{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 3.0,
  "eval_steps": 100,
  "global_step": 1065,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "dpo_losses": 0.6931471824645996,
      "epoch": 0.0,
      "grad_norm": 1.6016450442000167,
      "learning_rate": 4.672897196261682e-08,
      "logits/chosen": -2.861618995666504,
      "logits/rejected": -2.8205904960632324,
      "logps/chosen": -271.06011962890625,
      "logps/rejected": -211.1704559326172,
      "loss": 0.6931,
      "positive_losses": 0.0,
      "rewards/accuracies": 0.0,
      "rewards/chosen": 0.0,
      "rewards/margins": 0.0,
      "rewards/margins_max": 0.0,
      "rewards/margins_min": 0.0,
      "rewards/margins_std": 0.0,
      "rewards/rejected": 0.0,
      "step": 1
    },
    {
      "dpo_losses": 0.6929657459259033,
      "epoch": 0.03,
      "grad_norm": 15.69888585504856,
      "learning_rate": 4.6728971962616824e-07,
      "logits/chosen": -2.833556890487671,
      "logits/rejected": -2.790858507156372,
      "logps/chosen": -325.010498046875,
      "logps/rejected": -274.9622802734375,
      "loss": 0.7,
      "positive_losses": 0.06880617141723633,
      "rewards/accuracies": 0.5277777910232544,
      "rewards/chosen": 0.00039824165287427604,
      "rewards/margins": 0.0003647865087259561,
      "rewards/margins_max": 0.002922259271144867,
      "rewards/margins_min": -0.0023222416639328003,
      "rewards/margins_std": 0.0023321988992393017,
      "rewards/rejected": 3.345518052810803e-05,
      "step": 10
    },
    {
      "dpo_losses": 0.6923663020133972,
      "epoch": 0.06,
      "grad_norm": 1.8387431033463484,
      "learning_rate": 9.345794392523365e-07,
      "logits/chosen": -2.7247300148010254,
      "logits/rejected": -2.706224203109741,
      "logps/chosen": -293.1662902832031,
      "logps/rejected": -215.27590942382812,
      "loss": 0.693,
      "positive_losses": 0.006913089659065008,
      "rewards/accuracies": 0.625,
      "rewards/chosen": 0.006596668157726526,
      "rewards/margins": 0.0015658324118703604,
      "rewards/margins_max": 0.005163406487554312,
      "rewards/margins_min": -0.0021486091427505016,
      "rewards/margins_std": 0.003277386771515012,
      "rewards/rejected": 0.0050308359786868095,
      "step": 20
    },
    {
      "dpo_losses": 0.6898607015609741,
      "epoch": 0.08,
      "grad_norm": 2.2348347437651173,
      "learning_rate": 1.4018691588785047e-06,
      "logits/chosen": -2.8186888694763184,
      "logits/rejected": -2.7495126724243164,
      "logps/chosen": -301.5869140625,
      "logps/rejected": -231.0554962158203,
      "loss": 0.6894,
      "positive_losses": 0.0,
      "rewards/accuracies": 0.7749999761581421,
      "rewards/chosen": 0.0211896151304245,
      "rewards/margins": 0.0066024260595440865,
      "rewards/margins_max": 0.015645453706383705,
      "rewards/margins_min": -0.0015332145849242806,
      "rewards/margins_std": 0.007755809463560581,
      "rewards/rejected": 0.014587189070880413,
      "step": 30
    },
    {
      "dpo_losses": 0.6871124505996704,
      "epoch": 0.11,
      "grad_norm": 1.7215571973895225,
      "learning_rate": 1.869158878504673e-06,
      "logits/chosen": -2.8395392894744873,
      "logits/rejected": -2.759063243865967,
      "logps/chosen": -273.9014587402344,
      "logps/rejected": -223.36782836914062,
      "loss": 0.6866,
      "positive_losses": 0.006247329525649548,
      "rewards/accuracies": 0.7749999761581421,
      "rewards/chosen": 0.035831518471241,
      "rewards/margins": 0.01216290332376957,
      "rewards/margins_max": 0.0269247405230999,
      "rewards/margins_min": -0.0010026793461292982,
      "rewards/margins_std": 0.012388748116791248,
      "rewards/rejected": 0.02366861142218113,
      "step": 40
    },
    {
      "dpo_losses": 0.677656352519989,
      "epoch": 0.14,
      "grad_norm": 9.32393645493281,
      "learning_rate": 2.3364485981308413e-06,
      "logits/chosen": -2.805560350418091,
      "logits/rejected": -2.733527660369873,
      "logps/chosen": -269.43572998046875,
      "logps/rejected": -230.8954620361328,
      "loss": 0.6785,
      "positive_losses": 0.0010559081565588713,
      "rewards/accuracies": 0.8999999761581421,
      "rewards/chosen": 0.061344046145677567,
      "rewards/margins": 0.03152567520737648,
      "rewards/margins_max": 0.06917257606983185,
      "rewards/margins_min": 0.003379008499905467,
      "rewards/margins_std": 0.030225109308958054,
      "rewards/rejected": 0.029818367213010788,
      "step": 50
    },
    {
      "dpo_losses": 0.6655401587486267,
      "epoch": 0.17,
      "grad_norm": 2.375582473000896,
      "learning_rate": 2.8037383177570094e-06,
      "logits/chosen": -2.7405285835266113,
      "logits/rejected": -2.7018980979919434,
      "logps/chosen": -304.9367370605469,
      "logps/rejected": -259.7385559082031,
      "loss": 0.669,
      "positive_losses": 0.0,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.08702905476093292,
      "rewards/margins": 0.056699346750974655,
      "rewards/margins_max": 0.10381220281124115,
      "rewards/margins_min": 0.013735351152718067,
      "rewards/margins_std": 0.04012778028845787,
      "rewards/rejected": 0.030329709872603416,
      "step": 60
    },
    {
      "dpo_losses": 0.6565508842468262,
      "epoch": 0.2,
      "grad_norm": 1.7065569395440294,
      "learning_rate": 3.2710280373831774e-06,
      "logits/chosen": -2.765432119369507,
      "logits/rejected": -2.7045130729675293,
      "logps/chosen": -311.462646484375,
      "logps/rejected": -231.845703125,
      "loss": 0.6576,
      "positive_losses": 0.046324919909238815,
      "rewards/accuracies": 0.9125000238418579,
      "rewards/chosen": 0.11398597806692123,
      "rewards/margins": 0.07602710276842117,
      "rewards/margins_max": 0.16103656589984894,
      "rewards/margins_min": 0.011467654258012772,
      "rewards/margins_std": 0.06779597699642181,
      "rewards/rejected": 0.03795887157320976,
      "step": 70
    },
    {
      "dpo_losses": 0.6376875042915344,
      "epoch": 0.23,
      "grad_norm": 1.780395034931213,
      "learning_rate": 3.738317757009346e-06,
      "logits/chosen": -2.797402858734131,
      "logits/rejected": -2.7417943477630615,
      "logps/chosen": -307.58355712890625,
      "logps/rejected": -284.42138671875,
      "loss": 0.6438,
      "positive_losses": 0.044156648218631744,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.13686935603618622,
      "rewards/margins": 0.1172962412238121,
      "rewards/margins_max": 0.24144157767295837,
      "rewards/margins_min": 0.01226846780627966,
      "rewards/margins_std": 0.10532160103321075,
      "rewards/rejected": 0.019573111087083817,
      "step": 80
    },
    {
      "dpo_losses": 0.6190480589866638,
      "epoch": 0.25,
      "grad_norm": 9.140888334214253,
      "learning_rate": 4.205607476635514e-06,
      "logits/chosen": -2.7015762329101562,
      "logits/rejected": -2.6621322631835938,
      "logps/chosen": -260.407958984375,
      "logps/rejected": -199.31021118164062,
      "loss": 0.6212,
      "positive_losses": 0.014925527386367321,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.15586820244789124,
      "rewards/margins": 0.159098282456398,
      "rewards/margins_max": 0.30938515067100525,
      "rewards/margins_min": 0.042563945055007935,
      "rewards/margins_std": 0.12488297373056412,
      "rewards/rejected": -0.0032300553284585476,
      "step": 90
    },
    {
      "dpo_losses": 0.6197101473808289,
      "epoch": 0.28,
      "grad_norm": 2.0684254233213344,
      "learning_rate": 4.6728971962616825e-06,
      "logits/chosen": -2.7176895141601562,
      "logits/rejected": -2.6833646297454834,
      "logps/chosen": -265.6207580566406,
      "logps/rejected": -262.1724548339844,
      "loss": 0.6302,
      "positive_losses": 0.42638903856277466,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": 0.15790995955467224,
      "rewards/margins": 0.15775997936725616,
      "rewards/margins_max": 0.31875520944595337,
      "rewards/margins_min": 0.026545410975813866,
      "rewards/margins_std": 0.13236147165298462,
      "rewards/rejected": 0.00014998484402894974,
      "step": 100
    },
    {
      "epoch": 0.28,
      "eval_dpo_losses": 0.6731910109519958,
      "eval_logits/chosen": -2.701474666595459,
      "eval_logits/rejected": -2.6654608249664307,
      "eval_logps/chosen": -275.82159423828125,
      "eval_logps/rejected": -254.3717041015625,
      "eval_loss": 0.8170425891876221,
      "eval_positive_losses": 1.2658016681671143,
      "eval_rewards/accuracies": 0.5920000076293945,
      "eval_rewards/chosen": 0.08771849423646927,
      "eval_rewards/margins": 0.045646362006664276,
      "eval_rewards/margins_max": 0.27169111371040344,
      "eval_rewards/margins_min": -0.14722861349582672,
      "eval_rewards/margins_std": 0.1389097422361374,
      "eval_rewards/rejected": 0.04207213222980499,
      "eval_runtime": 428.4375,
      "eval_samples_per_second": 4.668,
      "eval_steps_per_second": 0.292,
      "step": 100
    },
    {
      "dpo_losses": 0.56014484167099,
      "epoch": 0.31,
      "grad_norm": 5.9261246784753006,
      "learning_rate": 4.999879018839288e-06,
      "logits/chosen": -2.756599187850952,
      "logits/rejected": -2.6693594455718994,
      "logps/chosen": -352.81890869140625,
      "logps/rejected": -294.49932861328125,
      "loss": 0.591,
      "positive_losses": 0.237422376871109,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.21786770224571228,
      "rewards/margins": 0.298523485660553,
      "rewards/margins_max": 0.5258058905601501,
      "rewards/margins_min": 0.09114833921194077,
      "rewards/margins_std": 0.19396927952766418,
      "rewards/rejected": -0.0806557908654213,
      "step": 110
    },
    {
      "dpo_losses": 0.5659720301628113,
      "epoch": 0.34,
      "grad_norm": 1.6908012876424043,
      "learning_rate": 4.99772856836941e-06,
      "logits/chosen": -2.7271599769592285,
      "logits/rejected": -2.6729092597961426,
      "logps/chosen": -322.3902893066406,
      "logps/rejected": -268.8707275390625,
      "loss": 0.5784,
      "positive_losses": 0.19832010567188263,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.2599591314792633,
      "rewards/margins": 0.2867242395877838,
      "rewards/margins_max": 0.5613428354263306,
      "rewards/margins_min": 0.091820128262043,
      "rewards/margins_std": 0.21394488215446472,
      "rewards/rejected": -0.02676514908671379,
      "step": 120
    },
    {
      "dpo_losses": 0.535868763923645,
      "epoch": 0.37,
      "grad_norm": 16.682848792926713,
      "learning_rate": 4.992892309373227e-06,
      "logits/chosen": -2.7010695934295654,
      "logits/rejected": -2.6261649131774902,
      "logps/chosen": -359.5436096191406,
      "logps/rejected": -313.9417419433594,
