trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.0,
  "eval_steps": 100,
  "global_step": 355,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.0,
      "grad_norm": 1.601592868423971,
      "learning_rate": 1.3888888888888887e-08,
      "logits/chosen": -2.861618995666504,
      "logits/rejected": -2.8205904960632324,
      "logps/chosen": -271.06011962890625,
      "logps/rejected": -211.1704559326172,
      "loss": 0.6931,
      "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
    },
    {
      "epoch": 0.03,
      "grad_norm": 9.671306242111815,
      "learning_rate": 1.3888888888888888e-07,
      "logits/chosen": -2.8328380584716797,
      "logits/rejected": -2.7900264263153076,
      "logps/chosen": -325.0357971191406,
      "logps/rejected": -275.0066833496094,
      "loss": 0.693,
      "rewards/accuracies": 0.4861111044883728,
      "rewards/chosen": 0.00014538393588736653,
      "rewards/margins": 0.000555534556042403,
      "rewards/margins_max": 0.003110126592218876,
      "rewards/margins_min": -0.0015574158169329166,
      "rewards/margins_std": 0.002125632017850876,
      "rewards/rejected": -0.0004101507074665278,
      "step": 10
    },
    {
      "epoch": 0.06,
      "grad_norm": 1.843946802013221,
      "learning_rate": 2.7777777777777776e-07,
      "logits/chosen": -2.72574782371521,
      "logits/rejected": -2.7075164318084717,
      "logps/chosen": -293.76934814453125,
      "logps/rejected": -215.7692108154297,
      "loss": 0.693,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": 0.0005661511677317321,
      "rewards/margins": 0.0004684908199124038,
      "rewards/margins_max": 0.003089509904384613,
      "rewards/margins_min": -0.0020298557355999947,
      "rewards/margins_std": 0.0022844516206532717,
      "rewards/rejected": 9.76602896116674e-05,
      "step": 20
    },
    {
      "epoch": 0.08,
      "grad_norm": 2.2773042809548794,
      "learning_rate": 4.1666666666666667e-07,
      "logits/chosen": -2.8187897205352783,
      "logits/rejected": -2.749154567718506,
      "logps/chosen": -303.4768371582031,
      "logps/rejected": -232.51852416992188,
      "loss": 0.6921,
      "rewards/accuracies": 0.7749999761581421,
      "rewards/chosen": 0.0022904500365257263,
      "rewards/margins": 0.002333464566618204,
      "rewards/margins_max": 0.0058710528537631035,
      "rewards/margins_min": -0.0009718179935589433,
      "rewards/margins_std": 0.003063285956159234,
      "rewards/rejected": -4.3014533730456606e-05,
      "step": 30
    },
    {
      "epoch": 0.11,
      "grad_norm": 1.731204115607293,
      "learning_rate": 4.998060489154965e-07,
      "logits/chosen": -2.8424630165100098,
      "logits/rejected": -2.762080669403076,
      "logps/chosen": -277.14581298828125,
      "logps/rejected": -225.66372680664062,
      "loss": 0.6915,
      "rewards/accuracies": 0.699999988079071,
      "rewards/chosen": 0.003388203214854002,
      "rewards/margins": 0.002678563352674246,
      "rewards/margins_max": 0.006867862306535244,
      "rewards/margins_min": -0.0013280096463859081,
      "rewards/margins_std": 0.0036745709367096424,
      "rewards/rejected": 0.0007096402114257216,
      "step": 40
    },
    {
      "epoch": 0.14,
      "grad_norm": 2.204111297815784,
      "learning_rate": 4.976275538042932e-07,
      "logits/chosen": -2.81162691116333,
      "logits/rejected": -2.739978551864624,
      "logps/chosen": -274.8628845214844,
      "logps/rejected": -233.8441162109375,
      "loss": 0.6899,
      "rewards/accuracies": 0.887499988079071,
      "rewards/chosen": 0.007072491105645895,
      "rewards/margins": 0.006740429904311895,
      "rewards/margins_max": 0.015335154719650745,
      "rewards/margins_min": 0.0001096004998544231,
