{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 1.592,
  "eval_steps": 50,
  "global_step": 100,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "entropy": 2.555550271272659,
      "epoch": 0.08,
      "grad_norm": 175.83624267578125,
      "learning_rate": 2.8571428571428573e-06,
      "logits/chosen": 0.41239441768472307,
      "logits/rejected": 0.4313717963372231,
      "logps/chosen": -157.81461162567138,
      "logps/rejected": -150.4321650505066,
      "loss": 0.6844395637512207,
      "mean_token_accuracy": 0.45540447160601616,
      "num_tokens": 63280.0,
      "rewards/accuracies": 0.55,
      "rewards/chosen": 0.09316946864128113,
      "rewards/margins": 0.030635011196136475,
      "rewards/rejected": 0.06253445744514466,
      "step": 5
    },
    {
      "entropy": 2.584770825505257,
      "epoch": 0.16,
      "grad_norm": 275.8374938964844,
      "learning_rate": 4.984280524733107e-06,
      "logits/chosen": 0.23018057449154425,
      "logits/rejected": 0.4137123508420314,
      "logps/chosen": -160.37082138061524,
      "logps/rejected": -154.9328401565552,
      "loss": 0.848213005065918,
      "mean_token_accuracy": 0.4679958797991276,
      "num_tokens": 128751.0,
      "rewards/accuracies": 0.4625,
      "rewards/chosen": 0.06873167753219604,
      "rewards/margins": -0.08100449442863464,
      "rewards/rejected": 0.1497361719608307,
      "step": 10
    },
    {
      "entropy": 2.485997956991196,
      "epoch": 0.24,
      "grad_norm": 273.0233154296875,
      "learning_rate": 4.809698831278217e-06,
      "logits/chosen": 0.1816923364201413,
      "logits/rejected": 0.23945612540369207,
      "logps/chosen": -151.38499155044556,
      "logps/rejected": -163.4970989227295,
      "loss": 0.8470439910888672,
      "mean_token_accuracy": 0.4733552895486355,
      "num_tokens": 194142.0,
      "rewards/accuracies": 0.45,
      "rewards/chosen": -0.32968448400497435,
      "rewards/margins": 0.06030769646167755,
      "rewards/rejected": -0.3899921804666519,
      "step": 15
    },
    {
      "entropy": 2.5318914532661436,
      "epoch": 0.32,
      "grad_norm": 234.37681579589844,
      "learning_rate": 4.454578706170075e-06,
      "logits/chosen": 0.07677000825058375,
      "logits/rejected": 0.18887137972641857,
      "logps/chosen": -135.0070219039917,
      "logps/rejected": -135.28493957519532,
      "loss": 0.9311827659606934,
      "mean_token_accuracy": 0.4595716767013073,
      "num_tokens": 254871.0,
      "rewards/accuracies": 0.5375,
      "rewards/chosen": -0.7181475728750228,
      "rewards/margins": 0.2766463726758957,
      "rewards/rejected": -0.9947939455509186,
      "step": 20
    },
    {
      "entropy": 2.462447080016136,
      "epoch": 0.4,
      "grad_norm": 250.9570770263672,
      "learning_rate": 3.946678240449515e-06,
      "logits/chosen": 0.20993061968672994,
      "logits/rejected": 0.20386893310256043,
      "logps/chosen": -145.39961681365966,
      "logps/rejected": -137.4315857887268,
      "loss": 0.9580174446105957,
      "mean_token_accuracy": 0.45824490338563917,
      "num_tokens": 317752.0,
      "rewards/accuracies": 0.58125,
      "rewards/chosen": -0.7164888560771943,
      "rewards/margins": 0.16717674732208251,
      "rewards/rejected": -0.8836656033992767,
      "step": 25
    },
    {
      "entropy": 2.4387545943260194,
      "epoch": 0.48,
      "grad_norm": 316.4103088378906,
      "learning_rate": 3.3256976548879183e-06,
      "logits/chosen": 0.21743581064338796,
      "logits/rejected": 0.2765391890690152,
      "logps/chosen": -148.74287595748902,
      "logps/rejected": -147.26203107833862,
      "loss": 0.9910862922668457,
      "mean_token_accuracy": 0.4532994642853737,
      "num_tokens": 387452.0,
      "rewards/accuracies": 0.5125,
