checkpoint-1100/trainer_state.json

{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9027492819039803,
  "eval_steps": 50,
  "global_step": 1100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.008206811653672548,
      "grad_norm": 0.07778492569923401,
      "learning_rate": 4.999451708687114e-06,
      "logits/chosen": -2.053281307220459,
      "logits/rejected": -2.495474338531494,
      "logps/chosen": -0.3126755356788635,
      "logps/rejected": -0.3312620520591736,
      "loss": 7.6211,
      "rewards/accuracies": 0.44999998807907104,
      "rewards/chosen": -0.4690132737159729,
      "rewards/margins": 0.027879873290657997,
      "rewards/rejected": -0.49689316749572754,
      "step": 10
    },
    {
      "epoch": 0.016413623307345096,
      "grad_norm": 0.07773654907941818,
      "learning_rate": 4.997807075247147e-06,
      "logits/chosen": -2.0624098777770996,
      "logits/rejected": -2.4424185752868652,
      "logps/chosen": -0.26926660537719727,
      "logps/rejected": -0.2978014051914215,
      "loss": 7.5195,
      "rewards/accuracies": 0.48750001192092896,
      "rewards/chosen": -0.4038998484611511,
      "rewards/margins": 0.04280223697423935,
      "rewards/rejected": -0.44670209288597107,
      "step": 20
    },
    {
      "epoch": 0.024620434961017644,
      "grad_norm": 0.07357177883386612,
      "learning_rate": 4.9950668210706795e-06,
      "logits/chosen": -2.068427562713623,
      "logits/rejected": -2.486642360687256,
      "logps/chosen": -0.29993391036987305,
      "logps/rejected": -0.34360918402671814,
      "loss": 7.4913,
      "rewards/accuracies": 0.4749999940395355,
      "rewards/chosen": -0.4499008059501648,
      "rewards/margins": 0.06551288068294525,
      "rewards/rejected": -0.5154137015342712,
      "step": 30
    },
    {
      "epoch": 0.03282724661469019,
      "grad_norm": 0.14212799072265625,
      "learning_rate": 4.9912321481237616e-06,
      "logits/chosen": -2.015650987625122,
      "logits/rejected": -2.3838727474212646,
      "logps/chosen": -0.2911723852157593,
      "logps/rejected": -0.30521970987319946,
      "loss": 7.5217,
      "rewards/accuracies": 0.48750001192092896,
      "rewards/chosen": -0.4367586076259613,
      "rewards/margins": 0.021070968359708786,
      "rewards/rejected": -0.4578295648097992,
      "step": 40
    },
    {
      "epoch": 0.04103405826836274,
      "grad_norm": 0.08107248693704605,
      "learning_rate": 4.986304738420684e-06,
      "logits/chosen": -2.1150989532470703,
      "logits/rejected": -2.4338631629943848,
      "logps/chosen": -0.26249754428863525,
      "logps/rejected": -0.3132360577583313,
      "loss": 7.519,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": -0.3937462866306305,
      "rewards/margins": 0.07610772550106049,
      "rewards/rejected": -0.4698540270328522,
      "step": 50
    },
    {
      "epoch": 0.04103405826836274,
      "eval_logits/chosen": -2.0232737064361572,
      "eval_logits/rejected": -2.4952735900878906,
      "eval_logps/chosen": -0.27974528074264526,
      "eval_logps/rejected": -0.3420677185058594,
      "eval_loss": 0.9291417598724365,
      "eval_rewards/accuracies": 0.49494948983192444,
      "eval_rewards/chosen": -0.41961798071861267,
      "eval_rewards/margins": 0.09348361939191818,
      "eval_rewards/rejected": -0.5131015777587891,
      "eval_runtime": 26.0563,
      "eval_samples_per_second": 30.242,
      "eval_steps_per_second": 3.799,
      "step": 50
    },
    {
      "epoch": 0.04924086992203529,
      "grad_norm": 0.06815352290868759,
      "learning_rate": 4.980286753286196e-06,
      "logits/chosen": -1.9890680313110352,
      "logits/rejected": -2.3848204612731934,
      "logps/chosen": -0.26213228702545166,
      "logps/rejected": -0.31342557072639465,
      "loss": 7.432,
      "rewards/accuracies": 0.512499988079071,
      "rewards/chosen": -0.3931984603404999,
      "rewards/margins": 0.0769399031996727,
      "rewards/rejected": -0.4701383709907532,
      "step": 60
    },
    {
      "epoch": 0.057447681575707836,
      "grad_norm": 0.06748568266630173,
      "learning_rate": 4.973180832407471e-06,
      "logits/chosen": -2.070542812347412,
      "logits/rejected": -2.3977038860321045,
      "logps/chosen": -0.24570491909980774,
      "logps/rejected": -0.3655605912208557,
      "loss": 7.35,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": -0.3685573935508728,
      "rewards/margins": 0.17978355288505554,
      "rewards/rejected": -0.548340916633606,
      "step": 70
    },
    {
      "epoch": 0.06565449322938038,
      "grad_norm": 0.10909309983253479,
      "learning_rate": 4.964990092676263e-06,
      "logits/chosen": -2.2012317180633545,
      "logits/rejected": -2.346029758453369,
      "logps/chosen": -0.2279246598482132,
      "logps/rejected": -0.35396742820739746,
      "loss": 7.5082,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -0.341886967420578,
      "rewards/margins": 0.18906418979167938,
      "rewards/rejected": -0.5309511423110962,
      "step": 80
    },
    {
      "epoch": 0.07386130488305294,
      "grad_norm": 0.05977805703878403,
      "learning_rate": 4.9557181268217225e-06,
      "logits/chosen": -2.0719449520111084,
      "logits/rejected": -2.4491190910339355,
      "logps/chosen": -0.2503294348716736,
      "logps/rejected": -0.29939892888069153,
      "loss": 7.5129,
      "rewards/accuracies": 0.4625000059604645,
      "rewards/chosen": -0.37549418210983276,
      "rewards/margins": 0.07360419631004333,
      "rewards/rejected": -0.4490983486175537,
      "step": 90
    },
    {
      "epoch": 0.08206811653672548,
      "grad_norm": 0.051751479506492615,
      "learning_rate": 4.9453690018345144e-06,
      "logits/chosen": -2.0634045600891113,
      "logits/rejected": -2.458428382873535,
      "logps/chosen": -0.24033495783805847,
      "logps/rejected": -0.29080909490585327,
      "loss": 7.4432,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": -0.3605024516582489,
      "rewards/margins": 0.07571124285459518,
      "rewards/rejected": -0.4362136721611023,
      "step": 100
    },
    {
      "epoch": 0.08206811653672548,
      "eval_logits/chosen": -2.0207154750823975,
      "eval_logits/rejected": -2.486215353012085,
      "eval_logps/chosen": -0.2376101016998291,
      "eval_logps/rejected": -0.32593628764152527,
      "eval_loss": 0.9085211753845215,
      "eval_rewards/accuracies": 0.5353535413742065,
      "eval_rewards/chosen": -0.35641518235206604,
      "eval_rewards/margins": 0.13248924911022186,
      "eval_rewards/rejected": -0.4889043867588043,
      "eval_runtime": 26.0119,
      "eval_samples_per_second": 30.294,
      "eval_steps_per_second": 3.806,
      "step": 100
    },
    {
      "epoch": 0.09027492819039803,
      "grad_norm": 0.06007291004061699,
      "learning_rate": 4.933947257182901e-06,
      "logits/chosen": -2.1248741149902344,
      "logits/rejected": -2.409808874130249,
      "logps/chosen": -0.2354653775691986,
      "logps/rejected": -0.30269068479537964,
      "loss": 7.317,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": -0.3531980812549591,
      "rewards/margins": 0.10083796828985214,
      "rewards/rejected": -0.45403605699539185,
      "step": 110
    },
    {
      "epoch": 0.09848173984407058,
      "grad_norm": 0.055738095194101334,
      "learning_rate": 4.921457902821578e-06,
      "logits/chosen": -2.0635311603546143,
      "logits/rejected": -2.4297730922698975,
      "logps/chosen": -0.2315257489681244,
      "logps/rejected": -0.33639490604400635,
      "loss": 7.2775,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -0.3472886383533478,
      "rewards/margins": 0.15730372071266174,
      "rewards/rejected": -0.5045923590660095,
      "step": 120
    },
    {
      "epoch": 0.10668855149774313,
      "grad_norm": 0.07971248030662537,
      "learning_rate": 4.907906416994146e-06,
      "logits/chosen": -2.07852840423584,
      "logits/rejected": -2.4043469429016113,
      "logps/chosen": -0.20596058666706085,
      "logps/rejected": -0.33416762948036194,
      "loss": 7.336,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -0.3089408874511719,
      "rewards/margins": 0.19231058657169342,
      "rewards/rejected": -0.5012514591217041,
      "step": 130
    },
    {
      "epoch": 0.11489536315141567,
      "grad_norm": 0.08581534773111343,
      "learning_rate": 4.893298743830168e-06,
      "logits/chosen": -2.115981340408325,
      "logits/rejected": -2.5363636016845703,
      "logps/chosen": -0.22111928462982178,
      "logps/rejected": -0.3136863708496094,
      "loss": 7.2892,
      "rewards/accuracies": 0.550000011920929,
      "rewards/chosen": -0.33167898654937744,
      "rewards/margins": 0.1388506144285202,
      "rewards/rejected": -0.47052955627441406,
      "step": 140
    },
    {
      "epoch": 0.12310217480508823,
      "grad_norm": 0.06293604522943497,