      "rewards/margins_std": 0.0068354858085513115,
      "rewards/rejected": 0.0003320608229842037,
      "step": 50
    },
    {
      "epoch": 0.17,
      "grad_norm": 2.4036561931966838,
      "learning_rate": 4.930493069997119e-07,
      "logits/chosen": -2.7489891052246094,
      "logits/rejected": -2.709763288497925,
      "logps/chosen": -312.48358154296875,
      "logps/rejected": -262.7074890136719,
      "loss": 0.688,
      "rewards/accuracies": 0.875,
      "rewards/chosen": 0.011560519225895405,
      "rewards/margins": 0.010920424945652485,
      "rewards/margins_max": 0.021653519943356514,
      "rewards/margins_min": 0.0011082137934863567,
      "rewards/margins_std": 0.009159913286566734,
      "rewards/rejected": 0.000640094222035259,
      "step": 60
    },
    {
      "epoch": 0.2,
      "grad_norm": 1.769456739730609,
      "learning_rate": 4.861156761634013e-07,
      "logits/chosen": -2.783851146697998,
      "logits/rejected": -2.7219314575195312,
      "logps/chosen": -321.22711181640625,
      "logps/rejected": -235.4314422607422,
      "loss": 0.6859,
      "rewards/accuracies": 0.875,
      "rewards/chosen": 0.016341406852006912,
      "rewards/margins": 0.014239917509257793,
      "rewards/margins_max": 0.031113797798752785,
      "rewards/margins_min": 0.0018715148326009512,
      "rewards/margins_std": 0.013196480460464954,
      "rewards/rejected": 0.0021014874801039696,
      "step": 70
    },
    {
      "epoch": 0.23,
      "grad_norm": 1.861795596875086,
      "learning_rate": 4.768938549177392e-07,
      "logits/chosen": -2.832259178161621,
      "logits/rejected": -2.775381565093994,
      "logps/chosen": -319.2157897949219,
      "logps/rejected": -286.1940002441406,
      "loss": 0.6844,
      "rewards/accuracies": 0.8500000238418579,
      "rewards/chosen": 0.020546909421682358,
      "rewards/margins": 0.01869961991906166,
      "rewards/margins_max": 0.04271908476948738,
      "rewards/margins_min": 0.0003298601950518787,
      "rewards/margins_std": 0.01946866512298584,
      "rewards/rejected": 0.0018472913652658463,
      "step": 80
    },
    {
      "epoch": 0.25,
      "grad_norm": 2.556237733247233,
      "learning_rate": 4.654732116743193e-07,
      "logits/chosen": -2.7664496898651123,
      "logits/rejected": -2.7240653038024902,
      "logps/chosen": -273.2864990234375,
      "logps/rejected": -198.71231079101562,
      "loss": 0.6806,
      "rewards/accuracies": 0.8500000238418579,
      "rewards/chosen": 0.02708309330046177,
      "rewards/margins": 0.02433420717716217,
      "rewards/margins_max": 0.052233658730983734,
      "rewards/margins_min": 0.0035423454828560352,
      "rewards/margins_std": 0.022480051964521408,
      "rewards/rejected": 0.002748883794993162,
      "step": 90
    },
    {
      "epoch": 0.28,
      "grad_norm": 1.897335973090835,
      "learning_rate": 4.519644235671752e-07,
      "logits/chosen": -2.8063488006591797,
      "logits/rejected": -2.768160581588745,
      "logps/chosen": -278.94610595703125,
      "logps/rejected": -261.92315673828125,
      "loss": 0.6803,
      "rewards/accuracies": 0.862500011920929,
      "rewards/chosen": 0.024656174704432487,
      "rewards/margins": 0.022013681009411812,
      "rewards/margins_max": 0.04718270152807236,
      "rewards/margins_min": 0.0019731284119188786,
      "rewards/margins_std": 0.020801713690161705,
      "rewards/rejected": 0.0026424946263432503,
      "step": 100
    },
    {
      "epoch": 0.28,
      "eval_logits/chosen": -2.803894519805908,
      "eval_logits/rejected": -2.7653636932373047,
      "eval_logps/chosen": -283.0562438964844,
      "eval_logps/rejected": -257.6935729980469,
      "eval_loss": 0.6907610297203064,
      "eval_rewards/accuracies": 0.5799999833106995,