      "rewards/chosen": -0.2023364007472992,
      "rewards/margins": 0.1426894724369049,
      "rewards/rejected": -0.3450258731842041,
      "step": 30
    },
    {
      "entropy": 2.469076004624367,
      "epoch": 0.56,
      "grad_norm": 256.6429748535156,
      "learning_rate": 2.6401761180929798e-06,
      "logits/chosen": 0.18299585651420763,
      "logits/rejected": 0.250682400907656,
      "logps/chosen": -178.9021216392517,
      "logps/rejected": -171.8313166618347,
      "loss": 0.8882085800170898,
      "mean_token_accuracy": 0.47426797077059746,
      "num_tokens": 454873.0,
      "rewards/accuracies": 0.53125,
      "rewards/chosen": 0.1502027690410614,
      "rewards/margins": 0.38659809827804564,
      "rewards/rejected": -0.23639532923698425,
      "step": 35
    },
    {
      "entropy": 2.4749947816133497,
      "epoch": 0.64,
      "grad_norm": 248.1305694580078,
      "learning_rate": 1.9436976651092143e-06,
      "logits/chosen": 0.27816104129128527,
      "logits/rejected": 0.37676442462284854,
      "logps/chosen": -157.81686353683472,
      "logps/rejected": -158.98215370178224,
      "loss": 1.0146113395690919,
      "mean_token_accuracy": 0.465312571823597,
      "num_tokens": 518527.0,
      "rewards/accuracies": 0.55625,
      "rewards/chosen": 0.05624905824661255,
      "rewards/margins": 0.09228863418102265,
      "rewards/rejected": -0.036039575934410095,
      "step": 40
    },
    {
      "entropy": 2.409931382536888,
      "epoch": 0.72,
      "grad_norm": 219.939697265625,
      "learning_rate": 1.2907027822369006e-06,
      "logits/chosen": 0.2784307162908892,
      "logits/rejected": 0.3215469827251049,
      "logps/chosen": -153.8018159866333,
      "logps/rejected": -159.60495262145997,
      "loss": 0.921458625793457,
      "mean_token_accuracy": 0.4758016161620617,
      "num_tokens": 586496.0,
      "rewards/accuracies": 0.525,
      "rewards/chosen": 0.1898827075958252,
      "rewards/margins": 0.46674283146858214,
      "rewards/rejected": -0.27686012387275694,
      "step": 45
    },
    {
      "entropy": 2.4379068195819853,
      "epoch": 0.8,
      "grad_norm": 235.46241760253906,
      "learning_rate": 7.322330470336314e-07,
      "logits/chosen": 0.28584471002719647,
      "logits/rejected": 0.3651064125555795,
      "logps/chosen": -163.8307611465454,
      "logps/rejected": -163.46687126159668,
      "loss": 0.979340648651123,
      "mean_token_accuracy": 0.4653128065168858,
      "num_tokens": 659345.0,
      "rewards/accuracies": 0.5,
      "rewards/chosen": 0.16986804008483886,
      "rewards/margins": 0.1913771450519562,
      "rewards/rejected": -0.02150910496711731,
      "step": 50
    },
    {
      "epoch": 0.8,
      "eval_entropy": 2.5069966416358946,
      "eval_logits/chosen": 0.29315065189355716,
      "eval_logits/rejected": 0.348470281243937,
      "eval_logps/chosen": -161.13176106262208,
      "eval_logps/rejected": -153.30834275054931,
      "eval_loss": 0.9624814391136169,
      "eval_mean_token_accuracy": 0.4618878421783447,
      "eval_num_tokens": 659345.0,
      "eval_rewards/accuracies": 0.531,
      "eval_rewards/chosen": 0.1580237216949463,
      "eval_rewards/margins": 0.2344533715248108,
      "eval_rewards/rejected": -0.0764296498298645,
      "eval_runtime": 203.9137,
      "eval_samples_per_second": 4.904,
      "eval_steps_per_second": 1.226,
      "step": 50
    },
    {
      "entropy": 2.47237910926342,
      "epoch": 0.88,
      "grad_norm": 254.37533569335938,
      "learning_rate": 3.119414452281158e-07,
      "logits/chosen": 0.31621869752899623,
      "logits/rejected": 0.438719071421892,
      "logps/chosen": -148.24002447128296,
      "logps/rejected": -142.998854637146,
      "loss": 0.9957857131958008,
      "mean_token_accuracy": 0.47029968202114103,