      "learning_rate": 4.8776412907378845e-06,
      "logits/chosen": -2.070842742919922,
      "logits/rejected": -2.4669342041015625,
      "logps/chosen": -0.20812074840068817,
      "logps/rejected": -0.29536327719688416,
      "loss": 7.26,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": -0.31218111515045166,
      "rewards/margins": 0.13086381554603577,
      "rewards/rejected": -0.4430449604988098,
      "step": 150
    },
    {
      "epoch": 0.12310217480508823,
      "eval_logits/chosen": -2.062544822692871,
      "eval_logits/rejected": -2.5318312644958496,
      "eval_logps/chosen": -0.2108660489320755,
      "eval_logps/rejected": -0.3196176588535309,
      "eval_loss": 0.8929102420806885,
      "eval_rewards/accuracies": 0.5555555820465088,
      "eval_rewards/chosen": -0.31629908084869385,
      "eval_rewards/margins": 0.1631273776292801,
      "eval_rewards/rejected": -0.47942644357681274,
      "eval_runtime": 26.0407,
      "eval_samples_per_second": 30.26,
      "eval_steps_per_second": 3.802,
      "step": 150
    },
    {
      "epoch": 0.13130898645876077,
      "grad_norm": 0.06755395233631134,
      "learning_rate": 4.860940925593703e-06,
      "logits/chosen": -2.187638998031616,
      "logits/rejected": -2.4928510189056396,
      "logps/chosen": -0.2070399969816208,
      "logps/rejected": -0.30727890133857727,
      "loss": 7.1947,
      "rewards/accuracies": 0.550000011920929,
      "rewards/chosen": -0.3105599880218506,
      "rewards/margins": 0.1503583937883377,
      "rewards/rejected": -0.4609183669090271,
      "step": 160
    },
    {
      "epoch": 0.1395157981124333,
      "grad_norm": 0.08956371247768402,
      "learning_rate": 4.84320497372973e-06,
      "logits/chosen": -2.0751285552978516,
      "logits/rejected": -2.478673219680786,
      "logps/chosen": -0.18197472393512726,
      "logps/rejected": -0.2756109833717346,
      "loss": 7.1774,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -0.2729620933532715,
      "rewards/margins": 0.14045441150665283,
      "rewards/rejected": -0.41341647505760193,
      "step": 170
    },
    {
      "epoch": 0.14772260976610588,
      "grad_norm": 0.07708129286766052,
      "learning_rate": 4.824441214720629e-06,
      "logits/chosen": -2.113537311553955,
      "logits/rejected": -2.530677556991577,
      "logps/chosen": -0.20599500834941864,
      "logps/rejected": -0.2911488711833954,
      "loss": 7.1722,
      "rewards/accuracies": 0.512499988079071,
      "rewards/chosen": -0.308992475271225,
      "rewards/margins": 0.1277308166027069,
      "rewards/rejected": -0.4367233216762543,
      "step": 180
    },
    {
      "epoch": 0.15592942141977842,
      "grad_norm": 0.0884585976600647,
      "learning_rate": 4.804657878971252e-06,
      "logits/chosen": -2.151444673538208,
      "logits/rejected": -2.559861898422241,
      "logps/chosen": -0.2093551605939865,
      "logps/rejected": -0.2878231108188629,
      "loss": 7.1186,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": -0.31403273344039917,
      "rewards/margins": 0.11770190298557281,
      "rewards/rejected": -0.4317346513271332,
      "step": 190
    },
    {
      "epoch": 0.16413623307345096,
      "grad_norm": 0.09445559978485107,
      "learning_rate": 4.783863644106502e-06,
      "logits/chosen": -2.278620481491089,
      "logits/rejected": -2.5897645950317383,
      "logps/chosen": -0.18631704151630402,
      "logps/rejected": -0.3201253116130829,
      "loss": 7.082,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -0.2794755697250366,
      "rewards/margins": 0.2007124423980713,
      "rewards/rejected": -0.4801879823207855,
      "step": 200
    },
    {
      "epoch": 0.16413623307345096,
      "eval_logits/chosen": -2.1718955039978027,
      "eval_logits/rejected": -2.6710257530212402,
      "eval_logps/chosen": -0.20382821559906006,
      "eval_logps/rejected": -0.3390556573867798,
      "eval_loss": 0.8775798678398132,
      "eval_rewards/accuracies": 0.5858585834503174,
      "eval_rewards/chosen": -0.3057423532009125,
      "eval_rewards/margins": 0.20284107327461243,
      "eval_rewards/rejected": -0.5085834264755249,
      "eval_runtime": 26.0531,
      "eval_samples_per_second": 30.246,
      "eval_steps_per_second": 3.8,
      "step": 200
    },
    {
      "epoch": 0.1723430447271235,
      "grad_norm": 0.11077430099248886,
      "learning_rate": 4.762067631165049e-06,
      "logits/chosen": -2.2566323280334473,
      "logits/rejected": -2.621065378189087,
      "logps/chosen": -0.18663282692432404,
      "logps/rejected": -0.290865957736969,
      "loss": 7.1321,
      "rewards/accuracies": 0.5,
      "rewards/chosen": -0.27994924783706665,
      "rewards/margins": 0.15634974837303162,
      "rewards/rejected": -0.43629899621009827,
      "step": 210
    },
    {
      "epoch": 0.18054985638079607,
      "grad_norm": 0.15500974655151367,
      "learning_rate": 4.7392794005985324e-06,
      "logits/chosen": -2.27781343460083,
      "logits/rejected": -2.700369358062744,
      "logps/chosen": -0.21367880702018738,
      "logps/rejected": -0.31559067964553833,
      "loss": 6.9886,
      "rewards/accuracies": 0.5874999761581421,
      "rewards/chosen": -0.3205181956291199,
      "rewards/margins": 0.15286779403686523,
      "rewards/rejected": -0.4733859896659851,
      "step": 220
    },
    {
      "epoch": 0.1887566680344686,
      "grad_norm": 0.12770676612854004,
      "learning_rate": 4.715508948078037e-06,
      "logits/chosen": -2.216815710067749,
      "logits/rejected": -2.759458541870117,
      "logps/chosen": -0.21546092629432678,
      "logps/rejected": -0.34664005041122437,
      "loss": 6.966,
      "rewards/accuracies": 0.574999988079071,
      "rewards/chosen": -0.32319143414497375,
      "rewards/margins": 0.196768656373024,
      "rewards/rejected": -0.5199600458145142,
      "step": 230
    },
    {
      "epoch": 0.19696347968814115,
      "grad_norm": 0.15062908828258514,
      "learning_rate": 4.690766700109659e-06,
      "logits/chosen": -2.2262110710144043,
      "logits/rejected": -2.7840607166290283,
      "logps/chosen": -0.2078159749507904,
      "logps/rejected": -0.4006090760231018,
      "loss": 7.012,
      "rewards/accuracies": 0.625,
      "rewards/chosen": -0.3117239773273468,
      "rewards/margins": 0.2891896665096283,
      "rewards/rejected": -0.6009136438369751,
      "step": 240
    },
    {
      "epoch": 0.2051702913418137,
      "grad_norm": 0.24995267391204834,
      "learning_rate": 4.665063509461098e-06,
      "logits/chosen": -2.4198288917541504,
      "logits/rejected": -2.8148205280303955,
      "logps/chosen": -0.23191122710704803,
      "logps/rejected": -0.38251757621765137,
      "loss": 6.7812,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": -0.34786686301231384,
      "rewards/margins": 0.2259095013141632,
      "rewards/rejected": -0.573776364326477,
      "step": 250
    },
    {
      "epoch": 0.2051702913418137,
      "eval_logits/chosen": -2.3532145023345947,
      "eval_logits/rejected": -2.9015841484069824,
      "eval_logps/chosen": -0.22620753943920135,
      "eval_logps/rejected": -0.4290919005870819,
      "eval_loss": 0.8434350490570068,
      "eval_rewards/accuracies": 0.5959596037864685,
      "eval_rewards/chosen": -0.33931130170822144,
      "eval_rewards/margins": 0.304326593875885,
      "eval_rewards/rejected": -0.6436378955841064,
      "eval_runtime": 26.012,
      "eval_samples_per_second": 30.294,
      "eval_steps_per_second": 3.806,
      "step": 250
    },
    {
      "epoch": 0.21337710299548626,
      "grad_norm": 0.22134838998317719,
      "learning_rate": 4.638410650401267e-06,
      "logits/chosen": -2.4054629802703857,
      "logits/rejected": -2.8117618560791016,
      "logps/chosen": -0.20882606506347656,
      "logps/rejected": -0.41059261560440063,
      "loss": 6.6024,
      "rewards/accuracies": 0.5625,
      "rewards/chosen": -0.31323909759521484,
      "rewards/margins": 0.3026497960090637,
      "rewards/rejected": -0.6158889532089233,
      "step": 260
    },
    {
      "epoch": 0.2215839146491588,
      "grad_norm": 0.23838171362876892,
      "learning_rate": 4.610819813755038e-06,
      "logits/chosen": -2.534034490585327,
      "logits/rejected": -2.846797227859497,
      "logps/chosen": -0.2372482568025589,
      "logps/rejected": -0.493452787399292,
      "loss": 6.7377,
      "rewards/accuracies": 0.612500011920929,
      "rewards/chosen": -0.35587236285209656,
      "rewards/margins": 0.38430681824684143,
      "rewards/rejected": -0.740179181098938,
      "step": 270
    },
    {
      "epoch": 0.22979072630283134,
      "grad_norm": 0.2903882563114166,
      "learning_rate": 4.582303101775249e-06,
      "logits/chosen": -2.4838695526123047,
      "logits/rejected": -2.8494999408721924,
      "logps/chosen": -0.23487380146980286,
