{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.11160714285714286,
  "eval_steps": 500,
  "global_step": 50,
  "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1969.0,
      "completions/mean_length": 670.640625,
      "completions/mean_terminated_length": 626.2096557617188,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.002232142857142857,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 5.688791751861572,
      "learning_rate": 0.0,
      "loss": 0.0,
      "num_tokens": 130921.0,
      "reward": 0.1523437350988388,
      "reward_std": 0.14437203109264374,
      "rewards/format_reward/mean": 0.4296875,
      "rewards/format_reward/std": 0.1751912236213684,
      "rewards/mcq_exact_match_reward/mean": 0.109375,
      "rewards/mcq_exact_match_reward/std": 0.3145764470100403,
      "step": 1
    },
    {
      "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": 0.0,
      "completions/max_length": 1241.0,
      "completions/max_terminated_length": 1241.0,
      "completions/mean_length": 485.5625,
      "completions/mean_terminated_length": 485.5625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.004464285714285714,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 14.286972999572754,
      "learning_rate": 2e-07,
      "loss": -0.0,
      "num_tokens": 239237.0,
      "reward": 0.25624996423721313,
      "reward_std": 0.3594309389591217,
      "rewards/format_reward/mean": 0.375,
      "rewards/format_reward/std": 0.26726123690605164,
      "rewards/mcq_exact_match_reward/mean": 0.21875,
      "rewards/mcq_exact_match_reward/std": 0.4166666865348816,
      "step": 2
    },
    {
      "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": 0.046875,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1409.0,
      "completions/mean_length": 619.875,
      "completions/mean_terminated_length": 549.6393432617188,
      "completions/min_length": 10.0,
      "completions/min_terminated_length": 10.0,
      "epoch": 0.006696428571428571,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.6942086219787598,
      "learning_rate": 4e-07,
      "loss": 0.0,
      "num_tokens": 364997.0,
      "reward": 0.2382812201976776,
      "reward_std": 0.3356662690639496,
      "rewards/format_reward/mean": 0.3515625,
      "rewards/format_reward/std": 0.26246222853660583,
      "rewards/mcq_exact_match_reward/mean": 0.203125,
      "rewards/mcq_exact_match_reward/std": 0.40550529956817627,
      "step": 3
    },
    {
      "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": 0.0,
      "completions/max_length": 1726.0,
      "completions/max_terminated_length": 1726.0,
      "completions/mean_length": 434.203125,
      "completions/mean_terminated_length": 434.203125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.008928571428571428,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 8.222421646118164,
      "learning_rate": 6e-07,
      "loss": -0.0,
      "num_tokens": 512530.0,
      "reward": 0.33203125,
      "reward_std": 0.25712186098098755,
      "rewards/format_reward/mean": 0.3515625,
      "rewards/format_reward/std": 0.2302463799715042,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "step": 4
    },
    {
      "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": 0.046875,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1684.0,
      "completions/mean_length": 542.75,
      "completions/mean_terminated_length": 468.7212829589844,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.011160714285714286,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.2874252796173096,
      "learning_rate": 8e-07,
      "loss": 0.0,
      "num_tokens": 636562.0,
      "reward": 0.2554687261581421,
      "reward_std": 0.31778302788734436,
      "rewards/format_reward/mean": 0.3671875,
      "rewards/format_reward/std": 0.2708333432674408,
      "rewards/mcq_exact_match_reward/mean": 0.21875,
      "rewards/mcq_exact_match_reward/std": 0.4166666865348816,
      "step": 5
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1934.0,
      "completions/mean_length": 570.921875,
      "completions/mean_terminated_length": 547.4761962890625,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.013392857142857142,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 16.40054702758789,
      "learning_rate": 1e-06,
      "loss": 0.0,
      "num_tokens": 770101.0,
      "reward": 0.3187499940395355,
