{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.125,
  "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": 0.0,
      "completions/max_length": 1593.0,
      "completions/max_terminated_length": 1593.0,
      "completions/mean_length": 529.984375,
      "completions/mean_terminated_length": 529.984375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0025,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.779510974884033,
      "learning_rate": 0.0,
      "loss": -0.0,
      "num_tokens": 126311.0,
      "reward": 0.40312495827674866,
      "reward_std": 0.3971395492553711,
      "rewards/format_reward/mean": 0.4375,
      "rewards/format_reward/std": 0.22712838649749756,
      "rewards/mcq_exact_match_reward/mean": 0.359375,
      "rewards/mcq_exact_match_reward/std": 0.4836103618144989,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1773.0,
      "completions/mean_length": 613.171875,
      "completions/mean_terminated_length": 590.3968505859375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.005,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 6.4733357429504395,
      "learning_rate": 2e-07,
      "loss": -0.0,
      "num_tokens": 257954.0,
      "reward": 0.3984374701976776,
      "reward_std": 0.3972335159778595,
      "rewards/format_reward/mean": 0.390625,
      "rewards/format_reward/std": 0.2592533528804779,
      "rewards/mcq_exact_match_reward/mean": 0.359375,
      "rewards/mcq_exact_match_reward/std": 0.4836103618144989,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1526.0,
      "completions/mean_length": 583.3125,
      "completions/mean_terminated_length": 560.0635375976562,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0075,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 4.108024597167969,
      "learning_rate": 4e-07,
      "loss": -0.0,
      "num_tokens": 387742.0,
      "reward": 0.29218748211860657,
      "reward_std": 0.3574400544166565,
      "rewards/format_reward/mean": 0.421875,
      "rewards/format_reward/std": 0.20351573824882507,
      "rewards/mcq_exact_match_reward/mean": 0.25,
      "rewards/mcq_exact_match_reward/std": 0.4364357888698578,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 821.0,
      "completions/mean_length": 359.453125,
      "completions/mean_terminated_length": 332.65081787109375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.01,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 10.06227970123291,
      "learning_rate": 6e-07,
      "loss": -0.0,
      "num_tokens": 469075.0,
      "reward": 0.3531249761581421,
      "reward_std": 0.38945531845092773,
      "rewards/format_reward/mean": 0.40625,
      "rewards/format_reward/std": 0.25,
      "rewards/mcq_exact_match_reward/mean": 0.3125,
      "rewards/mcq_exact_match_reward/std": 0.467176616191864,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1862.0,
      "completions/mean_length": 538.828125,
      "completions/mean_terminated_length": 490.1451416015625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0125,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 5.812922477722168,
      "learning_rate": 8e-07,
      "loss": -0.0,
      "num_tokens": 583136.0,
      "reward": 0.2945312261581421,
      "reward_std": 0.2819061279296875,
      "rewards/format_reward/mean": 0.2890625,
      "rewards/format_reward/std": 0.2928335666656494,
      "rewards/mcq_exact_match_reward/mean": 0.265625,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "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": 1669.0,
      "completions/mean_length": 634.6875,
      "completions/mean_terminated_length": 612.2540283203125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.015,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 3.530496597290039,
      "learning_rate": 1e-06,
      "loss": 0.0,
      "num_tokens": 716252.0,
      "reward": 0.34531253576278687,
      "reward_std": 0.3087073266506195,
      "rewards/format_reward/mean": 0.328125,
      "rewards/format_reward/std": 0.27048972249031067,
      "rewards/mcq_exact_match_reward/mean": 0.3125,
      "rewards/mcq_exact_match_reward/std": 0.467176616191864,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1684.0,
      "completions/mean_length": 653.84375,
      "completions/mean_terminated_length": 608.8709716796875,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "epoch": 0.0175,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 1.750030279159546,
