{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.8076923076923077,
  "eval_steps": 500,
  "global_step": 252,
  "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": 0.0015625,
      "completions/max_length": 2199.8,
      "completions/max_terminated_length": 1639.9,
      "completions/mean_length": 412.37578125,
      "completions/mean_terminated_length": 406.5489105224609,
      "completions/min_length": 119.4,
      "completions/min_terminated_length": 119.4,
      "entropy": 0.19867916740477085,
      "epoch": 0.03205128205128205,
      "frac_reward_zero_std": 0.028125,
      "grad_norm": 0.002149249856337326,
      "learning_rate": 1.4285714285714286e-06,
      "loss": 0.0001,
      "num_tokens": 2250657.0,
      "reward": -0.275947505235672,
      "reward_std": 0.1295723870396614,
      "rewards/grpo_reward_function/mean": -0.2759475141763687,
      "rewards/grpo_reward_function/std": 0.424668425321579,
      "sampling/importance_sampling_ratio/max": 0.27481397837400434,
      "sampling/importance_sampling_ratio/mean": 0.010644285473972559,
      "sampling/importance_sampling_ratio/min": 7.943800001636386e-18,
      "sampling/sampling_logp_difference/max": 1.52898166179657,
      "sampling/sampling_logp_difference/mean": 0.04394495449960232,
      "step": 10,
      "step_time": 108.14200488887727
    },
    {
      "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": 0.00234375,
      "completions/max_length": 2391.5,
      "completions/max_terminated_length": 1909.9,
      "completions/mean_length": 404.9625,
      "completions/mean_terminated_length": 396.33917541503905,
      "completions/min_length": 97.9,
      "completions/min_terminated_length": 97.9,
      "entropy": 0.19242343455553054,
      "epoch": 0.0641025641025641,
      "frac_reward_zero_std": 0.059375,
      "grad_norm": 0.0011151177348417973,
      "learning_rate": 3.015873015873016e-06,
      "loss": -0.0008,
      "num_tokens": 4484137.0,
      "reward": -0.24521568268537522,
      "reward_std": 0.1464571863412857,
      "rewards/grpo_reward_function/mean": -0.24521567821502685,
      "rewards/grpo_reward_function/std": 0.4808669984340668,
      "sampling/importance_sampling_ratio/max": 0.33791280463337897,
      "sampling/importance_sampling_ratio/mean": 0.015166364191100002,
      "sampling/importance_sampling_ratio/min": 2.5371388141412726e-20,
      "sampling/sampling_logp_difference/max": 1.3974398732185365,
      "sampling/sampling_logp_difference/mean": 0.042889106273651126,
      "step": 20,
      "step_time": 109.47327837087214
    },
    {
      "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": 0.01015625,
      "completions/max_length": 3805.9,
      "completions/max_terminated_length": 2067.3,
      "completions/mean_length": 453.64921875,
      "completions/mean_terminated_length": 416.3313415527344,
      "completions/min_length": 110.3,
      "completions/min_terminated_length": 110.3,
      "entropy": 0.1974952444434166,
      "epoch": 0.09615384615384616,
      "frac_reward_zero_std": 0.034375,
      "grad_norm": 8.499528216137695e-05,
      "learning_rate": 4.603174603174604e-06,
      "loss": -0.0002,
      "num_tokens": 6759408.0,
      "reward": -0.2948721319437027,
      "reward_std": 0.15282833948731422,
      "rewards/grpo_reward_function/mean": -0.29487212747335434,
      "rewards/grpo_reward_function/std": 0.47507801949977874,
      "sampling/importance_sampling_ratio/max": 0.13380551273003222,
      "sampling/importance_sampling_ratio/mean": 0.005529391323216259,
      "sampling/importance_sampling_ratio/min": 9.182118521107166e-23,
      "sampling/sampling_logp_difference/max": 1.5520962238311768,
      "sampling/sampling_logp_difference/mean": 0.043160675838589665,
      "step": 30,
      "step_time": 116.19033445902168
    },
    {
      "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": 0.0015625,