      "loss": 0.5857,
      "positive_losses": 0.6961414217948914,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.22820620238780975,
      "rewards/margins": 0.36331456899642944,
      "rewards/margins_max": 0.6659263372421265,
      "rewards/margins_min": 0.10675536096096039,
      "rewards/margins_std": 0.2517603039741516,
      "rewards/rejected": -0.1351083517074585,
      "step": 130
    },
    {
      "dpo_losses": 0.5288053750991821,
      "epoch": 0.39,
      "grad_norm": 1.8145271015177298,
      "learning_rate": 4.985375442281969e-06,
      "logits/chosen": -2.686540126800537,
      "logits/rejected": -2.6495883464813232,
      "logps/chosen": -291.7132263183594,
      "logps/rejected": -282.44207763671875,
      "loss": 0.5941,
      "positive_losses": 0.06369924545288086,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.26700812578201294,
      "rewards/margins": 0.39293378591537476,
      "rewards/margins_max": 0.7747925519943237,
      "rewards/margins_min": 0.09062640368938446,
      "rewards/margins_std": 0.31326359510421753,
      "rewards/rejected": -0.1259256899356842,
      "step": 140
    },
    {
      "dpo_losses": 0.5246155858039856,
      "epoch": 0.42,
      "grad_norm": 2.0475968114674195,
      "learning_rate": 4.9751860499858175e-06,
      "logits/chosen": -2.6848747730255127,
      "logits/rejected": -2.6494317054748535,
      "logps/chosen": -267.29241943359375,
      "logps/rejected": -231.824462890625,
      "loss": 0.5752,
      "positive_losses": 0.4476150572299957,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.21392588317394257,
      "rewards/margins": 0.4013427197933197,
      "rewards/margins_max": 0.7649358510971069,
      "rewards/margins_min": 0.10850365459918976,
      "rewards/margins_std": 0.3050392270088196,
      "rewards/rejected": -0.18741685152053833,
      "step": 150
    },
    {
      "dpo_losses": 0.4928368031978607,
      "epoch": 0.45,
      "grad_norm": 2.1220867158103944,
      "learning_rate": 4.962335089142376e-06,
      "logits/chosen": -2.6444039344787598,
      "logits/rejected": -2.595534324645996,
      "logps/chosen": -275.7962341308594,
      "logps/rejected": -258.319091796875,
      "loss": 0.5748,
      "positive_losses": 0.8500987887382507,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.2475060522556305,
      "rewards/margins": 0.486764132976532,
      "rewards/margins_max": 0.9087164998054504,
      "rewards/margins_min": 0.14835166931152344,
      "rewards/margins_std": 0.3423914611339569,
      "rewards/rejected": -0.23925814032554626,
      "step": 160
    },
    {
      "dpo_losses": 0.46527940034866333,
      "epoch": 0.48,
      "grad_norm": 10.044301131959962,
      "learning_rate": 4.946836378394967e-06,
      "logits/chosen": -2.6870791912078857,
      "logits/rejected": -2.6393063068389893,
      "logps/chosen": -299.3938293457031,
      "logps/rejected": -323.72540283203125,
      "loss": 0.5046,
      "positive_losses": 0.22114285826683044,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.30227792263031006,
      "rewards/margins": 0.5768753886222839,
      "rewards/margins_max": 0.9806014895439148,
      "rewards/margins_min": 0.16012534499168396,
      "rewards/margins_std": 0.3640903830528259,
      "rewards/rejected": -0.27459749579429626,
      "step": 170
    },
    {
      "dpo_losses": 0.47749489545822144,
      "epoch": 0.51,
      "grad_norm": 9.115623778041357,
      "learning_rate": 4.928706583513441e-06,
      "logits/chosen": -2.668745517730713,
      "logits/rejected": -2.58906626701355,
      "logps/chosen": -306.95440673828125,
      "logps/rejected": -323.16754150390625,
      "loss": 0.5888,
      "positive_losses": 0.8010058403015137,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.24946419894695282,
      "rewards/margins": 0.5392904281616211,
      "rewards/margins_max": 0.944929301738739,
      "rewards/margins_min": 0.2653813660144806,
      "rewards/margins_std": 0.31065526604652405,
      "rewards/rejected": -0.28982624411582947,
      "step": 180
    },
    {
      "dpo_losses": 0.4789065420627594,
      "epoch": 0.54,
      "grad_norm": 2.6157504107491274,
      "learning_rate": 4.907965199473471e-06,
      "logits/chosen": -2.6501259803771973,
      "logits/rejected": -2.5630240440368652,
      "logps/chosen": -287.6591491699219,
      "logps/rejected": -258.0931396484375,
      "loss": 0.5597,
      "positive_losses": 0.7399072647094727,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.34301725029945374,
      "rewards/margins": 0.5490720272064209,
      "rewards/margins_max": 1.0513516664505005,
      "rewards/margins_min": 0.12944109737873077,
      "rewards/margins_std": 0.4154055714607239,
      "rewards/rejected": -0.20605473220348358,
      "step": 190
    },
    {
      "dpo_losses": 0.45651349425315857,
      "epoch": 0.56,
      "grad_norm": 29.762351156901545,
      "learning_rate": 4.884634529493591e-06,
      "logits/chosen": -2.755312919616699,
      "logits/rejected": -2.6784257888793945,
      "logps/chosen": -310.8017883300781,
      "logps/rejected": -297.34014892578125,
      "loss": 0.5709,
      "positive_losses": 0.8813858032226562,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.2873106598854065,
      "rewards/margins": 0.6047379970550537,
      "rewards/margins_max": 1.1215251684188843,
      "rewards/margins_min": 0.15512080490589142,
      "rewards/margins_std": 0.433703750371933,
      "rewards/rejected": -0.3174273669719696,
      "step": 200
    },
    {
      "epoch": 0.56,
      "eval_dpo_losses": 0.6517772674560547,
      "eval_logits/chosen": -2.6989214420318604,
      "eval_logits/rejected": -2.659100294113159,
      "eval_logps/chosen": -293.91009521484375,
      "eval_logps/rejected": -279.2876892089844,
      "eval_loss": 2.1527481079101562,
      "eval_positive_losses": 14.134076118469238,
      "eval_rewards/accuracies": 0.6359999775886536,
      "eval_rewards/chosen": -0.09316667169332504,
      "eval_rewards/margins": 0.11392085999250412,
      "eval_rewards/margins_max": 0.6302202343940735,
      "eval_rewards/margins_min": -0.36466744542121887,
      "eval_rewards/margins_std": 0.3297393321990967,
      "eval_rewards/rejected": -0.20708754658699036,
      "eval_runtime": 428.0951,
      "eval_samples_per_second": 4.672,
      "eval_steps_per_second": 0.292,
      "step": 200
    },
    {
      "dpo_losses": 0.44806918501853943,
      "epoch": 0.59,
      "grad_norm": 11.98930454454778,
      "learning_rate": 4.858739661052539e-06,
      "logits/chosen": -2.669461488723755,
      "logits/rejected": -2.639768123626709,
      "logps/chosen": -288.4258728027344,
      "logps/rejected": -290.136962890625,
      "loss": 0.5443,
      "positive_losses": 1.0154649019241333,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.31758418679237366,
      "rewards/margins": 0.6547064185142517,
      "rewards/margins_max": 1.1989976167678833,
      "rewards/margins_min": 0.1595531553030014,
      "rewards/margins_std": 0.4659876823425293,
      "rewards/rejected": -0.33712226152420044,
      "step": 210
    },
    {
      "dpo_losses": 0.4408543109893799,
      "epoch": 0.62,
      "grad_norm": 12.374805795706262,
      "learning_rate": 4.830308438912687e-06,
      "logits/chosen": -2.7841882705688477,
      "logits/rejected": -2.6910300254821777,
      "logps/chosen": -319.4044494628906,
      "logps/rejected": -305.32757568359375,
      "loss": 0.6095,
      "positive_losses": 1.0382084846496582,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.3405179977416992,
      "rewards/margins": 0.6505187749862671,
      "rewards/margins_max": 1.1662898063659668,
      "rewards/margins_min": 0.17229273915290833,
      "rewards/margins_std": 0.4555598199367523,
      "rewards/rejected": -0.3100007474422455,
      "step": 220
    },
    {
      "dpo_losses": 0.44871968030929565,
      "epoch": 0.65,
      "grad_norm": 13.933794700856799,
      "learning_rate": 4.799371435178544e-06,
      "logits/chosen": -2.777639389038086,
      "logits/rejected": -2.6834521293640137,
      "logps/chosen": -340.2105407714844,
      "logps/rejected": -293.09320068359375,
      "loss": 0.5182,
      "positive_losses": 1.168283224105835,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.31465137004852295,
      "rewards/margins": 0.6246488094329834,
      "rewards/margins_max": 1.0976186990737915,
      "rewards/margins_min": 0.19674204289913177,
      "rewards/margins_std": 0.3980305790901184,
      "rewards/rejected": -0.3099974989891052,
      "step": 230
    },
    {
      "dpo_losses": 0.46423768997192383,
      "epoch": 0.68,
      "grad_norm": 13.472463144636682,
      "learning_rate": 4.765961916422575e-06,
      "logits/chosen": -2.7154135704040527,
      "logits/rejected": -2.647381067276001,
      "logps/chosen": -298.61419677734375,
      "logps/rejected": -314.968994140625,
      "loss": 0.6605,
      "positive_losses": 2.0480997562408447,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.2618282437324524,
      "rewards/margins": 0.5797696113586426,
      "rewards/margins_max": 1.0939089059829712,
      "rewards/margins_min": 0.2062896192073822,
      "rewards/margins_std": 0.39808210730552673,
      "rewards/rejected": -0.3179413676261902,
      "step": 240
    },
    {
      "dpo_losses": 0.46544280648231506,
      "epoch": 0.7,
      "grad_norm": 5.034375090927607,
      "learning_rate": 4.730115807913627e-06,
      "logits/chosen": -2.714775562286377,
      "logits/rejected": -2.642279863357544,
      "logps/chosen": -279.231201171875,
      "logps/rejected": -278.4535827636719,
      "loss": 0.5103,
      "positive_losses": 0.625244140625,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3204992711544037,
      "rewards/margins": 0.5945923924446106,
      "rewards/margins_max": 1.134671688079834,
      "rewards/margins_min": 0.1570335179567337,
      "rewards/margins_std": 0.4401992857456207,
      "rewards/rejected": -0.2740930914878845,
      "step": 250
    },
    {
      "dpo_losses": 0.4729411005973816,
      "epoch": 0.73,
      "grad_norm": 1.9876146679314552,
      "learning_rate": 4.691871654986485e-06,
      "logits/chosen": -2.7730844020843506,
      "logits/rejected": -2.6877987384796143,
      "logps/chosen": -267.6090393066406,
      "logps/rejected": -289.9382019042969,
      "loss": 0.6159,
      "positive_losses": 1.5114115476608276,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.2448449432849884,
      "rewards/margins": 0.5549170970916748,
      "rewards/margins_max": 1.0365012884140015,
      "rewards/margins_min": 0.1394290030002594,
      "rewards/margins_std": 0.4102291464805603,
      "rewards/rejected": -0.310072124004364,
      "step": 260
    },
    {
      "dpo_losses": 0.48496007919311523,
      "epoch": 0.76,
      "grad_norm": 2.1171697073889124,
      "learning_rate": 4.651270581594054e-06,