      "learning_rate": 9.99726628670463e-07,
      "loss": 0.0,
      "num_tokens": 858650.0,
      "reward": 0.3554687201976776,
      "reward_std": 0.3550691604614258,
      "rewards/format_reward/mean": 0.4296875,
      "rewards/format_reward/std": 0.1751912236213684,
      "rewards/mcq_exact_match_reward/mean": 0.3125,
      "rewards/mcq_exact_match_reward/std": 0.467176616191864,
      "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": 0.0,
      "completions/max_length": 1871.0,
      "completions/max_terminated_length": 1871.0,
      "completions/mean_length": 489.21875,
      "completions/mean_terminated_length": 489.21875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.02,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.0936737060546875,
      "learning_rate": 9.989068136093872e-07,
      "loss": 0.0,
      "num_tokens": 994856.0,
      "reward": 0.19218748807907104,
      "reward_std": 0.2725487947463989,
      "rewards/format_reward/mean": 0.359375,
      "rewards/format_reward/std": 0.2592533528804779,
      "rewards/mcq_exact_match_reward/mean": 0.15625,
      "rewards/mcq_exact_match_reward/std": 0.36596253514289856,
      "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": 0.0,
      "completions/max_length": 1867.0,
      "completions/max_terminated_length": 1867.0,
      "completions/mean_length": 436.765625,
      "completions/mean_terminated_length": 436.765625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0225,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 7.894036769866943,
      "learning_rate": 9.975414512725056e-07,
      "loss": -0.0,
      "num_tokens": 1124009.0,
      "reward": 0.37812498211860657,
      "reward_std": 0.3744494318962097,
      "rewards/format_reward/mean": 0.34375,
      "rewards/format_reward/std": 0.233588308095932,
      "rewards/mcq_exact_match_reward/mean": 0.34375,
      "rewards/mcq_exact_match_reward/std": 0.4787135720252991,
      "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": 0.046875,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1663.0,
      "completions/mean_length": 643.390625,
      "completions/mean_terminated_length": 574.3114624023438,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.025,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.2380053997039795,
      "learning_rate": 9.956320346634875e-07,
      "loss": 0.0,
      "num_tokens": 1252090.0,
      "reward": 0.2578125,
      "reward_std": 0.34344518184661865,
      "rewards/format_reward/mean": 0.390625,
      "rewards/format_reward/std": 0.2741328477859497,
      "rewards/mcq_exact_match_reward/mean": 0.21875,
      "rewards/mcq_exact_match_reward/std": 0.4166666865348816,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1463.0,
      "completions/mean_length": 586.25,
      "completions/mean_terminated_length": 563.0476684570312,
      "completions/min_length": 138.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.0275,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 1.0758506059646606,
      "learning_rate": 9.931806517013612e-07,
      "loss": 0.0,
      "num_tokens": 1367146.0,
      "reward": 0.33124998211860657,
      "reward_std": 0.20131203532218933,
      "rewards/format_reward/mean": 0.34375,
      "rewards/format_reward/std": 0.25,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1568.0,
      "completions/mean_length": 631.640625,
      "completions/mean_terminated_length": 585.9515991210938,
      "completions/min_length": 83.0,
      "completions/min_terminated_length": 83.0,
      "epoch": 0.03,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.425425410270691,
      "learning_rate": 9.901899829374047e-07,
      "loss": 0.0,
      "num_tokens": 1507715.0,
      "reward": 0.11171875149011612,
      "reward_std": 0.17002923786640167,
      "rewards/format_reward/mean": 0.3359375,
      "rewards/format_reward/std": 0.2824873626232147,
      "rewards/mcq_exact_match_reward/mean": 0.078125,
      "rewards/mcq_exact_match_reward/std": 0.27048972249031067,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1735.0,
      "completions/mean_length": 558.578125,
      "completions/mean_terminated_length": 510.5322570800781,
      "completions/min_length": 14.0,
      "completions/min_terminated_length": 14.0,
      "epoch": 0.0325,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.187204599380493,
