{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.00592,
  "eval_steps": 500,
  "global_step": 37,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3875,
      "completions/max_length": 512.0,
      "completions/max_terminated_length": 333.2,
      "completions/mean_length": 232.9,
      "completions/mean_terminated_length": 161.20882263183594,
      "completions/min_length": 19.0,
      "completions/min_terminated_length": 19.0,
      "entropy": 0.6788227945566178,
      "epoch": 0.0008,
      "frac_reward_zero_std": 0.45,
      "grad_norm": 1.7250084051184943e-17,
      "kl": 0.0,
      "learning_rate": 8.529119999999999e-07,
      "loss": -3.329563695879303e-17,
      "num_tokens": 139158.0,
      "reward": 0.006625001598149538,
      "reward_std": 0.02386485398747027,
      "rewards/env_goofspiel_reward/mean": 0.006625001598149538,
      "rewards/env_goofspiel_reward/std": 0.060312052723020315,
      "sampling/importance_sampling_ratio/max": 1.5548742481149117e-15,
      "sampling/importance_sampling_ratio/mean": 5.537038872768731e-17,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 34.56931686401367,
      "sampling/sampling_logp_difference/mean": 0.48103601932525636,
      "step": 5,
      "step_time": 26.902466528200147
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3625,
      "completions/max_length": 512.0,
      "completions/max_terminated_length": 380.0,
      "completions/mean_length": 208.325,
      "completions/mean_terminated_length": 130.9656951904297,
      "completions/min_length": 15.6,
      "completions/min_terminated_length": 15.6,
      "entropy": 0.8194768905639649,
      "epoch": 0.0016,
      "frac_reward_zero_std": 0.3875,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 1.919052e-06,
      "loss": -5.5036242336365485e-21,
      "num_tokens": 272885.0,
      "reward": -0.001062498986721039,
      "reward_std": 0.015114407148212194,
      "rewards/env_goofspiel_reward/mean": -0.001062498986721039,
      "rewards/env_goofspiel_reward/std": 0.04447857290506363,
      "sampling/importance_sampling_ratio/max": 2.6002960348423844e-15,
      "sampling/importance_sampling_ratio/mean": 8.147796120515316e-17,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 33.71624984741211,
      "sampling/sampling_logp_difference/mean": 0.5505338490009308,
      "step": 10,
      "step_time": 25.23802887940001
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.375,
      "completions/max_length": 512.0,
      "completions/max_terminated_length": 365.2,
      "completions/mean_length": 213.05625,
      "completions/mean_terminated_length": 134.1561492919922,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 0.7374677076935768,
      "epoch": 0.0024,
      "frac_reward_zero_std": 0.3,
      "grad_norm": 6.240696039252212e-21,
      "kl": 0.0,
      "learning_rate": 2.985192e-06,
      "loss": -2.649282052416546e-13,
      "num_tokens": 407112.0,
      "reward": -0.0010624994523823262,
      "reward_std": 0.015467961691319943,
      "rewards/env_goofspiel_reward/mean": -0.0010624994523823262,
      "rewards/env_goofspiel_reward/std": 0.044856540579348805,
      "sampling/importance_sampling_ratio/max": 6.0797169833741606e-12,
      "sampling/importance_sampling_ratio/mean": 1.9001603404032331e-13,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 33.525834274291995,
      "sampling/sampling_logp_difference/mean": 0.5196602582931519,
      "step": 15,
      "step_time": 25.776172687000233
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.35625,
      "completions/max_length": 512.0,
      "completions/max_terminated_length": 430.0,
      "completions/mean_length": 222.38125,
      "completions/mean_terminated_length": 144.48629455566407,
      "completions/min_length": 16.0,
      "completions/min_terminated_length": 16.0,
      "entropy": 0.7416900217533111,
      "epoch": 0.0032,
      "frac_reward_zero_std": 0.3,
      "grad_norm": 0.0,
      "kl": 0.0,
      "learning_rate": 4.051332e-06,
      "loss": 2.170032172889042e-15,
      "num_tokens": 543515.0,
      "reward": 0.0008125004358589649,