      "logits/chosen": -2.744439125061035,
      "logits/rejected": -2.6912388801574707,
      "logps/chosen": -275.9270935058594,
      "logps/rejected": -281.0453796386719,
      "loss": 0.5495,
      "positive_losses": 0.27067431807518005,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.3318844437599182,
      "rewards/margins": 0.5173426866531372,
      "rewards/margins_max": 0.9215744733810425,
      "rewards/margins_min": 0.1685103327035904,
      "rewards/margins_std": 0.33360785245895386,
      "rewards/rejected": -0.1854582577943802,
      "step": 270
    },
    {
      "dpo_losses": 0.46318453550338745,
      "epoch": 0.79,
      "grad_norm": 13.974021787306237,
      "learning_rate": 4.6083562460867545e-06,
      "logits/chosen": -2.7355339527130127,
      "logits/rejected": -2.6963067054748535,
      "logps/chosen": -275.7181091308594,
      "logps/rejected": -305.1938171386719,
      "loss": 0.5698,
      "positive_losses": 1.3025602102279663,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.27738872170448303,
      "rewards/margins": 0.6033554673194885,
      "rewards/margins_max": 1.1320282220840454,
      "rewards/margins_min": 0.15280218422412872,
      "rewards/margins_std": 0.4500836431980133,
      "rewards/rejected": -0.32596665620803833,
      "step": 280
    },
    {
      "dpo_losses": 0.47502636909484863,
      "epoch": 0.82,
      "grad_norm": 2.522662820391045,
      "learning_rate": 4.563174794266684e-06,
      "logits/chosen": -2.7417492866516113,
      "logits/rejected": -2.6863653659820557,
      "logps/chosen": -257.1270446777344,
      "logps/rejected": -290.1141662597656,
      "loss": 0.5746,
      "positive_losses": 1.490474820137024,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.2978917956352234,
      "rewards/margins": 0.558850884437561,
      "rewards/margins_max": 1.0547770261764526,
      "rewards/margins_min": 0.12949053943157196,
      "rewards/margins_std": 0.41951704025268555,
      "rewards/rejected": -0.26095908880233765,
      "step": 290
    },
    {
      "dpo_losses": 0.40710002183914185,
      "epoch": 0.85,
      "grad_norm": 7.662581585547771,
      "learning_rate": 4.5157748097670125e-06,
      "logits/chosen": -2.8059568405151367,
      "logits/rejected": -2.7222626209259033,
      "logps/chosen": -308.61895751953125,
      "logps/rejected": -326.9398498535156,
      "loss": 0.4758,
      "positive_losses": 0.8625243902206421,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.37894096970558167,
      "rewards/margins": 0.7540054321289062,
      "rewards/margins_max": 1.2992969751358032,
      "rewards/margins_min": 0.2781727910041809,
      "rewards/margins_std": 0.44917410612106323,
      "rewards/rejected": -0.3750644326210022,
      "step": 300
    },
    {
      "epoch": 0.85,
      "eval_dpo_losses": 0.6395776271820068,
      "eval_logits/chosen": -2.7293992042541504,
      "eval_logits/rejected": -2.68656063079834,
      "eval_logps/chosen": -292.8875427246094,
      "eval_logps/rejected": -281.82305908203125,
      "eval_loss": 2.250778913497925,
      "eval_positive_losses": 15.010305404663086,
      "eval_rewards/accuracies": 0.6589999794960022,
      "eval_rewards/chosen": -0.08294105529785156,
      "eval_rewards/margins": 0.14950056374073029,
      "eval_rewards/margins_max": 0.7146967649459839,
      "eval_rewards/margins_min": -0.4138467013835907,
      "eval_rewards/margins_std": 0.3812854588031769,
      "eval_rewards/rejected": -0.23244161903858185,
      "eval_runtime": 427.9931,
      "eval_samples_per_second": 4.673,
      "eval_steps_per_second": 0.292,
      "step": 300
    },
    {
      "dpo_losses": 0.45537829399108887,
      "epoch": 0.87,
      "grad_norm": 2.9848370419083263,
      "learning_rate": 4.466207261809989e-06,
      "logits/chosen": -2.788119316101074,
      "logits/rejected": -2.68644118309021,
      "logps/chosen": -267.47833251953125,
      "logps/rejected": -245.789306640625,
      "loss": 0.5975,
      "positive_losses": 2.0721895694732666,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.30337491631507874,
      "rewards/margins": 0.6190284490585327,
      "rewards/margins_max": 1.1597942113876343,
      "rewards/margins_min": 0.1627883017063141,
      "rewards/margins_std": 0.44780150055885315,
      "rewards/rejected": -0.3156535029411316,
      "step": 310
    },
    {
      "dpo_losses": 0.43933314085006714,
      "epoch": 0.9,
      "grad_norm": 19.682504946692188,
      "learning_rate": 4.414525450399713e-06,
      "logits/chosen": -2.736013650894165,
      "logits/rejected": -2.6776223182678223,
      "logps/chosen": -278.270751953125,
      "logps/rejected": -298.3285827636719,
      "loss": 0.5788,
      "positive_losses": 1.849744439125061,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.30002424120903015,
      "rewards/margins": 0.6766095757484436,
      "rewards/margins_max": 1.2602808475494385,
      "rewards/margins_min": 0.15154330432415009,
      "rewards/margins_std": 0.5036176443099976,
      "rewards/rejected": -0.37658533453941345,
      "step": 320
    },
    {
      "dpo_losses": 0.40220436453819275,
      "epoch": 0.93,
      "grad_norm": 21.985706194149614,
      "learning_rate": 4.360784949008615e-06,
      "logits/chosen": -2.803253650665283,
      "logits/rejected": -2.7153828144073486,
      "logps/chosen": -317.44683837890625,
      "logps/rejected": -311.2424011230469,
      "loss": 0.5348,
      "positive_losses": 0.7708458304405212,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3687300980091095,
      "rewards/margins": 0.8117581605911255,
      "rewards/margins_max": 1.4448413848876953,
      "rewards/margins_min": 0.24920454621315002,
      "rewards/margins_std": 0.5343087315559387,
      "rewards/rejected": -0.44302797317504883,
      "step": 330
    },
    {
      "dpo_losses": 0.42684751749038696,
      "epoch": 0.96,
      "grad_norm": 2.568835729945823,
      "learning_rate": 4.30504354481929e-06,
      "logits/chosen": -2.6699423789978027,
      "logits/rejected": -2.6169559955596924,
      "logps/chosen": -264.12371826171875,
      "logps/rejected": -287.60223388671875,
      "loss": 0.5126,
      "positive_losses": 1.0956456661224365,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.35721784830093384,
      "rewards/margins": 0.7215437293052673,
      "rewards/margins_max": 1.3981475830078125,
      "rewards/margins_min": 0.2521105110645294,
      "rewards/margins_std": 0.5147643089294434,
      "rewards/rejected": -0.3643258213996887,
      "step": 340
    },
    {
      "dpo_losses": 0.4029777944087982,
      "epoch": 0.99,
      "grad_norm": 33.74665962925535,
      "learning_rate": 4.247361176585904e-06,
      "logits/chosen": -2.6383445262908936,
      "logits/rejected": -2.59466290473938,
      "logps/chosen": -289.7620849609375,
      "logps/rejected": -337.9070739746094,
      "loss": 0.6815,
      "positive_losses": 2.7246575355529785,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.2519703507423401,
      "rewards/margins": 0.7665218710899353,
      "rewards/margins_max": 1.250725507736206,
      "rewards/margins_min": 0.27144742012023926,
      "rewards/margins_std": 0.44426339864730835,
      "rewards/rejected": -0.5145514607429504,
      "step": 350
    },
    {
      "dpo_losses": 0.403492271900177,
      "epoch": 1.01,
      "grad_norm": 18.92215966058587,
      "learning_rate": 4.187799870182038e-06,
      "logits/chosen": -2.672948122024536,
      "logits/rejected": -2.608213186264038,
      "logps/chosen": -267.0905456542969,
      "logps/rejected": -298.3930969238281,
      "loss": 0.459,
      "positive_losses": 0.638231635093689,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.33020204305648804,
      "rewards/margins": 0.7850592136383057,
      "rewards/margins_max": 1.3037402629852295,
      "rewards/margins_min": 0.2894684672355652,
      "rewards/margins_std": 0.4459232687950134,
      "rewards/rejected": -0.4548572599887848,
      "step": 360
    },
    {
      "dpo_losses": 0.37545788288116455,
      "epoch": 1.04,
      "grad_norm": 10.055654910482506,
      "learning_rate": 4.1264236719042365e-06,
      "logits/chosen": -2.6889281272888184,
      "logits/rejected": -2.6120352745056152,
      "logps/chosen": -311.17535400390625,
      "logps/rejected": -290.3823547363281,
      "loss": 0.4,
      "positive_losses": 0.17851710319519043,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.42974215745925903,
      "rewards/margins": 0.893783450126648,
      "rewards/margins_max": 1.454681158065796,
      "rewards/margins_min": 0.2657306492328644,
      "rewards/margins_std": 0.5430375933647156,
      "rewards/rejected": -0.4640413224697113,
      "step": 370
    },
    {
      "dpo_losses": 0.3519030511379242,
      "epoch": 1.07,
      "grad_norm": 2.081074851057877,
      "learning_rate": 4.063298579603001e-06,
      "logits/chosen": -2.6433703899383545,
      "logits/rejected": -2.5456502437591553,
      "logps/chosen": -250.69223022460938,
      "logps/rejected": -297.0175476074219,
      "loss": 0.3906,
      "positive_losses": 0.5173467397689819,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.3462817668914795,
      "rewards/margins": 0.9600400924682617,
      "rewards/margins_max": 1.5482779741287231,
      "rewards/margins_min": 0.45695358514785767,
      "rewards/margins_std": 0.49505558609962463,
      "rewards/rejected": -0.6137582063674927,
      "step": 380
    },
    {
      "dpo_losses": 0.3597998023033142,
      "epoch": 1.1,
      "grad_norm": 26.625294571800634,
      "learning_rate": 3.998492471715272e-06,
      "logits/chosen": -2.6424202919006348,
      "logits/rejected": -2.612208366394043,
      "logps/chosen": -267.8326721191406,
      "logps/rejected": -324.8028259277344,
      "loss": 0.4637,
      "positive_losses": 0.54234778881073,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.335815966129303,
      "rewards/margins": 0.9628133773803711,
      "rewards/margins_max": 1.653159499168396,
      "rewards/margins_min": 0.35895422101020813,
      "rewards/margins_std": 0.5844250321388245,
      "rewards/rejected": -0.6269975304603577,
      "step": 390
    },
    {
      "dpo_losses": 0.3521117568016052,
      "epoch": 1.13,
      "grad_norm": 21.15586052540318,
      "learning_rate": 3.932075034274723e-06,
      "logits/chosen": -2.6310527324676514,
      "logits/rejected": -2.5649473667144775,
      "logps/chosen": -264.017578125,
      "logps/rejected": -315.89727783203125,
      "loss": 0.4857,
      "positive_losses": 0.9953087568283081,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.2821255922317505,
      "rewards/margins": 0.9958402514457703,
      "rewards/margins_max": 1.7143990993499756,
      "rewards/margins_min": 0.42567577958106995,
      "rewards/margins_std": 0.5622113347053528,