      "logps/rejected": -0.5447143316268921,
      "loss": 6.6463,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.3523106873035431,
      "rewards/margins": 0.4647606909275055,
      "rewards/rejected": -0.8170714378356934,
      "step": 280
    },
    {
      "epoch": 0.23799753795650389,
      "grad_norm": 0.2977660596370697,
      "learning_rate": 4.55287302283426e-06,
      "logits/chosen": -2.539785861968994,
      "logits/rejected": -2.922631025314331,
      "logps/chosen": -0.2749824821949005,
      "logps/rejected": -0.5817859768867493,
      "loss": 6.2424,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": -0.41247373819351196,
      "rewards/margins": 0.4602052569389343,
      "rewards/rejected": -0.8726789355278015,
      "step": 290
    },
    {
      "epoch": 0.24620434961017645,
      "grad_norm": 0.388954222202301,
      "learning_rate": 4.522542485937369e-06,
      "logits/chosen": -2.6039352416992188,
      "logits/rejected": -2.8011627197265625,
      "logps/chosen": -0.37122753262519836,
      "logps/rejected": -0.6376734972000122,
      "loss": 6.3458,
      "rewards/accuracies": 0.5375000238418579,
      "rewards/chosen": -0.5568413138389587,
      "rewards/margins": 0.39966899156570435,
      "rewards/rejected": -0.9565103650093079,
      "step": 300
    },
    {
      "epoch": 0.24620434961017645,
      "eval_logits/chosen": -2.448641061782837,
      "eval_logits/rejected": -2.8639307022094727,
      "eval_logps/chosen": -0.35992980003356934,
      "eval_logps/rejected": -0.7961164712905884,
      "eval_loss": 0.7593368887901306,
      "eval_rewards/accuracies": 0.6060606241226196,
      "eval_rewards/chosen": -0.5398945808410645,
      "eval_rewards/margins": 0.6542800664901733,
      "eval_rewards/rejected": -1.1941747665405273,
      "eval_runtime": 26.0378,
      "eval_samples_per_second": 30.264,
      "eval_steps_per_second": 3.802,
      "step": 300
    },
    {
      "epoch": 0.254411161263849,
      "grad_norm": 0.41667941212654114,
      "learning_rate": 4.491324795060491e-06,
      "logits/chosen": -2.596318483352661,
      "logits/rejected": -2.8140273094177246,
      "logps/chosen": -0.3550270199775696,
      "logps/rejected": -0.673725426197052,
      "loss": 6.2201,
      "rewards/accuracies": 0.512499988079071,
      "rewards/chosen": -0.532540500164032,
      "rewards/margins": 0.47804751992225647,
      "rewards/rejected": -1.0105879306793213,
      "step": 310
    },
    {
      "epoch": 0.26261797291752154,
      "grad_norm": 0.5356656312942505,
      "learning_rate": 4.4592336433146e-06,
      "logits/chosen": -2.5131359100341797,
      "logits/rejected": -2.7125701904296875,
      "logps/chosen": -0.4790285527706146,
      "logps/rejected": -0.9030188322067261,
      "loss": 5.9656,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -0.7185428738594055,
      "rewards/margins": 0.6359853744506836,
      "rewards/rejected": -1.3545281887054443,
      "step": 320
    },
    {
      "epoch": 0.2708247845711941,
      "grad_norm": 0.6395894885063171,
      "learning_rate": 4.426283106939474e-06,
      "logits/chosen": -2.408463954925537,
      "logits/rejected": -2.784302234649658,
      "logps/chosen": -0.43364158272743225,
      "logps/rejected": -0.8759375810623169,
      "loss": 5.6972,
      "rewards/accuracies": 0.6000000238418579,
      "rewards/chosen": -0.6504624485969543,
      "rewards/margins": 0.6634438037872314,
      "rewards/rejected": -1.313906192779541,
      "step": 330
    },
    {
      "epoch": 0.2790315962248666,
      "grad_norm": 0.5482347011566162,
      "learning_rate": 4.3924876391293915e-06,
      "logits/chosen": -2.4591078758239746,
      "logits/rejected": -2.7352993488311768,
      "logps/chosen": -0.4786604344844818,
      "logps/rejected": -1.337053894996643,
      "loss": 5.5123,
      "rewards/accuracies": 0.6625000238418579,
      "rewards/chosen": -0.7179905772209167,
      "rewards/margins": 1.2875900268554688,
      "rewards/rejected": -2.0055809020996094,
      "step": 340
    },
    {
      "epoch": 0.2872384078785392,
      "grad_norm": 0.44611233472824097,
      "learning_rate": 4.357862063693486e-06,
      "logits/chosen": -2.426030397415161,
      "logits/rejected": -2.7272305488586426,
      "logps/chosen": -0.5190738439559937,
      "logps/rejected": -1.3207228183746338,
      "loss": 5.242,
      "rewards/accuracies": 0.6875,
      "rewards/chosen": -0.7786108255386353,
      "rewards/margins": 1.2024734020233154,
      "rewards/rejected": -1.9810841083526611,
      "step": 350
    },
    {
      "epoch": 0.2872384078785392,
      "eval_logits/chosen": -2.5548582077026367,
      "eval_logits/rejected": -2.8742663860321045,
      "eval_logps/chosen": -0.5429244637489319,
      "eval_logps/rejected": -1.556882619857788,
      "eval_loss": 0.6406257748603821,
      "eval_rewards/accuracies": 0.6464646458625793,
      "eval_rewards/chosen": -0.8143868446350098,
      "eval_rewards/margins": 1.520937204360962,
      "eval_rewards/rejected": -2.335324287414551,
      "eval_runtime": 26.0367,
      "eval_samples_per_second": 30.265,
      "eval_steps_per_second": 3.802,
      "step": 350
    },
    {
      "epoch": 0.29544521953221176,
      "grad_norm": 0.6435768008232117,
      "learning_rate": 4.322421568553529e-06,
      "logits/chosen": -2.5521063804626465,
      "logits/rejected": -2.8884310722351074,
      "logps/chosen": -0.5124669075012207,
      "logps/rejected": -1.4227123260498047,
      "loss": 5.2755,
      "rewards/accuracies": 0.6499999761581421,
      "rewards/chosen": -0.768700361251831,
      "rewards/margins": 1.365368127822876,
      "rewards/rejected": -2.134068489074707,
      "step": 360
    },
    {
      "epoch": 0.3036520311858843,
      "grad_norm": 0.7270930409431458,
      "learning_rate": 4.286181699082008e-06,
      "logits/chosen": -2.613312244415283,
      "logits/rejected": -2.8482322692871094,
      "logps/chosen": -0.7397282719612122,
      "logps/rejected": -1.7584373950958252,
      "loss": 4.6592,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -1.1095924377441406,
      "rewards/margins": 1.528063416481018,
      "rewards/rejected": -2.6376559734344482,
      "step": 370
    },
    {
      "epoch": 0.31185884283955684,
      "grad_norm": 0.417350172996521,
      "learning_rate": 4.249158351283414e-06,
      "logits/chosen": -2.6574740409851074,
      "logits/rejected": -2.8720998764038086,
      "logps/chosen": -0.7095499634742737,
      "logps/rejected": -2.2104392051696777,
      "loss": 4.4512,
      "rewards/accuracies": 0.637499988079071,
      "rewards/chosen": -1.0643248558044434,
      "rewards/margins": 2.2513341903686523,
      "rewards/rejected": -3.3156590461730957,
      "step": 380
    },
    {
      "epoch": 0.3200656544932294,
      "grad_norm": 0.5288704633712769,
      "learning_rate": 4.211367764821722e-06,
      "logits/chosen": -2.7438926696777344,
      "logits/rejected": -2.9578990936279297,
      "logps/chosen": -0.7221090793609619,
      "logps/rejected": -2.169327735900879,
      "loss": 4.8545,
      "rewards/accuracies": 0.5249999761581421,
      "rewards/chosen": -1.0831634998321533,
      "rewards/margins": 2.170828104019165,
      "rewards/rejected": -3.2539916038513184,
      "step": 390
    },
    {
      "epoch": 0.3282724661469019,
      "grad_norm": 0.5342369675636292,
      "learning_rate": 4.172826515897146e-06,
      "logits/chosen": -2.451097011566162,
      "logits/rejected": -2.8298330307006836,
      "logps/chosen": -0.6682878732681274,
      "logps/rejected": -2.4527621269226074,
      "loss": 4.5987,
      "rewards/accuracies": 0.7124999761581421,
      "rewards/chosen": -1.0024317502975464,
      "rewards/margins": 2.676711320877075,
      "rewards/rejected": -3.679143190383911,
      "step": 400
    },
    {
      "epoch": 0.3282724661469019,
      "eval_logits/chosen": -2.486518621444702,
      "eval_logits/rejected": -2.884521722793579,
      "eval_logps/chosen": -0.6234225630760193,
      "eval_logps/rejected": -2.3772060871124268,
      "eval_loss": 0.5354036688804626,
      "eval_rewards/accuracies": 0.6464646458625793,
      "eval_rewards/chosen": -0.9351338744163513,
      "eval_rewards/margins": 2.6306753158569336,
      "eval_rewards/rejected": -3.5658092498779297,
      "eval_runtime": 26.0392,
      "eval_samples_per_second": 30.262,
      "eval_steps_per_second": 3.802,
      "step": 400
    },
    {
      "epoch": 0.33647927780057446,
      "grad_norm": 0.3894485831260681,
      "learning_rate": 4.133551509975264e-06,
      "logits/chosen": -2.498328685760498,
      "logits/rejected": -2.832383394241333,
      "logps/chosen": -0.8072039484977722,
      "logps/rejected": -3.0143685340881348,
      "loss": 4.1825,
      "rewards/accuracies": 0.737500011920929,
      "rewards/chosen": -1.210806131362915,
      "rewards/margins": 3.310746669769287,