      "completions/max_length": 2842.5,
      "completions/max_terminated_length": 2502.3,
      "completions/mean_length": 414.8203125,
      "completions/mean_terminated_length": 409.07032470703126,
      "completions/min_length": 104.5,
      "completions/min_terminated_length": 104.5,
      "entropy": 0.19993907809257508,
      "epoch": 0.1282051282051282,
      "frac_reward_zero_std": 0.040625,
      "grad_norm": 0.0006745797790728429,
      "learning_rate": 6.1904761904761914e-06,
      "loss": -0.0002,
      "num_tokens": 9021210.0,
      "reward": -0.3394616931676865,
      "reward_std": 0.18811093866825104,
      "rewards/grpo_reward_function/mean": -0.3394616901874542,
      "rewards/grpo_reward_function/std": 0.4581590205430984,
      "sampling/importance_sampling_ratio/max": 0.13625358305871488,
      "sampling/importance_sampling_ratio/mean": 0.006088873354019597,
      "sampling/importance_sampling_ratio/min": 7.88066005403204e-16,
      "sampling/sampling_logp_difference/max": 1.499702548980713,
      "sampling/sampling_logp_difference/mean": 0.04408838003873825,
      "step": 40,
      "step_time": 108.74849968068301
    },
    {
      "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": 0.00546875,
      "completions/max_length": 3194.0,
      "completions/max_terminated_length": 2308.3,
      "completions/mean_length": 425.12578125,
      "completions/mean_terminated_length": 404.89239196777345,
      "completions/min_length": 100.8,
      "completions/min_terminated_length": 100.8,
      "entropy": 0.1986594047397375,
      "epoch": 0.16025641025641027,
      "frac_reward_zero_std": 0.028125,
      "grad_norm": 0.0010311153051996479,
      "learning_rate": 7.77777777777778e-06,
      "loss": -0.0001,
      "num_tokens": 11280859.0,
      "reward": -0.2975210875272751,
      "reward_std": 0.1474170058965683,
      "rewards/grpo_reward_function/mean": -0.29752107709646225,
      "rewards/grpo_reward_function/std": 0.4570117324590683,
      "sampling/importance_sampling_ratio/max": 0.29515470042824743,
      "sampling/importance_sampling_ratio/mean": 0.0092799658421427,
      "sampling/importance_sampling_ratio/min": 5.108517233618296e-19,
      "sampling/sampling_logp_difference/max": 1.4333376169204712,
      "sampling/sampling_logp_difference/mean": 0.04367562681436539,
      "step": 50,
      "step_time": 128.00819000415504
    },
    {
      "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": 0.0046875,
      "completions/max_length": 3292.6,
      "completions/max_terminated_length": 2423.1,
      "completions/mean_length": 442.92421875,
      "completions/mean_terminated_length": 425.74745788574216,
      "completions/min_length": 114.5,
      "completions/min_terminated_length": 114.5,
      "entropy": 0.19451429434120654,
      "epoch": 0.19230769230769232,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.00034187396484116245,
      "learning_rate": 9.365079365079366e-06,
      "loss": 0.0005,
      "num_tokens": 13606514.0,
      "reward": -0.291646933555603,
      "reward_std": 0.18595448434352874,
      "rewards/grpo_reward_function/mean": -0.2916469365358353,
      "rewards/grpo_reward_function/std": 0.500500214099884,
      "sampling/importance_sampling_ratio/max": 0.3140833295881748,
      "sampling/importance_sampling_ratio/mean": 0.008379129139939323,
      "sampling/importance_sampling_ratio/min": 1.013470538854885e-16,
      "sampling/sampling_logp_difference/max": 1.369999098777771,
      "sampling/sampling_logp_difference/mean": 0.04265468046069145,
      "step": 60,
      "step_time": 124.51363873668015
    },
    {
      "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": 0.00625,
      "completions/max_length": 3630.2,
      "completions/max_terminated_length": 2578.8,
      "completions/mean_length": 446.13671875,
      "completions/mean_terminated_length": 423.1569488525391,
      "completions/min_length": 114.2,