      "learning_rate": 9.866632986240029e-07,
      "loss": -0.0,
      "num_tokens": 1635680.0,
      "reward": 0.37031248211860657,
      "reward_std": 0.3377445340156555,
      "rewards/format_reward/mean": 0.421875,
      "rewards/format_reward/std": 0.25539806485176086,
      "rewards/mcq_exact_match_reward/mean": 0.328125,
      "rewards/mcq_exact_match_reward/std": 0.4732423722743988,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1957.0,
      "completions/mean_length": 501.96875,
      "completions/mean_terminated_length": 477.4285888671875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.035,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 8.544635772705078,
      "learning_rate": 9.826044551386742e-07,
      "loss": -0.0,
      "num_tokens": 1792854.0,
      "reward": 0.3460937440395355,
      "reward_std": 0.24423694610595703,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.20877929031848907,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "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": 1655.0,
      "completions/mean_length": 515.015625,
      "completions/mean_terminated_length": 490.68255615234375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0375,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 18.065509796142578,
      "learning_rate": 9.780178907671788e-07,
      "loss": 0.0,
      "num_tokens": 1943167.0,
      "reward": 0.22734373807907104,
      "reward_std": 0.24496060609817505,
      "rewards/format_reward/mean": 0.3984375,
      "rewards/format_reward/std": 0.20275264978408813,
      "rewards/mcq_exact_match_reward/mean": 0.1875,
      "rewards/mcq_exact_match_reward/std": 0.39339789748191833,
      "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": 1861.0,
      "completions/mean_length": 524.578125,
      "completions/mean_terminated_length": 475.4354553222656,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.04,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 5.3649091720581055,
      "learning_rate": 9.729086208503173e-07,
      "loss": 0.0,
      "num_tokens": 2081644.0,
      "reward": 0.2523437440395355,
      "reward_std": 0.2687857747077942,
      "rewards/format_reward/mean": 0.3359375,
      "rewards/format_reward/std": 0.25283610820770264,
      "rewards/mcq_exact_match_reward/mean": 0.21875,
      "rewards/mcq_exact_match_reward/std": 0.4166666865348816,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1615.0,
      "completions/mean_length": 474.5,
      "completions/mean_terminated_length": 423.7419128417969,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.0425,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 12.277502059936523,
      "learning_rate": 9.672822322997304e-07,
      "loss": 0.0,
      "num_tokens": 2189796.0,
      "reward": 0.2874999940395355,
      "reward_std": 0.38616982102394104,
      "rewards/format_reward/mean": 0.375,
      "rewards/format_reward/std": 0.29546841979026794,
      "rewards/mcq_exact_match_reward/mean": 0.25,
      "rewards/mcq_exact_match_reward/std": 0.4364357888698578,
      "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": 1229.0,
      "completions/max_terminated_length": 1229.0,
      "completions/mean_length": 512.828125,
      "completions/mean_terminated_length": 512.828125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.045,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.5680460929870605,
      "learning_rate": 9.611448774886923e-07,
      "loss": 0.0,
      "num_tokens": 2299697.0,
      "reward": 0.2945312261581421,
      "reward_std": 0.41407567262649536,
      "rewards/format_reward/mean": 0.4453125,
      "rewards/format_reward/std": 0.20152568817138672,
      "rewards/mcq_exact_match_reward/mean": 0.25,
      "rewards/mcq_exact_match_reward/std": 0.4364357888698578,
      "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": 1012.0,
      "completions/mean_length": 435.46875,
      "completions/mean_terminated_length": 409.873046875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0475,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 14.071467399597168,
      "learning_rate": 9.545032675245813e-07,
      "loss": 0.0,
      "num_tokens": 2418383.0,
      "reward": 0.2304687350988388,
      "reward_std": 0.37052351236343384,
      "rewards/format_reward/mean": 0.4296875,
      "rewards/format_reward/std": 0.1751912236213684,