      "reward_std": 0.3231881856918335,
      "rewards/format_reward/mean": 0.375,
      "rewards/format_reward/std": 0.2519763112068176,
      "rewards/mcq_exact_match_reward/mean": 0.28125,
      "rewards/mcq_exact_match_reward/std": 0.4531635046005249,
      "step": 6
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1421.0,
      "completions/mean_length": 540.3125,
      "completions/mean_terminated_length": 516.3809814453125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.015625,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.9286885261535645,
      "learning_rate": 9.99726628670463e-07,
      "loss": -0.0,
      "num_tokens": 896985.0,
      "reward": 0.2382812350988388,
      "reward_std": 0.3018285930156708,
      "rewards/format_reward/mean": 0.3515625,
      "rewards/format_reward/std": 0.2302463799715042,
      "rewards/mcq_exact_match_reward/mean": 0.203125,
      "rewards/mcq_exact_match_reward/std": 0.40550529956817627,
      "step": 7
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1182.0,
      "completions/mean_length": 408.03125,
      "completions/mean_terminated_length": 382.0000305175781,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.017857142857142856,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 10.685583114624023,
      "learning_rate": 9.989068136093872e-07,
      "loss": 0.0,
      "num_tokens": 1038179.0,
      "reward": 0.31874996423721313,
      "reward_std": 0.3212849497795105,
      "rewards/format_reward/mean": 0.375,
      "rewards/format_reward/std": 0.2182178944349289,
      "rewards/mcq_exact_match_reward/mean": 0.28125,
      "rewards/mcq_exact_match_reward/std": 0.4531635046005249,
      "step": 8
    },
    {
      "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": 0.046875,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1608.0,
      "completions/mean_length": 585.390625,
      "completions/mean_terminated_length": 513.458984375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.020089285714285716,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 6.234512805938721,
      "learning_rate": 9.975414512725056e-07,
      "loss": 0.0,
      "num_tokens": 1191276.0,
      "reward": 0.18984374403953552,
      "reward_std": 0.23865148425102234,
      "rewards/format_reward/mean": 0.3359375,
      "rewards/format_reward/std": 0.2824873626232147,
      "rewards/mcq_exact_match_reward/mean": 0.15625,
      "rewards/mcq_exact_match_reward/std": 0.36596253514289856,
      "step": 9
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1298.0,
      "completions/mean_length": 510.359375,
      "completions/mean_terminated_length": 485.9524230957031,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.022321428571428572,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 8.3804349899292,
      "learning_rate": 9.956320346634875e-07,
      "loss": 0.0,
      "num_tokens": 1355963.0,
      "reward": 0.19999998807907104,
      "reward_std": 0.24215617775917053,
      "rewards/format_reward/mean": 0.4375,
      "rewards/format_reward/std": 0.1666666716337204,
      "rewards/mcq_exact_match_reward/mean": 0.15625,
      "rewards/mcq_exact_match_reward/std": 0.36596253514289856,
      "step": 10
    },
    {
      "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": 0.0625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1759.0,
      "completions/mean_length": 546.875,
      "completions/mean_terminated_length": 446.8000183105469,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.024553571428571428,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 9.587211608886719,
      "learning_rate": 9.931806517013612e-07,
      "loss": 0.0,
      "num_tokens": 1510651.0,
      "reward": 0.18046873807907104,
      "reward_std": 0.2516200840473175,
      "rewards/format_reward/mean": 0.3984375,
      "rewards/format_reward/std": 0.20275264978408813,
      "rewards/mcq_exact_match_reward/mean": 0.140625,
      "rewards/mcq_exact_match_reward/std": 0.3503824472427368,
      "step": 11
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1622.0,
      "completions/mean_length": 468.15625,
      "completions/mean_terminated_length": 443.0793762207031,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.026785714285714284,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 9.15842342376709,
      "learning_rate": 9.901899829374047e-07,
      "loss": 0.0,
      "num_tokens": 1639349.0,
