{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.1404494382022472,
  "eval_steps": 500,
  "global_step": 100,
  "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": 0.35,
      "completions/max_length": 6342.6,
      "completions/max_terminated_length": 4389.8,
      "completions/mean_length": 5336.85,
      "completions/mean_terminated_length": 3317.43818359375,
      "completions/min_length": 3839.2,
      "completions/min_terminated_length": 2220.0,
      "entropy": 0.13346208892762662,
      "epoch": 0.007022471910112359,
      "frac_reward_zero_std": 0.6,
      "grad_norm": 9.64939933512099e-18,
      "learning_rate": 5e-07,
      "loss": -5.502499733292571e-18,
      "num_tokens": 217434.0,
      "reward": 0.23524999916553496,
      "reward_std": 0.33813109397888186,
      "rewards/compute_reward/mean": 0.23524999916553496,
      "rewards/compute_reward/std": 0.33813111782073973,
      "sampling/importance_sampling_ratio/max": 0.600124711072749,
      "sampling/importance_sampling_ratio/mean": 0.35002241265671047,
      "sampling/importance_sampling_ratio/min": 0.20000003157700258,
      "sampling/sampling_logp_difference/max": 0.40465545654296875,
      "sampling/sampling_logp_difference/mean": 0.012414481490850449,
      "step": 5,
      "step_time": 78.7855021980009
    },
    {
      "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": 0.5,
      "completions/max_length": 8082.8,
      "completions/max_terminated_length": 5149.0,
      "completions/mean_length": 6318.725,
      "completions/mean_terminated_length": 4152.404833984375,
      "completions/min_length": 5055.0,
      "completions/min_terminated_length": 3438.4,
      "entropy": 0.1376650758087635,
      "epoch": 0.014044943820224719,
      "frac_reward_zero_std": 0.2,
      "grad_norm": 4.5075054799781356e-14,
      "learning_rate": 5e-07,
      "loss": -4.503626627183621e-09,
      "num_tokens": 474551.0,
      "reward": 0.08315625190734863,
      "reward_std": 0.45229817628860475,
      "rewards/compute_reward/mean": 0.08315625190734863,
      "rewards/compute_reward/std": 0.4522981971502304,
      "sampling/importance_sampling_ratio/max": 1.0,
      "sampling/importance_sampling_ratio/mean": 0.5000000178813935,
      "sampling/importance_sampling_ratio/min": 0.2,
      "sampling/sampling_logp_difference/max": 0.4511141300201416,
      "sampling/sampling_logp_difference/mean": 0.01602336037904024,
      "step": 10,
      "step_time": 101.2872915075859
    },
    {
      "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": 0.25,
      "completions/max_length": 6691.0,
      "completions/max_terminated_length": 6335.8,
      "completions/mean_length": 5185.35,
      "completions/mean_terminated_length": 4696.5650390625,
      "completions/min_length": 3638.4,
      "completions/min_terminated_length": 3638.4,
      "entropy": 0.1449942274019122,
      "epoch": 0.021067415730337078,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.2945383787155151e-07,
      "learning_rate": 5e-07,
      "loss": -1.8636472304933705e-06,
      "num_tokens": 686789.0,
      "reward": 0.3246250033378601,
      "reward_std": 0.5596808314323425,
      "rewards/compute_reward/mean": 0.3246250033378601,
      "rewards/compute_reward/std": 0.5596808195114136,
      "sampling/importance_sampling_ratio/max": 0.4000196443371508,
      "sampling/importance_sampling_ratio/mean": 0.2500068536591781,
      "sampling/importance_sampling_ratio/min": 4.741177590247785e-12,
      "sampling/sampling_logp_difference/max": 0.5730626344680786,
      "sampling/sampling_logp_difference/mean": 0.01585924383252859,
      "step": 15,
      "step_time": 76.25853070278535
    },
    {
      "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": 0.525,
      "completions/max_length": 6157.0,
      "completions/max_terminated_length": 2626.2,
      "completions/mean_length": 5558.75,
      "completions/mean_terminated_length": 2014.99169921875,
      "completions/min_length": 4721.0,
