{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9945822640433419,
  "eval_steps": 500,
  "global_step": 109,
  "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": -5.0,
      "completions/max_length": 982.0,
      "completions/max_terminated_length": 982.0,
      "completions/mean_length": 520.5390625,
      "completions/mean_terminated_length": 520.5390625,
      "completions/min_length": 302.0,
      "completions/min_terminated_length": 302.0,
      "entropy": 0.114654541015625,
      "epoch": 0.009124607927003137,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.0243235994130373,
      "kl": 0.00023239850997924805,
      "learning_rate": 2e-05,
      "loss": -0.0022,
      "num_tokens": 408495.0,
      "reward": 0.963108241558075,
      "reward_std": 0.2168295979499817,
      "rewards/cross_solution_unittest_reward/mean": 0.48155416548252106,
      "rewards/cross_solution_unittest_reward/std": 0.22725073248147964,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1246.0,
      "completions/max_terminated_length": 1246.0,
      "completions/mean_length": 493.14324951171875,
      "completions/mean_terminated_length": 493.14324951171875,
      "completions/min_length": 200.0,
      "completions/min_terminated_length": 200.0,
      "entropy": 0.1129150390625,
      "epoch": 0.018249215854006275,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02810436487197876,
      "kl": 0.00023671984672546387,
      "learning_rate": 1.9995921928281893e-05,
      "loss": -0.004,
      "num_tokens": 819398.0,
      "reward": 0.7592691779136658,
      "reward_std": 0.20917575061321259,
      "rewards/cross_solution_unittest_reward/mean": 0.3796345815062523,
      "rewards/cross_solution_unittest_reward/std": 0.2533634677529335,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1210.0,
      "completions/max_terminated_length": 1210.0,
      "completions/mean_length": 584.3802490234375,
      "completions/mean_terminated_length": 584.3802490234375,
      "completions/min_length": 240.0,
      "completions/min_terminated_length": 240.0,
      "entropy": 0.09423828125,
      "epoch": 0.02737382378100941,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.024342378601431847,
      "kl": 0.00020760297775268555,
      "learning_rate": 1.9983691039261358e-05,
      "loss": -0.0025,
      "num_tokens": 1264568.0,
      "reward": 0.8243939280509949,
      "reward_std": 0.19790497422218323,
      "rewards/cross_solution_unittest_reward/mean": 0.41219695657491684,
      "rewards/cross_solution_unittest_reward/std": 0.20426170527935028,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.96875,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 902.0,
      "completions/mean_length": 510.72137451171875,
      "completions/mean_terminated_length": 494.6308898925781,
      "completions/min_length": 293.0,
      "completions/min_terminated_length": 293.0,
      "entropy": 0.12127685546875,
      "epoch": 0.03649843170801255,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02658909745514393,
      "kl": 0.00026023387908935547,
      "learning_rate": 1.9963317308626916e-05,
      "loss": -0.0008,
      "num_tokens": 1707901.0,
      "reward": 0.6146944165229797,
      "reward_std": 0.20338332653045654,
      "rewards/cross_solution_unittest_reward/mean": 0.30734721571207047,
      "rewards/cross_solution_unittest_reward/std": 0.1820686012506485,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1694.0,
      "completions/mean_length": 495.9921875,
      "completions/mean_terminated_length": 487.92950439453125,
      "completions/min_length": 228.0,
      "completions/min_terminated_length": 228.0,
      "entropy": 0.1229400634765625,
      "epoch": 0.04562303963501568,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03194279596209526,
      "kl": 0.0002792775630950928,
      "learning_rate": 1.99348173534855e-05,
      "loss": 0.0193,
      "num_tokens": 2092986.0,
      "reward": 0.8980245590209961,
      "reward_std": 0.2468685805797577,
      "rewards/cross_solution_unittest_reward/mean": 0.44901224970817566,
      "rewards/cross_solution_unittest_reward/std": 0.2793443500995636,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.96875,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1340.0,
      "completions/mean_length": 548.34375,
      "completions/mean_terminated_length": 532.4502563476562,
      "completions/min_length": 297.0,
      "completions/min_terminated_length": 297.0,
      "entropy": 0.1123809814453125,
      "epoch": 0.05474764756201882,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.024234920740127563,
      "kl": 0.0002339184284210205,
      "learning_rate": 1.989821441880933e-05,
      "loss": 0.0085,
      "num_tokens": 2517798.0,
      "reward": 0.8548812866210938,
      "reward_std": 0.16416706144809723,
      "rewards/cross_solution_unittest_reward/mean": 0.42744066566228867,
      "rewards/cross_solution_unittest_reward/std": 0.24477335810661316,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 3107.0,
      "completions/max_terminated_length": 3107.0,
      "completions/mean_length": 531.8958740234375,
      "completions/mean_terminated_length": 531.8958740234375,
      "completions/min_length": 232.0,
      "completions/min_terminated_length": 232.0,
      "entropy": 0.1142120361328125,
      "epoch": 0.06387225548902195,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.025173544883728027,
      "kl": 0.00025781989097595215,
      "learning_rate": 1.9853538358476933e-05,
      "loss": -0.0049,
      "num_tokens": 2962550.0,
      "reward": 0.8058516383171082,
      "reward_std": 0.26573851704597473,
      "rewards/cross_solution_unittest_reward/mean": 0.4029258117079735,
      "rewards/cross_solution_unittest_reward/std": 0.2175045683979988,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1891.0,
      "completions/max_terminated_length": 1891.0,
      "completions/mean_length": 522.03125,
      "completions/mean_terminated_length": 522.03125,
      "completions/min_length": 214.0,
      "completions/min_terminated_length": 214.0,
      "entropy": 0.1124114990234375,
      "epoch": 0.0729968634160251,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.025463413447141647,
      "kl": 0.00025027990341186523,
      "learning_rate": 1.9800825610923937e-05,
      "loss": 0.0018,
      "num_tokens": 3417666.0,
      "reward": 0.7429351210594177,
      "reward_std": 0.21776002645492554,