      "rewards/rejected": -0.7137147188186646,
      "step": 400
    },
    {
      "epoch": 1.13,
      "eval_dpo_losses": 0.6294958591461182,
      "eval_logits/chosen": -2.623967170715332,
      "eval_logits/rejected": -2.5809848308563232,
      "eval_logps/chosen": -299.22857666015625,
      "eval_logps/rejected": -293.3138732910156,
      "eval_loss": 2.841282844543457,
      "eval_positive_losses": 20.44219398498535,
      "eval_rewards/accuracies": 0.6539999842643738,
      "eval_rewards/chosen": -0.1463514119386673,
      "eval_rewards/margins": 0.20099812746047974,
      "eval_rewards/margins_max": 0.960481584072113,
      "eval_rewards/margins_min": -0.5524086356163025,
      "eval_rewards/margins_std": 0.5025739669799805,
      "eval_rewards/rejected": -0.3473495543003082,
      "eval_runtime": 428.0191,
      "eval_samples_per_second": 4.673,
      "eval_steps_per_second": 0.292,
      "step": 400
    },
    {
      "dpo_losses": 0.3926815390586853,
      "epoch": 1.15,
      "grad_norm": 3.4016552654438508,
      "learning_rate": 3.864117685978339e-06,
      "logits/chosen": -2.6535449028015137,
      "logits/rejected": -2.587183952331543,
      "logps/chosen": -264.6826171875,
      "logps/rejected": -305.70025634765625,
      "loss": 0.4393,
      "positive_losses": 0.919623076915741,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.3199400305747986,
      "rewards/margins": 0.8818240165710449,
      "rewards/margins_max": 1.6596267223358154,
      "rewards/margins_min": 0.17792963981628418,
      "rewards/margins_std": 0.6805464625358582,
      "rewards/rejected": -0.5618840456008911,
      "step": 410
    },
    {
      "dpo_losses": 0.34169021248817444,
      "epoch": 1.18,
      "grad_norm": 12.095094906646299,
      "learning_rate": 3.794693501389861e-06,
      "logits/chosen": -2.631622314453125,
      "logits/rejected": -2.5662035942077637,
      "logps/chosen": -282.8951416015625,
      "logps/rejected": -335.8517761230469,
      "loss": 0.4747,
      "positive_losses": 1.60855233669281,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.35081517696380615,
      "rewards/margins": 1.0359344482421875,
      "rewards/margins_max": 1.7826063632965088,
      "rewards/margins_min": 0.3784196078777313,
      "rewards/margins_std": 0.6319265365600586,
      "rewards/rejected": -0.6851192712783813,
      "step": 420
    },
    {
      "dpo_losses": 0.34610480070114136,
      "epoch": 1.21,
      "grad_norm": 36.67211799557329,
      "learning_rate": 3.7238771323626822e-06,
      "logits/chosen": -2.6721343994140625,
      "logits/rejected": -2.5792922973632812,
      "logps/chosen": -304.9643859863281,
      "logps/rejected": -327.1585693359375,
      "loss": 0.5169,
      "positive_losses": 2.1769745349884033,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.38255181908607483,
      "rewards/margins": 1.0523736476898193,
      "rewards/margins_max": 1.7099113464355469,
      "rewards/margins_min": 0.37860894203186035,
      "rewards/margins_std": 0.6167958378791809,
      "rewards/rejected": -0.6698219180107117,
      "step": 430
    },
    {
      "dpo_losses": 0.34532538056373596,
      "epoch": 1.24,
      "grad_norm": 53.272107903374255,
      "learning_rate": 3.651744727766676e-06,
      "logits/chosen": -2.6618311405181885,
      "logits/rejected": -2.587277889251709,
      "logps/chosen": -258.0617980957031,
      "logps/rejected": -312.40118408203125,
      "loss": 0.4213,
      "positive_losses": 1.2293035984039307,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3312895894050598,
      "rewards/margins": 1.0052523612976074,
      "rewards/margins_max": 1.6782900094985962,
      "rewards/margins_min": 0.3929985761642456,
      "rewards/margins_std": 0.5877587199211121,
      "rewards/rejected": -0.6739627122879028,
      "step": 440
    },
    {
      "dpo_losses": 0.36276787519454956,
      "epoch": 1.27,
      "grad_norm": 37.36635037871641,
      "learning_rate": 3.57837385160529e-06,
      "logits/chosen": -2.603156089782715,
      "logits/rejected": -2.536878824234009,
      "logps/chosen": -278.26312255859375,
      "logps/rejected": -322.89599609375,
      "loss": 0.4124,
      "positive_losses": 1.1274974346160889,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.3354751467704773,
      "rewards/margins": 0.9750736355781555,
      "rewards/margins_max": 1.7054522037506104,
      "rewards/margins_min": 0.23124954104423523,
      "rewards/margins_std": 0.675701916217804,
      "rewards/rejected": -0.6395983695983887,
      "step": 450
    },
    {
      "dpo_losses": 0.30381661653518677,
      "epoch": 1.3,
      "grad_norm": 12.67365701095189,
      "learning_rate": 3.503843399610941e-06,
      "logits/chosen": -2.6802544593811035,
      "logits/rejected": -2.607789993286133,
      "logps/chosen": -313.99456787109375,
      "logps/rejected": -374.64959716796875,
      "loss": 0.3767,
      "positive_losses": 0.6000474691390991,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.4133946895599365,
      "rewards/margins": 1.1892845630645752,
      "rewards/margins_max": 1.8352104425430298,
      "rewards/margins_min": 0.4562619626522064,
      "rewards/margins_std": 0.6288319826126099,
      "rewards/rejected": -0.7758898735046387,
      "step": 460
    },
    {
      "dpo_losses": 0.30051547288894653,
      "epoch": 1.32,
      "grad_norm": 32.69630841523818,
      "learning_rate": 3.4282335144083985e-06,
      "logits/chosen": -2.6007492542266846,
      "logits/rejected": -2.5776376724243164,
      "logps/chosen": -279.6783447265625,
      "logps/rejected": -333.8772888183594,
      "loss": 0.6002,
      "positive_losses": 2.7248969078063965,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.36538442969322205,
      "rewards/margins": 1.2326692342758179,
      "rewards/margins_max": 1.9220224618911743,
      "rewards/margins_min": 0.5394710302352905,
      "rewards/margins_std": 0.6206193566322327,
      "rewards/rejected": -0.8672847747802734,
      "step": 470
    },
    {
      "dpo_losses": 0.30322200059890747,
      "epoch": 1.35,
      "grad_norm": 23.908435357630577,
      "learning_rate": 3.351625499337395e-06,
      "logits/chosen": -2.6286513805389404,
      "logits/rejected": -2.5465543270111084,
      "logps/chosen": -289.9736328125,
      "logps/rejected": -319.82757568359375,
      "loss": 0.4283,
      "positive_losses": 1.0004392862319946,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.41263842582702637,
      "rewards/margins": 1.192806601524353,
      "rewards/margins_max": 1.9165725708007812,
      "rewards/margins_min": 0.5354443788528442,
      "rewards/margins_std": 0.6207983493804932,
      "rewards/rejected": -0.7801682353019714,
      "step": 480
    },
    {
      "dpo_losses": 0.3252703547477722,
      "epoch": 1.38,
      "grad_norm": 3.3829438710417796,
      "learning_rate": 3.2741017310271056e-06,
      "logits/chosen": -2.5804061889648438,
      "logits/rejected": -2.5292675495147705,
      "logps/chosen": -297.5831298828125,
      "logps/rejected": -357.71923828125,
      "loss": 0.4025,
      "positive_losses": 0.23716692626476288,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.4139462113380432,
      "rewards/margins": 1.1517314910888672,
      "rewards/margins_max": 1.8896795511245728,
      "rewards/margins_min": 0.39722785353660583,
      "rewards/margins_std": 0.6692711710929871,
      "rewards/rejected": -0.737785279750824,
      "step": 490
    },
    {
      "dpo_losses": 0.3372390866279602,
      "epoch": 1.41,
      "grad_norm": 28.960920640000488,
      "learning_rate": 3.195745570816532e-06,
      "logits/chosen": -2.574615001678467,
      "logits/rejected": -2.5533788204193115,
      "logps/chosen": -265.92535400390625,
      "logps/rejected": -319.3916015625,
      "loss": 0.6015,
      "positive_losses": 2.730269432067871,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.31839805841445923,
      "rewards/margins": 1.0484989881515503,
      "rewards/margins_max": 1.724321961402893,
      "rewards/margins_min": 0.428377240896225,
      "rewards/margins_std": 0.5665310621261597,
      "rewards/rejected": -0.7301009893417358,
      "step": 500
    },
    {
      "epoch": 1.41,
      "eval_dpo_losses": 0.6214867234230042,
      "eval_logits/chosen": -2.6245663166046143,
      "eval_logits/rejected": -2.580287218093872,
      "eval_logps/chosen": -292.57025146484375,
      "eval_logps/rejected": -288.6871337890625,
      "eval_loss": 2.4296762943267822,
      "eval_positive_losses": 16.24717140197754,
      "eval_rewards/accuracies": 0.6660000085830688,
      "eval_rewards/chosen": -0.0797680914402008,
      "eval_rewards/margins": 0.2213137298822403,
      "eval_rewards/margins_max": 0.9834086894989014,
      "eval_rewards/margins_min": -0.5416373610496521,
      "eval_rewards/margins_std": 0.5124924182891846,
      "eval_rewards/rejected": -0.3010818362236023,
      "eval_runtime": 427.7973,
      "eval_samples_per_second": 4.675,
      "eval_steps_per_second": 0.292,
      "step": 500
    },
    {
      "dpo_losses": 0.3315432071685791,
      "epoch": 1.44,
      "grad_norm": 4.95221805391932,
      "learning_rate": 3.116641275116018e-06,
      "logits/chosen": -2.60896372795105,
      "logits/rejected": -2.5090444087982178,
      "logps/chosen": -299.6880798339844,
      "logps/rejected": -325.5179748535156,
      "loss": 0.3921,
      "positive_losses": 0.49366044998168945,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.40952691435813904,
      "rewards/margins": 1.0708227157592773,
      "rewards/margins_max": 1.7833086252212524,
      "rewards/margins_min": 0.4603918194770813,
      "rewards/margins_std": 0.5918254852294922,
      "rewards/rejected": -0.6612958908081055,
      "step": 510
    },
    {
      "dpo_losses": 0.3335661292076111,
      "epoch": 1.46,
      "grad_norm": 268.46332093509534,
      "learning_rate": 3.0368739048062956e-06,
      "logits/chosen": -2.6480813026428223,
      "logits/rejected": -2.555812358856201,
      "logps/chosen": -274.15338134765625,
      "logps/rejected": -321.23468017578125,
      "loss": 0.7909,
      "positive_losses": 6.10087776184082,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.2626419961452484,
      "rewards/margins": 1.086563229560852,
      "rewards/margins_max": 1.8405935764312744,
      "rewards/margins_min": 0.38931983709335327,
      "rewards/margins_std": 0.6585657000541687,
      "rewards/rejected": -0.8239210844039917,
      "step": 520
    },
    {
      "dpo_losses": 0.31472882628440857,
      "epoch": 1.49,
      "grad_norm": 12.57072460383369,
      "learning_rate": 2.956529233772492e-06,
      "logits/chosen": -2.66115140914917,
      "logits/rejected": -2.599170684814453,
      "logps/chosen": -295.13287353515625,
      "logps/rejected": -370.9165954589844,
      "loss": 0.4347,