      "rewards/mcq_exact_match_reward/mean": 0.1875,
      "rewards/mcq_exact_match_reward/std": 0.39339789748191833,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1618.0,
      "completions/mean_length": 563.5,
      "completions/mean_terminated_length": 539.9365234375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.05,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 5.617673873901367,
      "learning_rate": 9.473646649103817e-07,
      "loss": 0.0,
      "num_tokens": 2543495.0,
      "reward": 0.22890624403953552,
      "reward_std": 0.3058883547782898,
      "rewards/format_reward/mean": 0.4140625,
      "rewards/format_reward/std": 0.2605654299259186,
      "rewards/mcq_exact_match_reward/mean": 0.1875,
      "rewards/mcq_exact_match_reward/std": 0.39339789748191833,
      "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": 1907.0,
      "completions/max_terminated_length": 1907.0,
      "completions/mean_length": 525.3125,
      "completions/mean_terminated_length": 525.3125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0525,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 6.890649318695068,
      "learning_rate": 9.397368756032444e-07,
      "loss": 0.0,
      "num_tokens": 2674187.0,
      "reward": 0.35859373211860657,
      "reward_std": 0.3056381344795227,
      "rewards/format_reward/mean": 0.4609375,
      "rewards/format_reward/std": 0.16194961965084076,
      "rewards/mcq_exact_match_reward/mean": 0.3125,
      "rewards/mcq_exact_match_reward/std": 0.467176616191864,
      "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": 1953.0,
      "completions/max_terminated_length": 1953.0,
      "completions/mean_length": 699.734375,
      "completions/mean_terminated_length": 699.734375,
      "completions/min_length": 9.0,
      "completions/min_terminated_length": 9.0,
      "epoch": 0.055,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 1.8531899452209473,
      "learning_rate": 9.316282404787869e-07,
      "loss": 0.0,
      "num_tokens": 2801594.0,
      "reward": 0.32656246423721313,
      "reward_std": 0.31934261322021484,
      "rewards/format_reward/mean": 0.453125,
      "rewards/format_reward/std": 0.23091863095760345,
      "rewards/mcq_exact_match_reward/mean": 0.28125,
      "rewards/mcq_exact_match_reward/std": 0.4531635046005249,
      "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": 1629.0,
      "completions/max_terminated_length": 1629.0,
      "completions/mean_length": 540.625,
      "completions/mean_terminated_length": 540.625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0575,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 3.7736644744873047,
      "learning_rate": 9.230476262104676e-07,
      "loss": -0.0,
      "num_tokens": 2918322.0,
      "reward": 0.48515623807907104,
      "reward_std": 0.3816661238670349,
      "rewards/format_reward/mean": 0.4765625,
      "rewards/format_reward/std": 0.13886408507823944,
      "rewards/mcq_exact_match_reward/mean": 0.4375,
      "rewards/mcq_exact_match_reward/std": 0.5,
      "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": 1274.0,
      "completions/max_terminated_length": 1274.0,
      "completions/mean_length": 436.390625,
      "completions/mean_terminated_length": 436.390625,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "epoch": 0.06,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 4.89503812789917,
      "learning_rate": 9.1400441557401e-07,
      "loss": -0.0,
      "num_tokens": 3018011.0,
      "reward": 0.3703124523162842,
      "reward_std": 0.3083305358886719,
      "rewards/format_reward/mean": 0.421875,
      "rewards/format_reward/std": 0.18298126757144928,
      "rewards/mcq_exact_match_reward/mean": 0.328125,
      "rewards/mcq_exact_match_reward/std": 0.4732423722743988,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1513.0,
      "completions/mean_length": 645.078125,
      "completions/mean_terminated_length": 622.8095703125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0625,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 3.7173430919647217,
      "learning_rate": 9.045084971874737e-07,
      "loss": -0.0,
      "num_tokens": 3163408.0,
      "reward": 0.39921870827674866,
      "reward_std": 0.38945621252059937,
      "rewards/format_reward/mean": 0.3984375,
      "rewards/format_reward/std": 0.22146137058734894,
      "rewards/mcq_exact_match_reward/mean": 0.359375,