      "rewards/cross_solution_unittest_reward/mean": 0.37146755307912827,
      "rewards/cross_solution_unittest_reward/std": 0.2737540230154991,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1064.0,
      "completions/mean_length": 495.703125,
      "completions/mean_terminated_length": 487.6396789550781,
      "completions/min_length": 250.0,
      "completions/min_terminated_length": 250.0,
      "entropy": 0.12152099609375,
      "epoch": 0.08212147134302823,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02767878770828247,
      "kl": 0.0002829134464263916,
      "learning_rate": 1.9740119169423337e-05,
      "loss": -0.0138,
      "num_tokens": 3811456.0,
      "reward": 0.8020304441452026,
      "reward_std": 0.24267295002937317,
      "rewards/cross_solution_unittest_reward/mean": 0.40101519227027893,
      "rewards/cross_solution_unittest_reward/std": 0.21552012115716934,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 994.0,
      "completions/max_terminated_length": 994.0,
      "completions/mean_length": 502.54168701171875,
      "completions/mean_terminated_length": 502.54168701171875,
      "completions/min_length": 250.0,
      "completions/min_terminated_length": 250.0,
      "entropy": 0.1065216064453125,
      "epoch": 0.09124607927003137,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.029218332841992378,
      "kl": 0.0002897977828979492,
      "learning_rate": 1.9671468547019575e-05,
      "loss": 0.0047,
      "num_tokens": 4254672.0,
      "reward": 0.8215359449386597,
      "reward_std": 0.20994165539741516,
      "rewards/cross_solution_unittest_reward/mean": 0.41076793521642685,
      "rewards/cross_solution_unittest_reward/std": 0.2297622635960579,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1138.0,
      "completions/max_terminated_length": 1138.0,
      "completions/mean_length": 546.2109375,
      "completions/mean_terminated_length": 546.2109375,
      "completions/min_length": 264.0,
      "completions/min_terminated_length": 264.0,
      "entropy": 0.1016082763671875,
      "epoch": 0.1003706871970345,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02588176354765892,
      "kl": 0.00028312206268310547,
      "learning_rate": 1.9594929736144978e-05,
      "loss": 0.0049,
      "num_tokens": 4680241.0,
      "reward": 0.829417884349823,
      "reward_std": 0.19322270154953003,
      "rewards/cross_solution_unittest_reward/mean": 0.4147089347243309,
      "rewards/cross_solution_unittest_reward/std": 0.20234351605176926,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.96875,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 907.0,
      "completions/mean_length": 506.40625,
      "completions/mean_terminated_length": 490.293212890625,
      "completions/min_length": 255.0,
      "completions/min_terminated_length": 255.0,
      "entropy": 0.1226959228515625,
      "epoch": 0.10949529512403763,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03448358178138733,
      "kl": 0.00037539005279541016,
      "learning_rate": 1.9510565162951538e-05,
      "loss": -0.0022,
      "num_tokens": 5112749.0,
      "reward": 0.6837499141693115,
      "reward_std": 0.23883536458015442,
      "rewards/cross_solution_unittest_reward/mean": 0.34187494963407516,
      "rewards/cross_solution_unittest_reward/std": 0.24100807309150696,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1035.0,
      "completions/max_terminated_length": 1035.0,
      "completions/mean_length": 534.4818115234375,
      "completions/mean_terminated_length": 534.4818115234375,
      "completions/min_length": 247.0,
      "completions/min_terminated_length": 247.0,
      "entropy": 0.1075286865234375,
      "epoch": 0.11861990305104078,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.027723070234060287,
      "kl": 0.0003420412540435791,
      "learning_rate": 1.941844363639525e-05,
      "loss": -0.0135,
      "num_tokens": 5525534.0,
      "reward": 0.8237327337265015,
      "reward_std": 0.19265763461589813,
      "rewards/cross_solution_unittest_reward/mean": 0.41186633706092834,
      "rewards/cross_solution_unittest_reward/std": 0.24469294399023056,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 2351.0,
      "completions/max_terminated_length": 2351.0,
      "completions/mean_length": 615.5104370117188,
      "completions/mean_terminated_length": 615.5104370117188,
      "completions/min_length": 281.0,
      "completions/min_terminated_length": 281.0,
      "entropy": 0.1039581298828125,
      "epoch": 0.1277445109780439,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.025678865611553192,
      "kl": 0.00036969780921936035,
      "learning_rate": 1.9318640292114526e-05,
      "loss": -0.0176,
      "num_tokens": 5993994.0,
      "reward": 0.8750783205032349,
      "reward_std": 0.16023778915405273,
      "rewards/cross_solution_unittest_reward/mean": 0.43753916025161743,
      "rewards/cross_solution_unittest_reward/std": 0.25390665233135223,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 826.0,
      "completions/mean_length": 467.890625,
      "completions/mean_terminated_length": 459.75457763671875,
      "completions/min_length": 211.0,
      "completions/min_terminated_length": 211.0,
      "entropy": 0.122222900390625,
      "epoch": 0.13686911890504705,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030205531045794487,
      "kl": 0.00045168399810791016,
      "learning_rate": 1.92112365311485e-05,
      "loss": 0.0082,
      "num_tokens": 6339752.0,
      "reward": 0.8696179389953613,
      "reward_std": 0.2131870836019516,
      "rewards/cross_solution_unittest_reward/mean": 0.43480896949768066,
      "rewards/cross_solution_unittest_reward/std": 0.2413550242781639,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 3274.0,
      "completions/max_terminated_length": 3274.0,
      "completions/mean_length": 536.6796875,
      "completions/mean_terminated_length": 536.6796875,
      "completions/min_length": 279.0,
      "completions/min_terminated_length": 279.0,
      "entropy": 0.1075592041015625,
      "epoch": 0.1459937268320502,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.027554774656891823,
      "kl": 0.0004489421844482422,
      "learning_rate": 1.9096319953545186e-05,
      "loss": 0.001,
      "num_tokens": 6744981.0,
      "reward": 0.8156441450119019,
      "reward_std": 0.22033701837062836,
      "rewards/cross_solution_unittest_reward/mean": 0.40782201290130615,