      "eval_rewards/chosen": 0.01537203136831522,
      "eval_rewards/margins": 0.00651828246191144,
      "eval_rewards/margins_max": 0.04056054726243019,
      "eval_rewards/margins_min": -0.02254258282482624,
      "eval_rewards/margins_std": 0.020977748557925224,
      "eval_rewards/rejected": 0.008853749372065067,
      "eval_runtime": 431.2136,
      "eval_samples_per_second": 4.638,
      "eval_steps_per_second": 0.29,
      "step": 100
    },
    {
      "epoch": 0.31,
      "grad_norm": 2.350716473544898,
      "learning_rate": 4.364984038837727e-07,
      "logits/chosen": -2.8714425563812256,
      "logits/rejected": -2.776231527328491,
      "logps/chosen": -370.3124084472656,
      "logps/rejected": -286.1990051269531,
      "loss": 0.6739,
      "rewards/accuracies": 0.8999999761581421,
      "rewards/chosen": 0.042933039367198944,
      "rewards/margins": 0.04058496281504631,
      "rewards/margins_max": 0.07831507176160812,
      "rewards/margins_min": 0.009233040735125542,
      "rewards/margins_std": 0.03092753328382969,
      "rewards/rejected": 0.0023480760864913464,
      "step": 110
    },
    {
      "epoch": 0.34,
      "grad_norm": 2.0267120141468924,
      "learning_rate": 4.1922503338800447e-07,
      "logits/chosen": -2.840080976486206,
      "logits/rejected": -2.7854738235473633,
      "logps/chosen": -343.77545166015625,
      "logps/rejected": -265.7751159667969,
      "loss": 0.6727,
      "rewards/accuracies": 0.8999999761581421,
      "rewards/chosen": 0.046107616275548935,
      "rewards/margins": 0.04191668704152107,
      "rewards/margins_max": 0.0861145481467247,
      "rewards/margins_min": 0.010150762274861336,
      "rewards/margins_std": 0.035071179270744324,
      "rewards/rejected": 0.004190927837044001,
      "step": 120
    },
    {
      "epoch": 0.37,
      "grad_norm": 1.7790024615351698,
      "learning_rate": 4.003117078299021e-07,
      "logits/chosen": -2.8337936401367188,
      "logits/rejected": -2.751962184906006,
      "logps/chosen": -377.0141906738281,
      "logps/rejected": -300.07891845703125,
      "loss": 0.6685,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.05350027605891228,
      "rewards/margins": 0.04998067393898964,
      "rewards/margins_max": 0.09481509029865265,
      "rewards/margins_min": 0.008272581733763218,
      "rewards/margins_std": 0.03825851529836655,
      "rewards/rejected": 0.0035195990931242704,
      "step": 130
    },
    {
      "epoch": 0.39,
      "grad_norm": 1.4853892826304387,
      "learning_rate": 3.799417157181075e-07,
      "logits/chosen": -2.786635398864746,
      "logits/rejected": -2.7476704120635986,
      "logps/chosen": -312.93548583984375,
      "logps/rejected": -269.5070495605469,
      "loss": 0.6703,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": 0.05478551238775253,
      "rewards/margins": 0.05136098712682724,
      "rewards/margins_max": 0.10997221618890762,
      "rewards/margins_min": 0.007977863773703575,
      "rewards/margins_std": 0.04762490838766098,
      "rewards/rejected": 0.003424527822062373,
      "step": 140
    },
    {
      "epoch": 0.42,
      "grad_norm": 1.9607644783681,
      "learning_rate": 3.583124620760659e-07,
      "logits/chosen": -2.788800001144409,
      "logits/rejected": -2.748765468597412,
      "logps/chosen": -283.2447814941406,
      "logps/rejected": -212.7663116455078,
      "loss": 0.6666,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.05440182611346245,
      "rewards/margins": 0.051237184554338455,
      "rewards/margins_max": 0.10327558219432831,
      "rewards/margins_min": 0.00882364995777607,
      "rewards/margins_std": 0.04341956600546837,
      "rewards/rejected": 0.0031646445859223604,
      "step": 150