      "rewards/mcq_exact_match_reward/std": 0.4836103618144989,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1519.0,
      "completions/mean_length": 522.21875,
      "completions/mean_terminated_length": 498.0000305175781,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.065,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 4.034616470336914,
      "learning_rate": 8.945702546981968e-07,
      "loss": -0.0,
      "num_tokens": 3294646.0,
      "reward": 0.33828121423721313,
      "reward_std": 0.45082423090934753,
      "rewards/format_reward/mean": 0.4140625,
      "rewards/format_reward/std": 0.2280818521976471,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1961.0,
      "completions/mean_length": 512.75,
      "completions/mean_terminated_length": 488.3809814453125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0675,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 8.771376609802246,
      "learning_rate": 8.842005554284295e-07,
      "loss": 0.0,
      "num_tokens": 3412702.0,
      "reward": 0.2867187261581421,
      "reward_std": 0.2777417004108429,
      "rewards/format_reward/mean": 0.5234375,
      "rewards/format_reward/std": 0.28770697116851807,
      "rewards/mcq_exact_match_reward/mean": 0.234375,
      "rewards/mcq_exact_match_reward/std": 0.42695629596710205,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1359.0,
      "completions/mean_length": 565.421875,
      "completions/mean_terminated_length": 517.5967407226562,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.07,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 4.636484146118164,
      "learning_rate": 8.734107384920769e-07,
      "loss": 0.0,
      "num_tokens": 3559089.0,
      "reward": 0.4468749761581421,
      "reward_std": 0.3345656991004944,
      "rewards/format_reward/mean": 0.40625,
      "rewards/format_reward/std": 0.21593283116817474,
      "rewards/mcq_exact_match_reward/mean": 0.40625,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1133.0,
      "completions/mean_length": 431.734375,
      "completions/mean_terminated_length": 406.0793762207031,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0725,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 2.271466016769409,
      "learning_rate": 8.622126023955445e-07,
      "loss": 0.0,
      "num_tokens": 3666240.0,
      "reward": 0.3164062201976776,
      "reward_std": 0.274751216173172,
      "rewards/format_reward/mean": 0.5078125,
      "rewards/format_reward/std": 0.20877929031848907,
      "rewards/mcq_exact_match_reward/mean": 0.265625,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "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": 1883.0,
      "completions/max_terminated_length": 1883.0,
      "completions/mean_length": 571.40625,
      "completions/mean_terminated_length": 571.40625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.075,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 5.261892318725586,
      "learning_rate": 8.506183921362442e-07,
      "loss": -0.0,
      "num_tokens": 3787794.0,
      "reward": 0.5640624761581421,
      "reward_std": 0.40072205662727356,
      "rewards/format_reward/mean": 0.484375,
      "rewards/format_reward/std": 0.125,
      "rewards/mcq_exact_match_reward/mean": 0.515625,
      "rewards/mcq_exact_match_reward/std": 0.5037065148353577,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1957.0,
      "completions/mean_length": 592.296875,
      "completions/mean_terminated_length": 569.1904907226562,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0775,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 2.3355727195739746,
      "learning_rate": 8.386407858128706e-07,
      "loss": 0.0,
      "num_tokens": 3928581.0,
      "reward": 0.1875,
      "reward_std": 0.2679750919342041,
      "rewards/format_reward/mean": 0.46875,
      "rewards/format_reward/std": 0.21593283116817474,
      "rewards/mcq_exact_match_reward/mean": 0.140625,
      "rewards/mcq_exact_match_reward/std": 0.3503824472427368,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1809.0,
      "completions/mean_length": 631.125,
      "completions/mean_terminated_length": 608.6349487304688,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.08,
      "frac_reward_zero_std": 0.25,
      "grad_norm": 13.961491584777832,
      "learning_rate": 8.262928807620843e-07,
      "loss": -0.0,