      "completions/min_terminated_length": 1498.2,
      "entropy": 0.08328314051032067,
      "epoch": 0.028089887640449437,
      "frac_reward_zero_std": 0.6,
      "grad_norm": 2.2411346435546875e-05,
      "learning_rate": 5e-07,
      "loss": -1.7270938769797795e-06,
      "num_tokens": 913419.0,
      "reward": 0.011406248807907105,
      "reward_std": 0.23478415608406067,
      "rewards/compute_reward/mean": 0.011406248807907105,
      "rewards/compute_reward/std": 0.23478415608406067,
      "sampling/importance_sampling_ratio/max": 0.6000377955271687,
      "sampling/importance_sampling_ratio/mean": 0.5250090071654995,
      "sampling/importance_sampling_ratio/min": 0.4000000000222174,
      "sampling/sampling_logp_difference/max": 0.26677918434143066,
      "sampling/sampling_logp_difference/mean": 0.0077647527679800986,
      "step": 20,
      "step_time": 86.48750274259365
    },
    {
      "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": 0.525,
      "completions/max_length": 7358.0,
      "completions/max_terminated_length": 5255.0,
      "completions/mean_length": 6298.35,
      "completions/mean_terminated_length": 3806.3400390625,
      "completions/min_length": 4295.4,
      "completions/min_terminated_length": 2676.4,
      "entropy": 0.11463551484048366,
      "epoch": 0.0351123595505618,
      "frac_reward_zero_std": 0.4,
      "grad_norm": 7.334165275096893e-09,
      "learning_rate": 5e-07,
      "loss": -4.845223333660442e-10,
      "num_tokens": 1168593.0,
      "reward": -0.027125003933906554,
      "reward_std": 0.5179957747459412,
      "rewards/compute_reward/mean": -0.027125003933906554,
      "rewards/compute_reward/std": 0.5179957985877991,
      "sampling/importance_sampling_ratio/max": 0.8000000000302949,
      "sampling/importance_sampling_ratio/mean": 0.5250000000044796,
      "sampling/importance_sampling_ratio/min": 0.2,
      "sampling/sampling_logp_difference/max": 0.4423918962478638,
      "sampling/sampling_logp_difference/mean": 0.013001743517816067,
      "step": 25,
      "step_time": 100.9442321440205
    },
    {
      "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": 7867.0,
      "completions/max_terminated_length": 6072.6,
      "completions/mean_length": 5969.625,
      "completions/mean_terminated_length": 3844.621484375,
      "completions/min_length": 3876.8,
      "completions/min_terminated_length": 2283.6,
      "entropy": 0.1684934377670288,
      "epoch": 0.042134831460674156,
      "frac_reward_zero_std": 0.4,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": -7.479580510505457e-10,
      "num_tokens": 1412506.0,
      "reward": 0.21940625309944153,
      "reward_std": 0.38065622448921205,
      "rewards/compute_reward/mean": 0.21940625309944153,
      "rewards/compute_reward/std": 0.38065624237060547,
      "sampling/importance_sampling_ratio/max": 0.8000000000015144,
      "sampling/importance_sampling_ratio/mean": 0.3750000059606567,
      "sampling/importance_sampling_ratio/min": 0.2,
      "sampling/sampling_logp_difference/max": 0.46221423149108887,
      "sampling/sampling_logp_difference/mean": 0.015006156265735626,
      "step": 30,
      "step_time": 92.77473875220167
    },
    {
      "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": 0.225,
      "completions/max_length": 5447.6,
      "completions/max_terminated_length": 4491.6,
      "completions/mean_length": 3979.525,
      "completions/mean_terminated_length": 3340.66669921875,
      "completions/min_length": 2628.0,
      "completions/min_terminated_length": 2628.0,
      "entropy": 0.15426228623837232,
      "epoch": 0.04915730337078652,
      "frac_reward_zero_std": 0.4,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": 5.5907651130837624e-08,
      "num_tokens": 1577087.0,
      "reward": 0.2894999995827675,
      "reward_std": 0.4346048951148987,
      "rewards/compute_reward/mean": 0.2894999995827675,
      "rewards/compute_reward/std": 0.43460490703582766,