      "rewards/cross_solution_unittest_reward/std": 0.19903528690338135,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1317.0,
      "completions/mean_length": 522.8880615234375,
      "completions/mean_terminated_length": 514.8955688476562,
      "completions/min_length": 253.0,
      "completions/min_terminated_length": 253.0,
      "entropy": 0.1123504638671875,
      "epoch": 0.15511833475905332,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.027508998289704323,
      "kl": 0.0005013346672058105,
      "learning_rate": 1.8973984286913584e-05,
      "loss": 0.0082,
      "num_tokens": 7159690.0,
      "reward": 0.9808885455131531,
      "reward_std": 0.23347638547420502,
      "rewards/cross_solution_unittest_reward/mean": 0.49044425785541534,
      "rewards/cross_solution_unittest_reward/std": 0.28279364109039307,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1142.0,
      "completions/max_terminated_length": 1142.0,
      "completions/mean_length": 522.3984375,
      "completions/mean_terminated_length": 522.3984375,
      "completions/min_length": 242.0,
      "completions/min_terminated_length": 242.0,
      "entropy": 0.1051788330078125,
      "epoch": 0.16424294268605646,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.027137286961078644,
      "kl": 0.0005218386650085449,
      "learning_rate": 1.8844329309978146e-05,
      "loss": -0.0001,
      "num_tokens": 7590731.0,
      "reward": 0.8454278707504272,
      "reward_std": 0.19084712862968445,
      "rewards/cross_solution_unittest_reward/mean": 0.42271392047405243,
      "rewards/cross_solution_unittest_reward/std": 0.22437505424022675,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 928.0,
      "completions/mean_length": 507.03125,
      "completions/mean_terminated_length": 498.9974060058594,
      "completions/min_length": 227.0,
      "completions/min_terminated_length": 227.0,
      "entropy": 0.1159210205078125,
      "epoch": 0.17336755061305958,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03020695596933365,
      "kl": 0.0006084442138671875,
      "learning_rate": 1.8707460771197773e-05,
      "loss": 0.006,
      "num_tokens": 8003919.0,
      "reward": 0.7663086652755737,
      "reward_std": 0.24537038803100586,
      "rewards/cross_solution_unittest_reward/mean": 0.38315432518720627,
      "rewards/cross_solution_unittest_reward/std": 0.2557295635342598,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1019.0,
      "completions/max_terminated_length": 1019.0,
      "completions/mean_length": 534.296875,
      "completions/mean_terminated_length": 534.296875,
      "completions/min_length": 272.0,
      "completions/min_terminated_length": 272.0,
      "entropy": 0.1091766357421875,
      "epoch": 0.18249215854006273,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03100455552339554,
      "kl": 0.0006772875785827637,
      "learning_rate": 1.856349030251589e-05,
      "loss": -0.0103,
      "num_tokens": 8459457.0,
      "reward": 0.7904119491577148,
      "reward_std": 0.23983600735664368,
      "rewards/cross_solution_unittest_reward/mean": 0.39520594477653503,
      "rewards/cross_solution_unittest_reward/std": 0.23401054739952087,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 846.0,
      "completions/max_terminated_length": 846.0,
      "completions/mean_length": 485.5390625,
      "completions/mean_terminated_length": 485.5390625,
      "completions/min_length": 236.0,
      "completions/min_terminated_length": 236.0,
      "entropy": 0.10430908203125,
      "epoch": 0.19161676646706588,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.032105084508657455,
      "kl": 0.0006991028785705566,
      "learning_rate": 1.8412535328311813e-05,
      "loss": -0.0052,
      "num_tokens": 8844608.0,
      "reward": 0.8521513938903809,
      "reward_std": 0.21744942665100098,
      "rewards/cross_solution_unittest_reward/mean": 0.4260757490992546,
      "rewards/cross_solution_unittest_reward/std": 0.21132270246744156,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1558.0,
      "completions/mean_length": 513.4479370117188,
      "completions/mean_terminated_length": 505.4308166503906,
      "completions/min_length": 242.0,
      "completions/min_terminated_length": 242.0,
      "entropy": 0.1094970703125,
      "epoch": 0.200741374394069,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.029629480093717575,
      "kl": 0.0006964206695556641,
      "learning_rate": 1.825471896962774e-05,
      "loss": -0.0126,
      "num_tokens": 9223148.0,
      "reward": 0.8136462569236755,
      "reward_std": 0.21329909563064575,
      "rewards/cross_solution_unittest_reward/mean": 0.4068231135606766,
      "rewards/cross_solution_unittest_reward/std": 0.24406656622886658,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1091.0,
      "completions/mean_length": 513.3724365234375,
      "completions/mean_terminated_length": 505.3551025390625,
      "completions/min_length": 243.0,
      "completions/min_terminated_length": 243.0,
      "entropy": 0.1093292236328125,
      "epoch": 0.20986598232107215,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.028564205393195152,
      "kl": 0.0008405447006225586,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 0.0105,
      "num_tokens": 9671827.0,
      "reward": 0.9153262376785278,
      "reward_std": 0.19782906770706177,
      "rewards/cross_solution_unittest_reward/mean": 0.45766307413578033,
      "rewards/cross_solution_unittest_reward/std": 0.23260264098644257,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 970.0,
      "completions/max_terminated_length": 970.0,
      "completions/mean_length": 499.43231201171875,
      "completions/mean_terminated_length": 499.43231201171875,
      "completions/min_length": 221.0,
      "completions/min_terminated_length": 221.0,
      "entropy": 0.1064300537109375,
      "epoch": 0.21899059024807527,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02926481142640114,
      "kl": 0.0008217096328735352,
      "learning_rate": 1.7919022459222754e-05,
      "loss": 0.0034,
      "num_tokens": 10091281.0,
      "reward": 0.7448365092277527,
      "reward_std": 0.2386973798274994,
      "rewards/cross_solution_unittest_reward/mean": 0.37241824716329575,
      "rewards/cross_solution_unittest_reward/std": 0.2582271620631218,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1083.0,