    },
    {
      "epoch": 0.45,
      "grad_norm": 1.6524576639358914,
      "learning_rate": 3.356335553954679e-07,
      "logits/chosen": -2.766443967819214,
      "logits/rejected": -2.712019681930542,
      "logps/chosen": -294.170654296875,
      "logps/rejected": -233.78359985351562,
      "loss": 0.664,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.06376218795776367,
      "rewards/margins": 0.05766482278704643,
      "rewards/margins_max": 0.11248371750116348,
      "rewards/margins_min": 0.00960196740925312,
      "rewards/margins_std": 0.046693481504917145,
      "rewards/rejected": 0.0060973623767495155,
      "step": 160
    },
    {
      "epoch": 0.48,
      "grad_norm": 1.829740022838758,
      "learning_rate": 3.121247763262235e-07,
      "logits/chosen": -2.8182921409606934,
      "logits/rejected": -2.7600901126861572,
      "logps/chosen": -322.2862854003906,
      "logps/rejected": -296.15496826171875,
      "loss": 0.6602,
      "rewards/accuracies": 0.9125000238418579,
      "rewards/chosen": 0.07335338741540909,
      "rewards/margins": 0.07224629819393158,
      "rewards/margins_max": 0.13527169823646545,
      "rewards/margins_min": 0.014563126489520073,
      "rewards/margins_std": 0.054327212274074554,
      "rewards/rejected": 0.001107088173739612,
      "step": 170
    },
    {
      "epoch": 0.51,
      "grad_norm": 1.954322630861037,
      "learning_rate": 2.880139477883347e-07,
      "logits/chosen": -2.8129706382751465,
      "logits/rejected": -2.72513484954834,
      "logps/chosen": -324.20794677734375,
      "logps/rejected": -293.87493896484375,
      "loss": 0.6567,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.07692877948284149,
      "rewards/margins": 0.07382867485284805,
      "rewards/margins_max": 0.14840444922447205,
      "rewards/margins_min": 0.021513966843485832,
      "rewards/margins_std": 0.05711384862661362,
      "rewards/rejected": 0.0031001190654933453,
      "step": 180
    },
    {
      "epoch": 0.54,
      "grad_norm": 2.352020928698477,
      "learning_rate": 2.635347271463544e-07,
      "logits/chosen": -2.7723231315612793,
      "logits/rejected": -2.6797118186950684,
      "logps/chosen": -313.4515686035156,
      "logps/rejected": -237.2364501953125,
      "loss": 0.6551,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": 0.08509302884340286,
      "rewards/margins": 0.08258074522018433,
      "rewards/margins_max": 0.16938893496990204,
      "rewards/margins_min": 0.015894772484898567,
      "rewards/margins_std": 0.07122750580310822,
      "rewards/rejected": 0.002512281993404031,
      "step": 190
    },
    {
      "epoch": 0.56,
      "grad_norm": 2.051556424366539,
      "learning_rate": 2.3892434184240534e-07,
      "logits/chosen": -2.8492965698242188,
      "logits/rejected": -2.7720444202423096,
      "logps/chosen": -330.9425048828125,
      "logps/rejected": -265.1010437011719,
      "loss": 0.6568,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.08590307831764221,
      "rewards/margins": 0.0809393972158432,
      "rewards/margins_max": 0.16236025094985962,
      "rewards/margins_min": 0.016894496977329254,
      "rewards/margins_std": 0.06359551846981049,
      "rewards/rejected": 0.004963680170476437,
      "step": 200
    },
    {
      "epoch": 0.56,
      "eval_logits/chosen": -2.787705421447754,
      "eval_logits/rejected": -2.749375343322754,
      "eval_logps/chosen": -280.6795959472656,
      "eval_logps/rejected": -256.39471435546875,
      "eval_loss": 0.6869410276412964,
      "eval_rewards/accuracies": 0.5960000157356262,
      "eval_rewards/chosen": 0.03913822025060654,
      "eval_rewards/margins": 0.01729637198150158,
      "eval_rewards/margins_max": 0.1024816706776619,