      "completions/min_terminated_length": 114.2,
      "entropy": 0.186833830550313,
      "epoch": 0.22435897435897437,
      "frac_reward_zero_std": 0.05625,
      "grad_norm": 0.0008922675305969227,
      "learning_rate": 9.985680226398261e-06,
      "loss": 0.0005,
      "num_tokens": 15864709.0,
      "reward": -0.23017083778977393,
      "reward_std": 0.17412283048033714,
      "rewards/grpo_reward_function/mean": -0.23017083778977393,
      "rewards/grpo_reward_function/std": 0.5827802836894989,
      "sampling/importance_sampling_ratio/max": 0.3794835552573204,
      "sampling/importance_sampling_ratio/mean": 0.01778233496006578,
      "sampling/importance_sampling_ratio/min": 3.8213540481902065e-23,
      "sampling/sampling_logp_difference/max": 1.4418152451515198,
      "sampling/sampling_logp_difference/mean": 0.04067011401057243,
      "step": 70,
      "step_time": 133.25019105784594
    },
    {
      "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": 0.00703125,
      "completions/max_length": 2734.7,
      "completions/max_terminated_length": 2040.6,
      "completions/mean_length": 462.809375,
      "completions/mean_terminated_length": 437.35008850097654,
      "completions/min_length": 132.1,
      "completions/min_terminated_length": 132.1,
      "entropy": 0.19860588386654854,
      "epoch": 0.2564102564102564,
      "frac_reward_zero_std": 0.046875,
      "grad_norm": 0.00016373889128419196,
      "learning_rate": 9.89846735808731e-06,
      "loss": -0.0002,
      "num_tokens": 18216537.0,
      "reward": -0.2901266008615494,
      "reward_std": 0.17708385214209557,
      "rewards/grpo_reward_function/mean": -0.2901266127824783,
      "rewards/grpo_reward_function/std": 0.47634183168411254,
      "sampling/importance_sampling_ratio/max": 0.21812310051172973,
      "sampling/importance_sampling_ratio/mean": 0.008993787248618901,
      "sampling/importance_sampling_ratio/min": 1.4824077907487356e-19,
      "sampling/sampling_logp_difference/max": 1.4244856238365173,
      "sampling/sampling_logp_difference/mean": 0.043855397030711174,
      "step": 80,
      "step_time": 124.79502142816781
    },
    {
      "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": 0.00625,
      "completions/max_length": 3060.3,
      "completions/max_terminated_length": 2223.8,
      "completions/mean_length": 443.69140625,
      "completions/mean_terminated_length": 420.85289306640624,
      "completions/min_length": 99.7,
      "completions/min_terminated_length": 99.7,
      "entropy": 0.19390448927879333,
      "epoch": 0.28846153846153844,
      "frac_reward_zero_std": 0.034375,
      "grad_norm": 0.0008450721128363838,
      "learning_rate": 9.733381816303395e-06,
      "loss": 0.0,
      "num_tokens": 20505158.0,
      "reward": -0.3168130427598953,
      "reward_std": 0.17042916193604468,
      "rewards/grpo_reward_function/mean": -0.3168130397796631,
      "rewards/grpo_reward_function/std": 0.44108698666095736,
      "sampling/importance_sampling_ratio/max": 0.15219400450587273,
      "sampling/importance_sampling_ratio/mean": 0.006058020202908665,
      "sampling/importance_sampling_ratio/min": 1.8109796575372013e-16,
      "sampling/sampling_logp_difference/max": 1.789087176322937,
      "sampling/sampling_logp_difference/mean": 0.04262695387005806,
      "step": 90,
      "step_time": 121.23764831908048
    },
    {
      "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": 0.009375,
      "completions/max_length": 3030.9,
      "completions/max_terminated_length": 2092.8,
      "completions/mean_length": 446.86796875,
      "completions/mean_terminated_length": 412.3775573730469,
      "completions/min_length": 129.0,
      "completions/min_terminated_length": 129.0,
      "entropy": 0.2053853865712881,
      "epoch": 0.32051282051282054,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.0017788003099368768,