      "sampling/importance_sampling_ratio/max": 0.4101849972048058,
      "sampling/importance_sampling_ratio/mean": 0.2273602893020211,
      "sampling/importance_sampling_ratio/min": 9.34835527009561e-05,
      "sampling/sampling_logp_difference/max": 0.5941443920135498,
      "sampling/sampling_logp_difference/mean": 0.014553416334092618,
      "step": 35,
      "step_time": 61.56992885797517
    },
    {
      "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": 0.7,
      "completions/max_length": 8080.2,
      "completions/max_terminated_length": 4724.0,
      "completions/mean_length": 7451.025,
      "completions/mean_terminated_length": 4068.96435546875,
      "completions/min_length": 6761.0,
      "completions/min_terminated_length": 3524.6,
      "entropy": 0.06469594687223434,
      "epoch": 0.056179775280898875,
      "frac_reward_zero_std": 0.4,
      "grad_norm": 1.546140993013978e-11,
      "learning_rate": 5e-07,
      "loss": 1.2250055136941995e-12,
      "num_tokens": 1879600.0,
      "reward": -0.5208124965429306,
      "reward_std": 0.28857092424295844,
      "rewards/compute_reward/mean": -0.5208124965429306,
      "rewards/compute_reward/std": 0.2885709242895246,
      "sampling/importance_sampling_ratio/max": 1.0,
      "sampling/importance_sampling_ratio/mean": 0.7,
      "sampling/importance_sampling_ratio/min": 0.4,
      "sampling/sampling_logp_difference/max": 0.2912771701812744,
      "sampling/sampling_logp_difference/mean": 0.010922013968229293,
      "step": 40,
      "step_time": 122.65326870220596
    },
    {
      "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": 0.475,
      "completions/max_length": 6348.2,
      "completions/max_terminated_length": 3134.0,
      "completions/mean_length": 5377.575,
      "completions/mean_terminated_length": 1950.0400390625,
      "completions/min_length": 4405.0,
      "completions/min_terminated_length": 1193.0,
      "entropy": 0.12274600118398667,
      "epoch": 0.06320224719101124,
      "frac_reward_zero_std": 0.6,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": 5.91442221775651e-05,
      "num_tokens": 2098759.0,
      "reward": -0.26150000691413877,
      "reward_std": 0.26170684695243834,
      "rewards/compute_reward/mean": -0.26150000691413877,
      "rewards/compute_reward/std": 0.2617068350315094,
      "sampling/importance_sampling_ratio/max": 0.6020115901934332,
      "sampling/importance_sampling_ratio/mean": 0.47526856281074287,
      "sampling/importance_sampling_ratio/min": 0.4000000296924043,
      "sampling/sampling_logp_difference/max": 0.2926159858703613,
      "sampling/sampling_logp_difference/mean": 0.010471996665000916,
      "step": 45,
      "step_time": 85.52473701598937
    },
    {
      "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": 0.125,
      "completions/max_length": 7601.6,
      "completions/max_terminated_length": 6804.0,
      "completions/mean_length": 5256.25,
      "completions/mean_terminated_length": 4877.31513671875,
      "completions/min_length": 3520.8,
      "completions/min_terminated_length": 3520.8,
      "entropy": 0.1676523458212614,
      "epoch": 0.0702247191011236,
      "frac_reward_zero_std": 0.2,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": -8.948970275923785e-09,
      "num_tokens": 2313249.0,
      "reward": 0.6231875061988831,
      "reward_std": 0.30267552062869074,
      "rewards/compute_reward/mean": 0.6231875061988831,
      "rewards/compute_reward/std": 0.302675536647439,
      "sampling/importance_sampling_ratio/max": 0.6000000545625805,
      "sampling/importance_sampling_ratio/mean": 0.12500003689434863,
      "sampling/importance_sampling_ratio/min": 1.509659897099789e-11,
      "sampling/sampling_logp_difference/max": 0.6264592289924622,
      "sampling/sampling_logp_difference/mean": 0.014229525998234748,
      "step": 50,
      "step_time": 76.5796675938298
    },
    {