      "positive_losses": 1.7203848361968994,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.3876939117908478,
      "rewards/margins": 1.1449024677276611,
      "rewards/margins_max": 1.7947896718978882,
      "rewards/margins_min": 0.4822443425655365,
      "rewards/margins_std": 0.5758861899375916,
      "rewards/rejected": -0.7572084069252014,
      "step": 530
    },
    {
      "dpo_losses": 0.3061717450618744,
      "epoch": 1.52,
      "grad_norm": 52.01331334980497,
      "learning_rate": 2.8756936566714317e-06,
      "logits/chosen": -2.641383647918701,
      "logits/rejected": -2.5458927154541016,
      "logps/chosen": -301.6697692871094,
      "logps/rejected": -321.75030517578125,
      "loss": 0.517,
      "positive_losses": 1.8778899908065796,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.39389947056770325,
      "rewards/margins": 1.17051100730896,
      "rewards/margins_max": 1.968420386314392,
      "rewards/margins_min": 0.5133339166641235,
      "rewards/margins_std": 0.6488900184631348,
      "rewards/rejected": -0.7766116261482239,
      "step": 540
    },
    {
      "dpo_losses": 0.28249287605285645,
      "epoch": 1.55,
      "grad_norm": 2.614844523196965,
      "learning_rate": 2.794454096031429e-06,
      "logits/chosen": -2.6530299186706543,
      "logits/rejected": -2.5759429931640625,
      "logps/chosen": -333.08905029296875,
      "logps/rejected": -405.0557861328125,
      "loss": 0.419,
      "positive_losses": 1.4631320238113403,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.3982090950012207,
      "rewards/margins": 1.2987768650054932,
      "rewards/margins_max": 2.1185762882232666,
      "rewards/margins_min": 0.46637725830078125,
      "rewards/margins_std": 0.7265578508377075,
      "rewards/rejected": -0.900567889213562,
      "step": 550
    },
    {
      "dpo_losses": 0.30363279581069946,
      "epoch": 1.58,
      "grad_norm": 1.8975860193228054,
      "learning_rate": 2.71289790878446e-06,
      "logits/chosen": -2.639392852783203,
      "logits/rejected": -2.566481113433838,
      "logps/chosen": -266.770263671875,
      "logps/rejected": -347.6555480957031,
      "loss": 0.4161,
      "positive_losses": 1.2152196168899536,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.38452547788619995,
      "rewards/margins": 1.20139741897583,
      "rewards/margins_max": 1.8644053936004639,
      "rewards/margins_min": 0.5276289582252502,
      "rewards/margins_std": 0.6155052781105042,
      "rewards/rejected": -0.8168719410896301,
      "step": 560
    },
    {
      "dpo_losses": 0.29614293575286865,
      "epoch": 1.61,
      "grad_norm": 13.926983753305802,
      "learning_rate": 2.6311127923312156e-06,
      "logits/chosen": -2.6632423400878906,
      "logits/rejected": -2.583941698074341,
      "logps/chosen": -297.9169006347656,
      "logps/rejected": -379.502685546875,
      "loss": 0.41,
      "positive_losses": 1.1672667264938354,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.37709641456604004,
      "rewards/margins": 1.2561410665512085,
      "rewards/margins_max": 1.988631248474121,
      "rewards/margins_min": 0.43355387449264526,
      "rewards/margins_std": 0.7217207551002502,
      "rewards/rejected": -0.8790447115898132,
      "step": 570
    },
    {
      "dpo_losses": 0.3167431652545929,
      "epoch": 1.63,
      "grad_norm": 19.967959106256718,
      "learning_rate": 2.549186690240057e-06,
      "logits/chosen": -2.588282346725464,
      "logits/rejected": -2.5570175647735596,
      "logps/chosen": -251.79891967773438,
      "logps/rejected": -347.29925537109375,
      "loss": 0.4115,
      "positive_losses": 1.0607062578201294,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.34643691778182983,
      "rewards/margins": 1.1596298217773438,
      "rewards/margins_max": 1.9422941207885742,
      "rewards/margins_min": 0.3299016058444977,
      "rewards/margins_std": 0.7123270630836487,
      "rewards/rejected": -0.8131929636001587,
      "step": 580
    },
    {
      "dpo_losses": 0.3140312135219574,
      "epoch": 1.66,
      "grad_norm": 15.856578320993615,
      "learning_rate": 2.4672076976812548e-06,
      "logits/chosen": -2.60188627243042,
      "logits/rejected": -2.5129995346069336,
      "logps/chosen": -274.2960510253906,
      "logps/rejected": -346.67181396484375,
      "loss": 0.5648,
      "positive_losses": 3.2257080078125,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.32442304491996765,
      "rewards/margins": 1.1771596670150757,
      "rewards/margins_max": 1.9481910467147827,
      "rewards/margins_min": 0.39941343665122986,
      "rewards/margins_std": 0.6915642619132996,
      "rewards/rejected": -0.8527366518974304,
      "step": 590
    },
    {
      "dpo_losses": 0.3060692846775055,
      "epoch": 1.69,
      "grad_norm": 3.4425530487605087,
      "learning_rate": 2.3852639666982218e-06,
      "logits/chosen": -2.625056743621826,
      "logits/rejected": -2.5557961463928223,
      "logps/chosen": -256.4402770996094,
      "logps/rejected": -375.25640869140625,
      "loss": 0.4849,
      "positive_losses": 0.7688635587692261,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3693460524082184,
      "rewards/margins": 1.2135040760040283,
      "rewards/margins_max": 1.9841582775115967,
      "rewards/margins_min": 0.4423845410346985,
      "rewards/margins_std": 0.6956911087036133,
      "rewards/rejected": -0.8441581726074219,
      "step": 600
    },
    {
      "epoch": 1.69,
      "eval_dpo_losses": 0.6153340935707092,
      "eval_logits/chosen": -2.612006664276123,
      "eval_logits/rejected": -2.5658888816833496,
      "eval_logps/chosen": -308.942138671875,
      "eval_logps/rejected": -310.1337585449219,
      "eval_loss": 3.8076722621917725,
      "eval_positive_losses": 30.076915740966797,
      "eval_rewards/accuracies": 0.6629999876022339,
      "eval_rewards/chosen": -0.24348701536655426,
      "eval_rewards/margins": 0.27206119894981384,
      "eval_rewards/margins_max": 1.1650824546813965,
      "eval_rewards/margins_min": -0.6778736710548401,
      "eval_rewards/margins_std": 0.6337091326713562,
      "eval_rewards/rejected": -0.5155482292175293,
      "eval_runtime": 428.1488,
      "eval_samples_per_second": 4.671,
      "eval_steps_per_second": 0.292,
      "step": 600
    },
    {
      "dpo_losses": 0.30735138058662415,
      "epoch": 1.72,
      "grad_norm": 3.0019110314936746,
      "learning_rate": 2.303443611417584e-06,
      "logits/chosen": -2.6273159980773926,
      "logits/rejected": -2.5301547050476074,
      "logps/chosen": -290.5046081542969,
      "logps/rejected": -325.96075439453125,
      "loss": 0.5284,
      "positive_losses": 2.10502028465271,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.3719572424888611,
      "rewards/margins": 1.2126883268356323,
      "rewards/margins_max": 1.9886195659637451,
      "rewards/margins_min": 0.4652635455131531,
      "rewards/margins_std": 0.6831039190292358,
      "rewards/rejected": -0.840731143951416,
      "step": 610
    },
    {
      "dpo_losses": 0.2810274362564087,
      "epoch": 1.75,
      "grad_norm": 3.0926491676565164,
      "learning_rate": 2.2218346133000264e-06,
      "logits/chosen": -2.6870579719543457,
      "logits/rejected": -2.6051993370056152,
      "logps/chosen": -304.4597473144531,
      "logps/rejected": -379.030029296875,
      "loss": 0.5267,
      "positive_losses": 3.393756866455078,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.37165260314941406,
      "rewards/margins": 1.3295249938964844,
      "rewards/margins_max": 2.1557211875915527,
      "rewards/margins_min": 0.4981165826320648,
      "rewards/margins_std": 0.7513960599899292,
      "rewards/rejected": -0.9578725695610046,
      "step": 620
    },
    {
      "dpo_losses": 0.27874812483787537,
      "epoch": 1.77,
      "grad_norm": 97.52454618902361,
      "learning_rate": 2.140524726533792e-06,
      "logits/chosen": -2.5727076530456543,
      "logits/rejected": -2.5339112281799316,
      "logps/chosen": -332.1949157714844,
      "logps/rejected": -376.3540954589844,
      "loss": 0.4514,
      "positive_losses": 1.4352390766143799,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.4186096787452698,
      "rewards/margins": 1.357527494430542,
      "rewards/margins_max": 2.143828868865967,
      "rewards/margins_min": 0.596580982208252,
      "rewards/margins_std": 0.7103701233863831,
      "rewards/rejected": -0.9389175176620483,
      "step": 630
    },
    {
      "dpo_losses": 0.30733007192611694,
      "epoch": 1.8,
      "grad_norm": 105.97606230300312,
      "learning_rate": 2.059601383672566e-06,
      "logits/chosen": -2.6715691089630127,
      "logits/rejected": -2.5807442665100098,
      "logps/chosen": -274.9263000488281,
      "logps/rejected": -315.69342041015625,
      "loss": 0.5271,
      "positive_losses": 2.661163091659546,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3693527579307556,
      "rewards/margins": 1.19617760181427,
      "rewards/margins_max": 1.8424479961395264,
      "rewards/margins_min": 0.5671444535255432,
      "rewards/margins_std": 0.5639213919639587,
      "rewards/rejected": -0.8268247842788696,
      "step": 640
    },
    {
      "dpo_losses": 0.30703574419021606,
      "epoch": 1.83,
      "grad_norm": 21.752524744582452,
      "learning_rate": 1.9791516016192214e-06,
      "logits/chosen": -2.6271615028381348,
      "logits/rejected": -2.5591979026794434,
      "logps/chosen": -293.38482666015625,
      "logps/rejected": -375.1172790527344,
      "loss": 0.3758,
      "positive_losses": 0.48926544189453125,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3680533170700073,
      "rewards/margins": 1.2065770626068115,
      "rewards/margins_max": 1.9407564401626587,
      "rewards/margins_min": 0.43435949087142944,
      "rewards/margins_std": 0.6759322881698608,
      "rewards/rejected": -0.8385237455368042,
      "step": 650
    },
    {
      "dpo_losses": 0.3295736312866211,
      "epoch": 1.86,
      "grad_norm": 5.6897035459178085,
      "learning_rate": 1.8992618880565039e-06,
      "logits/chosen": -2.5168254375457764,
      "logits/rejected": -2.4887542724609375,
      "logps/chosen": -222.0303955078125,
      "logps/rejected": -301.1322326660156,
      "loss": 0.6442,
      "positive_losses": 5.43920373916626,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.24137599766254425,
      "rewards/margins": 1.1015827655792236,
      "rewards/margins_max": 1.7651869058609009,
      "rewards/margins_min": 0.37374579906463623,
      "rewards/margins_std": 0.6391667723655701,
      "rewards/rejected": -0.8602069020271301,
      "step": 660
    },
    {
      "dpo_losses": 0.2765592634677887,
      "epoch": 1.89,
      "grad_norm": 20.92314981646376,
      "learning_rate": 1.8200181484252888e-06,
      "logits/chosen": -2.6348729133605957,