      "learning_rate": 9.493048024473413e-06,
      "loss": 0.0,
      "num_tokens": 22735201.0,
      "reward": -0.2926711246371269,
      "reward_std": 0.13454800248146057,
      "rewards/grpo_reward_function/mean": -0.29267111867666246,
      "rewards/grpo_reward_function/std": 0.4685827106237411,
      "sampling/importance_sampling_ratio/max": 0.3497620947659016,
      "sampling/importance_sampling_ratio/mean": 0.010224268934689463,
      "sampling/importance_sampling_ratio/min": 1.3665770084452744e-22,
      "sampling/sampling_logp_difference/max": 1.469244658946991,
      "sampling/sampling_logp_difference/mean": 0.045334973558783534,
      "step": 100,
      "step_time": 121.1782845430076
    },
    {
      "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": 0.0109375,
      "completions/max_length": 3420.3,
      "completions/max_terminated_length": 2431.0,
      "completions/mean_length": 500.421875,
      "completions/mean_terminated_length": 460.7878082275391,
      "completions/min_length": 122.6,
      "completions/min_terminated_length": 122.6,
      "entropy": 0.20378807857632636,
      "epoch": 0.3525641025641026,
      "frac_reward_zero_std": 0.025,
      "grad_norm": 1.6932425828206626e-05,
      "learning_rate": 9.18128665415186e-06,
      "loss": -0.0006,
      "num_tokens": 25112061.0,
      "reward": -0.29770597368478774,
      "reward_std": 0.17742747887969018,
      "rewards/grpo_reward_function/mean": -0.29770597964525225,
      "rewards/grpo_reward_function/std": 0.4508321911096573,
      "sampling/importance_sampling_ratio/max": 0.32459819614887236,
      "sampling/importance_sampling_ratio/mean": 0.009665063163265587,
      "sampling/importance_sampling_ratio/min": 1.5782984274003976e-25,
      "sampling/sampling_logp_difference/max": 1.4484178304672242,
      "sampling/sampling_logp_difference/mean": 0.0441218126565218,
      "step": 110,
      "step_time": 140.20600229538977
    },
    {
      "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": 0.01328125,
      "completions/max_length": 3741.9,
      "completions/max_terminated_length": 2664.0,
      "completions/mean_length": 498.21796875,
      "completions/mean_terminated_length": 449.88751220703125,
      "completions/min_length": 126.2,
      "completions/min_terminated_length": 126.2,
      "entropy": 0.20669106915593147,
      "epoch": 0.38461538461538464,
      "frac_reward_zero_std": 0.053125,
      "grad_norm": 0.0012622156269559955,
      "learning_rate": 8.803053886449644e-06,
      "loss": -0.0001,
      "num_tokens": 27466064.0,
      "reward": -0.28830415159463885,
      "reward_std": 0.1518045909702778,
      "rewards/grpo_reward_function/mean": -0.28830414414405825,
      "rewards/grpo_reward_function/std": 0.4679163545370102,
      "sampling/importance_sampling_ratio/max": 0.3135061163455248,
      "sampling/importance_sampling_ratio/mean": 0.011271886434406042,
      "sampling/importance_sampling_ratio/min": 4.960477138108862e-24,
      "sampling/sampling_logp_difference/max": 1.522000241279602,
      "sampling/sampling_logp_difference/mean": 0.04500752612948418,
      "step": 120,
      "step_time": 139.78347074612975
    },
    {
      "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": 0.0125,
      "completions/max_length": 3804.1,
      "completions/max_terminated_length": 2468.0,
      "completions/mean_length": 492.92265625,
      "completions/mean_terminated_length": 447.16551513671874,
      "completions/min_length": 119.7,
      "completions/min_terminated_length": 119.7,
      "entropy": 0.20307930894196033,
      "epoch": 0.4166666666666667,
      "frac_reward_zero_std": 0.040625,
      "grad_norm": 0.0014322418381552205,
      "learning_rate": 8.364362621864595e-06,
      "loss": 0.0001,
      "num_tokens": 29859033.0,
      "reward": -0.30168216228485106,
      "reward_std": 0.16489436179399491,