      "num_tokens": 4040245.0,
      "reward": 0.6007812023162842,
      "reward_std": 0.30588001012802124,
      "rewards/format_reward/mean": 0.5390625,
      "rewards/format_reward/std": 0.29953256249427795,
      "rewards/mcq_exact_match_reward/mean": 0.546875,
      "rewards/mcq_exact_match_reward/std": 0.501733124256134,
      "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": 0.046875,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1644.0,
      "completions/mean_length": 785.359375,
      "completions/mean_terminated_length": 723.2622680664062,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0825,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 9.326311111450195,
      "learning_rate": 8.135881792367685e-07,
      "loss": -0.0,
      "num_tokens": 4174364.0,
      "reward": 0.31484371423721313,
      "reward_std": 0.46168267726898193,
      "rewards/format_reward/mean": 0.4921875,
      "rewards/format_reward/std": 0.24384792149066925,
      "rewards/mcq_exact_match_reward/mean": 0.265625,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "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": 1089.0,
      "completions/max_terminated_length": 1089.0,
      "completions/mean_length": 250.96875,
      "completions/mean_terminated_length": 250.96875,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.085,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 17.84054946899414,
      "learning_rate": 8.005405736415125e-07,
      "loss": -0.0,
      "num_tokens": 4269002.0,
      "reward": 0.6109374761581421,
      "reward_std": 0.33161211013793945,
      "rewards/format_reward/mean": 0.484375,
      "rewards/format_reward/std": 0.1534975916147232,
      "rewards/mcq_exact_match_reward/mean": 0.5625,
      "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": 1405.0,
      "completions/max_terminated_length": 1405.0,
      "completions/mean_length": 385.0,
      "completions/mean_terminated_length": 385.0,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0875,
      "frac_reward_zero_std": 0.25,
      "grad_norm": 9.34335994720459,
      "learning_rate": 7.871643313414718e-07,
      "loss": 0.0,
      "num_tokens": 4384322.0,
      "reward": 0.3531249761581421,
      "reward_std": 0.3210226893424988,
      "rewards/format_reward/mean": 0.5625,
      "rewards/format_reward/std": 0.208927720785141,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 843.0,
      "completions/mean_length": 298.609375,
      "completions/mean_terminated_length": 242.1774139404297,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.09,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 9.821782112121582,
      "learning_rate": 7.734740790612136e-07,
      "loss": -0.0,
      "num_tokens": 4496857.0,
      "reward": 0.4273437261581421,
      "reward_std": 0.36625146865844727,
      "rewards/format_reward/mean": 0.5234375,
      "rewards/format_reward/std": 0.1875000149011612,
      "rewards/mcq_exact_match_reward/mean": 0.375,
      "rewards/mcq_exact_match_reward/std": 0.48795005679130554,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1628.0,
      "completions/mean_length": 588.8125,
      "completions/mean_terminated_length": 541.741943359375,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.0925,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 45.789329528808594,
      "learning_rate": 7.594847868906076e-07,
      "loss": 0.0,
      "num_tokens": 4618269.0,
      "reward": 0.4976562261581421,
      "reward_std": 0.46462389826774597,
      "rewards/format_reward/mean": 0.4453125,
      "rewards/format_reward/std": 0.15728822350502014,
      "rewards/mcq_exact_match_reward/mean": 0.453125,
      "rewards/mcq_exact_match_reward/std": 0.501733124256134,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1805.0,
      "completions/mean_length": 506.859375,
      "completions/mean_terminated_length": 482.3968505859375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.095,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 11.930426597595215,
      "learning_rate": 7.452117519152541e-07,
      "loss": 0.0,
      "num_tokens": 4767964.0,
      "reward": 0.26484376192092896,
      "reward_std": 0.33171117305755615,
      "rewards/format_reward/mean": 0.4609375,
      "rewards/format_reward/std": 0.13524486124515533,
      "rewards/mcq_exact_match_reward/mean": 0.21875,