      "logits/rejected": -2.5967307090759277,
      "logps/chosen": -266.20751953125,
      "logps/rejected": -333.8230285644531,
      "loss": 0.4481,
      "positive_losses": 0.6337646245956421,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.39696359634399414,
      "rewards/margins": 1.3200980424880981,
      "rewards/margins_max": 2.1041204929351807,
      "rewards/margins_min": 0.6077708005905151,
      "rewards/margins_std": 0.6646975874900818,
      "rewards/rejected": -0.9231343269348145,
      "step": 670
    },
    {
      "dpo_losses": 0.27144327759742737,
      "epoch": 1.92,
      "grad_norm": 12.752481854801099,
      "learning_rate": 1.7415055935504234e-06,
      "logits/chosen": -2.647383213043213,
      "logits/rejected": -2.569831371307373,
      "logps/chosen": -275.3692932128906,
      "logps/rejected": -349.95391845703125,
      "loss": 0.3911,
      "positive_losses": 1.6107314825057983,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3534119725227356,
      "rewards/margins": 1.331969976425171,
      "rewards/margins_max": 1.9460681676864624,
      "rewards/margins_min": 0.6937587857246399,
      "rewards/margins_std": 0.5510456562042236,
      "rewards/rejected": -0.9785581827163696,
      "step": 680
    },
    {
      "dpo_losses": 0.31174060702323914,
      "epoch": 1.94,
      "grad_norm": 170.68997160422168,
      "learning_rate": 1.6638086480134954e-06,
      "logits/chosen": -2.568826198577881,
      "logits/rejected": -2.5301690101623535,
      "logps/chosen": -205.6598358154297,
      "logps/rejected": -288.9075927734375,
      "loss": 0.4249,
      "positive_losses": 1.04084050655365,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.3269308805465698,
      "rewards/margins": 1.1985323429107666,
      "rewards/margins_max": 1.9826653003692627,
      "rewards/margins_min": 0.42818838357925415,
      "rewards/margins_std": 0.7044063806533813,
      "rewards/rejected": -0.871601402759552,
      "step": 690
    },
    {
      "dpo_losses": 0.3133111298084259,
      "epoch": 1.97,
      "grad_norm": 20.526148796689263,
      "learning_rate": 1.5870108593710473e-06,
      "logits/chosen": -2.510403871536255,
      "logits/rejected": -2.4163358211517334,
      "logps/chosen": -260.0558166503906,
      "logps/rejected": -288.7188720703125,
      "loss": 0.4012,
      "positive_losses": 0.6993476152420044,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.39475178718566895,
      "rewards/margins": 1.1830546855926514,
      "rewards/margins_max": 1.8831546306610107,
      "rewards/margins_min": 0.4869031012058258,
      "rewards/margins_std": 0.6312520503997803,
      "rewards/rejected": -0.7883028984069824,
      "step": 700
    },
    {
      "epoch": 1.97,
      "eval_dpo_losses": 0.6160134673118591,
      "eval_logits/chosen": -2.5672061443328857,
      "eval_logits/rejected": -2.5207982063293457,
      "eval_logps/chosen": -316.204345703125,
      "eval_logps/rejected": -318.60394287109375,
      "eval_loss": 4.435854434967041,
      "eval_positive_losses": 36.7813835144043,
      "eval_rewards/accuracies": 0.6660000085830688,
      "eval_rewards/chosen": -0.3161095082759857,
      "eval_rewards/margins": 0.2841406464576721,
      "eval_rewards/margins_max": 1.2285473346710205,
      "eval_rewards/margins_min": -0.7319729328155518,
      "eval_rewards/margins_std": 0.6758950352668762,
      "eval_rewards/rejected": -0.6002501249313354,
      "eval_runtime": 428.0426,
      "eval_samples_per_second": 4.672,
      "eval_steps_per_second": 0.292,
      "step": 700
    },
    {
      "dpo_losses": 0.2887941598892212,
      "epoch": 2.0,
      "grad_norm": 4.587904967849547,
      "learning_rate": 1.511194808315853e-06,
      "logits/chosen": -2.50614595413208,
      "logits/rejected": -2.467355251312256,
      "logps/chosen": -220.85116577148438,
      "logps/rejected": -320.2411804199219,
      "loss": 0.3757,
      "positive_losses": 1.769884705543518,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.302704393863678,
      "rewards/margins": 1.2528822422027588,
      "rewards/margins_max": 1.8332780599594116,
      "rewards/margins_min": 0.6378909945487976,
      "rewards/margins_std": 0.5379610061645508,
      "rewards/rejected": -0.950177788734436,
      "step": 710
    },
    {
      "dpo_losses": 0.24147991836071014,
      "epoch": 2.03,
      "grad_norm": 23.216671918698182,
      "learning_rate": 1.4364420198778662e-06,
      "logits/chosen": -2.6771039962768555,
      "logits/rejected": -2.5842394828796387,
      "logps/chosen": -323.87799072265625,
      "logps/rejected": -410.9130859375,
      "loss": 0.4475,
      "positive_losses": 0.7118709683418274,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.401747465133667,
      "rewards/margins": 1.474344253540039,
      "rewards/margins_max": 2.140209674835205,
      "rewards/margins_min": 0.758759081363678,
      "rewards/margins_std": 0.6255311965942383,
      "rewards/rejected": -1.072596788406372,
      "step": 720
    },
    {
      "dpo_losses": 0.2333422601222992,
      "epoch": 2.06,
      "grad_norm": 3.3962167784716084,
      "learning_rate": 1.3628328757603243e-06,
      "logits/chosen": -2.596156597137451,
      "logits/rejected": -2.5163309574127197,
      "logps/chosen": -296.4609069824219,
      "logps/rejected": -384.97747802734375,
      "loss": 0.246,
      "positive_losses": 0.371989905834198,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.4179394245147705,
      "rewards/margins": 1.5231090784072876,
      "rewards/margins_max": 2.151522397994995,
      "rewards/margins_min": 0.7981916666030884,
      "rewards/margins_std": 0.6200301647186279,
      "rewards/rejected": -1.1051695346832275,
      "step": 730
    },
    {
      "dpo_losses": 0.2638745903968811,
      "epoch": 2.08,
      "grad_norm": 43.270622849201196,
      "learning_rate": 1.2904465279052725e-06,
      "logits/chosen": -2.5701091289520264,
      "logits/rejected": -2.4919536113739014,
      "logps/chosen": -284.53875732421875,
      "logps/rejected": -374.70904541015625,
      "loss": 0.4198,
      "positive_losses": 2.1527743339538574,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.35879984498023987,
      "rewards/margins": 1.408245325088501,
      "rewards/margins_max": 2.17195725440979,
      "rewards/margins_min": 0.658360481262207,
      "rewards/margins_std": 0.7011166214942932,
      "rewards/rejected": -1.0494455099105835,
      "step": 740
    },
    {
      "dpo_losses": 0.2600082457065582,
      "epoch": 2.11,
      "grad_norm": 7.673688188705642,
      "learning_rate": 1.219360813381446e-06,
      "logits/chosen": -2.529010057449341,
      "logits/rejected": -2.5041661262512207,
      "logps/chosen": -248.3673553466797,
      "logps/rejected": -341.32403564453125,
      "loss": 0.3417,
      "positive_losses": 0.5860393643379211,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.38368460536003113,
      "rewards/margins": 1.4487890005111694,
      "rewards/margins_max": 2.2007343769073486,
      "rewards/margins_min": 0.7243565320968628,
      "rewards/margins_std": 0.6694687604904175,
      "rewards/rejected": -1.065104603767395,
      "step": 750
    },
    {
      "dpo_losses": 0.24842312932014465,
      "epoch": 2.14,
      "grad_norm": 97.34256118856746,
      "learning_rate": 1.1496521706860392e-06,
      "logits/chosen": -2.495732307434082,
      "logits/rejected": -2.436852216720581,
      "logps/chosen": -251.5,
      "logps/rejected": -371.1795654296875,
      "loss": 0.3739,
      "positive_losses": 1.177607536315918,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.3628827929496765,
      "rewards/margins": 1.469103217124939,
      "rewards/margins_max": 2.268505096435547,
      "rewards/margins_min": 0.7123178243637085,
      "rewards/margins_std": 0.6905183792114258,
      "rewards/rejected": -1.1062204837799072,
      "step": 760
    },
    {
      "dpo_losses": 0.23375526070594788,
      "epoch": 2.17,
      "grad_norm": 20.95642655523226,
      "learning_rate": 1.0813955575503588e-06,
      "logits/chosen": -2.5345511436462402,
      "logits/rejected": -2.4596359729766846,
      "logps/chosen": -265.13592529296875,
      "logps/rejected": -327.0379333496094,
      "loss": 0.4462,
      "positive_losses": 2.026779890060425,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.39701634645462036,
      "rewards/margins": 1.5435272455215454,
      "rewards/margins_max": 2.2371368408203125,
      "rewards/margins_min": 0.8194782137870789,
      "rewards/margins_std": 0.6429990530014038,
      "rewards/rejected": -1.1465108394622803,
      "step": 770
    },
    {
      "dpo_losses": 0.25705593824386597,
      "epoch": 2.2,
      "grad_norm": 6.445962325196414,
      "learning_rate": 1.0146643703377488e-06,
      "logits/chosen": -2.586146593093872,
      "logits/rejected": -2.476118564605713,
      "logps/chosen": -282.3785095214844,
      "logps/rejected": -333.51214599609375,
      "loss": 0.4776,
      "positive_losses": 2.5824341773986816,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.3519349992275238,
      "rewards/margins": 1.417069673538208,
      "rewards/margins_max": 2.1547417640686035,
      "rewards/margins_min": 0.6326209902763367,
      "rewards/margins_std": 0.6806846857070923,
      "rewards/rejected": -1.0651346445083618,
      "step": 780
    },
    {
      "dpo_losses": 0.23853960633277893,
      "epoch": 2.23,
      "grad_norm": 41.08497708161555,
      "learning_rate": 9.495303651204496e-07,
      "logits/chosen": -2.541194200515747,
      "logits/rejected": -2.441011428833008,
      "logps/chosen": -287.2564697265625,
      "logps/rejected": -357.0499572753906,
      "loss": 0.2993,
      "positive_losses": 1.2920652627944946,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.35095715522766113,
      "rewards/margins": 1.4691743850708008,
      "rewards/margins_max": 2.056344509124756,
      "rewards/margins_min": 0.7152417898178101,
      "rewards/margins_std": 0.5961428880691528,
      "rewards/rejected": -1.1182173490524292,
      "step": 790
    },
    {
      "dpo_losses": 0.21679171919822693,
      "epoch": 2.25,
      "grad_norm": 56.83316198119941,
      "learning_rate": 8.860635805202616e-07,
      "logits/chosen": -2.567403554916382,
      "logits/rejected": -2.521256923675537,
      "logps/chosen": -284.0165710449219,
      "logps/rejected": -438.0687561035156,
      "loss": 0.3245,
      "positive_losses": 0.24880298972129822,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.37534329295158386,
      "rewards/margins": 1.6406404972076416,
      "rewards/margins_max": 2.3722925186157227,
      "rewards/margins_min": 0.8349443674087524,
      "rewards/margins_std": 0.6923807263374329,
      "rewards/rejected": -1.2652971744537354,
      "step": 800
    },
    {
      "epoch": 2.25,
      "eval_dpo_losses": 0.6122725009918213,