      "rewards/grpo_reward_function/mean": -0.3016821652650833,
      "rewards/grpo_reward_function/std": 0.4505169212818146,
      "sampling/importance_sampling_ratio/max": 0.22667273543775082,
      "sampling/importance_sampling_ratio/mean": 0.0076055907527916135,
      "sampling/importance_sampling_ratio/min": 1.894551322523181e-19,
      "sampling/sampling_logp_difference/max": 1.4547302842140197,
      "sampling/sampling_logp_difference/mean": 0.043515678495168686,
      "step": 130,
      "step_time": 133.21774347238244
    },
    {
      "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": 0.00859375,
      "completions/max_length": 3465.7,
      "completions/max_terminated_length": 2434.9,
      "completions/mean_length": 484.92421875,
      "completions/mean_terminated_length": 453.5516418457031,
      "completions/min_length": 119.5,
      "completions/min_terminated_length": 119.5,
      "entropy": 0.20659017004072666,
      "epoch": 0.44871794871794873,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.0002746249901608735,
      "learning_rate": 7.872186891068997e-06,
      "loss": -0.0008,
      "num_tokens": 32161808.0,
      "reward": -0.2935184553265572,
      "reward_std": 0.16844813525676727,
      "rewards/grpo_reward_function/mean": -0.29351845383644104,
      "rewards/grpo_reward_function/std": 0.4532062470912933,
      "sampling/importance_sampling_ratio/max": 0.3149314503185451,
      "sampling/importance_sampling_ratio/mean": 0.011943908949615435,
      "sampling/importance_sampling_ratio/min": 3.97985328172928e-23,
      "sampling/sampling_logp_difference/max": 1.4437684297561646,
      "sampling/sampling_logp_difference/mean": 0.04485268816351891,
      "step": 140,
      "step_time": 142.53533354103564
    },
    {
      "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": 0.009375,
      "completions/max_length": 3581.1,
      "completions/max_terminated_length": 2453.6,
      "completions/mean_length": 461.7703125,
      "completions/mean_terminated_length": 427.3534698486328,
      "completions/min_length": 105.6,
      "completions/min_terminated_length": 105.6,
      "entropy": 0.19936109744012356,
      "epoch": 0.4807692307692308,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.00022260655992366576,
      "learning_rate": 7.3343509862697295e-06,
      "loss": -0.0005,
      "num_tokens": 34443498.0,
      "reward": -0.2657527238130569,
      "reward_std": 0.15306396633386612,
      "rewards/grpo_reward_function/mean": -0.2657527312636375,
      "rewards/grpo_reward_function/std": 0.4356934979557991,
      "sampling/importance_sampling_ratio/max": 0.3762046877294779,
      "sampling/importance_sampling_ratio/mean": 0.012305672373622656,
      "sampling/importance_sampling_ratio/min": 5.441343613780519e-26,
      "sampling/sampling_logp_difference/max": 1.4880928158760072,
      "sampling/sampling_logp_difference/mean": 0.04350205473601818,
      "step": 150,
      "step_time": 124.75091602019965
    },
    {
      "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": 0.01171875,
      "completions/max_length": 3721.0,
      "completions/max_terminated_length": 2522.3,
      "completions/mean_length": 473.11796875,
      "completions/mean_terminated_length": 430.37310180664065,
      "completions/min_length": 126.0,
      "completions/min_terminated_length": 126.0,
      "entropy": 0.19293792247772218,
      "epoch": 0.5128205128205128,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.0002647318948475709,
      "learning_rate": 6.759405075659165e-06,
      "loss": -0.0014,
      "num_tokens": 36770557.0,
      "reward": -0.2993745982646942,
      "reward_std": 0.14573606997728347,
      "rewards/grpo_reward_function/mean": -0.2993745982646942,
      "rewards/grpo_reward_function/std": 0.4381078839302063,
      "sampling/importance_sampling_ratio/max": 0.24595369808375836,