      "completions/mean_length": 496.18231201171875,
      "completions/mean_terminated_length": 488.1201171875,
      "completions/min_length": 278.0,
      "completions/min_terminated_length": 278.0,
      "entropy": 0.11749267578125,
      "epoch": 0.22811519817507842,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03824177011847496,
      "kl": 0.001001119613647461,
      "learning_rate": 1.7741416106390828e-05,
      "loss": 0.0011,
      "num_tokens": 10495487.0,
      "reward": 0.8449365496635437,
      "reward_std": 0.20980952680110931,
      "rewards/cross_solution_unittest_reward/mean": 0.42246825993061066,
      "rewards/cross_solution_unittest_reward/std": 0.21189714968204498,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 848.0,
      "completions/max_terminated_length": 848.0,
      "completions/mean_length": 497.1458435058594,
      "completions/mean_terminated_length": 497.1458435058594,
      "completions/min_length": 214.0,
      "completions/min_terminated_length": 214.0,
      "entropy": 0.11029052734375,
      "epoch": 0.23723980610208156,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.028777848929166794,
      "kl": 0.0008819103240966797,
      "learning_rate": 1.7557495743542586e-05,
      "loss": -0.0032,
      "num_tokens": 10853559.0,
      "reward": 0.7521219253540039,
      "reward_std": 0.18529602885246277,
      "rewards/cross_solution_unittest_reward/mean": 0.37606097012758255,
      "rewards/cross_solution_unittest_reward/std": 0.24145866930484772,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 766.0,
      "completions/max_terminated_length": 766.0,
      "completions/mean_length": 468.05731201171875,
      "completions/mean_terminated_length": 468.05731201171875,
      "completions/min_length": 253.0,
      "completions/min_terminated_length": 253.0,
      "entropy": 0.1114501953125,
      "epoch": 0.24636441402908468,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02834751084446907,
      "kl": 0.0009758472442626953,
      "learning_rate": 1.736741137876405e-05,
      "loss": -0.0084,
      "num_tokens": 11230181.0,
      "reward": 0.8723708987236023,
      "reward_std": 0.19247429072856903,
      "rewards/cross_solution_unittest_reward/mean": 0.4361855015158653,
      "rewards/cross_solution_unittest_reward/std": 0.21063780039548874,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1015.0,
      "completions/mean_length": 507.0989685058594,
      "completions/mean_terminated_length": 499.0652770996094,
      "completions/min_length": 234.0,
      "completions/min_terminated_length": 234.0,
      "entropy": 0.1055908203125,
      "epoch": 0.2554890219560878,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.029102731496095657,
      "kl": 0.0011304616928100586,
      "learning_rate": 1.7171318047589637e-05,
      "loss": -0.0081,
      "num_tokens": 11663099.0,
      "reward": 0.9932507276535034,
      "reward_std": 0.18286040425300598,
      "rewards/cross_solution_unittest_reward/mean": 0.4966253191232681,
      "rewards/cross_solution_unittest_reward/std": 0.2659049928188324,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1139.0,
      "completions/mean_length": 489.8515625,
      "completions/mean_terminated_length": 481.7728576660156,
      "completions/min_length": 275.0,
      "completions/min_terminated_length": 275.0,
      "entropy": 0.113128662109375,
      "epoch": 0.26461362988309095,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030337095260620117,
      "kl": 0.0015281438827514648,
      "learning_rate": 1.696937568655294e-05,
      "loss": 0.0098,
      "num_tokens": 12059194.0,
      "reward": 0.8777666091918945,
      "reward_std": 0.2015705108642578,
      "rewards/cross_solution_unittest_reward/mean": 0.43888331949710846,
      "rewards/cross_solution_unittest_reward/std": 0.289621040225029,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 863.0,
      "completions/max_terminated_length": 863.0,
      "completions/mean_length": 518.9349365234375,
      "completions/mean_terminated_length": 518.9349365234375,
      "completions/min_length": 249.0,
      "completions/min_terminated_length": 249.0,
      "entropy": 0.10247802734375,
      "epoch": 0.2737382378100941,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.028208184987306595,
      "kl": 0.001240849494934082,
      "learning_rate": 1.6761749002740195e-05,
      "loss": -0.0072,
      "num_tokens": 12477345.0,
      "reward": 0.7518314123153687,
      "reward_std": 0.223491370677948,
      "rewards/cross_solution_unittest_reward/mean": 0.37591565400362015,
      "rewards/cross_solution_unittest_reward/std": 0.2666970565915108,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 766.0,
      "completions/mean_length": 470.55731201171875,
      "completions/mean_terminated_length": 462.4281921386719,
      "completions/min_length": 248.0,
      "completions/min_terminated_length": 248.0,
      "entropy": 0.10906982421875,
      "epoch": 0.28286284573709725,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03271934762597084,
      "kl": 0.0014473199844360352,
      "learning_rate": 1.6548607339452853e-05,
      "loss": 0.0095,
      "num_tokens": 12841471.0,
      "reward": 0.9323596954345703,
      "reward_std": 0.1906483769416809,
      "rewards/cross_solution_unittest_reward/mean": 0.46617985516786575,
      "rewards/cross_solution_unittest_reward/std": 0.21004272997379303,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 906.0,
      "completions/mean_length": 483.4505310058594,
      "completions/mean_terminated_length": 475.3551025390625,
      "completions/min_length": 223.0,
      "completions/min_terminated_length": 223.0,
      "entropy": 0.105499267578125,
      "epoch": 0.2919874536641004,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.045950278639793396,
      "kl": 0.00164031982421875,
      "learning_rate": 1.6330124538088705e-05,
      "loss": 0.0251,
      "num_tokens": 13295500.0,
      "reward": 0.9082147479057312,
      "reward_std": 0.3082047402858734,
      "rewards/cross_solution_unittest_reward/mean": 0.4541073888540268,
      "rewards/cross_solution_unittest_reward/std": 0.24801740050315857,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 778.0,
      "completions/mean_length": 476.5130310058594,
      "completions/mean_terminated_length": 468.39947509765625,
      "completions/min_length": 273.0,
      "completions/min_terminated_length": 273.0,