      "rewards/rejected": -4.521553039550781,
      "step": 410
    },
    {
      "epoch": 0.344686089454247,
      "grad_norm": 1.1384832859039307,
      "learning_rate": 4.093559974371725e-06,
      "logits/chosen": -2.3939032554626465,
      "logits/rejected": -2.736551284790039,
      "logps/chosen": -0.8623531460762024,
      "logps/rejected": -3.228794813156128,
      "loss": 4.0383,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -1.293529748916626,
      "rewards/margins": 3.5496623516082764,
      "rewards/rejected": -4.843192100524902,
      "step": 420
    },
    {
      "epoch": 0.3528929011079196,
      "grad_norm": 0.5636156797409058,
      "learning_rate": 4.052869450695776e-06,
      "logits/chosen": -2.466431140899658,
      "logits/rejected": -2.7423882484436035,
      "logps/chosen": -0.867216944694519,
      "logps/rejected": -2.9713022708892822,
      "loss": 3.9764,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -1.3008254766464233,
      "rewards/margins": 3.156127691268921,
      "rewards/rejected": -4.456953525543213,
      "step": 430
    },
    {
      "epoch": 0.36109971276159214,
      "grad_norm": 1.3254886865615845,
      "learning_rate": 4.011497787155938e-06,
      "logits/chosen": -2.438335418701172,
      "logits/rejected": -2.716845989227295,
      "logps/chosen": -0.9805113673210144,
      "logps/rejected": -3.646573543548584,
      "loss": 4.0643,
      "rewards/accuracies": 0.737500011920929,
      "rewards/chosen": -1.4707671403884888,
      "rewards/margins": 3.9990932941436768,
      "rewards/rejected": -5.469860553741455,
      "step": 440
    },
    {
      "epoch": 0.3693065244152647,
      "grad_norm": 1.0398495197296143,
      "learning_rate": 3.969463130731183e-06,
      "logits/chosen": -2.3812689781188965,
      "logits/rejected": -2.7526910305023193,
      "logps/chosen": -1.0975981950759888,
      "logps/rejected": -3.9439749717712402,
      "loss": 3.9749,
      "rewards/accuracies": 0.75,
      "rewards/chosen": -1.6463972330093384,
      "rewards/margins": 4.269565105438232,
      "rewards/rejected": -5.915962219238281,
      "step": 450
    },
    {
      "epoch": 0.3693065244152647,
      "eval_logits/chosen": -2.4038827419281006,
      "eval_logits/rejected": -2.7936673164367676,
      "eval_logps/chosen": -1.0481945276260376,
      "eval_logps/rejected": -3.5816102027893066,
      "eval_loss": 0.47537824511528015,
      "eval_rewards/accuracies": 0.7373737096786499,
      "eval_rewards/chosen": -1.5722918510437012,
      "eval_rewards/margins": 3.800123691558838,
      "eval_rewards/rejected": -5.372415542602539,
      "eval_runtime": 26.0116,
      "eval_samples_per_second": 30.294,
      "eval_steps_per_second": 3.806,
      "step": 450
    },
    {
      "epoch": 0.3775133360689372,
      "grad_norm": 1.091323971748352,
      "learning_rate": 3.92678391921108e-06,
      "logits/chosen": -2.5687508583068848,
      "logits/rejected": -2.766624927520752,
      "logps/chosen": -1.1868960857391357,
      "logps/rejected": -3.4050726890563965,
      "loss": 4.114,
      "rewards/accuracies": 0.762499988079071,
      "rewards/chosen": -1.780344009399414,
      "rewards/margins": 3.3272647857666016,
      "rewards/rejected": -5.107609272003174,
      "step": 460
    },
    {
      "epoch": 0.38572014772260976,
      "grad_norm": 0.8932163715362549,
      "learning_rate": 3.88347887310836e-06,
      "logits/chosen": -2.438856601715088,
      "logits/rejected": -2.8660335540771484,
      "logps/chosen": -1.420062780380249,
      "logps/rejected": -4.232975006103516,
      "loss": 3.9707,
      "rewards/accuracies": 0.7749999761581421,
      "rewards/chosen": -2.130094051361084,
      "rewards/margins": 4.219367980957031,
      "rewards/rejected": -6.349462032318115,
      "step": 470
    },
    {
      "epoch": 0.3939269593762823,
      "grad_norm": 1.110568642616272,
      "learning_rate": 3.839566987447492e-06,
      "logits/chosen": -2.5447945594787598,
      "logits/rejected": -2.720705032348633,
      "logps/chosen": -1.9743589162826538,
      "logps/rejected": -3.963395595550537,
      "loss": 3.7559,
      "rewards/accuracies": 0.8125,
      "rewards/chosen": -2.961538314819336,
      "rewards/margins": 2.983555316925049,
      "rewards/rejected": -5.945094108581543,
      "step": 480
    },
    {
      "epoch": 0.40213377102995485,
      "grad_norm": 1.1167631149291992,
      "learning_rate": 3.795067523432826e-06,
      "logits/chosen": -2.6660027503967285,
      "logits/rejected": -2.839933395385742,
      "logps/chosen": -2.3371267318725586,
      "logps/rejected": -4.315327167510986,
      "loss": 3.5333,
      "rewards/accuracies": 0.7875000238418579,
      "rewards/chosen": -3.505690097808838,
      "rewards/margins": 2.9673006534576416,
      "rewards/rejected": -6.4729905128479,
      "step": 490
    },
    {
      "epoch": 0.4103405826836274,
      "grad_norm": 1.7681843042373657,
      "learning_rate": 3.7500000000000005e-06,
      "logits/chosen": -2.4737250804901123,
      "logits/rejected": -2.7522482872009277,
      "logps/chosen": -2.802852153778076,
      "logps/rejected": -5.4331955909729,
      "loss": 3.1958,
      "rewards/accuracies": 0.800000011920929,
      "rewards/chosen": -4.204278469085693,
      "rewards/margins": 3.9455153942108154,
      "rewards/rejected": -8.14979362487793,
      "step": 500
    },
    {
      "epoch": 0.4103405826836274,
      "eval_logits/chosen": -2.464381456375122,
      "eval_logits/rejected": -2.8693907260894775,
      "eval_logps/chosen": -2.83821177482605,
      "eval_logps/rejected": -5.77597188949585,
      "eval_loss": 0.3702296018600464,
      "eval_rewards/accuracies": 0.8383838534355164,
      "eval_rewards/chosen": -4.257318019866943,
      "eval_rewards/margins": 4.406640529632568,
      "eval_rewards/rejected": -8.663958549499512,
      "eval_runtime": 26.0401,
      "eval_samples_per_second": 30.261,
      "eval_steps_per_second": 3.802,
      "step": 500
    },
    {
      "epoch": 0.4185473943373,
      "grad_norm": 2.1794912815093994,
      "learning_rate": 3.7043841852542884e-06,
      "logits/chosen": -2.4379305839538574,
      "logits/rejected": -2.823613166809082,
      "logps/chosen": -3.3340702056884766,
      "logps/rejected": -6.568157196044922,
      "loss": 2.8418,
      "rewards/accuracies": 0.8999999761581421,
      "rewards/chosen": -5.001105308532715,
      "rewards/margins": 4.851131439208984,
      "rewards/rejected": -9.8522367477417,
      "step": 510
    },
    {
      "epoch": 0.4267542059909725,
      "grad_norm": 2.4649722576141357,
      "learning_rate": 3.658240087799655e-06,
      "logits/chosen": -2.627397298812866,
      "logits/rejected": -2.9164626598358154,
      "logps/chosen": -5.047720909118652,
      "logps/rejected": -7.8882904052734375,
      "loss": 2.73,
      "rewards/accuracies": 0.8374999761581421,
      "rewards/chosen": -7.5715813636779785,
      "rewards/margins": 4.2608537673950195,
      "rewards/rejected": -11.832433700561523,
      "step": 520
    },
    {
      "epoch": 0.43496101764464506,
      "grad_norm": 2.53389835357666,
      "learning_rate": 3.611587947962319e-06,
      "logits/chosen": -2.4901440143585205,
      "logits/rejected": -2.8462088108062744,
      "logps/chosen": -4.257150173187256,
      "logps/rejected": -7.8700971603393555,
      "loss": 2.5886,
      "rewards/accuracies": 0.8999999761581421,
      "rewards/chosen": -6.385725498199463,
      "rewards/margins": 5.419419288635254,
      "rewards/rejected": -11.805145263671875,
      "step": 530
    },
    {
      "epoch": 0.4431678292983176,
      "grad_norm": 5.214587688446045,
      "learning_rate": 3.564448228912682e-06,
      "logits/chosen": -2.4020960330963135,
      "logits/rejected": -2.8110060691833496,
      "logps/chosen": -4.833271503448486,
      "logps/rejected": -8.95389461517334,
      "loss": 2.5918,
      "rewards/accuracies": 0.875,
      "rewards/chosen": -7.24990701675415,
      "rewards/margins": 6.180935859680176,
      "rewards/rejected": -13.4308443069458,
      "step": 540
    },
    {
      "epoch": 0.45137464095199015,
      "grad_norm": 2.7313215732574463,
      "learning_rate": 3.516841607689501e-06,
      "logits/chosen": -2.5269761085510254,
      "logits/rejected": -2.780574321746826,
      "logps/chosen": -5.635412693023682,
      "logps/rejected": -9.682252883911133,
      "loss": 2.3387,
      "rewards/accuracies": 0.8999999761581421,
      "rewards/chosen": -8.453118324279785,
      "rewards/margins": 6.070260524749756,
      "rewards/rejected": -14.5233793258667,
      "step": 550
    },
    {
      "epoch": 0.45137464095199015,
      "eval_logits/chosen": -2.4869625568389893,
      "eval_logits/rejected": -2.8525495529174805,
      "eval_logps/chosen": -5.279529094696045,
      "eval_logps/rejected": -9.829012870788574,
      "eval_loss": 0.28920263051986694,