      "sampling/importance_sampling_ratio/mean": 0.00787243063095957,
      "sampling/importance_sampling_ratio/min": 6.3163868467048435e-22,
      "sampling/sampling_logp_difference/max": 1.4162023305892943,
      "sampling/sampling_logp_difference/mean": 0.042249183356761935,
      "step": 160,
      "step_time": 125.13825652077794
    },
    {
      "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": 0.0109375,
      "completions/max_length": 3383.8,
      "completions/max_terminated_length": 2437.6,
      "completions/mean_length": 493.4984375,
      "completions/mean_terminated_length": 453.9416534423828,
      "completions/min_length": 125.1,
      "completions/min_terminated_length": 125.1,
      "entropy": 0.18732221610844135,
      "epoch": 0.5448717948717948,
      "frac_reward_zero_std": 0.053125,
      "grad_norm": 4.303013523235627e-05,
      "learning_rate": 6.156489278357967e-06,
      "loss": 0.0003,
      "num_tokens": 39161855.0,
      "reward": -0.2639135167002678,
      "reward_std": 0.14633900821208953,
      "rewards/grpo_reward_function/mean": -0.2639135167002678,
      "rewards/grpo_reward_function/std": 0.4198122411966324,
      "sampling/importance_sampling_ratio/max": 0.31537908464670183,
      "sampling/importance_sampling_ratio/mean": 0.016544956981670113,
      "sampling/importance_sampling_ratio/min": 3.4869657448903444e-20,
      "sampling/sampling_logp_difference/max": 1.4604986667633058,
      "sampling/sampling_logp_difference/mean": 0.04075679816305637,
      "step": 170,
      "step_time": 129.28650456257165
    },
    {
      "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": 0.0109375,
      "completions/max_length": 3635.7,
      "completions/max_terminated_length": 2598.9,
      "completions/mean_length": 502.08984375,
      "completions/mean_terminated_length": 462.4349060058594,
      "completions/min_length": 115.7,
      "completions/min_terminated_length": 115.7,
      "entropy": 0.20257815159857273,
      "epoch": 0.5769230769230769,
      "frac_reward_zero_std": 0.034375,
      "grad_norm": 0.00110081427350525,
      "learning_rate": 5.535188360698687e-06,
      "loss": 0.0001,
      "num_tokens": 41584806.0,
      "reward": -0.2823033004999161,
      "reward_std": 0.15869370326399804,
      "rewards/grpo_reward_function/mean": -0.28230329751968386,
      "rewards/grpo_reward_function/std": 0.42252050116658213,
      "sampling/importance_sampling_ratio/max": 0.33239962328225375,
      "sampling/importance_sampling_ratio/mean": 0.01243584465701133,
      "sampling/importance_sampling_ratio/min": 4.5339507161993353e-20,
      "sampling/sampling_logp_difference/max": 1.5050785779953002,
      "sampling/sampling_logp_difference/mean": 0.04326198622584343,
      "step": 180,
      "step_time": 138.22612802386283
    },
    {
      "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": 0.0125,
      "completions/max_length": 3394.0,
      "completions/max_terminated_length": 2141.9,
      "completions/mean_length": 499.40859375,
      "completions/mean_terminated_length": 454.2876892089844,
      "completions/min_length": 130.1,
      "completions/min_terminated_length": 130.1,
      "entropy": 0.20578568913042544,
      "epoch": 0.6089743589743589,
      "frac_reward_zero_std": 0.04375,
      "grad_norm": 0.00043071462756502963,
      "learning_rate": 4.905379363794907e-06,
      "loss": -0.0004,
      "num_tokens": 43902245.0,
      "reward": -0.31633972823619844,
      "reward_std": 0.15992135927081108,
      "rewards/grpo_reward_function/mean": -0.31633972078561784,
      "rewards/grpo_reward_function/std": 0.39677430093288424,
      "sampling/importance_sampling_ratio/max": 0.27164736688137053,
      "sampling/importance_sampling_ratio/mean": 0.011083718878217042,
      "sampling/importance_sampling_ratio/min": 1.0615744153343375e-25,
      "sampling/sampling_logp_difference/max": 1.5213817000389098,