      "entropy": 0.11260986328125,
      "epoch": 0.3011120615911035,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03122815489768982,
      "kl": 0.001780867576599121,
      "learning_rate": 1.6106478796354382e-05,
      "loss": 0.0068,
      "num_tokens": 13676457.0,
      "reward": 0.8194860816001892,
      "reward_std": 0.1633000671863556,
      "rewards/cross_solution_unittest_reward/mean": 0.4097430929541588,
      "rewards/cross_solution_unittest_reward/std": 0.2260184958577156,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 794.0,
      "completions/max_terminated_length": 794.0,
      "completions/mean_length": 480.32293701171875,
      "completions/mean_terminated_length": 480.32293701171875,
      "completions/min_length": 266.0,
      "completions/min_terminated_length": 266.0,
      "entropy": 0.10528564453125,
      "epoch": 0.31023666951810663,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03177294880151749,
      "kl": 0.002082347869873047,
      "learning_rate": 1.5877852522924733e-05,
      "loss": -0.0068,
      "num_tokens": 14100165.0,
      "reward": 0.9499843120574951,
      "reward_std": 0.18478095531463623,
      "rewards/cross_solution_unittest_reward/mean": 0.47499214112758636,
      "rewards/cross_solution_unittest_reward/std": 0.23216352611780167,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1548.0,
      "completions/max_terminated_length": 1548.0,
      "completions/mean_length": 490.5208435058594,
      "completions/mean_terminated_length": 490.5208435058594,
      "completions/min_length": 222.0,
      "completions/min_terminated_length": 222.0,
      "entropy": 0.109344482421875,
      "epoch": 0.3193612774451098,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03219935670495033,
      "kl": 0.001997709274291992,
      "learning_rate": 1.5644432188667695e-05,
      "loss": 0.0101,
      "num_tokens": 14479845.0,
      "reward": 0.8249267339706421,
      "reward_std": 0.22756871581077576,
      "rewards/cross_solution_unittest_reward/mean": 0.41246332228183746,
      "rewards/cross_solution_unittest_reward/std": 0.249651201069355,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1074.0,
      "completions/max_terminated_length": 1074.0,
      "completions/mean_length": 508.5546875,
      "completions/mean_terminated_length": 508.5546875,
      "completions/min_length": 254.0,
      "completions/min_terminated_length": 254.0,
      "entropy": 0.110809326171875,
      "epoch": 0.32848588537211293,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.033212192356586456,
      "kl": 0.0023381710052490234,
      "learning_rate": 1.5406408174555978e-05,
      "loss": -0.007,
      "num_tokens": 14897138.0,
      "reward": 0.7724038362503052,
      "reward_std": 0.2212439626455307,
      "rewards/cross_solution_unittest_reward/mean": 0.3862019032239914,
      "rewards/cross_solution_unittest_reward/std": 0.2340112403035164,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_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": -5.0,
      "completions/max_length": 957.0,
      "completions/max_terminated_length": 957.0,
      "completions/mean_length": 499.2239685058594,
      "completions/mean_terminated_length": 499.2239685058594,
      "completions/min_length": 244.0,
      "completions/min_terminated_length": 244.0,
      "entropy": 0.0943603515625,
      "epoch": 0.3376104932991161,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.032685279846191406,
      "kl": 0.0021796226501464844,
      "learning_rate": 1.5163974616389621e-05,
      "loss": -0.004,
      "num_tokens": 15299728.0,
      "reward": 0.8877171874046326,
      "reward_std": 0.1671483963727951,
      "rewards/cross_solution_unittest_reward/mean": 0.4438585937023163,
      "rewards/cross_solution_unittest_reward/std": 0.2307201847434044,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 1086.0,
      "completions/max_terminated_length": 1086.0,
      "completions/mean_length": 479.34375,
      "completions/mean_terminated_length": 479.34375,
      "completions/min_length": 190.0,
      "completions/min_terminated_length": 190.0,
      "entropy": 0.1061859130859375,
      "epoch": 0.34673510122611917,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.031533192843198776,
      "kl": 0.0025458335876464844,
      "learning_rate": 1.491732924645604e-05,
      "loss": 0.0057,
      "num_tokens": 15681980.0,
      "reward": 0.8841136693954468,
      "reward_std": 0.18882207572460175,
      "rewards/cross_solution_unittest_reward/mean": 0.442056804895401,
      "rewards/cross_solution_unittest_reward/std": 0.23417016118764877,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 975.0,
      "completions/mean_length": 462.5520935058594,
      "completions/mean_terminated_length": 454.402099609375,
      "completions/min_length": 202.0,
      "completions/min_terminated_length": 202.0,
      "entropy": 0.11846923828125,
      "epoch": 0.3558597091531223,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.09556223452091217,
      "kl": 0.0063495635986328125,
      "learning_rate": 1.4666673232256738e-05,
      "loss": 0.0078,
      "num_tokens": 15989736.0,
      "reward": 0.7538800239562988,
      "reward_std": 0.2889000177383423,
      "rewards/cross_solution_unittest_reward/mean": 0.3769400045275688,
      "rewards/cross_solution_unittest_reward/std": 0.2265937775373459,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1821.0,
      "completions/mean_length": 492.40625,
      "completions/mean_terminated_length": 484.3341979980469,
      "completions/min_length": 233.0,
      "completions/min_terminated_length": 233.0,
      "entropy": 0.1049957275390625,
      "epoch": 0.36498431708012546,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.10226094722747803,
      "kl": 0.006191730499267578,
      "learning_rate": 1.4412211012432213e-05,
      "loss": -0.0045,
      "num_tokens": 16399484.0,
      "reward": 0.7214807271957397,
      "reward_std": 0.20936036109924316,
      "rewards/cross_solution_unittest_reward/mean": 0.3607403114438057,
      "rewards/cross_solution_unittest_reward/std": 0.286220021545887,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.953125,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1305.0,
      "completions/mean_length": 521.59375,
      "completions/mean_terminated_length": 497.4803161621094,
      "completions/min_length": 272.0,