      "reward": 0.18203124403953552,
      "reward_std": 0.2767041325569153,
      "rewards/format_reward/mean": 0.4140625,
      "rewards/format_reward/std": 0.19012710452079773,
      "rewards/mcq_exact_match_reward/mean": 0.140625,
      "rewards/mcq_exact_match_reward/std": 0.3503824472427368,
      "step": 12
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1434.0,
      "completions/mean_length": 361.546875,
      "completions/mean_terminated_length": 334.7778015136719,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.029017857142857144,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 90.891357421875,
      "learning_rate": 9.866632986240029e-07,
      "loss": -0.0,
      "num_tokens": 1770720.0,
      "reward": 0.20937499403953552,
      "reward_std": 0.32802659273147583,
      "rewards/format_reward/mean": 0.375,
      "rewards/format_reward/std": 0.2182178944349289,
      "rewards/mcq_exact_match_reward/mean": 0.171875,
      "rewards/mcq_exact_match_reward/std": 0.38025420904159546,
      "step": 13
    },
    {
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1516.0,
      "completions/mean_length": 462.453125,
      "completions/mean_terminated_length": 411.3064270019531,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.03125,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 14.163143157958984,
      "learning_rate": 9.826044551386742e-07,
      "loss": 0.0,
      "num_tokens": 1887861.0,
      "reward": 0.18359375,
      "reward_std": 0.241878479719162,
      "rewards/format_reward/mean": 0.4296875,
      "rewards/format_reward/std": 0.2798410654067993,
      "rewards/mcq_exact_match_reward/mean": 0.140625,
      "rewards/mcq_exact_match_reward/std": 0.3503824472427368,
      "step": 14
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 2009.0,
      "completions/mean_length": 462.859375,
      "completions/mean_terminated_length": 437.69842529296875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.033482142857142856,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 15.210722923278809,
      "learning_rate": 9.780178907671788e-07,
      "loss": 0.0,
      "num_tokens": 2004364.0,
      "reward": 0.43437498807907104,
      "reward_std": 0.32749149203300476,
      "rewards/format_reward/mean": 0.4375,
      "rewards/format_reward/std": 0.1666666716337204,
      "rewards/mcq_exact_match_reward/mean": 0.390625,
      "rewards/mcq_exact_match_reward/std": 0.4917473793029785,
      "step": 15
    },
    {
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1953.0,
      "completions/mean_length": 307.4375,
      "completions/mean_terminated_length": 251.29031372070312,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.03571428571428571,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 36.37540054321289,
      "learning_rate": 9.729086208503173e-07,
      "loss": -0.0,
      "num_tokens": 2123304.0,
      "reward": 0.30781248211860657,
      "reward_std": 0.27473184466362,
      "rewards/format_reward/mean": 0.421875,
      "rewards/format_reward/std": 0.20351573824882507,
      "rewards/mcq_exact_match_reward/mean": 0.265625,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "step": 16
    },
    {
      "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": 0.0,
      "completions/max_length": 1886.0,
      "completions/max_terminated_length": 1886.0,
      "completions/mean_length": 273.625,
      "completions/mean_terminated_length": 273.625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.03794642857142857,
      "frac_reward_zero_std": 0.5,
      "grad_norm": 9.184857368469238,
      "learning_rate": 9.672822322997304e-07,
      "loss": -0.0,
      "num_tokens": 2229352.0,
      "reward": 0.3453124761581421,
      "reward_std": 0.23332780599594116,
      "rewards/format_reward/mean": 0.484375,
      "rewards/format_reward/std": 0.1985812783241272,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "step": 17
    },
    {
      "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": 0.0,
      "completions/max_length": 1820.0,
      "completions/max_terminated_length": 1820.0,
      "completions/mean_length": 198.578125,
      "completions/mean_terminated_length": 198.578125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.04017857142857143,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 13.390080451965332,
      "learning_rate": 9.611448774886923e-07,
      "loss": -0.0,
