{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.6230098297106466,
  "eval_steps": 1000,
  "global_step": 9000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.06922331441229405,
      "grad_norm": 52.84375,
      "learning_rate": 2.9483376313193374e-05,
      "logits/chosen": -0.7172003388404846,
      "logits/rejected": -0.8999207019805908,
      "logps/chosen": -304.1668701171875,
      "logps/rejected": -216.46609497070312,
      "loss": 0.6793,
      "rewards/accuracies": 0.5619999766349792,
      "rewards/chosen": 0.10503997653722763,
      "rewards/margins": 0.037437956780195236,
      "rewards/rejected": 0.06760203093290329,
      "step": 1000
    },
    {
      "epoch": 0.06922331441229405,
      "eval_logits/chosen": -0.6456733345985413,
      "eval_logits/rejected": -0.8373212218284607,
      "eval_logps/chosen": -304.64984130859375,
      "eval_logps/rejected": -217.35638427734375,
      "eval_loss": 0.6695967316627502,
      "eval_rewards/accuracies": 0.5669527053833008,
      "eval_rewards/chosen": 0.20719687640666962,
      "eval_rewards/margins": 0.07671476900577545,
      "eval_rewards/rejected": 0.13048213720321655,
      "eval_runtime": 1809.5718,
      "eval_samples_per_second": 3.992,
      "eval_steps_per_second": 0.998,
      "step": 1000
    },
    {
      "epoch": 0.1384466288245881,
      "grad_norm": 20.1875,
      "learning_rate": 2.8964156527458075e-05,
      "logits/chosen": -0.6381000280380249,
      "logits/rejected": -0.8281131386756897,
      "logps/chosen": -298.145263671875,
      "logps/rejected": -213.81825256347656,
      "loss": 0.6633,
      "rewards/accuracies": 0.5932499766349792,
      "rewards/chosen": 0.25646814703941345,
      "rewards/margins": 0.09272526204586029,
      "rewards/rejected": 0.16374288499355316,
      "step": 2000
    },
    {
      "epoch": 0.1384466288245881,
      "eval_logits/chosen": -0.5784724354743958,
      "eval_logits/rejected": -0.774167001247406,
      "eval_logps/chosen": -304.0080871582031,
      "eval_logps/rejected": -217.06524658203125,
      "eval_loss": 0.6561135649681091,
      "eval_rewards/accuracies": 0.5975452661514282,
      "eval_rewards/chosen": 0.27137282490730286,
      "eval_rewards/margins": 0.11177627742290497,
      "eval_rewards/rejected": 0.1595965474843979,
      "eval_runtime": 1809.9391,
      "eval_samples_per_second": 3.991,
      "eval_steps_per_second": 0.998,
      "step": 2000
    },
    {
      "epoch": 0.20766994323688218,
      "grad_norm": 45.59375,
      "learning_rate": 2.8444936741722772e-05,
      "logits/chosen": -0.6119475960731506,
      "logits/rejected": -0.8187024593353271,
      "logps/chosen": -306.91265869140625,
      "logps/rejected": -214.48626708984375,
      "loss": 0.6522,
      "rewards/accuracies": 0.6052500009536743,
      "rewards/chosen": 0.26545238494873047,
      "rewards/margins": 0.13211578130722046,
      "rewards/rejected": 0.13333660364151,
      "step": 3000
    },
    {
      "epoch": 0.20766994323688218,
      "eval_logits/chosen": -0.5946438908576965,
      "eval_logits/rejected": -0.7924441695213318,
      "eval_logps/chosen": -303.99310302734375,
      "eval_logps/rejected": -217.3607177734375,
      "eval_loss": 0.6483559608459473,
      "eval_rewards/accuracies": 0.6118955016136169,
      "eval_rewards/chosen": 0.27287325263023376,
      "eval_rewards/margins": 0.14282457530498505,
      "eval_rewards/rejected": 0.13004866242408752,
      "eval_runtime": 1808.1656,
      "eval_samples_per_second": 3.995,
      "eval_steps_per_second": 0.999,