      "completions/min_terminated_length": 272.0,
      "entropy": 0.1065673828125,
      "epoch": 0.3741089250071286,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03374212980270386,
      "kl": 0.003753662109375,
      "learning_rate": 1.4154150130018867e-05,
      "loss": 0.0217,
      "num_tokens": 16819632.0,
      "reward": 0.6985855102539062,
      "reward_std": 0.17247408628463745,
      "rewards/cross_solution_unittest_reward/mean": 0.34929272532463074,
      "rewards/cross_solution_unittest_reward/std": 0.21424289792776108,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 777.0,
      "completions/mean_length": 498.4192810058594,
      "completions/mean_terminated_length": 490.3629455566406,
      "completions/min_length": 229.0,
      "completions/min_terminated_length": 229.0,
      "entropy": 0.0918121337890625,
      "epoch": 0.38323353293413176,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.0317884236574173,
      "kl": 0.00311279296875,
      "learning_rate": 1.3892701063173917e-05,
      "loss": 0.0088,
      "num_tokens": 17261449.0,
      "reward": 0.8251139521598816,
      "reward_std": 0.16213150322437286,
      "rewards/cross_solution_unittest_reward/mean": 0.4125569462776184,
      "rewards/cross_solution_unittest_reward/std": 0.23735850304365158,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 758.0,
      "completions/max_terminated_length": 758.0,
      "completions/mean_length": 464.19012451171875,
      "completions/mean_terminated_length": 464.19012451171875,
      "completions/min_length": 256.0,
      "completions/min_terminated_length": 256.0,
      "entropy": 0.103759765625,
      "epoch": 0.39235814086113485,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.033187512308359146,
      "kl": 0.0038928985595703125,
      "learning_rate": 1.362807705350641e-05,
      "loss": 0.0018,
      "num_tokens": 17594050.0,
      "reward": 0.9705511927604675,
      "reward_std": 0.15792036056518555,
      "rewards/cross_solution_unittest_reward/mean": 0.48527561128139496,
      "rewards/cross_solution_unittest_reward/std": 0.25453296303749084,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1812.0,
      "completions/mean_length": 511.6640625,
      "completions/mean_terminated_length": 503.6423034667969,
      "completions/min_length": 238.0,
      "completions/min_terminated_length": 238.0,
      "entropy": 0.0930633544921875,
      "epoch": 0.401482748788138,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02915688417851925,
      "kl": 0.0028166770935058594,
      "learning_rate": 1.3360493932154301e-05,
      "loss": 0.0031,
      "num_tokens": 18001433.0,
      "reward": 0.6709164977073669,
      "reward_std": 0.17882965505123138,
      "rewards/cross_solution_unittest_reward/mean": 0.3354582153260708,
      "rewards/cross_solution_unittest_reward/std": 0.20766809582710266,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 855.0,
      "completions/mean_length": 456.7265625,
      "completions/mean_terminated_length": 448.5613708496094,
      "completions/min_length": 205.0,
      "completions/min_terminated_length": 205.0,
      "entropy": 0.1132659912109375,
      "epoch": 0.41060735671514115,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.038488611578941345,
      "kl": 0.004986763000488281,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.0137,
      "num_tokens": 18379368.0,
      "reward": 0.9005228281021118,
      "reward_std": 0.19103503227233887,
      "rewards/cross_solution_unittest_reward/mean": 0.4502613842487335,
      "rewards/cross_solution_unittest_reward/std": 0.25590047985315323,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 3564.0,
      "completions/max_terminated_length": 3564.0,
      "completions/mean_length": 498.2630310058594,
      "completions/mean_terminated_length": 498.2630310058594,
      "completions/min_length": 244.0,
      "completions/min_terminated_length": 244.0,
      "entropy": 0.098541259765625,
      "epoch": 0.4197319646421443,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.034151654690504074,
      "kl": 0.0035920143127441406,
      "learning_rate": 1.2817325568414299e-05,
      "loss": -0.0086,
      "num_tokens": 18790933.0,
      "reward": 0.7998638153076172,
      "reward_std": 0.26316848397254944,
      "rewards/cross_solution_unittest_reward/mean": 0.3999319151043892,
      "rewards/cross_solution_unittest_reward/std": 0.2370620146393776,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 908.0,
      "completions/max_terminated_length": 908.0,
      "completions/mean_length": 479.18231201171875,
      "completions/mean_terminated_length": 479.18231201171875,
      "completions/min_length": 219.0,
      "completions/min_terminated_length": 219.0,
      "entropy": 0.11773681640625,
      "epoch": 0.42885657256914744,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.037232693284749985,
      "kl": 0.004299640655517578,
      "learning_rate": 1.2542183341934873e-05,
      "loss": -0.0027,
      "num_tokens": 19200115.0,
      "reward": 0.667292058467865,
      "reward_std": 0.18406903743743896,
      "rewards/cross_solution_unittest_reward/mean": 0.3336460068821907,
      "rewards/cross_solution_unittest_reward/std": 0.25224795937538147,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 866.0,
      "completions/max_terminated_length": 866.0,
      "completions/mean_length": 480.40625,
      "completions/mean_terminated_length": 480.40625,
      "completions/min_length": 245.0,
      "completions/min_terminated_length": 245.0,
      "entropy": 0.1033172607421875,
      "epoch": 0.43798118049615054,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.0333259254693985,
      "kl": 0.003824949264526367,
      "learning_rate": 1.2264967674257647e-05,
      "loss": -0.0009,
      "num_tokens": 19586559.0,
      "reward": 0.8440192937850952,
      "reward_std": 0.23113533854484558,
      "rewards/cross_solution_unittest_reward/mean": 0.4220096617937088,
      "rewards/cross_solution_unittest_reward/std": 0.2280796319246292,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 998.0,
      "completions/mean_length": 486.1458435058594,
      "completions/mean_terminated_length": 478.05743408203125,
      "completions/min_length": 267.0,
      "completions/min_terminated_length": 267.0,
      "entropy": 0.09979248046875,
      "epoch": 0.4471057884231537,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03385407105088234,