      "eval_rewards/accuracies": 0.8888888955116272,
      "eval_rewards/chosen": -7.91929292678833,
      "eval_rewards/margins": 6.824225425720215,
      "eval_rewards/rejected": -14.743520736694336,
      "eval_runtime": 26.0459,
      "eval_samples_per_second": 30.254,
      "eval_steps_per_second": 3.801,
      "step": 550
    },
    {
      "epoch": 0.4595814526056627,
      "grad_norm": 7.664855003356934,
      "learning_rate": 3.4687889661302577e-06,
      "logits/chosen": -2.517958164215088,
      "logits/rejected": -2.821098804473877,
      "logps/chosen": -6.172144889831543,
      "logps/rejected": -11.623320579528809,
      "loss": 2.1915,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -9.258216857910156,
      "rewards/margins": 8.176764488220215,
      "rewards/rejected": -17.434982299804688,
      "step": 560
    },
    {
      "epoch": 0.46778826425933523,
      "grad_norm": 5.298526287078857,
      "learning_rate": 3.4203113817116955e-06,
      "logits/chosen": -2.4968509674072266,
      "logits/rejected": -2.7755746841430664,
      "logps/chosen": -6.602853298187256,
      "logps/rejected": -12.261907577514648,
      "loss": 1.9401,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -9.904280662536621,
      "rewards/margins": 8.488581657409668,
      "rewards/rejected": -18.39286231994629,
      "step": 570
    },
    {
      "epoch": 0.47599507591300777,
      "grad_norm": 3.2574403285980225,
      "learning_rate": 3.3714301183045382e-06,
      "logits/chosen": -2.510072708129883,
      "logits/rejected": -2.7066006660461426,
      "logps/chosen": -6.970999240875244,
      "logps/rejected": -12.280068397521973,
      "loss": 2.201,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -10.456499099731445,
      "rewards/margins": 7.963601589202881,
      "rewards/rejected": -18.420101165771484,
      "step": 580
    },
    {
      "epoch": 0.48420188756668037,
      "grad_norm": 7.752432346343994,
      "learning_rate": 3.3221666168464584e-06,
      "logits/chosen": -2.4650096893310547,
      "logits/rejected": -2.7430062294006348,
      "logps/chosen": -7.387778282165527,
      "logps/rejected": -12.886815071105957,
      "loss": 1.7232,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -11.081666946411133,
      "rewards/margins": 8.248555183410645,
      "rewards/rejected": -19.330224990844727,
      "step": 590
    },
    {
      "epoch": 0.4924086992203529,
      "grad_norm": 7.162052154541016,
      "learning_rate": 3.272542485937369e-06,
      "logits/chosen": -2.502263069152832,
      "logits/rejected": -2.775761365890503,
      "logps/chosen": -7.409277439117432,
      "logps/rejected": -12.631124496459961,
      "loss": 1.7381,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": -11.11391544342041,
      "rewards/margins": 7.832771301269531,
      "rewards/rejected": -18.946685791015625,
      "step": 600
    },
    {
      "epoch": 0.4924086992203529,
      "eval_logits/chosen": -2.414571523666382,
      "eval_logits/rejected": -2.7700724601745605,
      "eval_logps/chosen": -6.709319591522217,
      "eval_logps/rejected": -12.794181823730469,
      "eval_loss": 0.22023409605026245,
      "eval_rewards/accuracies": 0.9494949579238892,
      "eval_rewards/chosen": -10.063980102539062,
      "eval_rewards/margins": 9.127293586730957,
      "eval_rewards/rejected": -19.191272735595703,
      "eval_runtime": 26.0239,
      "eval_samples_per_second": 30.28,
      "eval_steps_per_second": 3.804,
      "step": 600
    },
    {
      "epoch": 0.5006155108740254,
      "grad_norm": 6.312295436859131,
      "learning_rate": 3.222579492361179e-06,
      "logits/chosen": -2.48272705078125,
      "logits/rejected": -2.7790474891662598,
      "logps/chosen": -7.929927825927734,
      "logps/rejected": -14.165933609008789,
      "loss": 1.8554,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -11.894890785217285,
      "rewards/margins": 9.354007720947266,
      "rewards/rejected": -21.248899459838867,
      "step": 610
    },
    {
      "epoch": 0.508822322527698,
      "grad_norm": 5.254099369049072,
      "learning_rate": 3.1722995515381644e-06,
      "logits/chosen": -2.4686760902404785,
      "logits/rejected": -2.7497403621673584,
      "logps/chosen": -6.739262580871582,
      "logps/rejected": -13.017419815063477,
      "loss": 1.5174,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": -10.108892440795898,
      "rewards/margins": 9.417237281799316,
      "rewards/rejected": -19.52613067626953,
      "step": 620
    },
    {
      "epoch": 0.5170291341813705,
      "grad_norm": 2.79254412651062,
      "learning_rate": 3.121724717912138e-06,
      "logits/chosen": -2.5906078815460205,
      "logits/rejected": -2.776789426803589,
      "logps/chosen": -8.108521461486816,
      "logps/rejected": -13.654635429382324,
      "loss": 1.8195,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": -12.162781715393066,
      "rewards/margins": 8.319172859191895,
      "rewards/rejected": -20.48195457458496,
      "step": 630
    },
    {
      "epoch": 0.5252359458350431,
      "grad_norm": 4.617709636688232,
      "learning_rate": 3.0708771752766397e-06,
      "logits/chosen": -2.4493868350982666,
      "logits/rejected": -2.7978568077087402,
      "logps/chosen": -7.94122314453125,
      "logps/rejected": -14.420333862304688,
      "loss": 1.7681,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -11.911835670471191,
      "rewards/margins": 9.718667984008789,
      "rewards/rejected": -21.630504608154297,
      "step": 640
    },
    {
      "epoch": 0.5334427574887156,
      "grad_norm": 15.788602828979492,
      "learning_rate": 3.019779227044398e-06,
      "logits/chosen": -2.495669364929199,
      "logits/rejected": -2.799513101577759,
      "logps/chosen": -7.9494948387146,
      "logps/rejected": -14.22314167022705,
      "loss": 1.7372,
      "rewards/accuracies": 0.9125000238418579,
      "rewards/chosen": -11.924242973327637,
      "rewards/margins": 9.410469055175781,
      "rewards/rejected": -21.3347110748291,
      "step": 650
    },
    {
      "epoch": 0.5334427574887156,
      "eval_logits/chosen": -2.4212491512298584,
      "eval_logits/rejected": -2.7830140590667725,
      "eval_logps/chosen": -6.399094581604004,
      "eval_logps/rejected": -13.696700096130371,
      "eval_loss": 0.19624173641204834,
      "eval_rewards/accuracies": 0.9595959782600403,
      "eval_rewards/chosen": -9.598641395568848,
      "eval_rewards/margins": 10.94640827178955,
      "eval_rewards/rejected": -20.5450496673584,
      "eval_runtime": 26.0246,
      "eval_samples_per_second": 30.279,
      "eval_steps_per_second": 3.804,
      "step": 650
    },
    {
      "epoch": 0.5416495691423882,
      "grad_norm": 1.8350448608398438,
      "learning_rate": 2.9684532864643123e-06,
      "logits/chosen": -2.4959726333618164,
      "logits/rejected": -2.7555196285247803,
      "logps/chosen": -7.877779483795166,
      "logps/rejected": -15.448628425598145,
      "loss": 1.2842,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -11.816670417785645,
      "rewards/margins": 11.35627269744873,
      "rewards/rejected": -23.172941207885742,
      "step": 660
    },
    {
      "epoch": 0.5498563807960607,
      "grad_norm": 5.017009735107422,
      "learning_rate": 2.9169218667902562e-06,
      "logits/chosen": -2.4682934284210205,
      "logits/rejected": -2.7640793323516846,
      "logps/chosen": -8.19635009765625,
      "logps/rejected": -16.366384506225586,
      "loss": 1.8156,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -12.294524192810059,
      "rewards/margins": 12.255053520202637,
      "rewards/rejected": -24.549579620361328,
      "step": 670
    },
    {
      "epoch": 0.5580631924497332,
      "grad_norm": 5.384292125701904,
      "learning_rate": 2.8652075714060296e-06,
      "logits/chosen": -2.381718873977661,
      "logits/rejected": -2.8303215503692627,
      "logps/chosen": -7.562445163726807,
      "logps/rejected": -16.368242263793945,
      "loss": 1.4923,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -11.343667984008789,
      "rewards/margins": 13.208694458007812,
      "rewards/rejected": -24.5523624420166,
      "step": 680
    },
    {
      "epoch": 0.5662700041034058,
      "grad_norm": 4.969016075134277,
      "learning_rate": 2.813333083910761e-06,
      "logits/chosen": -2.449165105819702,
      "logits/rejected": -2.7401328086853027,
      "logps/chosen": -7.27236795425415,
      "logps/rejected": -15.723406791687012,
      "loss": 1.464,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -10.908552169799805,
      "rewards/margins": 12.676557540893555,
      "rewards/rejected": -23.58510971069336,
      "step": 690
    },
    {
      "epoch": 0.5744768157570784,
      "grad_norm": 7.394704818725586,
      "learning_rate": 2.761321158169134e-06,
      "logits/chosen": -2.392939805984497,