      "eval_rewards/margins_min": -0.05720587447285652,
      "eval_rewards/margins_std": 0.05308415740728378,
      "eval_rewards/rejected": 0.021841851994395256,
      "eval_runtime": 429.0815,
      "eval_samples_per_second": 4.661,
      "eval_steps_per_second": 0.291,
      "step": 200
    },
    {
      "epoch": 0.59,
      "grad_norm": 1.9346913930945893,
      "learning_rate": 2.1442129043167873e-07,
      "logits/chosen": -2.7502646446228027,
      "logits/rejected": -2.7227656841278076,
      "logps/chosen": -310.6435852050781,
      "logps/rejected": -256.34246826171875,
      "loss": 0.6507,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.09540757536888123,
      "rewards/margins": 0.09458459913730621,
      "rewards/margins_max": 0.1914738416671753,
      "rewards/margins_min": 0.01557054091244936,
      "rewards/margins_std": 0.0801827609539032,
      "rewards/rejected": 0.0008229793747887015,
      "step": 210
    },
    {
      "epoch": 0.62,
      "grad_norm": 1.8397256282784018,
      "learning_rate": 1.9026303129961048e-07,
      "logits/chosen": -2.849534511566162,
      "logits/rejected": -2.7543559074401855,
      "logps/chosen": -343.22222900390625,
      "logps/rejected": -274.1776428222656,
      "loss": 0.6479,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.10234014689922333,
      "rewards/margins": 0.10084100067615509,
      "rewards/margins_max": 0.1773754209280014,
      "rewards/margins_min": 0.02473507821559906,
      "rewards/margins_std": 0.06906573474407196,
      "rewards/rejected": 0.001499147154390812,
      "step": 220
    },
    {
      "epoch": 0.65,
      "grad_norm": 1.9517251453168138,
      "learning_rate": 1.6668368145931396e-07,
      "logits/chosen": -2.8495993614196777,
      "logits/rejected": -2.756133556365967,
      "logps/chosen": -361.55810546875,
      "logps/rejected": -262.0633850097656,
      "loss": 0.6422,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.1011759415268898,
      "rewards/margins": 0.10087559372186661,
      "rewards/margins_max": 0.19324660301208496,
      "rewards/margins_min": 0.022996682673692703,
      "rewards/margins_std": 0.07676897943019867,
      "rewards/rejected": 0.0003003578167408705,
      "step": 230
    },
    {
      "epoch": 0.68,
      "grad_norm": 1.5814942001733312,
      "learning_rate": 1.4391174773015834e-07,
      "logits/chosen": -2.7951114177703857,
      "logits/rejected": -2.724926233291626,
      "logps/chosen": -316.5010070800781,
      "logps/rejected": -283.17291259765625,
      "loss": 0.6514,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": 0.0829603299498558,
      "rewards/margins": 0.08294094353914261,
      "rewards/margins_max": 0.16664841771125793,
      "rewards/margins_min": 0.015887869521975517,
      "rewards/margins_std": 0.06911344826221466,
      "rewards/rejected": 1.939358116942458e-05,
      "step": 240
    },
    {
      "epoch": 0.7,
      "grad_norm": 1.9477304039773469,
      "learning_rate": 1.2216791228457775e-07,
      "logits/chosen": -2.795609951019287,
      "logits/rejected": -2.7185869216918945,
      "logps/chosen": -302.67822265625,
      "logps/rejected": -251.4832000732422,
      "loss": 0.6494,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": 0.0860290601849556,
      "rewards/margins": 0.090418241918087,
      "rewards/margins_max": 0.18044039607048035,
      "rewards/margins_min": 0.016120051965117455,
      "rewards/margins_std": 0.07592571526765823,
      "rewards/rejected": -0.0043891845270991325,
      "step": 250
    },
    {
      "epoch": 0.73,
      "grad_norm": 1.7138528589589788,
      "learning_rate": 1.0166289402331391e-07,
      "logits/chosen": -2.8433239459991455,
      "logits/rejected": -2.760486125946045,
