{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1250231410058624,
  "eval_steps": 1013,
  "global_step": 2026,
  "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": 0.005916666820645332,
      "completions/max_length": 1048.664,
      "completions/max_terminated_length": 887.636,
      "completions/mean_length": 740.2631862792969,
      "completions/mean_terminated_length": 737.0485073242188,
      "completions/min_length": 516.448,
      "completions/min_terminated_length": 516.448,
      "entropy": 0.7370091029206912,
      "epoch": 0.015427337241592102,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.82421875,
      "learning_rate": 1e-06,
      "loss": -0.0191,
      "num_tokens": 55870796.0,
      "reward": -0.6514818173646927,
      "reward_std": 0.7432788764238357,
      "rewards/ADERawReward/mean": -6.492442960739136,
      "rewards/ADERawReward/std": 5.876542773246765,
      "rewards/StrictFormatReward/mean": 0.5888750042319297,
      "rewards/StrictFormatReward/std": 1.9825107951164245,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.061124999895691874,
      "rewards/TrajectoryFaithfulnessReward/std": 0.23321482253074646,
      "sampling/importance_sampling_ratio/max": 2.6557672066688536,
      "sampling/importance_sampling_ratio/mean": 0.4120183044672012,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 4.071227742671967,
      "sampling/sampling_logp_difference/mean": 0.02530779618769884,
      "step": 250,
      "step_time": 102.83658894309168
    },
    {
      "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": 0.005083333484828472,
      "completions/max_length": 1042.992,
      "completions/max_terminated_length": 883.088,
      "completions/mean_length": 735.9094775390626,
      "completions/mean_terminated_length": 733.132591796875,
      "completions/min_length": 557.044,
      "completions/min_terminated_length": 557.044,
      "entropy": 0.7054843534032503,
      "epoch": 0.030854674483184203,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.431640625,
      "learning_rate": 1e-06,
      "loss": -0.0206,
      "num_tokens": 111634735.0,
      "reward": -0.5281950639486312,
      "reward_std": 0.5776918355226517,
      "rewards/ADERawReward/mean": -5.5796587896347045,
      "rewards/ADERawReward/std": 4.435033079147339,
      "rewards/StrictFormatReward/mean": 0.7227083352506161,
      "rewards/StrictFormatReward/std": 1.5988679203987122,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.042500000093132255,
      "rewards/TrajectoryFaithfulnessReward/std": 0.1935399295091629,
      "sampling/importance_sampling_ratio/max": 2.6669081573486326,
      "sampling/importance_sampling_ratio/mean": 0.424478932261467,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 3.480512677669525,
      "sampling/sampling_logp_difference/mean": 0.024822900630533697,
      "step": 500,
      "step_time": 102.76567207844415
    },
    {
      "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": 0.005833333492279053,
      "completions/max_length": 1037.572,
      "completions/max_terminated_length": 866.908,
      "completions/mean_length": 734.3183942871094,
      "completions/mean_terminated_length": 731.1217890625,
      "completions/min_length": 581.092,
      "completions/min_terminated_length": 581.092,
      "entropy": 0.6821570705374082,
      "epoch": 0.046282011724776306,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.3046875,
      "learning_rate": 1e-06,
      "loss": -0.0167,
      "num_tokens": 167358632.0,
      "reward": -0.46010204815864564,
      "reward_std": 0.5086695972681046,
      "rewards/ADERawReward/mean": -5.064103653907776,
      "rewards/ADERawReward/std": 3.8888480224609374,
      "rewards/StrictFormatReward/mean": 0.7910000030398369,
      "rewards/StrictFormatReward/std": 1.3112610425949096,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.032791666954755785,
      "rewards/TrajectoryFaithfulnessReward/std": 0.16757118993997575,
      "sampling/importance_sampling_ratio/max": 2.6628545818328857,
      "sampling/importance_sampling_ratio/mean": 0.416588748216629,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 3.5062717790603637,
      "sampling/sampling_logp_difference/mean": 0.02463912143558264,
      "step": 750,
      "step_time": 103.00564296311047
    },
    {
      "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": 0.005541666816920042,
      "completions/max_length": 1038.12,
      "completions/max_terminated_length": 863.428,
      "completions/mean_length": 730.9473972167968,
      "completions/mean_terminated_length": 727.8880185546875,
      "completions/min_length": 577.432,
      "completions/min_terminated_length": 577.432,
      "entropy": 0.6596990981002648,
      "epoch": 0.061709348966368406,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.419921875,
      "learning_rate": 1e-06,
      "loss": -0.0099,
      "num_tokens": 223006825.0,
      "reward": -0.42084186458587647,
      "reward_std": 0.46201441729068754,
      "rewards/ADERawReward/mean": -4.784501824378967,
      "rewards/ADERawReward/std": 3.564310556411743,
      "rewards/StrictFormatReward/mean": 0.8240000032186509,
      "rewards/StrictFormatReward/std": 1.1784188017845154,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.024791666973382233,
      "rewards/TrajectoryFaithfulnessReward/std": 0.1419000388979912,
      "sampling/importance_sampling_ratio/max": 2.6621846599578856,
      "sampling/importance_sampling_ratio/mean": 0.41088387966156004,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 3.7157851243019104,
      "sampling/sampling_logp_difference/mean": 0.024803065180778502,
      "step": 1000,
      "step_time": 102.76144676574134
    },
    {
      "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": 0.004541666798293591,
      "completions/max_length": 1014.732,