      "logits/rejected": -2.800342559814453,
      "logps/chosen": -7.558518409729004,
      "logps/rejected": -17.065820693969727,
      "loss": 1.2698,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -11.337777137756348,
      "rewards/margins": 14.260955810546875,
      "rewards/rejected": -25.598730087280273,
      "step": 700
    },
    {
      "epoch": 0.5744768157570784,
      "eval_logits/chosen": -2.3927063941955566,
      "eval_logits/rejected": -2.72641921043396,
      "eval_logps/chosen": -6.722636699676514,
      "eval_logps/rejected": -15.053749084472656,
      "eval_loss": 0.16624146699905396,
      "eval_rewards/accuracies": 0.9595959782600403,
      "eval_rewards/chosen": -10.083953857421875,
      "eval_rewards/margins": 12.496668815612793,
      "eval_rewards/rejected": -22.580623626708984,
      "eval_runtime": 26.0296,
      "eval_samples_per_second": 30.273,
      "eval_steps_per_second": 3.803,
      "step": 700
    },
    {
      "epoch": 0.582683627410751,
      "grad_norm": 2.325347423553467,
      "learning_rate": 2.70919460833079e-06,
      "logits/chosen": -2.3837151527404785,
      "logits/rejected": -2.718639850616455,
      "logps/chosen": -7.265778541564941,
      "logps/rejected": -16.18191909790039,
      "loss": 1.4758,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -10.898667335510254,
      "rewards/margins": 13.374208450317383,
      "rewards/rejected": -24.272876739501953,
      "step": 710
    },
    {
      "epoch": 0.5908904390644235,
      "grad_norm": 8.78543758392334,
      "learning_rate": 2.6569762988232838e-06,
      "logits/chosen": -2.3132755756378174,
      "logits/rejected": -2.718923807144165,
      "logps/chosen": -7.462459564208984,
      "logps/rejected": -16.323883056640625,
      "loss": 1.4369,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -11.193689346313477,
      "rewards/margins": 13.292137145996094,
      "rewards/rejected": -24.485824584960938,
      "step": 720
    },
    {
      "epoch": 0.599097250718096,
      "grad_norm": 5.5666584968566895,
      "learning_rate": 2.604689134322999e-06,
      "logits/chosen": -2.3744144439697266,
      "logits/rejected": -2.6114182472229004,
      "logps/chosen": -7.262286186218262,
      "logps/rejected": -15.194799423217773,
      "loss": 1.3424,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -10.893428802490234,
      "rewards/margins": 11.898767471313477,
      "rewards/rejected": -22.79219627380371,
      "step": 730
    },
    {
      "epoch": 0.6073040623717686,
      "grad_norm": 2.3607444763183594,
      "learning_rate": 2.5523560497083927e-06,
      "logits/chosen": -2.391589641571045,
      "logits/rejected": -2.649766445159912,
      "logps/chosen": -8.12696647644043,
      "logps/rejected": -17.4465274810791,
      "loss": 1.2588,
      "rewards/accuracies": 1.0,
      "rewards/chosen": -12.190450668334961,
      "rewards/margins": 13.979342460632324,
      "rewards/rejected": -26.169790267944336,
      "step": 740
    },
    {
      "epoch": 0.6155108740254411,
      "grad_norm": 3.9366135597229004,
      "learning_rate": 2.5e-06,
      "logits/chosen": -2.4100446701049805,
      "logits/rejected": -2.6920006275177,
      "logps/chosen": -8.519265174865723,
      "logps/rejected": -16.66709327697754,
      "loss": 1.3798,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": -12.778898239135742,
      "rewards/margins": 12.221744537353516,
      "rewards/rejected": -25.000640869140625,
      "step": 750
    },
    {
      "epoch": 0.6155108740254411,
      "eval_logits/chosen": -2.377652406692505,
      "eval_logits/rejected": -2.7032172679901123,
      "eval_logps/chosen": -7.426075458526611,
      "eval_logps/rejected": -15.338685035705566,
      "eval_loss": 0.1500895917415619,
      "eval_rewards/accuracies": 0.9494949579238892,
      "eval_rewards/chosen": -11.139113426208496,
      "eval_rewards/margins": 11.868914604187012,
      "eval_rewards/rejected": -23.008026123046875,
      "eval_runtime": 26.0362,
      "eval_samples_per_second": 30.266,
      "eval_steps_per_second": 3.802,
      "step": 750
    },
    {
      "epoch": 0.6237176856791137,
      "grad_norm": 6.827406406402588,
      "learning_rate": 2.447643950291608e-06,
      "logits/chosen": -2.3275043964385986,
      "logits/rejected": -2.6344006061553955,
      "logps/chosen": -7.848365783691406,
      "logps/rejected": -17.661056518554688,
      "loss": 0.8437,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -11.77254867553711,
      "rewards/margins": 14.719035148620605,
      "rewards/rejected": -26.4915828704834,
      "step": 760
    },
    {
      "epoch": 0.6319244973327862,
      "grad_norm": 3.160689353942871,
      "learning_rate": 2.3953108656770018e-06,
      "logits/chosen": -2.480649709701538,
      "logits/rejected": -2.686522960662842,
      "logps/chosen": -9.342796325683594,
      "logps/rejected": -18.42390251159668,
      "loss": 1.1501,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -14.014195442199707,
      "rewards/margins": 13.62165641784668,
      "rewards/rejected": -27.635854721069336,
      "step": 770
    },
    {
      "epoch": 0.6401313089864588,
      "grad_norm": 3.751577854156494,
      "learning_rate": 2.3430237011767166e-06,
      "logits/chosen": -2.405439853668213,
      "logits/rejected": -2.7555787563323975,
      "logps/chosen": -7.906872749328613,
      "logps/rejected": -19.13258171081543,
      "loss": 1.1781,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -11.860309600830078,
      "rewards/margins": 16.838565826416016,
      "rewards/rejected": -28.69887351989746,
      "step": 780
    },
    {
      "epoch": 0.6483381206401313,
      "grad_norm": 3.2225842475891113,
      "learning_rate": 2.290805391669212e-06,
      "logits/chosen": -2.48250412940979,
      "logits/rejected": -2.7523207664489746,
      "logps/chosen": -9.760643005371094,
      "logps/rejected": -18.688955307006836,
      "loss": 0.8935,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -14.640963554382324,
      "rewards/margins": 13.392468452453613,
      "rewards/rejected": -28.033432006835938,
      "step": 790
    },
    {
      "epoch": 0.6565449322938038,
      "grad_norm": 6.352996826171875,
      "learning_rate": 2.238678841830867e-06,
      "logits/chosen": -2.417715072631836,
      "logits/rejected": -2.649390697479248,
      "logps/chosen": -9.004033088684082,
      "logps/rejected": -18.346654891967773,
      "loss": 0.908,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": -13.506048202514648,
      "rewards/margins": 14.013933181762695,
      "rewards/rejected": -27.51997947692871,
      "step": 800
    },
    {
      "epoch": 0.6565449322938038,
      "eval_logits/chosen": -2.3531436920166016,
      "eval_logits/rejected": -2.679354667663574,
      "eval_logps/chosen": -8.068766593933105,
      "eval_logps/rejected": -18.179340362548828,
      "eval_loss": 0.13774718344211578,
      "eval_rewards/accuracies": 0.9595959782600403,
      "eval_rewards/chosen": -12.1031494140625,
      "eval_rewards/margins": 15.165861129760742,
      "eval_rewards/rejected": -27.26900863647461,
      "eval_runtime": 26.0374,
      "eval_samples_per_second": 30.264,
      "eval_steps_per_second": 3.802,
      "step": 800
    },
    {
      "epoch": 0.6647517439474764,
      "grad_norm": 24.157991409301758,
      "learning_rate": 2.186666916089239e-06,
      "logits/chosen": -2.443021297454834,
      "logits/rejected": -2.726386308670044,
      "logps/chosen": -9.991727828979492,
      "logps/rejected": -19.389759063720703,
      "loss": 0.8971,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": -14.987591743469238,
      "rewards/margins": 14.0970458984375,
      "rewards/rejected": -29.084636688232422,
      "step": 810
    },
    {
      "epoch": 0.6729585556011489,
      "grad_norm": 1.581506371498108,
      "learning_rate": 2.134792428593971e-06,
      "logits/chosen": -2.352402687072754,
      "logits/rejected": -2.5905702114105225,
      "logps/chosen": -8.19101619720459,
      "logps/rejected": -18.939958572387695,
      "loss": 1.0462,
      "rewards/accuracies": 1.0,
      "rewards/chosen": -12.286526679992676,
      "rewards/margins": 16.123409271240234,
      "rewards/rejected": -28.409936904907227,
      "step": 820
    },
    {
      "epoch": 0.6811653672548215,
      "grad_norm": 0.910886824131012,
      "learning_rate": 2.0830781332097446e-06,
      "logits/chosen": -2.4748313426971436,
      "logits/rejected": -2.702345371246338,
      "logps/chosen": -9.464627265930176,
      "logps/rejected": -18.847036361694336,
      "loss": 1.0372,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -14.196942329406738,
      "rewards/margins": 14.07361125946045,
      "rewards/rejected": -28.270553588867188,
      "step": 830
    },
    {
      "epoch": 0.689372178908494,