      "sampling/sampling_logp_difference/mean": 0.04432792365550995,
      "step": 190,
      "step_time": 128.48162812702358
    },
    {
      "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": 0.0078125,
      "completions/max_length": 3664.2,
      "completions/max_terminated_length": 2662.7,
      "completions/mean_length": 472.95703125,
      "completions/mean_terminated_length": 444.4935272216797,
      "completions/min_length": 123.6,
      "completions/min_terminated_length": 123.6,
      "entropy": 0.2034649882465601,
      "epoch": 0.6410256410256411,
      "frac_reward_zero_std": 0.04375,
      "grad_norm": 0.0005240383166085926,
      "learning_rate": 4.277074584714447e-06,
      "loss": -0.0011,
      "num_tokens": 46234190.0,
      "reward": -0.3208511009812355,
      "reward_std": 0.15701716020703316,
      "rewards/grpo_reward_function/mean": -0.3208511024713516,
      "rewards/grpo_reward_function/std": 0.45050418078899385,
      "sampling/importance_sampling_ratio/max": 0.328475890122354,
      "sampling/importance_sampling_ratio/mean": 0.00983516314299777,
      "sampling/importance_sampling_ratio/min": 5.287639308021113e-21,
      "sampling/sampling_logp_difference/max": 1.5113076210021972,
      "sampling/sampling_logp_difference/mean": 0.04435745738446713,
      "step": 200,
      "step_time": 129.30136452838778
    },
    {
      "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": 0.0109375,
      "completions/max_length": 3303.2,
      "completions/max_terminated_length": 2150.1,
      "completions/mean_length": 471.20078125,
      "completions/mean_terminated_length": 431.7250091552734,
      "completions/min_length": 108.1,
      "completions/min_terminated_length": 108.1,
      "entropy": 0.2041917782276869,
      "epoch": 0.6730769230769231,
      "frac_reward_zero_std": 0.04375,
      "grad_norm": 0.00012037056894249982,
      "learning_rate": 3.6602624074407354e-06,
      "loss": -0.0001,
      "num_tokens": 48535751.0,
      "reward": -0.30149365663528443,
      "reward_std": 0.1691875860095024,
      "rewards/grpo_reward_function/mean": -0.3014936536550522,
      "rewards/grpo_reward_function/std": 0.47970321476459504,
      "sampling/importance_sampling_ratio/max": 0.2858625270426273,
      "sampling/importance_sampling_ratio/mean": 0.009694790339563043,
      "sampling/importance_sampling_ratio/min": 2.2194460385427903e-23,
      "sampling/sampling_logp_difference/max": 1.5242174863815308,
      "sampling/sampling_logp_difference/mean": 0.04446354694664478,
      "step": 210,
      "step_time": 133.9577619012445
    },
    {
      "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": 0.0125,
      "completions/max_length": 3120.7,
      "completions/max_terminated_length": 2390.0,
      "completions/mean_length": 491.26171875,
      "completions/mean_terminated_length": 445.72156982421876,
      "completions/min_length": 127.1,
      "completions/min_terminated_length": 127.1,
      "entropy": 0.19974812418222426,
      "epoch": 0.7051282051282052,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.00018642290477094414,
      "learning_rate": 3.0647485139889145e-06,
      "loss": 0.0001,
      "num_tokens": 50902950.0,
      "reward": -0.2944794222712517,
      "reward_std": 0.1628885895013809,
      "rewards/grpo_reward_function/mean": -0.2944794237613678,
      "rewards/grpo_reward_function/std": 0.4895193099975586,
      "sampling/importance_sampling_ratio/max": 0.240592747554183,
      "sampling/importance_sampling_ratio/mean": 0.008309419581200928,
      "sampling/importance_sampling_ratio/min": 1.3308400815697736e-22,
      "sampling/sampling_logp_difference/max": 1.5919574618339538,
      "sampling/sampling_logp_difference/mean": 0.04345178902149201,
      "step": 220,
      "step_time": 126.08353825174272
    },
    {
      "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": 0.01171875,