      "num_tokens": 2336037.0,
      "reward": 0.47109371423721313,
      "reward_std": 0.28497010469436646,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.421875,
      "rewards/mcq_exact_match_reward/std": 0.49776285886764526,
      "step": 18
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1105.0,
      "completions/mean_length": 160.203125,
      "completions/mean_terminated_length": 130.23809814453125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.04241071428571429,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 18.958669662475586,
      "learning_rate": 9.545032675245813e-07,
      "loss": -0.0,
      "num_tokens": 2460450.0,
      "reward": 0.5328124761581421,
      "reward_std": 0.31479179859161377,
      "rewards/format_reward/mean": 0.484375,
      "rewards/format_reward/std": 0.1534975916147232,
      "rewards/mcq_exact_match_reward/mean": 0.484375,
      "rewards/mcq_exact_match_reward/std": 0.5037065148353577,
      "step": 19
    },
    {
      "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": 0.0,
      "completions/max_length": 1464.0,
      "completions/max_terminated_length": 1464.0,
      "completions/mean_length": 201.640625,
      "completions/mean_terminated_length": 201.640625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.044642857142857144,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 12.093040466308594,
      "learning_rate": 9.473646649103817e-07,
      "loss": -0.0,
      "num_tokens": 2565795.0,
      "reward": 0.18906249105930328,
      "reward_std": 0.18418481945991516,
      "rewards/format_reward/mean": 0.484375,
      "rewards/format_reward/std": 0.1534975916147232,
      "rewards/mcq_exact_match_reward/mean": 0.140625,
      "rewards/mcq_exact_match_reward/std": 0.3503824472427368,
      "step": 20
    },
    {
      "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": 0.0,
      "completions/max_length": 1520.0,
      "completions/max_terminated_length": 1520.0,
      "completions/mean_length": 83.59375,
      "completions/mean_terminated_length": 83.59375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.046875,
      "frac_reward_zero_std": 0.25,
      "grad_norm": 22.807132720947266,
      "learning_rate": 9.397368756032444e-07,
      "loss": -0.0,
      "num_tokens": 2669537.0,
      "reward": 0.4078124761581421,
      "reward_std": 0.326413094997406,
      "rewards/format_reward/mean": 0.484375,
      "rewards/format_reward/std": 0.08768405020236969,
      "rewards/mcq_exact_match_reward/mean": 0.359375,
      "rewards/mcq_exact_match_reward/std": 0.4836103618144989,
      "step": 21
    },
    {
      "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": 0.0,
      "completions/max_length": 1720.0,
      "completions/max_terminated_length": 1720.0,
      "completions/mean_length": 65.453125,
      "completions/mean_terminated_length": 65.453125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.049107142857142856,
      "frac_reward_zero_std": 0.625,
      "grad_norm": 13.282039642333984,
      "learning_rate": 9.316282404787869e-07,
      "loss": -0.0,
      "num_tokens": 2762814.0,
      "reward": 0.4093749523162842,
      "reward_std": 0.16249822080135345,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.08908708393573761,
      "rewards/mcq_exact_match_reward/mean": 0.359375,
      "rewards/mcq_exact_match_reward/std": 0.4836103618144989,
      "step": 22
    },
    {
      "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": 0.0,
      "completions/max_length": 1298.0,
      "completions/max_terminated_length": 1298.0,
      "completions/mean_length": 105.953125,
      "completions/mean_terminated_length": 105.953125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.05133928571428571,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 23.15970802307129,
      "learning_rate": 9.230476262104676e-07,
      "loss": 0.0,
      "num_tokens": 2870939.0,
      "reward": 0.5195311903953552,
      "reward_std": 0.32276755571365356,
      "rewards/format_reward/mean": 0.5078125,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.46875,
      "rewards/mcq_exact_match_reward/std": 0.5029674172401428,
      "step": 23
    },
    {
      "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": 0.0,
      "completions/max_length": 1328.0,
      "completions/max_terminated_length": 1328.0,
      "completions/mean_length": 102.25,