      "grad_norm": 1.8233784437179565,
      "learning_rate": 2.031546713535688e-06,
      "logits/chosen": -2.460084915161133,
      "logits/rejected": -2.678401470184326,
      "logps/chosen": -9.879974365234375,
      "logps/rejected": -18.783390045166016,
      "loss": 0.9703,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -14.819961547851562,
      "rewards/margins": 13.355127334594727,
      "rewards/rejected": -28.17508888244629,
      "step": 840
    },
    {
      "epoch": 0.6975789905621665,
      "grad_norm": 3.4594898223876953,
      "learning_rate": 1.9802207729556023e-06,
      "logits/chosen": -2.3830952644348145,
      "logits/rejected": -2.6068103313446045,
      "logps/chosen": -9.286922454833984,
      "logps/rejected": -18.89388656616211,
      "loss": 1.0259,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -13.930383682250977,
      "rewards/margins": 14.410444259643555,
      "rewards/rejected": -28.340829849243164,
      "step": 850
    },
    {
      "epoch": 0.6975789905621665,
      "eval_logits/chosen": -2.3181838989257812,
      "eval_logits/rejected": -2.6507365703582764,
      "eval_logps/chosen": -7.670541286468506,
      "eval_logps/rejected": -18.20172119140625,
      "eval_loss": 0.12306927889585495,
      "eval_rewards/accuracies": 0.9696969985961914,
      "eval_rewards/chosen": -11.505812644958496,
      "eval_rewards/margins": 15.796772956848145,
      "eval_rewards/rejected": -27.302583694458008,
      "eval_runtime": 26.0288,
      "eval_samples_per_second": 30.274,
      "eval_steps_per_second": 3.803,
      "step": 850
    },
    {
      "epoch": 0.7057858022158392,
      "grad_norm": 11.1419038772583,
      "learning_rate": 1.9291228247233607e-06,
      "logits/chosen": -2.438063859939575,
      "logits/rejected": -2.653733730316162,
      "logps/chosen": -9.637109756469727,
      "logps/rejected": -20.197423934936523,
      "loss": 0.781,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -14.455665588378906,
      "rewards/margins": 15.840471267700195,
      "rewards/rejected": -30.296138763427734,
      "step": 860
    },
    {
      "epoch": 0.7139926138695117,
      "grad_norm": 8.1745023727417,
      "learning_rate": 1.8782752820878636e-06,
      "logits/chosen": -2.292649269104004,
      "logits/rejected": -2.5739500522613525,
      "logps/chosen": -7.215020656585693,
      "logps/rejected": -18.043460845947266,
      "loss": 0.845,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -10.822530746459961,
      "rewards/margins": 16.242664337158203,
      "rewards/rejected": -27.065195083618164,
      "step": 870
    },
    {
      "epoch": 0.7221994255231843,
      "grad_norm": 49.25035858154297,
      "learning_rate": 1.827700448461836e-06,
      "logits/chosen": -2.3827719688415527,
      "logits/rejected": -2.6681952476501465,
      "logps/chosen": -8.750313758850098,
      "logps/rejected": -17.976327896118164,
      "loss": 0.8561,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -13.125471115112305,
      "rewards/margins": 13.839022636413574,
      "rewards/rejected": -26.964492797851562,
      "step": 880
    },
    {
      "epoch": 0.7304062371768568,
      "grad_norm": 5.7622504234313965,
      "learning_rate": 1.7774205076388207e-06,
      "logits/chosen": -2.3274879455566406,
      "logits/rejected": -2.6361851692199707,
      "logps/chosen": -6.857377052307129,
      "logps/rejected": -16.838680267333984,
      "loss": 0.8304,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -10.286066055297852,
      "rewards/margins": 14.971952438354492,
      "rewards/rejected": -25.258018493652344,
      "step": 890
    },
    {
      "epoch": 0.7386130488305294,
      "grad_norm": 8.557459831237793,
      "learning_rate": 1.7274575140626318e-06,
      "logits/chosen": -2.393751859664917,
      "logits/rejected": -2.641291379928589,
      "logps/chosen": -7.221099853515625,
      "logps/rejected": -17.681804656982422,
      "loss": 0.7794,
      "rewards/accuracies": 0.925000011920929,
      "rewards/chosen": -10.831649780273438,
      "rewards/margins": 15.691058158874512,
      "rewards/rejected": -26.522708892822266,
      "step": 900
    },
    {
      "epoch": 0.7386130488305294,
      "eval_logits/chosen": -2.3039753437042236,
      "eval_logits/rejected": -2.6469638347625732,
      "eval_logps/chosen": -6.395583152770996,
      "eval_logps/rejected": -16.935155868530273,
      "eval_loss": 0.11225885152816772,
      "eval_rewards/accuracies": 0.9797979593276978,
      "eval_rewards/chosen": -9.593375205993652,
      "eval_rewards/margins": 15.809358596801758,
      "eval_rewards/rejected": -25.402732849121094,
      "eval_runtime": 26.0286,
      "eval_samples_per_second": 30.274,
      "eval_steps_per_second": 3.804,
      "step": 900
    },
    {
      "epoch": 0.7468198604842019,
      "grad_norm": 1.0288090705871582,
      "learning_rate": 1.677833383153542e-06,
      "logits/chosen": -2.3569540977478027,
      "logits/rejected": -2.668043613433838,
      "logps/chosen": -8.015088081359863,
      "logps/rejected": -19.568172454833984,
      "loss": 0.8783,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -12.022631645202637,
      "rewards/margins": 17.32962989807129,
      "rewards/rejected": -29.35226058959961,
      "step": 910
    },
    {
      "epoch": 0.7550266721378744,
      "grad_norm": 30.93399429321289,
      "learning_rate": 1.6285698816954626e-06,
      "logits/chosen": -2.2761807441711426,
      "logits/rejected": -2.577350378036499,
      "logps/chosen": -7.287099361419678,
      "logps/rejected": -18.970417022705078,
      "loss": 0.88,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": -10.930647850036621,
      "rewards/margins": 17.524978637695312,
      "rewards/rejected": -28.45562744140625,
      "step": 920
    },
    {
      "epoch": 0.763233483791547,
      "grad_norm": 4.3290534019470215,
      "learning_rate": 1.5796886182883053e-06,
      "logits/chosen": -2.391355037689209,
      "logits/rejected": -2.6769471168518066,
      "logps/chosen": -7.963728427886963,
      "logps/rejected": -18.105741500854492,
      "loss": 0.8804,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -11.945592880249023,
      "rewards/margins": 15.213022232055664,
      "rewards/rejected": -27.158611297607422,
      "step": 930
    },
    {
      "epoch": 0.7714402954452195,
      "grad_norm": 26.385705947875977,
      "learning_rate": 1.5312110338697427e-06,
      "logits/chosen": -2.4121921062469482,
      "logits/rejected": -2.68375825881958,
      "logps/chosen": -9.157960891723633,
      "logps/rejected": -20.01449203491211,
      "loss": 0.9232,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -13.73694133758545,
      "rewards/margins": 16.28479766845703,
      "rewards/rejected": -30.021739959716797,
      "step": 940
    },
    {
      "epoch": 0.7796471070988921,
      "grad_norm": 3.784264087677002,
      "learning_rate": 1.4831583923105e-06,
      "logits/chosen": -2.466296672821045,
      "logits/rejected": -2.634974956512451,
      "logps/chosen": -9.769882202148438,
      "logps/rejected": -18.733013153076172,
      "loss": 0.6239,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -14.654825210571289,
      "rewards/margins": 13.444696426391602,
      "rewards/rejected": -28.09952163696289,
      "step": 950
    },
    {
      "epoch": 0.7796471070988921,
      "eval_logits/chosen": -2.3149495124816895,
      "eval_logits/rejected": -2.655733346939087,
      "eval_logps/chosen": -7.496860027313232,
      "eval_logps/rejected": -18.8516788482666,
      "eval_loss": 0.09784827381372452,
      "eval_rewards/accuracies": 0.9696969985961914,
      "eval_rewards/chosen": -11.24528980255127,
      "eval_rewards/margins": 17.032228469848633,
      "eval_rewards/rejected": -28.27751922607422,
      "eval_runtime": 25.9982,
      "eval_samples_per_second": 30.31,
      "eval_steps_per_second": 3.808,
      "step": 950
    },
    {
      "epoch": 0.7878539187525646,
      "grad_norm": 6.634799003601074,
      "learning_rate": 1.4355517710873184e-06,
      "logits/chosen": -2.4064090251922607,
      "logits/rejected": -2.7013659477233887,
      "logps/chosen": -9.801457405090332,
      "logps/rejected": -20.174358367919922,
      "loss": 0.6526,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -14.702186584472656,
      "rewards/margins": 15.559350967407227,
      "rewards/rejected": -30.261539459228516,
      "step": 960
    },
    {
      "epoch": 0.7960607304062371,
      "grad_norm": 46.21657943725586,
      "learning_rate": 1.388412052037682e-06,
      "logits/chosen": -2.356576681137085,
      "logits/rejected": -2.6283116340637207,
      "logps/chosen": -9.053842544555664,
      "logps/rejected": -21.279687881469727,
      "loss": 0.9765,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -13.580764770507812,
      "rewards/margins": 18.33877182006836,