      "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": 0.6,
      "completions/max_length": 7477.6,
      "completions/max_terminated_length": 5252.6,
      "completions/mean_length": 6872.8,
      "completions/mean_terminated_length": 4782.11669921875,
      "completions/min_length": 6005.6,
      "completions/min_terminated_length": 4387.0,
      "entropy": 0.12369837798178196,
      "epoch": 0.07724719101123595,
      "frac_reward_zero_std": 0.4,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": -2.9906460688816355e-19,
      "num_tokens": 2592889.0,
      "reward": -0.4731250017881393,
      "reward_std": 0.21400290131568908,
      "rewards/compute_reward/mean": -0.4731250017881393,
      "rewards/compute_reward/std": 0.21400290429592134,
      "sampling/importance_sampling_ratio/max": 0.8000000000400606,
      "sampling/importance_sampling_ratio/mean": 0.6000000000050764,
      "sampling/importance_sampling_ratio/min": 0.2,
      "sampling/sampling_logp_difference/max": 0.37660536766052244,
      "sampling/sampling_logp_difference/mean": 0.01498084533959627,
      "step": 55,
      "step_time": 109.11029338700465
    },
    {
      "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": 0.3,
      "completions/max_length": 6866.0,
      "completions/max_terminated_length": 6220.2,
      "completions/mean_length": 5375.525,
      "completions/mean_terminated_length": 4569.51435546875,
      "completions/min_length": 3334.8,
      "completions/min_terminated_length": 3334.8,
      "entropy": 0.14546145126223564,
      "epoch": 0.08426966292134831,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 8.715250743307479e-15,
      "learning_rate": 5e-07,
      "loss": -0.0002411100547760725,
      "num_tokens": 2811574.0,
      "reward": 0.1729999989271164,
      "reward_std": 0.6171267032623291,
      "rewards/compute_reward/mean": 0.1729999989271164,
      "rewards/compute_reward/std": 0.6171267211437226,
      "sampling/importance_sampling_ratio/max": 0.8053813435137271,
      "sampling/importance_sampling_ratio/mean": 0.30101747438311577,
      "sampling/importance_sampling_ratio/min": 3.9959536479727876e-08,
      "sampling/sampling_logp_difference/max": 0.5940536499023438,
      "sampling/sampling_logp_difference/mean": 0.015674336068332195,
      "step": 60,
      "step_time": 82.90924808238633
    },
    {
      "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": 0.25,
      "completions/max_length": 6714.4,
      "completions/max_terminated_length": 5924.6,
      "completions/mean_length": 4677.0,
      "completions/mean_terminated_length": 4329.3583984375,
      "completions/min_length": 3018.8,
      "completions/min_terminated_length": 3018.8,
      "entropy": 0.15218620076775552,
      "epoch": 0.09129213483146068,
      "frac_reward_zero_std": 0.4,
      "grad_norm": 7.748603820800781e-07,
      "learning_rate": 5e-07,
      "loss": -1.951258497001618e-08,
      "num_tokens": 3004854.0,
      "reward": 0.4383750021457672,
      "reward_std": 0.43163261413574217,
      "rewards/compute_reward/mean": 0.4383750021457672,
      "rewards/compute_reward/std": 0.43163261413574217,
      "sampling/importance_sampling_ratio/max": 0.6001360186763123,
      "sampling/importance_sampling_ratio/mean": 0.2500173559734094,
      "sampling/importance_sampling_ratio/min": 8.312385355119191e-16,
      "sampling/sampling_logp_difference/max": 0.5871267795562745,
      "sampling/sampling_logp_difference/mean": 0.014853741601109504,
      "step": 65,
      "step_time": 72.67464896581369
    },
    {
      "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": 0.5,
      "completions/max_length": 7633.4,
      "completions/max_terminated_length": 5260.2,
      "completions/mean_length": 6409.45,
      "completions/mean_terminated_length": 4188.1984375,
      "completions/min_length": 4947.8,
      "completions/min_terminated_length": 3327.2,
      "entropy": 0.10757671780884266,
      "epoch": 0.09831460674157304,