      "completions/mean_terminated_length": 102.25,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.05357142857142857,
      "frac_reward_zero_std": 0.625,
      "grad_norm": 12.401535987854004,
      "learning_rate": 9.1400441557401e-07,
      "loss": -0.0,
      "num_tokens": 2973171.0,
      "reward": 0.4414061903953552,
      "reward_std": 0.19047920405864716,
      "rewards/format_reward/mean": 0.5078125,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.390625,
      "rewards/mcq_exact_match_reward/std": 0.4917473793029785,
      "step": 24
    },
    {
      "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": 0.0,
      "completions/max_length": 2043.0,
      "completions/max_terminated_length": 2043.0,
      "completions/mean_length": 89.15625,
      "completions/mean_terminated_length": 89.15625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.05580357142857143,
      "frac_reward_zero_std": 0.875,
      "grad_norm": 6.303321361541748,
      "learning_rate": 9.045084971874737e-07,
      "loss": -0.0,
      "num_tokens": 3078101.0,
      "reward": 0.5968749523162842,
      "reward_std": 0.0646936446428299,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.546875,
      "rewards/mcq_exact_match_reward/std": 0.501733124256134,
      "step": 25
    },
    {
      "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": 0.0,
      "completions/max_length": 955.0,
      "completions/max_terminated_length": 955.0,
      "completions/mean_length": 68.046875,
      "completions/mean_terminated_length": 68.046875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.05803571428571429,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 32.51689529418945,
      "learning_rate": 8.945702546981968e-07,
      "loss": -0.0,
      "num_tokens": 3169840.0,
      "reward": 0.46953120827674866,
      "reward_std": 0.429858535528183,
      "rewards/format_reward/mean": 0.4765625,
      "rewards/format_reward/std": 0.10652101784944534,
      "rewards/mcq_exact_match_reward/mean": 0.421875,
      "rewards/mcq_exact_match_reward/std": 0.49776285886764526,
      "step": 26
    },
    {
      "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": 0.0,
      "completions/max_length": 800.0,
      "completions/max_terminated_length": 800.0,
      "completions/mean_length": 113.171875,
      "completions/mean_terminated_length": 113.171875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.060267857142857144,
      "frac_reward_zero_std": 0.25,
      "grad_norm": 34.18963623046875,
      "learning_rate": 8.842005554284295e-07,
      "loss": -0.0,
      "num_tokens": 3263939.0,
      "reward": 0.5031249523162842,
      "reward_std": 0.368122935295105,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.08908708393573761,
      "rewards/mcq_exact_match_reward/mean": 0.453125,
      "rewards/mcq_exact_match_reward/std": 0.501733124256134,
      "step": 27
    },
    {
      "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": 0.0,
      "completions/max_length": 365.0,
      "completions/max_terminated_length": 365.0,
      "completions/mean_length": 11.796875,
      "completions/mean_terminated_length": 11.796875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0625,
      "frac_reward_zero_std": 0.875,
      "grad_norm": 14.970887184143066,
      "learning_rate": 8.734107384920769e-07,
      "loss": -0.0,
      "num_tokens": 3366678.0,
      "reward": 0.6437499523162842,
      "reward_std": 0.0578637532889843,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.59375,
      "rewards/mcq_exact_match_reward/std": 0.49501484632492065,
      "step": 28
    },
    {
      "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": 0.0,
      "completions/max_length": 1756.0,
      "completions/max_terminated_length": 1756.0,
      "completions/mean_length": 133.453125,
      "completions/mean_terminated_length": 133.453125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.06473214285714286,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 24.315492630004883,
      "learning_rate": 8.622126023955445e-07,
      "loss": -0.0,
      "num_tokens": 3497851.0,
      "reward": 0.38749998807907104,
      "reward_std": 0.29143062233924866,
      "rewards/format_reward/mean": 0.4375,
      "rewards/format_reward/std": 0.1666666716337204,
      "rewards/mcq_exact_match_reward/mean": 0.34375,
      "rewards/mcq_exact_match_reward/std": 0.4787135720252991,
      "step": 29