      "rewards/mcq_exact_match_reward/std": 0.4166666865348816,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1677.0,
      "completions/mean_length": 461.21875,
      "completions/mean_terminated_length": 410.0322570800781,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.0975,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 7.486371994018555,
      "learning_rate": 7.306705814893439e-07,
      "loss": -0.0,
      "num_tokens": 4896738.0,
      "reward": 0.38203123211860657,
      "reward_std": 0.38516172766685486,
      "rewards/format_reward/mean": 0.5390625,
      "rewards/format_reward/std": 0.33694106340408325,
      "rewards/mcq_exact_match_reward/mean": 0.328125,
      "rewards/mcq_exact_match_reward/std": 0.4732423722743988,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 2025.0,
      "completions/mean_length": 707.28125,
      "completions/mean_terminated_length": 664.0322265625,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "epoch": 0.1,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.6023790836334229,
      "learning_rate": 7.158771761692464e-07,
      "loss": -0.0,
      "num_tokens": 5045348.0,
      "reward": 0.3242187201976776,
      "reward_std": 0.32094645500183105,
      "rewards/format_reward/mean": 0.5859375,
      "rewards/format_reward/std": 0.27537402510643005,
      "rewards/mcq_exact_match_reward/mean": 0.265625,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "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": 0.03125,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1170.0,
      "completions/mean_length": 496.59375,
      "completions/mean_terminated_length": 446.5483703613281,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.1025,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 5.383449554443359,
      "learning_rate": 7.008477123264847e-07,
      "loss": 0.0,
      "num_tokens": 5171682.0,
      "reward": 0.4585937559604645,
      "reward_std": 0.3189367353916168,
      "rewards/format_reward/mean": 0.5234375,
      "rewards/format_reward/std": 0.28770697116851807,
      "rewards/mcq_exact_match_reward/mean": 0.40625,
      "rewards/mcq_exact_match_reward/std": 0.49501484632492065,
      "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": 1085.0,
      "completions/mean_length": 292.515625,
      "completions/mean_terminated_length": 264.65081787109375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.105,
      "frac_reward_zero_std": 0.25,
      "grad_norm": 7.30802583694458,
      "learning_rate": 6.855986244591103e-07,
      "loss": 0.0,
      "num_tokens": 5266507.0,
      "reward": 0.2679687440395355,
      "reward_std": 0.21786025166511536,
      "rewards/format_reward/mean": 0.6484375,
      "rewards/format_reward/std": 0.24688033759593964,
      "rewards/mcq_exact_match_reward/mean": 0.203125,
      "rewards/mcq_exact_match_reward/std": 0.40550529956817627,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1765.0,
      "completions/mean_length": 466.4375,
      "completions/mean_terminated_length": 441.3333740234375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.1075,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 8.825201034545898,
      "learning_rate": 6.701465872208216e-07,
      "loss": -0.0,
      "num_tokens": 5418023.0,
      "reward": 0.4687499701976776,
      "reward_std": 0.44940704107284546,
      "rewards/format_reward/mean": 0.46875,
      "rewards/format_reward/std": 0.279951810836792,
      "rewards/mcq_exact_match_reward/mean": 0.421875,
      "rewards/mcq_exact_match_reward/std": 0.49776285886764526,
      "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": 1946.0,
      "completions/max_terminated_length": 1946.0,
      "completions/mean_length": 516.765625,
      "completions/mean_terminated_length": 516.765625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.11,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 12.140402793884277,
      "learning_rate": 6.545084971874736e-07,
      "loss": 0.0,
      "num_tokens": 5564008.0,
      "reward": 0.2554687261581421,
      "reward_std": 0.34813106060028076,
      "rewards/format_reward/mean": 0.5234375,
      "rewards/format_reward/std": 0.22589658200740814,
      "rewards/mcq_exact_match_reward/mean": 0.203125,
      "rewards/mcq_exact_match_reward/std": 0.40550529956817627,
      "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": 1668.0,