      "step": 3000
    },
    {
      "epoch": 0.2768932576491762,
      "grad_norm": 15.4453125,
      "learning_rate": 2.7925716955987473e-05,
      "logits/chosen": -0.6306295394897461,
      "logits/rejected": -0.8517826795578003,
      "logps/chosen": -307.3712463378906,
      "logps/rejected": -212.0769500732422,
      "loss": 0.6386,
      "rewards/accuracies": 0.6212499737739563,
      "rewards/chosen": 0.32602864503860474,
      "rewards/margins": 0.17370422184467316,
      "rewards/rejected": 0.15232445299625397,
      "step": 4000
    },
    {
      "epoch": 0.2768932576491762,
      "eval_logits/chosen": -0.6721797585487366,
      "eval_logits/rejected": -0.8792731165885925,
      "eval_logps/chosen": -303.22210693359375,
      "eval_logps/rejected": -216.90435791015625,
      "eval_loss": 0.6395601630210876,
      "eval_rewards/accuracies": 0.6196474432945251,
      "eval_rewards/chosen": 0.3499698042869568,
      "eval_rewards/margins": 0.17428545653820038,
      "eval_rewards/rejected": 0.1756843477487564,
      "eval_runtime": 1809.8644,
      "eval_samples_per_second": 3.991,
      "eval_steps_per_second": 0.998,
      "step": 4000
    },
    {
      "epoch": 0.3461165720614703,
      "grad_norm": 57.375,
      "learning_rate": 2.7406497170252167e-05,
      "logits/chosen": -0.70070880651474,
      "logits/rejected": -0.9215654730796814,
      "logps/chosen": -302.41265869140625,
      "logps/rejected": -212.0404052734375,
      "loss": 0.6371,
      "rewards/accuracies": 0.625249981880188,
      "rewards/chosen": 0.38654881715774536,
      "rewards/margins": 0.1891336739063263,
      "rewards/rejected": 0.19741514325141907,
      "step": 5000
    },
    {
      "epoch": 0.3461165720614703,
      "eval_logits/chosen": -0.6983235478401184,
      "eval_logits/rejected": -0.9088946580886841,
      "eval_logps/chosen": -302.4079895019531,
      "eval_logps/rejected": -216.43348693847656,
      "eval_loss": 0.6330362558364868,
      "eval_rewards/accuracies": 0.6294757723808289,
      "eval_rewards/chosen": 0.43138742446899414,
      "eval_rewards/margins": 0.20861481130123138,
      "eval_rewards/rejected": 0.22277262806892395,
      "eval_runtime": 1811.3565,
      "eval_samples_per_second": 3.988,
      "eval_steps_per_second": 0.997,
      "step": 5000
    },
    {
      "epoch": 0.41533988647376435,
      "grad_norm": 68.1875,
      "learning_rate": 2.6887277384516867e-05,
      "logits/chosen": -0.7243251800537109,
      "logits/rejected": -0.9372639656066895,
      "logps/chosen": -298.91192626953125,
      "logps/rejected": -214.60647583007812,
      "loss": 0.6348,
      "rewards/accuracies": 0.6307500004768372,
      "rewards/chosen": 0.4019148051738739,
      "rewards/margins": 0.20437340438365936,
      "rewards/rejected": 0.19754138588905334,
      "step": 6000
    },
    {
      "epoch": 0.41533988647376435,
      "eval_logits/chosen": -0.7277015447616577,
      "eval_logits/rejected": -0.9341850280761719,
      "eval_logps/chosen": -302.8192138671875,
      "eval_logps/rejected": -216.93202209472656,
      "eval_loss": 0.626986563205719,
      "eval_rewards/accuracies": 0.6369508504867554,
      "eval_rewards/chosen": 0.3902588188648224,
      "eval_rewards/margins": 0.21733936667442322,
      "eval_rewards/rejected": 0.17291945219039917,
      "eval_runtime": 1811.3554,
      "eval_samples_per_second": 3.988,
      "eval_steps_per_second": 0.997,
      "step": 6000
    },
    {
      "epoch": 0.4845632008860584,
      "grad_norm": 42.96875,