    },
    {
      "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": 0.0,
      "completions/max_length": 10.0,
      "completions/max_terminated_length": 10.0,
      "completions/mean_length": 6.0625,
      "completions/mean_terminated_length": 6.0625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.06696428571428571,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 17.23012924194336,
      "learning_rate": 8.506183921362442e-07,
      "loss": -0.0,
      "num_tokens": 3587679.0,
      "reward": 0.48671871423721313,
      "reward_std": 0.2717972695827484,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.4375,
      "rewards/mcq_exact_match_reward/std": 0.5,
      "step": 30
    },
    {
      "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": 0.0,
      "completions/max_length": 953.0,
      "completions/max_terminated_length": 953.0,
      "completions/mean_length": 90.21875,
      "completions/mean_terminated_length": 90.21875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.06919642857142858,
      "frac_reward_zero_std": 0.625,
      "grad_norm": 19.001739501953125,
      "learning_rate": 8.386407858128706e-07,
      "loss": -0.0,
      "num_tokens": 3679245.0,
      "reward": 0.23671874403953552,
      "reward_std": 0.1466214656829834,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.1875,
      "rewards/mcq_exact_match_reward/std": 0.39339789748191833,
      "step": 31
    },
    {
      "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": 0.0,
      "completions/max_length": 10.0,
      "completions/max_terminated_length": 10.0,
      "completions/mean_length": 6.0625,
      "completions/mean_terminated_length": 6.0625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.07142857142857142,
      "frac_reward_zero_std": 0.625,
      "grad_norm": 22.32570457458496,
      "learning_rate": 8.262928807620843e-07,
      "loss": -0.0,
      "num_tokens": 3758209.0,
      "reward": 0.33124998211860657,
      "reward_std": 0.17570310831069946,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.28125,
      "rewards/mcq_exact_match_reward/std": 0.4531635046005249,
      "step": 32
    },
    {
      "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": 0.0,
      "completions/max_length": 767.0,
      "completions/max_terminated_length": 767.0,
      "completions/mean_length": 41.234375,
      "completions/mean_terminated_length": 41.234375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.07366071428571429,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 20.529081344604492,
      "learning_rate": 8.135881792367685e-07,
      "loss": -0.0,
      "num_tokens": 3854136.0,
      "reward": 0.5812499523162842,
      "reward_std": 0.3072218596935272,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.53125,
      "rewards/mcq_exact_match_reward/std": 0.5029674172401428,
      "step": 33
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.07589285714285714,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 28.661266326904297,
      "learning_rate": 8.005405736415125e-07,
      "loss": -0.0,
      "num_tokens": 3962232.0,
      "reward": 0.4874999523162842,
      "reward_std": 0.3298332095146179,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.4375,
      "rewards/mcq_exact_match_reward/std": 0.5,
      "step": 34
    },
    {
      "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": 0.0,
      "completions/max_length": 494.0,
      "completions/max_terminated_length": 494.0,
      "completions/mean_length": 19.140625,
      "completions/mean_terminated_length": 19.140625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.078125,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 31.662900924682617,
      "learning_rate": 7.871643313414718e-07,
      "loss": -0.0,
      "num_tokens": 4071625.0,
      "reward": 0.40859371423721313,
      "reward_std": 0.28999006748199463,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.359375,
      "rewards/mcq_exact_match_reward/std": 0.4836103618144989,
      "step": 35
    },
    {
      "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": 0.0,
      "completions/max_length": 557.0,
      "completions/max_terminated_length": 557.0,
      "completions/mean_length": 28.25,
      "completions/mean_terminated_length": 28.25,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.08035714285714286,
      "frac_reward_zero_std": 0.875,