      "logps/chosen": -282.75128173828125,
      "logps/rejected": -259.15283203125,
      "loss": 0.6432,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": 0.09342271834611893,
      "rewards/margins": 0.09564103186130524,
      "rewards/margins_max": 0.19372263550758362,
      "rewards/margins_min": 0.020370684564113617,
      "rewards/margins_std": 0.08056275546550751,
      "rewards/rejected": -0.0022183198016136885,
      "step": 260
    },
    {
      "epoch": 0.76,
      "grad_norm": 2.0016114180555578,
      "learning_rate": 8.259540650444734e-08,
      "logits/chosen": -2.7903430461883545,
      "logits/rejected": -2.7373478412628174,
      "logps/chosen": -300.1326599121094,
      "logps/rejected": -262.3350830078125,
      "loss": 0.6496,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": 0.08982918411493301,
      "rewards/margins": 0.08818400651216507,
      "rewards/margins_max": 0.16805505752563477,
      "rewards/margins_min": 0.026935186237096786,
      "rewards/margins_std": 0.06377857178449631,
      "rewards/rejected": 0.0016451816773042083,
      "step": 270
    },
    {
      "epoch": 0.79,
      "grad_norm": 1.7516715377367251,
      "learning_rate": 6.515023221586721e-08,
      "logits/chosen": -2.766188144683838,
      "logits/rejected": -2.7269351482391357,
      "logps/chosen": -294.2494812011719,
      "logps/rejected": -272.6824951171875,
      "loss": 0.648,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": 0.09207499772310257,
      "rewards/margins": 0.09292837232351303,
      "rewards/margins_max": 0.17264562845230103,
      "rewards/margins_min": 0.023857418447732925,
      "rewards/margins_std": 0.06797114759683609,
      "rewards/rejected": -0.0008533738437108696,
      "step": 280
    },
    {
      "epoch": 0.82,
      "grad_norm": 1.7538431067028422,
      "learning_rate": 4.949643185335287e-08,
      "logits/chosen": -2.771496295928955,
      "logits/rejected": -2.721856117248535,
      "logps/chosen": -278.96405029296875,
      "logps/rejected": -264.3961181640625,
      "loss": 0.653,
      "rewards/accuracies": 0.887499988079071,
      "rewards/chosen": 0.07952173054218292,
      "rewards/margins": 0.08330051600933075,
      "rewards/margins_max": 0.18094156682491302,
      "rewards/margins_min": 0.007182260509580374,
      "rewards/margins_std": 0.07998403161764145,
      "rewards/rejected": -0.0037787805777043104,
      "step": 290
    },
    {
      "epoch": 0.85,
      "grad_norm": 1.9243414334438866,
      "learning_rate": 3.578570595810274e-08,
      "logits/chosen": -2.8379387855529785,
      "logits/rejected": -2.7606096267700195,
      "logps/chosen": -335.54931640625,
      "logps/rejected": -289.7465515136719,
      "loss": 0.6398,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.10963740199804306,
      "rewards/margins": 0.11276860535144806,
      "rewards/margins_max": 0.21979229152202606,
      "rewards/margins_min": 0.021601524204015732,
      "rewards/margins_std": 0.08902540057897568,
      "rewards/rejected": -0.0031312075443565845,
      "step": 300
    },
    {
      "epoch": 0.85,
      "eval_logits/chosen": -2.781425952911377,
      "eval_logits/rejected": -2.743434190750122,
      "eval_logps/chosen": -280.510498046875,
      "eval_logps/rejected": -256.65869140625,
      "eval_loss": 0.6854214668273926,
      "eval_rewards/accuracies": 0.5899999737739563,
      "eval_rewards/chosen": 0.04082921892404556,
      "eval_rewards/margins": 0.021626880392432213,
      "eval_rewards/margins_max": 0.12814800441265106,
      "eval_rewards/margins_min": -0.07126501202583313,
      "eval_rewards/margins_std": 0.06649063527584076,
      "eval_rewards/rejected": 0.0192023366689682,
      "eval_runtime": 428.7677,