      "frac_reward_zero_std": 0.2,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": -2.9429700343364117e-11,
      "num_tokens": 3265376.0,
      "reward": 0.09690623879432678,
      "reward_std": 0.5412219107151032,
      "rewards/compute_reward/mean": 0.09690623879432678,
      "rewards/compute_reward/std": 0.54122194647789,
      "sampling/importance_sampling_ratio/max": 0.8000000000121613,
      "sampling/importance_sampling_ratio/mean": 0.5000000000016139,
      "sampling/importance_sampling_ratio/min": 0.2,
      "sampling/sampling_logp_difference/max": 0.5057941436767578,
      "sampling/sampling_logp_difference/mean": 0.01242091991007328,
      "step": 70,
      "step_time": 100.18304894539179
    },
    {
      "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": 0.475,
      "completions/max_length": 7485.6,
      "completions/max_terminated_length": 4652.0,
      "completions/mean_length": 6131.35,
      "completions/mean_terminated_length": 3532.04501953125,
      "completions/min_length": 4299.8,
      "completions/min_terminated_length": 2692.0,
      "entropy": 0.12216376923024655,
      "epoch": 0.10533707865168539,
      "frac_reward_zero_std": 0.2,
      "grad_norm": 1.7136335372924805e-07,
      "learning_rate": 5e-07,
      "loss": -1.2361371659608266e-08,
      "num_tokens": 3514990.0,
      "reward": 0.02084375321865082,
      "reward_std": 0.6247592151165009,
      "rewards/compute_reward/mean": 0.02084375321865082,
      "rewards/compute_reward/std": 0.6247592031955719,
      "sampling/importance_sampling_ratio/max": 0.8000003093028454,
      "sampling/importance_sampling_ratio/mean": 0.475000045385957,
      "sampling/importance_sampling_ratio/min": 0.20000000000001467,
      "sampling/sampling_logp_difference/max": 0.46574991941452026,
      "sampling/sampling_logp_difference/mean": 0.013754993863403796,
      "step": 75,
      "step_time": 96.75799002298155
    },
    {
      "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": 0.2,
      "completions/max_length": 5894.8,
      "completions/max_terminated_length": 5214.2,
      "completions/mean_length": 4056.425,
      "completions/mean_terminated_length": 3393.43935546875,
      "completions/min_length": 2331.4,
      "completions/min_terminated_length": 2331.4,
      "entropy": 0.14181112498044968,
      "epoch": 0.11235955056179775,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.00015544891357421875,
      "learning_rate": 5e-07,
      "loss": 4.291571385692805e-05,
      "num_tokens": 3682831.0,
      "reward": 0.1677812457084656,
      "reward_std": 0.41842788457870483,
      "rewards/compute_reward/mean": 0.1677812457084656,
      "rewards/compute_reward/std": 0.4184278786182404,
      "sampling/importance_sampling_ratio/max": 0.4010849003840121,
      "sampling/importance_sampling_ratio/mean": 0.20015338080880837,
      "sampling/importance_sampling_ratio/min": 6.377647438378011e-10,
      "sampling/sampling_logp_difference/max": 0.5928982496261597,
      "sampling/sampling_logp_difference/mean": 0.01402966920286417,
      "step": 80,
      "step_time": 62.0829234021774
    },
    {
      "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": 0.725,
      "completions/max_length": 8033.8,
      "completions/max_terminated_length": 5586.4,
      "completions/mean_length": 7458.45,
      "completions/mean_terminated_length": 5082.3,
      "completions/min_length": 6264.2,
      "completions/min_terminated_length": 4664.6,
      "entropy": 0.049185825511813164,
      "epoch": 0.11938202247191011,
      "frac_reward_zero_std": 0.2,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": -1.0684439999408823e-09,
      "num_tokens": 3987497.0,
      "reward": -0.36359376907348634,
      "reward_std": 0.35454786419868467,
      "rewards/compute_reward/mean": -0.36359376907348634,
      "rewards/compute_reward/std": 0.35454787611961364,
      "sampling/importance_sampling_ratio/max": 1.0,
      "sampling/importance_sampling_ratio/mean": 0.725,