      "completions/max_length": 3751.5,
      "completions/max_terminated_length": 2523.9,
      "completions/mean_length": 474.57265625,
      "completions/mean_terminated_length": 431.8814666748047,
      "completions/min_length": 114.4,
      "completions/min_terminated_length": 114.4,
      "entropy": 0.1985881496220827,
      "epoch": 0.7371794871794872,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.0026658825070435413,
      "learning_rate": 2.5000000000000015e-06,
      "loss": -0.0012,
      "num_tokens": 53179911.0,
      "reward": -0.2606832191348076,
      "reward_std": 0.14023807421326637,
      "rewards/grpo_reward_function/mean": -0.2606832206249237,
      "rewards/grpo_reward_function/std": 0.44046649634838103,
      "sampling/importance_sampling_ratio/max": 0.4130809709429741,
      "sampling/importance_sampling_ratio/mean": 0.01407341011799872,
      "sampling/importance_sampling_ratio/min": 5.8113626282032196e-24,
      "sampling/sampling_logp_difference/max": 1.5074282884597778,
      "sampling/sampling_logp_difference/mean": 0.043073756620287895,
      "step": 230,
      "step_time": 135.25556658878924
    },
    {
      "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": 0.0078125,
      "completions/max_length": 3329.5,
      "completions/max_terminated_length": 1825.4,
      "completions/mean_length": 442.0640625,
      "completions/mean_terminated_length": 413.4823333740234,
      "completions/min_length": 124.7,
      "completions/min_terminated_length": 124.7,
      "entropy": 0.1891279250383377,
      "epoch": 0.7692307692307693,
      "frac_reward_zero_std": 0.05,
      "grad_norm": 0.000754428644898323,
      "learning_rate": 1.9749948729627188e-06,
      "loss": -0.0004,
      "num_tokens": 55444105.0,
      "reward": -0.25499732717871665,
      "reward_std": 0.15963515415787696,
      "rewards/grpo_reward_function/mean": -0.2549973249435425,
      "rewards/grpo_reward_function/std": 0.45704702734947206,
      "sampling/importance_sampling_ratio/max": 0.7155711248517036,
      "sampling/importance_sampling_ratio/mean": 0.023246456822380423,
      "sampling/importance_sampling_ratio/min": 3.208456082463682e-19,
      "sampling/sampling_logp_difference/max": 1.402456557750702,
      "sampling/sampling_logp_difference/mean": 0.041706265136599543,
      "step": 240,
      "step_time": 131.66669817045332
    },
    {
      "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": 0.00859375,
      "completions/max_length": 3810.7,
      "completions/max_terminated_length": 2189.7,
      "completions/mean_length": 461.63203125,
      "completions/mean_terminated_length": 430.09210510253905,
      "completions/min_length": 128.3,
      "completions/min_terminated_length": 128.3,
      "entropy": 0.2066630445420742,
      "epoch": 0.8012820512820513,
      "frac_reward_zero_std": 0.0375,
      "grad_norm": 0.00013277689397189554,
      "learning_rate": 1.4980793256432474e-06,
      "loss": 0.0007,
      "num_tokens": 57751202.0,
      "reward": -0.2818883016705513,
      "reward_std": 0.13387203291058541,
      "rewards/grpo_reward_function/mean": -0.28188829869031906,
      "rewards/grpo_reward_function/std": 0.4075877174735069,
      "sampling/importance_sampling_ratio/max": 0.4012782000005245,
      "sampling/importance_sampling_ratio/mean": 0.014056763611733913,
      "sampling/importance_sampling_ratio/min": 3.294229030138832e-22,
      "sampling/sampling_logp_difference/max": 1.4437347650527954,
      "sampling/sampling_logp_difference/mean": 0.04458657465875149,
      "step": 250,
      "step_time": 136.0185191631317
    }
  ],
  "logging_steps": 10,
  "max_steps": 312,
  "num_input_tokens_seen": 58200744,
  "num_train_epochs": 1,
  "save_steps": 63,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": false
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 4,
  "trial_name": null,
  "trial_params": null
}