      "num_tokens": 731475.0,
      "rewards/accuracies": 0.59375,
      "rewards/chosen": 0.2521739423274994,
      "rewards/margins": 0.1846893012523651,
      "rewards/rejected": 0.06748464107513427,
      "step": 55
    },
    {
      "entropy": 2.4306333363056183,
      "epoch": 0.96,
      "grad_norm": 226.28863525390625,
      "learning_rate": 6.268021954544095e-08,
      "logits/chosen": 0.27655188516120244,
      "logits/rejected": 0.40121284602477314,
      "logps/chosen": -156.78616828918456,
      "logps/rejected": -163.7671283721924,
      "loss": 0.8905390739440918,
      "mean_token_accuracy": 0.4723482772707939,
      "num_tokens": 797757.0,
      "rewards/accuracies": 0.5125,
      "rewards/chosen": 0.2607013940811157,
      "rewards/margins": 0.30302430391311647,
      "rewards/rejected": -0.04232290983200073,
      "step": 60
    },
    {
      "epoch": 1.0,
      "eval_entropy": 2.5108700938224793,
      "eval_logits/chosen": 0.2960853689074952,
      "eval_logits/rejected": 0.35102944673757674,
      "eval_logps/chosen": -161.2097156829834,
      "eval_logps/rejected": -153.38722127532958,
      "eval_loss": 0.9581051468849182,
      "eval_mean_token_accuracy": 0.4621651805639267,
      "eval_num_tokens": 829784.0,
      "eval_rewards/accuracies": 0.532,
      "eval_rewards/chosen": 0.11904637241363525,
      "eval_rewards/margins": 0.23491560649871826,
      "eval_rewards/rejected": -0.11586923408508301,
      "eval_runtime": 204.736,
      "eval_samples_per_second": 4.884,
      "eval_steps_per_second": 1.221,
      "step": 63
    },
    {
      "entropy": 2.4772405922412872,
      "epoch": 1.032,
      "grad_norm": 70.3385238647461,
      "learning_rate": 4.184239109116393e-06,
      "logits/chosen": 0.29999179148367466,
      "logits/rejected": 0.3366013686931765,
      "logps/chosen": -145.2724313735962,
      "logps/rejected": -168.38544821739197,
      "loss": 0.20300202071666718,
      "mean_token_accuracy": 0.4766918160021305,
      "num_tokens": 27180.0,
      "rewards/accuracies": 0.84375,
      "rewards/chosen": 3.185014322400093,
      "rewards/margins": 6.363163933157921,
      "rewards/rejected": -3.17814964056015,
      "step": 65
    },
    {
      "entropy": 2.5059203624725344,
      "epoch": 1.112,
      "grad_norm": 46.46247482299805,
      "learning_rate": 4.006586590948141e-06,
      "logits/chosen": 0.30248451255827935,
      "logits/rejected": 0.4369742989935694,
      "logps/chosen": -151.66757192611695,
      "logps/rejected": -152.82527809143068,
      "loss": 0.16016263961791993,
      "mean_token_accuracy": 0.48107142075896264,
      "num_tokens": 86611.0,
      "rewards/accuracies": 0.93125,
      "rewards/chosen": 3.108510234951973,
      "rewards/margins": 6.6764221549034115,
      "rewards/rejected": -3.5679119408130644,
      "step": 70
    },
    {
      "entropy": 2.5288314819335938,
      "epoch": 1.192,
      "grad_norm": 112.07616424560547,
      "learning_rate": 3.81608040719339e-06,
      "logits/chosen": 0.3381024272346584,
      "logits/rejected": 0.3919709001428062,
      "logps/chosen": -151.99142513275146,
      "logps/rejected": -158.68587474823,
      "loss": 0.1984729766845703,
      "mean_token_accuracy": 0.47099064812064173,
      "num_tokens": 153056.0,
      "rewards/accuracies": 0.88125,
      "rewards/chosen": 2.1314487934112547,
      "rewards/margins": 5.91470921933651,
      "rewards/rejected": -3.7832603842020034,
      "step": 75
    },
    {
      "entropy": 2.5066609025001525,
      "epoch": 1.272,
      "grad_norm": 47.97988510131836,
      "learning_rate": 3.6143458894413463e-06,
      "logits/chosen": 0.25372562736194504,
      "logits/rejected": 0.3085220085200435,
      "logps/chosen": -127.87930164337158,
      "logps/rejected": -152.39403762817383,