      "sampling/importance_sampling_ratio/min": 0.2000000000015946,
      "sampling/sampling_logp_difference/max": 0.4322138071060181,
      "sampling/sampling_logp_difference/mean": 0.011536732874810695,
      "step": 85,
      "step_time": 121.9961906942015
    },
    {
      "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": 0.7,
      "completions/max_length": 8083.6,
      "completions/max_terminated_length": 5480.0,
      "completions/mean_length": 7200.725,
      "completions/mean_terminated_length": 4447.6,
      "completions/min_length": 5399.0,
      "completions/min_terminated_length": 3780.4,
      "entropy": 0.06885569505393505,
      "epoch": 0.12640449438202248,
      "frac_reward_zero_std": 0.2,
      "grad_norm": 6.502887117676437e-11,
      "learning_rate": 5e-07,
      "loss": 3.4291680606202136e-12,
      "num_tokens": 4279862.0,
      "reward": -0.3385625034570694,
      "reward_std": 0.6242223262786866,
      "rewards/compute_reward/mean": -0.3385625034570694,
      "rewards/compute_reward/std": 0.6242223501205444,
      "sampling/importance_sampling_ratio/max": 1.0,
      "sampling/importance_sampling_ratio/mean": 0.7,
      "sampling/importance_sampling_ratio/min": 0.20000000000054166,
      "sampling/sampling_logp_difference/max": 0.36994409561157227,
      "sampling/sampling_logp_difference/mean": 0.012296585179865361,
      "step": 90,
      "step_time": 117.92537313302746
    },
    {
      "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": 0.4,
      "completions/max_length": 6249.8,
      "completions/max_terminated_length": 4446.6,
      "completions/mean_length": 5328.1,
      "completions/mean_terminated_length": 3106.85,
      "completions/min_length": 3716.6,
      "completions/min_terminated_length": 2106.2,
      "entropy": 0.11026849336922169,
      "epoch": 0.13342696629213482,
      "frac_reward_zero_std": 0.6,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": -5.5089893757909183e-11,
      "num_tokens": 4497202.0,
      "reward": 0.19196875393390656,
      "reward_std": 0.35176035165786745,
      "rewards/compute_reward/mean": 0.19196875393390656,
      "rewards/compute_reward/std": 0.35176035165786745,
      "sampling/importance_sampling_ratio/max": 0.603415791407131,
      "sampling/importance_sampling_ratio/mean": 0.4010292052433563,
      "sampling/importance_sampling_ratio/min": 0.20000481952592963,
      "sampling/sampling_logp_difference/max": 0.38380826711654664,
      "sampling/sampling_logp_difference/mean": 0.010995614714920521,
      "step": 95,
      "step_time": 80.9261963216064
    },
    {
      "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": 0.825,
      "completions/max_length": 8078.4,
      "completions/max_terminated_length": 2545.0,
      "completions/mean_length": 7691.95,
      "completions/mean_terminated_length": 2276.2333984375,
      "completions/min_length": 6652.6,
      "completions/min_terminated_length": 1790.2,
      "entropy": 0.04812453463673592,
      "epoch": 0.1404494382022472,
      "frac_reward_zero_std": 0.2,
      "grad_norm": 0.0,
      "learning_rate": 5e-07,
      "loss": 1.299868734295816e-19,
      "num_tokens": 4809424.0,
      "reward": -0.5054375052452087,
      "reward_std": 0.3607128977775574,
      "rewards/compute_reward/mean": -0.5054375052452087,
      "rewards/compute_reward/std": 0.36071292757987977,
      "sampling/importance_sampling_ratio/max": 1.0,
      "sampling/importance_sampling_ratio/mean": 0.825,
      "sampling/importance_sampling_ratio/min": 0.6,
      "sampling/sampling_logp_difference/max": 0.1978625774383545,
      "sampling/sampling_logp_difference/mean": 0.0077566314488649365,
      "step": 100,
      "step_time": 127.10719606040512
    }
  ],
  "logging_steps": 5,
  "max_steps": 712,
  "num_input_tokens_seen": 4809424,
  "num_train_epochs": 1,
  "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": 0.0,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}