      "eval_logits/chosen": -2.5420889854431152,
      "eval_logits/rejected": -2.495191812515259,
      "eval_logps/chosen": -322.1156311035156,
      "eval_logps/rejected": -328.4566955566406,
      "eval_loss": 4.987257957458496,
      "eval_positive_losses": 41.807315826416016,
      "eval_rewards/accuracies": 0.6660000085830688,
      "eval_rewards/chosen": -0.37522199749946594,
      "eval_rewards/margins": 0.32355546951293945,
      "eval_rewards/margins_max": 1.3767565488815308,
      "eval_rewards/margins_min": -0.8213516473770142,
      "eval_rewards/margins_std": 0.7506230473518372,
      "eval_rewards/rejected": -0.6987775564193726,
      "eval_runtime": 427.7547,
      "eval_samples_per_second": 4.676,
      "eval_steps_per_second": 0.292,
      "step": 800
    },
    {
      "dpo_losses": 0.23845510184764862,
      "epoch": 2.28,
      "grad_norm": 45.330745249154575,
      "learning_rate": 8.24332262395994e-07,
      "logits/chosen": -2.6028668880462646,
      "logits/rejected": -2.5543370246887207,
      "logps/chosen": -261.5367736816406,
      "logps/rejected": -386.88311767578125,
      "loss": 0.322,
      "positive_losses": 0.35725727677345276,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.33237648010253906,
      "rewards/margins": 1.502893328666687,
      "rewards/margins_max": 2.2335424423217773,
      "rewards/margins_min": 0.7811623811721802,
      "rewards/margins_std": 0.6627609133720398,
      "rewards/rejected": -1.1705169677734375,
      "step": 810
    },
    {
      "dpo_losses": 0.2232833206653595,
      "epoch": 2.31,
      "grad_norm": 125.42519869926457,
      "learning_rate": 7.644027904586587e-07,
      "logits/chosen": -2.522122383117676,
      "logits/rejected": -2.4744601249694824,
      "logps/chosen": -291.8446350097656,
      "logps/rejected": -407.2629699707031,
      "loss": 0.4412,
      "positive_losses": 1.212071418762207,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.38060301542282104,
      "rewards/margins": 1.6479623317718506,
      "rewards/margins_max": 2.4151339530944824,
      "rewards/margins_min": 0.8389755487442017,
      "rewards/margins_std": 0.7126222848892212,
      "rewards/rejected": -1.2673594951629639,
      "step": 820
    },
    {
      "dpo_losses": 0.19962266087532043,
      "epoch": 2.34,
      "grad_norm": 1.2244041063978244,
      "learning_rate": 7.06339606893347e-07,
      "logits/chosen": -2.5285003185272217,
      "logits/rejected": -2.4152493476867676,
      "logps/chosen": -344.0813903808594,
      "logps/rejected": -399.686279296875,
      "loss": 0.2693,
      "positive_losses": 0.9188915491104126,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.4107728898525238,
      "rewards/margins": 1.713904619216919,
      "rewards/margins_max": 2.3771450519561768,
      "rewards/margins_min": 1.0773948431015015,
      "rewards/margins_std": 0.5749639868736267,
      "rewards/rejected": -1.3031316995620728,
      "step": 830
    },
    {
      "dpo_losses": 0.23649196326732635,
      "epoch": 2.37,
      "grad_norm": 26.912374085283705,
      "learning_rate": 6.502051470645149e-07,
      "logits/chosen": -2.5514166355133057,
      "logits/rejected": -2.4797396659851074,
      "logps/chosen": -291.35321044921875,
      "logps/rejected": -364.1729431152344,
      "loss": 0.2924,
      "positive_losses": 0.6994352340698242,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.34709078073501587,
      "rewards/margins": 1.5194952487945557,
      "rewards/margins_max": 2.184370279312134,
      "rewards/margins_min": 0.7820354700088501,
      "rewards/margins_std": 0.625164270401001,
      "rewards/rejected": -1.172404408454895,
      "step": 840
    },
    {
      "dpo_losses": 0.24174419045448303,
      "epoch": 2.39,
      "grad_norm": 3.693058286081406,
      "learning_rate": 5.960597723792194e-07,
      "logits/chosen": -2.5517287254333496,
      "logits/rejected": -2.4682512283325195,
      "logps/chosen": -261.80987548828125,
      "logps/rejected": -360.38836669921875,
      "loss": 0.3853,
      "positive_losses": 1.5493557453155518,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.34312885999679565,
      "rewards/margins": 1.5497392416000366,
      "rewards/margins_max": 2.327932119369507,
      "rewards/margins_min": 0.7185215353965759,
      "rewards/margins_std": 0.7230452299118042,
      "rewards/rejected": -1.2066104412078857,
      "step": 850
    },
    {
      "dpo_losses": 0.2026406079530716,
      "epoch": 2.42,
      "grad_norm": 2.222119311757867,
      "learning_rate": 5.43961705380465e-07,
      "logits/chosen": -2.5729174613952637,
      "logits/rejected": -2.5054492950439453,
      "logps/chosen": -306.1053771972656,
      "logps/rejected": -388.9463806152344,
      "loss": 0.3711,
      "positive_losses": 0.9561820030212402,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.41304224729537964,
      "rewards/margins": 1.733215093612671,
      "rewards/margins_max": 2.5817983150482178,
      "rewards/margins_min": 0.8288100957870483,
      "rewards/margins_std": 0.7729387879371643,
      "rewards/rejected": -1.320172905921936,
      "step": 860
    },
    {
      "dpo_losses": 0.21425673365592957,
      "epoch": 2.45,
      "grad_norm": 66.44963288790616,
      "learning_rate": 4.939669671404871e-07,
      "logits/chosen": -2.5059876441955566,
      "logits/rejected": -2.4485790729522705,
      "logps/chosen": -279.19970703125,
      "logps/rejected": -408.12322998046875,
      "loss": 0.3434,
      "positive_losses": 1.1921786069869995,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.33581799268722534,
      "rewards/margins": 1.6330255270004272,
      "rewards/margins_max": 2.296076536178589,
      "rewards/margins_min": 0.9167500734329224,
      "rewards/margins_std": 0.6031621098518372,
      "rewards/rejected": -1.2972075939178467,
      "step": 870
    },
    {
      "dpo_losses": 0.25025516748428345,
      "epoch": 2.48,
      "grad_norm": 123.5361354333723,
      "learning_rate": 4.461293170212644e-07,
      "logits/chosen": -2.6370673179626465,
      "logits/rejected": -2.532200813293457,
      "logps/chosen": -293.84722900390625,
      "logps/rejected": -376.1434020996094,
      "loss": 0.4365,
      "positive_losses": 2.6627707481384277,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.3460078835487366,
      "rewards/margins": 1.482655644416809,
      "rewards/margins_max": 2.2597808837890625,
      "rewards/margins_min": 0.659686267375946,
      "rewards/margins_std": 0.7317071557044983,
      "rewards/rejected": -1.1366477012634277,
      "step": 880
    },
    {
      "dpo_losses": 0.23249849677085876,
      "epoch": 2.51,
      "grad_norm": 1.9172572041643672,
      "learning_rate": 4.005001948670606e-07,
      "logits/chosen": -2.6003735065460205,
      "logits/rejected": -2.5068199634552,
      "logps/chosen": -309.77880859375,
      "logps/rejected": -387.4895324707031,
      "loss": 0.2846,
      "positive_losses": 1.0592377185821533,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.34349530935287476,
      "rewards/margins": 1.5754855871200562,
      "rewards/margins_max": 2.404482126235962,
      "rewards/margins_min": 0.7314172983169556,
      "rewards/margins_std": 0.7336990833282471,
      "rewards/rejected": -1.2319902181625366,
      "step": 890
    },
    {
      "dpo_losses": 0.21542592346668243,
      "epoch": 2.54,
      "grad_norm": 4.647113412719633,
      "learning_rate": 3.571286656911377e-07,
      "logits/chosen": -2.5903160572052,
      "logits/rejected": -2.48293399810791,
      "logps/chosen": -291.11273193359375,
      "logps/rejected": -415.226318359375,
      "loss": 0.3018,
      "positive_losses": 1.301671028137207,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.37233322858810425,
      "rewards/margins": 1.68588125705719,
      "rewards/margins_max": 2.4951331615448,
      "rewards/margins_min": 0.911718487739563,
      "rewards/margins_std": 0.7062323093414307,
      "rewards/rejected": -1.3135478496551514,
      "step": 900
    },
    {
      "epoch": 2.54,
      "eval_dpo_losses": 0.6084067225456238,
      "eval_logits/chosen": -2.527609348297119,
      "eval_logits/rejected": -2.4804465770721436,
      "eval_logps/chosen": -322.69512939453125,
      "eval_logps/rejected": -330.5147399902344,
      "eval_loss": 5.034203052520752,
      "eval_positive_losses": 42.12240219116211,
      "eval_rewards/accuracies": 0.6679999828338623,
      "eval_rewards/chosen": -0.38101711869239807,
      "eval_rewards/margins": 0.33834153413772583,
      "eval_rewards/margins_max": 1.4141428470611572,
      "eval_rewards/margins_min": -0.8335634469985962,
      "eval_rewards/margins_std": 0.7645081877708435,
      "eval_rewards/rejected": -0.7193585634231567,
      "eval_runtime": 428.4002,
      "eval_samples_per_second": 4.669,
      "eval_steps_per_second": 0.292,
      "step": 900
    },
    {
      "dpo_losses": 0.23167657852172852,
      "epoch": 2.56,
      "grad_norm": 126.34610358307454,
      "learning_rate": 3.1606136691612555e-07,
      "logits/chosen": -2.546440601348877,
      "logits/rejected": -2.4756686687469482,
      "logps/chosen": -282.63519287109375,
      "logps/rejected": -346.7596435546875,
      "loss": 0.2519,
      "positive_losses": 0.3456798493862152,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.4117736220359802,
      "rewards/margins": 1.565037488937378,
      "rewards/margins_max": 2.315796375274658,
      "rewards/margins_min": 0.9116442799568176,
      "rewards/margins_std": 0.6392379999160767,
      "rewards/rejected": -1.1532639265060425,
      "step": 910
    },
    {
      "dpo_losses": 0.24891960620880127,
      "epoch": 2.59,
      "grad_norm": 11.00927447430954,
      "learning_rate": 2.773424582247844e-07,
      "logits/chosen": -2.516782522201538,
      "logits/rejected": -2.4053750038146973,
      "logps/chosen": -279.5264587402344,
      "logps/rejected": -330.05859375,
      "loss": 0.5011,
      "positive_losses": 4.046810150146484,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.33952027559280396,
      "rewards/margins": 1.5328131914138794,
      "rewards/margins_max": 2.476053237915039,
      "rewards/margins_min": 0.6028861403465271,
      "rewards/margins_std": 0.8224283456802368,
      "rewards/rejected": -1.1932929754257202,
      "step": 920
    },
    {
      "dpo_losses": 0.2301318347454071,
      "epoch": 2.62,
      "grad_norm": 148.93749344313875,
      "learning_rate": 2.410135740750821e-07,
      "logits/chosen": -2.549276351928711,
      "logits/rejected": -2.4753193855285645,
      "logps/chosen": -247.61831665039062,
      "logps/rejected": -352.13970947265625,
      "loss": 0.3682,
      "positive_losses": 1.7548551559448242,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.34124690294265747,
      "rewards/margins": 1.578126311302185,