      "eval_samples_per_second": 4.665,
      "eval_steps_per_second": 0.292,
      "step": 300
    },
    {
      "epoch": 0.87,
      "grad_norm": 1.7722376322831996,
      "learning_rate": 2.415092479103503e-08,
      "logits/chosen": -2.835430860519409,
      "logits/rejected": -2.740483522415161,
      "logps/chosen": -287.6938171386719,
      "logps/rejected": -214.74563598632812,
      "loss": 0.64,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.10122032463550568,
      "rewards/margins": 0.10643725097179413,
      "rewards/margins_max": 0.21550527215003967,
      "rewards/margins_min": 0.02140904776751995,
      "rewards/margins_std": 0.08673234283924103,
      "rewards/rejected": -0.005216927733272314,
      "step": 310
    },
    {
      "epoch": 0.9,
      "grad_norm": 1.8570363058712267,
      "learning_rate": 1.4704840690808656e-08,
      "logits/chosen": -2.80118727684021,
      "logits/rejected": -2.743530035018921,
      "logps/chosen": -298.8092956542969,
      "logps/rejected": -260.87152099609375,
      "loss": 0.6445,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 0.09463869780302048,
      "rewards/margins": 0.09665326029062271,
      "rewards/margins_max": 0.20018891990184784,
      "rewards/margins_min": 0.01689526066184044,
      "rewards/margins_std": 0.08293472975492477,
      "rewards/rejected": -0.0020145513117313385,
      "step": 320
    },
    {
      "epoch": 0.93,
      "grad_norm": 1.9373896782753002,
      "learning_rate": 7.538995394063995e-09,
      "logits/chosen": -2.8751907348632812,
      "logits/rejected": -2.7946133613586426,
      "logps/chosen": -343.7359619140625,
      "logps/rejected": -267.89105224609375,
      "loss": 0.6408,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": 0.10583887249231339,
      "rewards/margins": 0.11535302549600601,
      "rewards/margins_max": 0.20861783623695374,
      "rewards/margins_min": 0.029701709747314453,
      "rewards/margins_std": 0.08170617371797562,
      "rewards/rejected": -0.009514153003692627,
      "step": 330
    },
    {
      "epoch": 0.96,
      "grad_norm": 1.9173076290433462,
      "learning_rate": 2.7228329070159705e-09,
      "logits/chosen": -2.762208938598633,
      "logits/rejected": -2.710221767425537,
      "logps/chosen": -289.7626037597656,
      "logps/rejected": -251.68887329101562,
      "loss": 0.6385,
      "rewards/accuracies": 0.9125000238418579,
      "rewards/chosen": 0.10082916170358658,
      "rewards/margins": 0.10602158308029175,
      "rewards/margins_max": 0.22432371973991394,
      "rewards/margins_min": 0.02705308236181736,
      "rewards/margins_std": 0.09002183377742767,
      "rewards/rejected": -0.005192426033318043,
      "step": 340
    },
    {
      "epoch": 0.99,
      "grad_norm": 5.134028042605605,
      "learning_rate": 3.0302652553296226e-10,
      "logits/chosen": -2.7456893920898438,
      "logits/rejected": -2.7021021842956543,
      "logps/chosen": -306.5563049316406,
      "logps/rejected": -287.2024230957031,
      "loss": 0.647,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": 0.08402810990810394,
      "rewards/margins": 0.09153296053409576,
      "rewards/margins_max": 0.18022149801254272,
      "rewards/margins_min": 0.017785457894206047,
      "rewards/margins_std": 0.07233723253011703,
      "rewards/rejected": -0.007504852022975683,
      "step": 350
    },
    {
      "epoch": 1.0,
      "step": 355,
      "total_flos": 0.0,
      "train_loss": 0.6630057381912017,
      "train_runtime": 4226.0112,
      "train_samples_per_second": 1.344,
      "train_steps_per_second": 0.084
    }
  ],
  "logging_steps": 10,
  "max_steps": 355,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 100,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}