      "kl": 0.003530740737915039,
      "learning_rate": 1.1985904666457455e-05,
      "loss": 0.0061,
      "num_tokens": 19995087.0,
      "reward": 0.6481320261955261,
      "reward_std": 0.26141294836997986,
      "rewards/cross_solution_unittest_reward/mean": 0.3240659758448601,
      "rewards/cross_solution_unittest_reward/std": 0.25914470851421356,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "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": -5.0,
      "completions/max_length": 966.0,
      "completions/max_terminated_length": 966.0,
      "completions/mean_length": 514.703125,
      "completions/mean_terminated_length": 514.703125,
      "completions/min_length": 251.0,
      "completions/min_terminated_length": 251.0,
      "entropy": 0.0972137451171875,
      "epoch": 0.45623039635015683,
      "frac_reward_zero_std": 0.0833333358168602,
      "grad_norm": 0.03325730189681053,
      "kl": 0.004256248474121094,
      "learning_rate": 1.170522192632624e-05,
      "loss": -0.0051,
      "num_tokens": 20408813.0,
      "reward": 0.8064446449279785,
      "reward_std": 0.22871556878089905,
      "rewards/cross_solution_unittest_reward/mean": 0.40322232991456985,
      "rewards/cross_solution_unittest_reward/std": 0.28535734862089157,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 50
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 834.0,
      "completions/mean_length": 510.9453125,
      "completions/mean_terminated_length": 502.92169189453125,
      "completions/min_length": 169.0,
      "completions/min_terminated_length": 169.0,
      "entropy": 0.0950927734375,
      "epoch": 0.46535500427716,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03309718891978264,
      "kl": 0.004229068756103516,
      "learning_rate": 1.1423148382732854e-05,
      "loss": 0.0142,
      "num_tokens": 20829704.0,
      "reward": 0.9345626831054688,
      "reward_std": 0.21369287371635437,
      "rewards/cross_solution_unittest_reward/mean": 0.46728135645389557,
      "rewards/cross_solution_unittest_reward/std": 0.2026936635375023,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 51
    },
    {
      "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": -5.0,
      "completions/max_length": 760.0,
      "completions/max_terminated_length": 760.0,
      "completions/mean_length": 485.2786560058594,
      "completions/mean_terminated_length": 485.2786560058594,
      "completions/min_length": 208.0,
      "completions/min_terminated_length": 208.0,
      "entropy": 0.0979461669921875,
      "epoch": 0.4744796122041631,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4690159559249878,
      "kl": 0.013956546783447266,
      "learning_rate": 1.1139914098905406e-05,
      "loss": 0.0024,
      "num_tokens": 21222691.0,
      "reward": 0.9599242210388184,
      "reward_std": 0.23365139961242676,
      "rewards/cross_solution_unittest_reward/mean": 0.479962095618248,
      "rewards/cross_solution_unittest_reward/std": 0.24594762176275253,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 52
    },
    {
      "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": -5.0,
      "completions/max_length": 891.0,
      "completions/max_terminated_length": 891.0,
      "completions/mean_length": 477.48699951171875,
      "completions/mean_terminated_length": 477.48699951171875,
      "completions/min_length": 233.0,
      "completions/min_terminated_length": 233.0,
      "entropy": 0.095062255859375,
      "epoch": 0.4836042201311662,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03301261365413666,
      "kl": 0.0033338069915771484,
      "learning_rate": 1.08557500847884e-05,
      "loss": 0.0043,
      "num_tokens": 21619614.0,
      "reward": 0.7612521052360535,
      "reward_std": 0.18586371839046478,
      "rewards/cross_solution_unittest_reward/mean": 0.38062604516744614,
      "rewards/cross_solution_unittest_reward/std": 0.22070425003767014,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 53
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 992.0,
      "completions/mean_length": 529.8698120117188,
      "completions/mean_terminated_length": 521.8955688476562,
      "completions/min_length": 272.0,
      "completions/min_terminated_length": 272.0,
      "entropy": 0.098876953125,
      "epoch": 0.49272882805816937,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.037023600190877914,
      "kl": 0.005710601806640625,
      "learning_rate": 1.0570888108627682e-05,
      "loss": -0.0049,
      "num_tokens": 22079444.0,
      "reward": 0.7765120267868042,
      "reward_std": 0.25535768270492554,
      "rewards/cross_solution_unittest_reward/mean": 0.3882560133934021,
      "rewards/cross_solution_unittest_reward/std": 0.21029234677553177,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 54
    },
    {
      "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": -5.0,
      "completions/max_length": 968.0,
      "completions/max_terminated_length": 968.0,
      "completions/mean_length": 480.6302185058594,
      "completions/mean_terminated_length": 480.6302185058594,
      "completions/min_length": 233.0,
      "completions/min_terminated_length": 233.0,
      "entropy": 0.1016998291015625,
      "epoch": 0.5018534359851725,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03228428214788437,
      "kl": 0.003750324249267578,
      "learning_rate": 1.0285560507936962e-05,
      "loss": -0.0011,
      "num_tokens": 22452846.0,
      "reward": 0.7638306617736816,
      "reward_std": 0.2091209590435028,
      "rewards/cross_solution_unittest_reward/mean": 0.38191530853509903,
      "rewards/cross_solution_unittest_reward/std": 0.21358844637870789,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 55
    },
    {
      "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": -5.0,
      "completions/max_length": 886.0,
      "completions/max_terminated_length": 886.0,
      "completions/mean_length": 487.8567810058594,
      "completions/mean_terminated_length": 487.8567810058594,
      "completions/min_length": 213.0,
      "completions/min_terminated_length": 213.0,
      "entropy": 0.1000823974609375,
      "epoch": 0.5109780439121756,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.032228607684373856,
      "kl": 0.0042340755462646484,
      "learning_rate": 1e-05,
      "loss": 0.0103,
      "num_tokens": 22825151.0,
      "reward": 0.826278567314148,