      "completions/max_terminated_length": 859.056,
      "completions/mean_length": 727.5094375,
      "completions/mean_terminated_length": 724.9905017089844,
      "completions/min_length": 584.332,
      "completions/min_terminated_length": 584.332,
      "entropy": 0.6467662775715192,
      "epoch": 0.0771366862079605,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.66796875,
      "learning_rate": 1e-06,
      "loss": -0.0215,
      "num_tokens": 278568875.0,
      "reward": -0.3954123160839081,
      "reward_std": 0.4379507850408554,
      "rewards/ADERawReward/mean": -4.579247982025146,
      "rewards/ADERawReward/std": 3.392575692653656,
      "rewards/StrictFormatReward/mean": 0.836375002503395,
      "rewards/StrictFormatReward/std": 1.1209392004013061,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.021125000294297935,
      "rewards/TrajectoryFaithfulnessReward/std": 0.12883992713689804,
      "sampling/importance_sampling_ratio/max": 2.699490216732025,
      "sampling/importance_sampling_ratio/mean": 0.41803274899721143,
      "sampling/importance_sampling_ratio/min": 1.6055192099884153e-06,
      "sampling/sampling_logp_difference/max": 4.016801550865173,
      "sampling/sampling_logp_difference/mean": 0.024675262413918972,
      "step": 1250,
      "step_time": 102.27029848615639
    },
    {
      "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": 0.002500000074505806,
      "completions/max_length": 945.28,
      "completions/max_terminated_length": 852.88,
      "completions/mean_length": 725.0985200195313,
      "completions/mean_terminated_length": 723.7116264648438,
      "completions/min_length": 596.484,
      "completions/min_terminated_length": 596.484,
      "entropy": 0.6350844318767389,
      "epoch": 0.09256402344955261,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.419921875,
      "learning_rate": 1e-06,
      "loss": -0.0189,
      "num_tokens": 334072887.0,
      "reward": -0.36599990248680114,
      "reward_std": 0.38822157627344134,
      "rewards/ADERawReward/mean": -4.372457194328308,
      "rewards/ADERawReward/std": 3.109687662124634,
      "rewards/StrictFormatReward/mean": 0.8803750035762787,
      "rewards/StrictFormatReward/std": 0.911559506893158,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.016791666973382236,
      "rewards/TrajectoryFaithfulnessReward/std": 0.11165152180194855,
      "sampling/importance_sampling_ratio/max": 2.6761970496177674,
      "sampling/importance_sampling_ratio/mean": 0.40891835248470304,
      "sampling/importance_sampling_ratio/min": 9.472505189478397e-06,
      "sampling/sampling_logp_difference/max": 3.42186851978302,
      "sampling/sampling_logp_difference/mean": 0.02459824702888727,
      "step": 1500,
      "step_time": 102.27085299471486
    },
    {
      "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": 0.0021666667275130748,
      "completions/max_length": 925.572,
      "completions/max_terminated_length": 840.94,
      "completions/mean_length": 722.0470627441406,
      "completions/mean_terminated_length": 720.8377180175781,
      "completions/min_length": 611.58,
      "completions/min_terminated_length": 611.58,
      "entropy": 0.6213784945507844,
      "epoch": 0.1079913606911447,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.58203125,
      "learning_rate": 1e-06,
      "loss": -0.012,
      "num_tokens": 389499072.0,
      "reward": -0.3459859914779663,
      "reward_std": 0.36564248079061507,
      "rewards/ADERawReward/mean": -4.219901427745819,
      "rewards/ADERawReward/std": 3.0195640125274656,
      "rewards/StrictFormatReward/mean": 0.8950416705608368,
      "rewards/StrictFormatReward/std": 0.8084545879364013,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.013500000279396773,
      "rewards/TrajectoryFaithfulnessReward/std": 0.09319193106889724,
      "sampling/importance_sampling_ratio/max": 2.6610923824310304,
      "sampling/importance_sampling_ratio/mean": 0.40683302807807925,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 3.4339644517898558,
      "sampling/sampling_logp_difference/mean": 0.02464620542526245,
      "step": 1750,
      "step_time": 99.183495038067
    },
    {
      "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": 0.0019166667200624943,
      "completions/max_length": 909.052,
      "completions/max_terminated_length": 835.544,
      "completions/mean_length": 720.742646484375,
      "completions/mean_terminated_length": 719.6681306152344,
      "completions/min_length": 584.152,
      "completions/min_terminated_length": 584.152,
      "entropy": 0.6136423728168011,
      "epoch": 0.12341869793273681,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.3359375,
      "learning_rate": 1e-06,
      "loss": -0.0141,
      "num_tokens": 444900063.0,
      "reward": -0.34251336628198625,
      "reward_std": 0.37447755241394043,
      "rewards/ADERawReward/mean": -4.1890502076148985,
      "rewards/ADERawReward/std": 3.077660232067108,
      "rewards/StrictFormatReward/mean": 0.9055833375453949,
      "rewards/StrictFormatReward/std": 0.7454135446548462,
      "rewards/TrajectoryFaithfulnessReward/mean": -0.014166666973382234,
      "rewards/TrajectoryFaithfulnessReward/std": 0.09863431739807128,
      "sampling/importance_sampling_ratio/max": 2.6561739139556884,
      "sampling/importance_sampling_ratio/mean": 0.4042038552761078,
      "sampling/importance_sampling_ratio/min": 0.0,
      "sampling/sampling_logp_difference/max": 3.3377611017227173,
      "sampling/sampling_logp_difference/mean": 0.02476299414038658,
      "step": 2000,
      "step_time": 98.3213485963922
    }
  ],
  "logging_steps": 250,
  "max_steps": 2026,
  "num_input_tokens_seen": 450655945,
  "num_train_epochs": 1,
  "save_steps": 1013,
  "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": 8,
  "trial_name": null,
  "trial_params": null
}