      "learning_rate": 2.6368057598781565e-05,
      "logits/chosen": -0.7278515100479126,
      "logits/rejected": -0.9392414093017578,
      "logps/chosen": -299.16754150390625,
      "logps/rejected": -211.17393493652344,
      "loss": 0.6228,
      "rewards/accuracies": 0.6457499861717224,
      "rewards/chosen": 0.4171576201915741,
      "rewards/margins": 0.23923394083976746,
      "rewards/rejected": 0.17792373895645142,
      "step": 7000
    },
    {
      "epoch": 0.4845632008860584,
      "eval_logits/chosen": -0.6790074706077576,
      "eval_logits/rejected": -0.8909901976585388,
      "eval_logps/chosen": -302.5709533691406,
      "eval_logps/rejected": -216.92449951171875,
      "eval_loss": 0.621570348739624,
      "eval_rewards/accuracies": 0.6430416703224182,
      "eval_rewards/chosen": 0.41509076952934265,
      "eval_rewards/margins": 0.24142169952392578,
      "eval_rewards/rejected": 0.17366909980773926,
      "eval_runtime": 1809.1079,
      "eval_samples_per_second": 3.993,
      "eval_steps_per_second": 0.998,
      "step": 7000
    },
    {
      "epoch": 0.5537865152983524,
      "grad_norm": 68.9375,
      "learning_rate": 2.5848837813046265e-05,
      "logits/chosen": -0.6841186285018921,
      "logits/rejected": -0.8984114527702332,
      "logps/chosen": -307.65289306640625,
      "logps/rejected": -220.9628448486328,
      "loss": 0.6251,
      "rewards/accuracies": 0.6430000066757202,
      "rewards/chosen": 0.38440626859664917,
      "rewards/margins": 0.23852017521858215,
      "rewards/rejected": 0.1458861082792282,
      "step": 8000
    },
    {
      "epoch": 0.5537865152983524,
      "eval_logits/chosen": -0.6851416230201721,
      "eval_logits/rejected": -0.9019606709480286,
      "eval_logps/chosen": -302.9433898925781,
      "eval_logps/rejected": -217.45289611816406,
      "eval_loss": 0.6197263598442078,
      "eval_rewards/accuracies": 0.6425802707672119,
      "eval_rewards/chosen": 0.3778453469276428,
      "eval_rewards/margins": 0.25701332092285156,
      "eval_rewards/rejected": 0.12083201110363007,
      "eval_runtime": 1812.2911,
      "eval_samples_per_second": 3.986,
      "eval_steps_per_second": 0.997,
      "step": 8000
    },
    {
      "epoch": 0.6230098297106466,
      "grad_norm": 23.046875,
      "learning_rate": 2.532961802731096e-05,
      "logits/chosen": -0.6993722915649414,
      "logits/rejected": -0.9128779172897339,
      "logps/chosen": -300.31256103515625,
      "logps/rejected": -215.20162963867188,
      "loss": 0.615,
      "rewards/accuracies": 0.640250027179718,
      "rewards/chosen": 0.4151590168476105,
      "rewards/margins": 0.2731047570705414,
      "rewards/rejected": 0.1420542448759079,
      "step": 9000
    },
    {
      "epoch": 0.6230098297106466,
      "eval_logits/chosen": -0.709079384803772,
      "eval_logits/rejected": -0.9235769510269165,
      "eval_logps/chosen": -302.36663818359375,
      "eval_logps/rejected": -217.17535400390625,
      "eval_loss": 0.6112583875656128,
      "eval_rewards/accuracies": 0.651024341583252,
      "eval_rewards/chosen": 0.43552058935165405,
      "eval_rewards/margins": 0.2869325280189514,
      "eval_rewards/rejected": 0.14858807623386383,
      "eval_runtime": 1804.2977,
      "eval_samples_per_second": 4.003,
      "eval_steps_per_second": 1.001,
      "step": 9000
    }
  ],
  "logging_steps": 1000,
  "max_steps": 57784,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 4,
  "save_steps": 1000,
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}