      "grad_norm": 4.329936504364014,
      "learning_rate": 7.734740790612136e-07,
      "loss": 0.0,
      "num_tokens": 4188697.0,
      "reward": 0.04921875149011612,
      "reward_std": 0.00220970856025815,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.0,
      "rewards/mcq_exact_match_reward/std": 0.0,
      "step": 36
    },
    {
      "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": 0.0,
      "completions/max_length": 281.0,
      "completions/max_terminated_length": 281.0,
      "completions/mean_length": 12.34375,
      "completions/mean_terminated_length": 12.34375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.08258928571428571,
      "frac_reward_zero_std": 0.625,
      "grad_norm": 29.80189323425293,
      "learning_rate": 7.594847868906076e-07,
      "loss": 0.0,
      "num_tokens": 4284655.0,
      "reward": 0.25312498211860657,
      "reward_std": 0.1530819833278656,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.203125,
      "rewards/mcq_exact_match_reward/std": 0.40550529956817627,
      "step": 37
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.08482142857142858,
      "frac_reward_zero_std": 0.25,
      "grad_norm": 23.103416442871094,
      "learning_rate": 7.452117519152541e-07,
      "loss": -0.0,
      "num_tokens": 4397455.0,
      "reward": 0.33124998211860657,
      "reward_std": 0.3335031270980835,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.28125,
      "rewards/mcq_exact_match_reward/std": 0.4531635046005249,
      "step": 38
    },
    {
      "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": 0.0,
      "completions/max_length": 519.0,
      "completions/max_terminated_length": 519.0,
      "completions/mean_length": 14.015625,
      "completions/mean_terminated_length": 14.015625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.08705357142857142,
      "frac_reward_zero_std": 0.625,
      "grad_norm": 25.02655601501465,
      "learning_rate": 7.306705814893439e-07,
      "loss": -0.0,
      "num_tokens": 4501840.0,
      "reward": 0.26874998211860657,
      "reward_std": 0.1735912561416626,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.21875,
      "rewards/mcq_exact_match_reward/std": 0.4166666865348816,
      "step": 39
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.08928571428571429,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 33.39609909057617,
      "learning_rate": 7.158771761692464e-07,
      "loss": -0.0,
      "num_tokens": 4589424.0,
      "reward": 0.5812499523162842,
      "reward_std": 0.30509039759635925,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.53125,
      "rewards/mcq_exact_match_reward/std": 0.5029674172401428,
      "step": 40
    },
    {
      "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": 0.0,
      "completions/max_length": 194.0,
      "completions/max_terminated_length": 194.0,
      "completions/mean_length": 10.65625,
      "completions/mean_terminated_length": 10.65625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.09151785714285714,
      "frac_reward_zero_std": 0.75,
      "grad_norm": 31.670530319213867,
      "learning_rate": 7.008477123264847e-07,
      "loss": -0.0,
      "num_tokens": 4678810.0,
      "reward": 0.28437498211860657,
      "reward_std": 0.1315089464187622,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.234375,
      "rewards/mcq_exact_match_reward/std": 0.42695629596710205,
      "step": 41
    },
    {
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 37.90625,
      "completions/mean_terminated_length": 6.000000476837158,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.09375,
      "frac_reward_zero_std": 0.75,
      "grad_norm": 11.955126762390137,
      "learning_rate": 6.855986244591103e-07,
      "loss": -0.0,
      "num_tokens": 4784756.0,
      "reward": 0.29921871423721313,
      "reward_std": 0.0887577086687088,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.0625,
      "rewards/mcq_exact_match_reward/mean": 0.25,
      "rewards/mcq_exact_match_reward/std": 0.4364357888698578,
      "step": 42
    },
    {
      "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": 0.0,
      "completions/max_length": 114.0,
      "completions/max_terminated_length": 114.0,
      "completions/mean_length": 7.6875,
      "completions/mean_terminated_length": 7.6875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.09598214285714286,