      "completions/max_terminated_length": 1668.0,
      "completions/mean_length": 471.59375,
      "completions/mean_terminated_length": 471.59375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.1125,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 8.025932312011719,
      "learning_rate": 6.387014543809223e-07,
      "loss": -0.0,
      "num_tokens": 5671670.0,
      "reward": 0.4429687261581421,
      "reward_std": 0.36176156997680664,
      "rewards/format_reward/mean": 0.6796875,
      "rewards/format_reward/std": 0.31329265236854553,
      "rewards/mcq_exact_match_reward/mean": 0.375,
      "rewards/mcq_exact_match_reward/std": 0.48795005679130554,
      "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": 1328.0,
      "completions/max_terminated_length": 1328.0,
      "completions/mean_length": 517.34375,
      "completions/mean_terminated_length": 517.34375,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.115,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 9.665616989135742,
      "learning_rate": 6.227427435703995e-07,
      "loss": 0.0,
      "num_tokens": 5811316.0,
      "reward": 0.3515625,
      "reward_std": 0.301347553730011,
      "rewards/format_reward/mean": 0.546875,
      "rewards/format_reward/std": 0.2630521357059479,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1520.0,
      "completions/mean_length": 422.984375,
      "completions/mean_terminated_length": 397.19049072265625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.1175,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 8.625753402709961,
      "learning_rate": 6.066498153718734e-07,
      "loss": 0.0,
      "num_tokens": 5928691.0,
      "reward": 0.33671873807907104,
      "reward_std": 0.3822405934333801,
      "rewards/format_reward/mean": 0.7109375,
      "rewards/format_reward/std": 0.3060849606990814,
      "rewards/mcq_exact_match_reward/mean": 0.265625,
      "rewards/mcq_exact_match_reward/std": 0.44515693187713623,
      "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": 0.015625,
      "completions/max_length": 2048.0,
      "completions/max_terminated_length": 1758.0,
      "completions/mean_length": 532.25,
      "completions/mean_terminated_length": 508.19049072265625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.12,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 3.9692816734313965,
      "learning_rate": 5.90440267166055e-07,
      "loss": -0.0,
      "num_tokens": 6060691.0,
      "reward": 0.22343748807907104,
      "reward_std": 0.3473894000053406,
      "rewards/format_reward/mean": 0.515625,
      "rewards/format_reward/std": 0.2812775671482086,
      "rewards/mcq_exact_match_reward/mean": 0.171875,
      "rewards/mcq_exact_match_reward/std": 0.38025420904159546,
      "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": 1096.0,
      "completions/max_terminated_length": 1096.0,
      "completions/mean_length": 323.53125,
      "completions/mean_terminated_length": 323.53125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.1225,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 10.702794075012207,
      "learning_rate": 5.741318238559209e-07,
      "loss": -0.0,
      "num_tokens": 6188885.0,
      "reward": 0.35078123211860657,
      "reward_std": 0.39584439992904663,
      "rewards/format_reward/mean": 0.5390625,
      "rewards/format_reward/std": 0.2236899733543396,
      "rewards/mcq_exact_match_reward/mean": 0.296875,
      "rewards/mcq_exact_match_reward/std": 0.4604927599430084,
      "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": 1617.0,
      "completions/max_terminated_length": 1617.0,
      "completions/mean_length": 418.078125,
      "completions/mean_terminated_length": 418.078125,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.125,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 6.304938316345215,
      "learning_rate": 5.577423184847931e-07,
      "loss": 0.0,
      "num_tokens": 6307218.0,
      "reward": 0.38359373807907104,
      "reward_std": 0.42584139108657837,
      "rewards/format_reward/mean": 0.7109375,
      "rewards/format_reward/std": 0.34277912974357605,
      "rewards/mcq_exact_match_reward/mean": 0.3125,
      "rewards/mcq_exact_match_reward/std": 0.467176616191864,
      "step": 50
    }
  ],
  "logging_steps": 1,
  "max_steps": 100,
  "num_input_tokens_seen": 6307218,
  "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
}