      "rewards/margins_max": 2.319645404815674,
      "rewards/margins_min": 0.9039157032966614,
      "rewards/margins_std": 0.6513063311576843,
      "rewards/rejected": -1.2368793487548828,
      "step": 930
    },
    {
      "dpo_losses": 0.24650125205516815,
      "epoch": 2.65,
      "grad_norm": 11.35506158970768,
      "learning_rate": 2.0711377893064182e-07,
      "logits/chosen": -2.600371837615967,
      "logits/rejected": -2.5059494972229004,
      "logps/chosen": -316.21978759765625,
      "logps/rejected": -412.1728515625,
      "loss": 0.3137,
      "positive_losses": 1.2988446950912476,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.3205033838748932,
      "rewards/margins": 1.5079448223114014,
      "rewards/margins_max": 2.3231558799743652,
      "rewards/margins_min": 0.7083476781845093,
      "rewards/margins_std": 0.7245646715164185,
      "rewards/rejected": -1.1874414682388306,
      "step": 940
    },
    {
      "dpo_losses": 0.2975192964076996,
      "epoch": 2.68,
      "grad_norm": 15.828059445816866,
      "learning_rate": 1.756795252547111e-07,
      "logits/chosen": -2.5015082359313965,
      "logits/rejected": -2.4382565021514893,
      "logps/chosen": -235.37344360351562,
      "logps/rejected": -310.40557861328125,
      "loss": 0.4552,
      "positive_losses": 1.4398473501205444,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.2754918038845062,
      "rewards/margins": 1.3022301197052002,
      "rewards/margins_max": 2.1911404132843018,
      "rewards/margins_min": 0.44522780179977417,
      "rewards/margins_std": 0.7827215194702148,
      "rewards/rejected": -1.0267384052276611,
      "step": 950
    },
    {
      "dpo_losses": 0.24521894752979279,
      "epoch": 2.7,
      "grad_norm": 1.9786682371307234,
      "learning_rate": 1.4674461431281013e-07,
      "logits/chosen": -2.611396312713623,
      "logits/rejected": -2.5341451168060303,
      "logps/chosen": -249.8568572998047,
      "logps/rejected": -331.5431213378906,
      "loss": 0.3377,
      "positive_losses": 0.8475897908210754,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.33382731676101685,
      "rewards/margins": 1.483553171157837,
      "rewards/margins_max": 2.2694802284240723,
      "rewards/margins_min": 0.754567563533783,
      "rewards/margins_std": 0.6848300695419312,
      "rewards/rejected": -1.1497260332107544,
      "step": 960
    },
    {
      "dpo_losses": 0.22631379961967468,
      "epoch": 2.73,
      "grad_norm": 7.689855199276502,
      "learning_rate": 1.2034015982622243e-07,
      "logits/chosen": -2.581221103668213,
      "logits/rejected": -2.4607036113739014,
      "logps/chosen": -314.9461669921875,
      "logps/rejected": -405.8033142089844,
      "loss": 0.2875,
      "positive_losses": 0.16267403960227966,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.374321848154068,
      "rewards/margins": 1.6238701343536377,
      "rewards/margins_max": 2.506998062133789,
      "rewards/margins_min": 0.6969125270843506,
      "rewards/margins_std": 0.7903028130531311,
      "rewards/rejected": -1.249548316001892,
      "step": 970
    },
    {
      "dpo_losses": 0.21151915192604065,
      "epoch": 2.76,
      "grad_norm": 31.254181360884537,
      "learning_rate": 9.649455451539419e-08,
      "logits/chosen": -2.498505115509033,
      "logits/rejected": -2.43967342376709,
      "logps/chosen": -256.50653076171875,
      "logps/rejected": -356.4197692871094,
      "loss": 0.3825,
      "positive_losses": 1.7091434001922607,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.3194945156574249,
      "rewards/margins": 1.6285871267318726,
      "rewards/margins_max": 2.3824660778045654,
      "rewards/margins_min": 0.9284858703613281,
      "rewards/margins_std": 0.6376523971557617,
      "rewards/rejected": -1.30909264087677,
      "step": 980
    },
    {
      "dpo_losses": 0.23487678170204163,
      "epoch": 2.79,
      "grad_norm": 1.7809863648318363,
      "learning_rate": 7.523343956923196e-08,
      "logits/chosen": -2.5769989490509033,
      "logits/rejected": -2.469118595123291,
      "logps/chosen": -281.84222412109375,
      "logps/rejected": -380.6399230957031,
      "loss": 0.4188,
      "positive_losses": 1.5458621978759766,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3675388693809509,
      "rewards/margins": 1.5846505165100098,
      "rewards/margins_max": 2.5048413276672363,
      "rewards/margins_min": 0.7771235108375549,
      "rewards/margins_std": 0.7780028581619263,
      "rewards/rejected": -1.217111587524414,
      "step": 990
    },
    {
      "dpo_losses": 0.25376203656196594,
      "epoch": 2.82,
      "grad_norm": 1.5753701970833003,
      "learning_rate": 5.657967707312195e-08,
      "logits/chosen": -2.519397020339966,
      "logits/rejected": -2.498129367828369,
      "logps/chosen": -262.6558532714844,
      "logps/rejected": -383.01519775390625,
      "loss": 0.4364,
      "positive_losses": 2.2228844165802,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.30776503682136536,
      "rewards/margins": 1.4387822151184082,
      "rewards/margins_max": 2.223803997039795,
      "rewards/margins_min": 0.6689078211784363,
      "rewards/margins_std": 0.686796247959137,
      "rewards/rejected": -1.1310170888900757,
      "step": 1000
    },
    {
      "epoch": 2.82,
      "eval_dpo_losses": 0.6098241806030273,
      "eval_logits/chosen": -2.5284554958343506,
      "eval_logits/rejected": -2.4815618991851807,
      "eval_logps/chosen": -323.31011962890625,
      "eval_logps/rejected": -330.9999694824219,
      "eval_loss": 5.09750509262085,
      "eval_positive_losses": 42.87459945678711,
      "eval_rewards/accuracies": 0.6679999828338623,
      "eval_rewards/chosen": -0.3871670663356781,
      "eval_rewards/margins": 0.33704349398612976,
      "eval_rewards/margins_max": 1.4157270193099976,
      "eval_rewards/margins_min": -0.8368792533874512,
      "eval_rewards/margins_std": 0.7695222496986389,
      "eval_rewards/rejected": -0.7242105007171631,
      "eval_runtime": 428.3697,
      "eval_samples_per_second": 4.669,
      "eval_steps_per_second": 0.292,
      "step": 1000
    },
    {
      "dpo_losses": 0.22761714458465576,
      "epoch": 2.85,
      "grad_norm": 73.10678525814994,
      "learning_rate": 4.055332542531959e-08,
      "logits/chosen": -2.6184372901916504,
      "logits/rejected": -2.5384011268615723,
      "logps/chosen": -288.6739501953125,
      "logps/rejected": -428.9107360839844,
      "loss": 0.4915,
      "positive_losses": 1.0805866718292236,
      "rewards/accuracies": 1.0,
      "rewards/chosen": 0.350180983543396,
      "rewards/margins": 1.5830507278442383,
      "rewards/margins_max": 2.3764092922210693,
      "rewards/margins_min": 0.8162466883659363,
      "rewards/margins_std": 0.6986463665962219,
      "rewards/rejected": -1.2328697443008423,
      "step": 1010
    },
    {
      "dpo_losses": 0.2362273633480072,
      "epoch": 2.87,
      "grad_norm": 3.7756318529859643,
      "learning_rate": 2.7171617768147472e-08,
      "logits/chosen": -2.5988545417785645,
      "logits/rejected": -2.529019355773926,
      "logps/chosen": -282.69207763671875,
      "logps/rejected": -383.0326232910156,
      "loss": 0.402,
      "positive_losses": 1.398983359336853,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3093782067298889,
      "rewards/margins": 1.4970686435699463,
      "rewards/margins_max": 2.100078821182251,
      "rewards/margins_min": 0.7865512371063232,
      "rewards/margins_std": 0.584915280342102,
      "rewards/rejected": -1.1876904964447021,
      "step": 1020
    },
    {
      "dpo_losses": 0.22967854142189026,
      "epoch": 2.9,
      "grad_norm": 7.332508849988266,
      "learning_rate": 1.6448943457189616e-08,
      "logits/chosen": -2.502159833908081,
      "logits/rejected": -2.438899278640747,
      "logps/chosen": -271.15960693359375,
      "logps/rejected": -378.86492919921875,
      "loss": 0.2637,
      "positive_losses": 0.03314323350787163,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.38446491956710815,
      "rewards/margins": 1.6038453578948975,
      "rewards/margins_max": 2.506626605987549,
      "rewards/margins_min": 0.7737765908241272,
      "rewards/margins_std": 0.7757932543754578,
      "rewards/rejected": -1.2193803787231445,
      "step": 1030
    },
    {
      "dpo_losses": 0.2706490755081177,
      "epoch": 2.93,
      "grad_norm": 128.69745590434178,
      "learning_rate": 8.39683258841123e-09,
      "logits/chosen": -2.45271635055542,
      "logits/rejected": -2.3681931495666504,
      "logps/chosen": -259.78302001953125,
      "logps/rejected": -344.37237548828125,
      "loss": 0.4338,
      "positive_losses": 2.175943613052368,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.3506382405757904,
      "rewards/margins": 1.4272511005401611,
      "rewards/margins_max": 2.363046646118164,
      "rewards/margins_min": 0.47343364357948303,
      "rewards/margins_std": 0.8201966285705566,
      "rewards/rejected": -1.0766130685806274,
      "step": 1040
    },
    {
      "dpo_losses": 0.2392924576997757,
      "epoch": 2.96,
      "grad_norm": 1.460037117185786,
      "learning_rate": 3.0239435998430376e-09,
      "logits/chosen": -2.5590803623199463,
      "logits/rejected": -2.4645559787750244,
      "logps/chosen": -264.31146240234375,
      "logps/rejected": -341.509521484375,
      "loss": 0.296,
      "positive_losses": 0.19127945601940155,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.3725205659866333,
      "rewards/margins": 1.5269343852996826,
      "rewards/margins_max": 2.308137893676758,
      "rewards/margins_min": 0.6272153854370117,
      "rewards/margins_std": 0.7376829385757446,
      "rewards/rejected": -1.1544138193130493,
      "step": 1050
    },
    {
      "dpo_losses": 0.22728697955608368,
      "epoch": 2.99,
      "grad_norm": 12.16631271335167,
      "learning_rate": 3.3605396115826695e-10,
      "logits/chosen": -2.488372564315796,
      "logits/rejected": -2.471071243286133,
      "logps/chosen": -243.7676239013672,
      "logps/rejected": -374.9264221191406,
      "loss": 0.2637,
      "positive_losses": 0.6191848516464233,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.35623699426651,
      "rewards/margins": 1.5707523822784424,
      "rewards/margins_max": 2.3378148078918457,
      "rewards/margins_min": 0.7644645571708679,
      "rewards/margins_std": 0.6974471807479858,
      "rewards/rejected": -1.2145154476165771,
      "step": 1060
    },
    {
      "epoch": 3.0,
      "step": 1065,
      "total_flos": 0.0,
      "train_loss": 0.4752508000029085,
      "train_runtime": 13210.775,
      "train_samples_per_second": 1.289,
      "train_steps_per_second": 0.081
    }
  ],
  "logging_steps": 10,
  "max_steps": 1065,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}