      "frac_reward_zero_std": 0.75,
      "grad_norm": 29.96390151977539,
      "learning_rate": 6.701465872208216e-07,
      "loss": -0.0,
      "num_tokens": 4861592.0,
      "reward": 0.5656249523162842,
      "reward_std": 0.10205792635679245,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.515625,
      "rewards/mcq_exact_match_reward/std": 0.5037065148353577,
      "step": 43
    },
    {
      "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": 0.0,
      "completions/max_length": 10.0,
      "completions/max_terminated_length": 10.0,
      "completions/mean_length": 6.0625,
      "completions/mean_terminated_length": 6.0625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.09821428571428571,
      "frac_reward_zero_std": 0.375,
      "grad_norm": 51.860042572021484,
      "learning_rate": 6.545084971874736e-07,
      "loss": -0.0,
      "num_tokens": 4939596.0,
      "reward": 0.42499998211860657,
      "reward_std": 0.3029785454273224,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.375,
      "rewards/mcq_exact_match_reward/std": 0.48795005679130554,
      "step": 44
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.10044642857142858,
      "frac_reward_zero_std": 0.625,
      "grad_norm": 23.78936195373535,
      "learning_rate": 6.387014543809223e-07,
      "loss": -0.0,
      "num_tokens": 5043372.0,
      "reward": 0.8624999523162842,
      "reward_std": 0.1462520956993103,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.8125,
      "rewards/mcq_exact_match_reward/std": 0.39339789748191833,
      "step": 45
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.10267857142857142,
      "frac_reward_zero_std": 0.75,
      "grad_norm": 24.635610580444336,
      "learning_rate": 6.227427435703995e-07,
      "loss": 0.0,
      "num_tokens": 5147556.0,
      "reward": 0.4718749523162842,
      "reward_std": 0.10205791890621185,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.421875,
      "rewards/mcq_exact_match_reward/std": 0.49776285886764526,
      "step": 46
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.10491071428571429,
      "frac_reward_zero_std": 0.875,
      "grad_norm": 7.54861307144165,
      "learning_rate": 6.066498153718734e-07,
      "loss": -0.0,
      "num_tokens": 5248228.0,
      "reward": 0.5343749523162842,
      "reward_std": 0.04419417306780815,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.484375,
      "rewards/mcq_exact_match_reward/std": 0.5037065148353577,
      "step": 47
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.10714285714285714,
      "frac_reward_zero_std": 0.75,
      "grad_norm": 82.69651794433594,
      "learning_rate": 5.90440267166055e-07,
      "loss": -0.0,
      "num_tokens": 5335820.0,
      "reward": 0.31562498211860657,
      "reward_std": 0.1315089464187622,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.265625,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "step": 48
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.109375,
      "frac_reward_zero_std": 0.75,
      "grad_norm": 21.174829483032227,
      "learning_rate": 5.741318238559209e-07,
      "loss": -0.0,
      "num_tokens": 5434012.0,
      "reward": 0.78125,
      "reward_std": 0.10342571139335632,
      "rewards/format_reward/mean": 0.46875,
      "rewards/format_reward/std": 0.12198751419782639,
      "rewards/mcq_exact_match_reward/mean": 0.734375,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "step": 49
    },
    {
      "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": 0.0,
      "completions/max_length": 6.0,
      "completions/max_terminated_length": 6.0,
      "completions/mean_length": 6.0,
      "completions/mean_terminated_length": 6.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.11160714285714286,
      "frac_reward_zero_std": 0.5,
      "grad_norm": 38.467594146728516,
      "learning_rate": 5.577423184847931e-07,
      "loss": -0.0,
      "num_tokens": 5531180.0,
      "reward": 0.39374998211860657,
      "reward_std": 0.21306739747524261,
      "rewards/format_reward/mean": 0.5,
      "rewards/format_reward/std": 0.0,
      "rewards/mcq_exact_match_reward/mean": 0.34375,
      "rewards/mcq_exact_match_reward/std": 0.4787135720252991,
      "step": 50
    }
  ],
  "logging_steps": 1,
  "max_steps": 100,
  "num_input_tokens_seen": 5531180,
  "num_train_epochs": 1,
  "save_steps": 50,
  "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
}