      "reward_std": 0.015114407520741224,
      "rewards/env_goofspiel_reward/mean": 0.0008125004358589649,
      "rewards/env_goofspiel_reward/std": 0.04427960952743888,
      "sampling/importance_sampling_ratio/max": 9.7079462209307e-14,
      "sampling/importance_sampling_ratio/mean": 3.0337332194254157e-15,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 33.49689865112305,
      "sampling/sampling_logp_difference/mean": 0.5143256843090057,
      "step": 20,
      "step_time": 26.41383430980004
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.3875,
      "completions/max_length": 480.2,
      "completions/max_terminated_length": 350.8,
      "completions/mean_length": 206.175,
      "completions/mean_terminated_length": 116.13554382324219,
      "completions/min_length": 15.2,
      "completions/min_terminated_length": 15.2,
      "entropy": 0.6977409303188324,
      "epoch": 0.004,
      "frac_reward_zero_std": 0.2875,
      "grad_norm": 2.0909891318276304e-18,
      "kl": 0.0,
      "learning_rate": 5.117472e-06,
      "loss": -5.2856475859179964e-17,
      "num_tokens": 680288.0,
      "reward": 0.020250003039836883,
      "reward_std": 0.044370952248573306,
      "rewards/env_goofspiel_reward/mean": 0.020250003039836883,
      "rewards/env_goofspiel_reward/std": 0.13559252247214318,
      "sampling/importance_sampling_ratio/max": 2.52013723800786e-15,
      "sampling/importance_sampling_ratio/mean": 7.877870563254164e-17,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 33.450994873046874,
      "sampling/sampling_logp_difference/mean": 0.5328368723392487,
      "step": 25,
      "step_time": 24.940973892599867
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.45625,
      "completions/max_length": 512.0,
      "completions/max_terminated_length": 404.6,
      "completions/mean_length": 238.2375,
      "completions/mean_terminated_length": 141.82444915771484,
      "completions/min_length": 16.6,
      "completions/min_terminated_length": 16.6,
      "entropy": 0.6496311470866203,
      "epoch": 0.0048,
      "frac_reward_zero_std": 0.3,
      "grad_norm": 3.2634135614775815e-22,
      "kl": 0.0,
      "learning_rate": 6.183612e-06,
      "loss": 5.1048653439855114e-17,
      "num_tokens": 818011.0,
      "reward": 0.007000001426786184,
      "reward_std": 0.024925515986979008,
      "rewards/env_goofspiel_reward/mean": 0.007000001426786184,
      "rewards/env_goofspiel_reward/std": 0.05999897662550211,
      "sampling/importance_sampling_ratio/max": 4.958355600719433e-15,
      "sampling/importance_sampling_ratio/mean": 2.02652089726915e-16,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 33.400308227539064,
      "sampling/sampling_logp_difference/mean": 0.46320154070854186,
      "step": 30,
      "step_time": 27.004615230400123
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.34375,
      "completions/max_length": 512.0,
      "completions/max_terminated_length": 394.2,
      "completions/mean_length": 227.01875,
      "completions/mean_terminated_length": 154.02509613037108,
      "completions/min_length": 13.8,
      "completions/min_terminated_length": 13.8,
      "entropy": 0.7401682421565056,
      "epoch": 0.0056,
      "frac_reward_zero_std": 0.3625,
      "grad_norm": 2.2770879894661554e-15,
      "kl": 0.0,
      "learning_rate": 7.249752e-06,
      "loss": 6.233609379277093e-16,
      "num_tokens": 955959.0,
      "reward": -0.00068749925121665,
      "reward_std": 0.014937631692737341,
      "rewards/env_goofspiel_reward/mean": -0.00068749925121665,
      "rewards/env_goofspiel_reward/std": 0.04448435390368104,
      "sampling/importance_sampling_ratio/max": 2.827083637381056e-14,
      "sampling/importance_sampling_ratio/mean": 8.835648350382244e-16,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 33.62417449951172,
      "sampling/sampling_logp_difference/mean": 0.48951873779296873,
      "step": 35,
      "step_time": 26.58530942660018
    }
  ],
  "logging_steps": 5,
  "max_steps": 18750,
  "num_input_tokens_seen": 1009125,
  "num_train_epochs": 3,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}