      "rewards/rejected": -31.91953468322754,
      "step": 970
    },
    {
      "epoch": 0.8042675420599097,
      "grad_norm": 3.6938939094543457,
      "learning_rate": 1.3417599122003464e-06,
      "logits/chosen": -2.4083611965179443,
      "logits/rejected": -2.6813862323760986,
      "logps/chosen": -9.064926147460938,
      "logps/rejected": -20.76726531982422,
      "loss": 0.8067,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -13.597389221191406,
      "rewards/margins": 17.553510665893555,
      "rewards/rejected": -31.15089988708496,
      "step": 980
    },
    {
      "epoch": 0.8124743537135822,
      "grad_norm": 9.791804313659668,
      "learning_rate": 1.2956158147457116e-06,
      "logits/chosen": -2.3875112533569336,
      "logits/rejected": -2.6877217292785645,
      "logps/chosen": -9.713172912597656,
      "logps/rejected": -20.51764488220215,
      "loss": 0.6454,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -14.5697603225708,
      "rewards/margins": 16.206707000732422,
      "rewards/rejected": -30.776464462280273,
      "step": 990
    },
    {
      "epoch": 0.8206811653672548,
      "grad_norm": 1.188546061515808,
      "learning_rate": 1.2500000000000007e-06,
      "logits/chosen": -2.400709629058838,
      "logits/rejected": -2.613481283187866,
      "logps/chosen": -9.234119415283203,
      "logps/rejected": -20.3234920501709,
      "loss": 0.7065,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -13.851181030273438,
      "rewards/margins": 16.63405990600586,
      "rewards/rejected": -30.485240936279297,
      "step": 1000
    },
    {
      "epoch": 0.8206811653672548,
      "eval_logits/chosen": -2.3037211894989014,
      "eval_logits/rejected": -2.6389148235321045,
      "eval_logps/chosen": -7.639871120452881,
      "eval_logps/rejected": -18.97751808166504,
      "eval_loss": 0.09854340553283691,
      "eval_rewards/accuracies": 0.9595959782600403,
      "eval_rewards/chosen": -11.459808349609375,
      "eval_rewards/margins": 17.006471633911133,
      "eval_rewards/rejected": -28.466276168823242,
      "eval_runtime": 26.0416,
      "eval_samples_per_second": 30.259,
      "eval_steps_per_second": 3.802,
      "step": 1000
    },
    {
      "epoch": 0.8288879770209274,
      "grad_norm": 34.27748107910156,
      "learning_rate": 1.204932476567175e-06,
      "logits/chosen": -2.3972585201263428,
      "logits/rejected": -2.6077466011047363,
      "logps/chosen": -9.558837890625,
      "logps/rejected": -21.304996490478516,
      "loss": 0.9793,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -14.338258743286133,
      "rewards/margins": 17.61923599243164,
      "rewards/rejected": -31.957494735717773,
      "step": 1010
    },
    {
      "epoch": 0.8370947886746,
      "grad_norm": 6.985296249389648,
      "learning_rate": 1.160433012552508e-06,
      "logits/chosen": -2.4298081398010254,
      "logits/rejected": -2.6570680141448975,
      "logps/chosen": -9.846342086791992,
      "logps/rejected": -20.0854434967041,
      "loss": 0.6918,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -14.769511222839355,
      "rewards/margins": 15.35865306854248,
      "rewards/rejected": -30.1281681060791,
      "step": 1020
    },
    {
      "epoch": 0.8453016003282725,
      "grad_norm": 16.403400421142578,
      "learning_rate": 1.11652112689164e-06,
      "logits/chosen": -2.3441762924194336,
      "logits/rejected": -2.5944087505340576,
      "logps/chosen": -9.49109172821045,
      "logps/rejected": -21.233768463134766,
      "loss": 0.8113,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": -14.236638069152832,
      "rewards/margins": 17.614015579223633,
      "rewards/rejected": -31.850650787353516,
      "step": 1030
    },
    {
      "epoch": 0.853508411981945,
      "grad_norm": 4.5160040855407715,
      "learning_rate": 1.073216080788921e-06,
      "logits/chosen": -2.4048399925231934,
      "logits/rejected": -2.6450343132019043,
      "logps/chosen": -8.749728202819824,
      "logps/rejected": -20.77089500427246,
      "loss": 0.8586,
      "rewards/accuracies": 0.949999988079071,
      "rewards/chosen": -13.124593734741211,
      "rewards/margins": 18.031749725341797,
      "rewards/rejected": -31.15634536743164,
      "step": 1040
    },
    {
      "epoch": 0.8617152236356176,
      "grad_norm": 5.133055210113525,
      "learning_rate": 1.0305368692688175e-06,
      "logits/chosen": -2.36543869972229,
      "logits/rejected": -2.5925793647766113,
      "logps/chosen": -9.336064338684082,
      "logps/rejected": -20.378089904785156,
      "loss": 0.8972,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -14.004096984863281,
      "rewards/margins": 16.563039779663086,
      "rewards/rejected": -30.5671329498291,
      "step": 1050
    },
    {
      "epoch": 0.8617152236356176,
      "eval_logits/chosen": -2.3072853088378906,
      "eval_logits/rejected": -2.6326565742492676,
      "eval_logps/chosen": -8.39708137512207,
      "eval_logps/rejected": -20.57996940612793,
      "eval_loss": 0.09786142408847809,
      "eval_rewards/accuracies": 0.9595959782600403,
      "eval_rewards/chosen": -12.595619201660156,
      "eval_rewards/margins": 18.27433204650879,
      "eval_rewards/rejected": -30.869949340820312,
      "eval_runtime": 26.0409,
      "eval_samples_per_second": 30.26,
      "eval_steps_per_second": 3.802,
      "step": 1050
    },
    {
      "epoch": 0.8699220352892901,
      "grad_norm": 24.745431900024414,
      "learning_rate": 9.88502212844063e-07,
      "logits/chosen": -2.269465923309326,
      "logits/rejected": -2.610417127609253,
      "logps/chosen": -7.833917140960693,
      "logps/rejected": -21.22043228149414,
      "loss": 0.8923,
      "rewards/accuracies": 1.0,
      "rewards/chosen": -11.750875473022461,
      "rewards/margins": 20.079769134521484,
      "rewards/rejected": -31.830646514892578,
      "step": 1060
    },
    {
      "epoch": 0.8781288469429627,
      "grad_norm": 1.3538175821304321,
      "learning_rate": 9.471305493042243e-07,
      "logits/chosen": -2.3404107093811035,
      "logits/rejected": -2.6831154823303223,
      "logps/chosen": -9.409693717956543,
      "logps/rejected": -22.07307243347168,
      "loss": 0.8694,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -14.114542007446289,
      "rewards/margins": 18.99506950378418,
      "rewards/rejected": -33.10961151123047,
      "step": 1070
    },
    {
      "epoch": 0.8863356585966352,
      "grad_norm": 3.945587635040283,
      "learning_rate": 9.064400256282757e-07,
      "logits/chosen": -2.402027130126953,
      "logits/rejected": -2.6573710441589355,
      "logps/chosen": -9.867189407348633,
      "logps/rejected": -21.119680404663086,
      "loss": 1.0965,
      "rewards/accuracies": 0.987500011920929,
      "rewards/chosen": -14.800783157348633,
      "rewards/margins": 16.878738403320312,
      "rewards/rejected": -31.679523468017578,
      "step": 1080
    },
    {
      "epoch": 0.8945424702503078,
      "grad_norm": 5.244990348815918,
      "learning_rate": 8.664484900247363e-07,
      "logits/chosen": -2.3451342582702637,
      "logits/rejected": -2.563744306564331,
      "logps/chosen": -8.859700202941895,
      "logps/rejected": -19.88092613220215,
      "loss": 0.6535,
      "rewards/accuracies": 0.9624999761581421,
      "rewards/chosen": -13.28955078125,
      "rewards/margins": 16.53183937072754,
      "rewards/rejected": -29.82139015197754,
      "step": 1090
    },
    {
      "epoch": 0.9027492819039803,
      "grad_norm": 2.167773962020874,
      "learning_rate": 8.271734841028553e-07,
      "logits/chosen": -2.306011199951172,
      "logits/rejected": -2.5858731269836426,
      "logps/chosen": -8.683083534240723,
      "logps/rejected": -20.459806442260742,
      "loss": 0.7775,
      "rewards/accuracies": 0.9750000238418579,
      "rewards/chosen": -13.024624824523926,
      "rewards/margins": 17.66508674621582,
      "rewards/rejected": -30.689708709716797,
      "step": 1100
    },
    {
      "epoch": 0.9027492819039803,
      "eval_logits/chosen": -2.2880752086639404,
      "eval_logits/rejected": -2.615443468093872,
      "eval_logps/chosen": -7.504647731781006,
      "eval_logps/rejected": -19.788803100585938,
      "eval_loss": 0.09024464339017868,
      "eval_rewards/accuracies": 0.9696969985961914,
      "eval_rewards/chosen": -11.25697135925293,
      "eval_rewards/margins": 18.42623519897461,
      "eval_rewards/rejected": -29.68320655822754,
      "eval_runtime": 26.0393,
      "eval_samples_per_second": 30.262,
      "eval_steps_per_second": 3.802,
      "step": 1100
    }
  ],
  "logging_steps": 10,
  "max_steps": 1500,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 2,
  "save_steps": 50,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 3.9998944549776916e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}