      "loss": 0.17381094694137572,
      "mean_token_accuracy": 0.47169380336999894,
      "num_tokens": 215920.0,
      "rewards/accuracies": 0.9375,
      "rewards/chosen": 2.709118887782097,
      "rewards/margins": 5.862504243850708,
      "rewards/rejected": -3.153385379910469,
      "step": 80
    },
    {
      "entropy": 2.42382150888443,
      "epoch": 1.3519999999999999,
      "grad_norm": 76.58686065673828,
      "learning_rate": 3.403104165467883e-06,
      "logits/chosen": 0.27046587254706894,
      "logits/rejected": 0.3554108506103396,
      "logps/chosen": -135.95888204574584,
      "logps/rejected": -156.9404288291931,
      "loss": 0.17821238040924073,
      "mean_token_accuracy": 0.49741546660661695,
      "num_tokens": 287127.0,
      "rewards/accuracies": 0.91875,
      "rewards/chosen": 3.4181615144014357,
      "rewards/margins": 7.317546558380127,
      "rewards/rejected": -3.8993850499391556,
      "step": 85
    },
    {
      "entropy": 2.5261152565479277,
      "epoch": 1.432,
      "grad_norm": 80.35875701904297,
      "learning_rate": 3.184157475180208e-06,
      "logits/chosen": 0.2524602257174569,
      "logits/rejected": 0.30296234191214516,
      "logps/chosen": -156.0525354385376,
      "logps/rejected": -149.48046531677247,
      "loss": 0.23036706447601318,
      "mean_token_accuracy": 0.46826066076755524,
      "num_tokens": 354068.0,
      "rewards/accuracies": 0.89375,
      "rewards/chosen": 3.04692747592926,
      "rewards/margins": 5.816668316721916,
      "rewards/rejected": -2.7697408348321915,
      "step": 90
    },
    {
      "entropy": 2.4211883395910263,
      "epoch": 1.512,
      "grad_norm": 53.557090759277344,
      "learning_rate": 2.9593737945414264e-06,
      "logits/chosen": 0.17738279076359578,
      "logits/rejected": 0.17275444698543332,
      "logps/chosen": -135.7138165473938,
      "logps/rejected": -157.41348686218262,
      "loss": 0.16570401191711426,
      "mean_token_accuracy": 0.4731420993804932,
      "num_tokens": 421426.0,
      "rewards/accuracies": 0.89375,
      "rewards/chosen": 2.8382038921117783,
      "rewards/margins": 6.569511443376541,
      "rewards/rejected": -3.731307566165924,
      "step": 95
    },
    {
      "entropy": 2.341481050848961,
      "epoch": 1.592,
      "grad_norm": 127.4275131225586,
      "learning_rate": 2.730670898658255e-06,
      "logits/chosen": -0.04766494180334789,
      "logits/rejected": 0.09169179261277503,
      "logps/chosen": -153.73596954345703,
      "logps/rejected": -152.93934383392335,
      "loss": 0.1649105429649353,
      "mean_token_accuracy": 0.4882186971604824,
      "num_tokens": 489806.0,
      "rewards/accuracies": 0.9125,
      "rewards/chosen": 3.46675723195076,
      "rewards/margins": 6.959583753347397,
      "rewards/rejected": -3.4928264021873474,
      "step": 100
    },
    {
      "epoch": 1.592,
      "eval_entropy": 2.3993507494926454,
      "eval_logits/chosen": -0.03254038471881061,
      "eval_logits/rejected": 0.02311345798919562,
      "eval_logps/chosen": -161.87789566040038,
      "eval_logps/rejected": -154.16960359954834,
      "eval_loss": 1.181304931640625,
      "eval_mean_token_accuracy": 0.4609604501724243,
      "eval_num_tokens": 489806.0,
      "eval_rewards/accuracies": 0.516,
      "eval_rewards/chosen": -0.21504417657852173,
      "eval_rewards/margins": 0.29201573705673217,
      "eval_rewards/rejected": -0.5070599136352539,
      "eval_runtime": 206.6722,
      "eval_samples_per_second": 4.839,
      "eval_steps_per_second": 1.21,
      "step": 100
    }
  ],
  "logging_steps": 5,
  "max_steps": 189,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 3,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 5237912290590720.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}