 5.589955806732178,
      "kl/avg_steps": 0.359375,
      "kl/beta": 0.009117788635194302,
      "kl/n_epsilon_steps": 0.3125,
      "kl/p_epsilon_steps": 0.671875,
      "learning_rate": 3.4965187065971735e-12,
      "logits/chosen": 0.051325783133506775,
      "logits/rejected": -0.18619848787784576,
      "logps/chosen": -144.16314697265625,
      "logps/ref_chosen": -83.48721313476562,
      "logps/ref_rejected": -106.11189270019531,
      "logps/rejected": -186.27435302734375,
      "loss": 1.2825,
      "rewards/accuracies": 0.671875,
      "rewards/chosen": -0.5530954599380493,
      "rewards/margins": 0.17370952665805817,
      "rewards/rejected": -0.7268049716949463,
      "step": 661
    },
    {
      "epoch": 0.999244142101285,
      "step": 661,
      "total_flos": 0.0,
      "train_loss": 1.1906072249390895,
      "train_runtime": 2288.3837,
      "train_samples_per_second": 18.5,
      "train_steps_per_second": 0.289
    }
  ],
  "logging_steps": 1,
  "max_steps": 661,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": false,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 8,
  "trial_name": null,
  "trial_params": null
}