      "reward_std": 0.19069427251815796,
      "rewards/cross_solution_unittest_reward/mean": 0.4131392240524292,
      "rewards/cross_solution_unittest_reward/std": 0.21790383756160736,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 56
    },
    {
      "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": -5.0,
      "completions/max_length": 1560.0,
      "completions/max_terminated_length": 1560.0,
      "completions/mean_length": 494.5520935058594,
      "completions/mean_terminated_length": 494.5520935058594,
      "completions/min_length": 277.0,
      "completions/min_terminated_length": 277.0,
      "entropy": 0.099639892578125,
      "epoch": 0.5201026518391788,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.5714826583862305,
      "kl": 0.06778192520141602,
      "learning_rate": 9.71443949206304e-06,
      "loss": 0.0054,
      "num_tokens": 23267683.0,
      "reward": 0.8670098185539246,
      "reward_std": 0.18901364505290985,
      "rewards/cross_solution_unittest_reward/mean": 0.4335048869252205,
      "rewards/cross_solution_unittest_reward/std": 0.2112448662519455,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 57
    },
    {
      "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": -4.953125,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 966.0,
      "completions/mean_length": 494.9739685058594,
      "completions/mean_terminated_length": 470.6509094238281,
      "completions/min_length": 238.0,
      "completions/min_terminated_length": 238.0,
      "entropy": 0.1165924072265625,
      "epoch": 0.5292272597661819,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03703828528523445,
      "kl": 0.004994392395019531,
      "learning_rate": 9.42911189137232e-06,
      "loss": 0.0108,
      "num_tokens": 23682969.0,
      "reward": 0.8357396125793457,
      "reward_std": 0.1958376169204712,
      "rewards/cross_solution_unittest_reward/mean": 0.41786976903676987,
      "rewards/cross_solution_unittest_reward/std": 0.28136271238327026,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 58
    },
    {
      "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": -5.0,
      "completions/max_length": 906.0,
      "completions/max_terminated_length": 906.0,
      "completions/mean_length": 464.0755310058594,
      "completions/mean_terminated_length": 464.0755310058594,
      "completions/min_length": 228.0,
      "completions/min_terminated_length": 228.0,
      "entropy": 0.1112823486328125,
      "epoch": 0.5383518676931851,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03699680417776108,
      "kl": 0.004660606384277344,
      "learning_rate": 9.144249915211605e-06,
      "loss": -0.0151,
      "num_tokens": 24082606.0,
      "reward": 0.8085569739341736,
      "reward_std": 0.1914159059524536,
      "rewards/cross_solution_unittest_reward/mean": 0.4042784869670868,
      "rewards/cross_solution_unittest_reward/std": 0.2565099969506264,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 59
    },
    {
      "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": -5.0,
      "completions/max_length": 1127.0,
      "completions/max_terminated_length": 1127.0,
      "completions/mean_length": 488.7890625,
      "completions/mean_terminated_length": 488.7890625,
      "completions/min_length": 258.0,
      "completions/min_terminated_length": 258.0,
      "entropy": 0.093719482421875,
      "epoch": 0.5474764756201882,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030238306149840355,
      "kl": 0.004515171051025391,
      "learning_rate": 8.860085901094595e-06,
      "loss": 0.003,
      "num_tokens": 24514333.0,
      "reward": 0.8814453482627869,
      "reward_std": 0.23693448305130005,
      "rewards/cross_solution_unittest_reward/mean": 0.44072266668081284,
      "rewards/cross_solution_unittest_reward/std": 0.22619559615850449,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 60
    },
    {
      "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": -5.0,
      "completions/max_length": 984.0,
      "completions/max_terminated_length": 984.0,
      "completions/mean_length": 526.2005615234375,
      "completions/mean_terminated_length": 526.2005615234375,
      "completions/min_length": 270.0,
      "completions/min_terminated_length": 270.0,
      "entropy": 0.0959625244140625,
      "epoch": 0.5566010835471913,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03212675452232361,
      "kl": 0.0042629241943359375,
      "learning_rate": 8.576851617267151e-06,
      "loss": -0.0098,
      "num_tokens": 24961698.0,
      "reward": 0.8686231374740601,
      "reward_std": 0.19413742423057556,
      "rewards/cross_solution_unittest_reward/mean": 0.43431156873703003,
      "rewards/cross_solution_unittest_reward/std": 0.26492878049612045,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 61
    },
    {
      "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": -5.0,
      "completions/max_length": 919.0,
      "completions/max_terminated_length": 919.0,
      "completions/mean_length": 520.0104370117188,
      "completions/mean_terminated_length": 520.0104370117188,
      "completions/min_length": 265.0,
      "completions/min_terminated_length": 265.0,
      "entropy": 0.1033935546875,
      "epoch": 0.5657256914741945,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.0341770313680172,
      "kl": 0.004704952239990234,
      "learning_rate": 8.294778073673762e-06,
      "loss": 0.0204,
      "num_tokens": 25386422.0,
      "reward": 0.747821033000946,
      "reward_std": 0.21126197278499603,
      "rewards/cross_solution_unittest_reward/mean": 0.37391049042344093,
      "rewards/cross_solution_unittest_reward/std": 0.19313736259937286,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 62
    },
    {
      "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": -5.0,
      "completions/max_length": 802.0,
      "completions/max_terminated_length": 802.0,
      "completions/mean_length": 457.11199951171875,
      "completions/mean_terminated_length": 457.11199951171875,
      "completions/min_length": 234.0,
      "completions/min_terminated_length": 234.0,
      "entropy": 0.1014862060546875,
      "epoch": 0.5748502994011976,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03489193320274353,
      "kl": 0.004260540008544922,
      "learning_rate": 8.014095333542548e-06,
      "loss": -0.0047,
      "num_tokens": 25751057.0,
      "reward": 0.7548073530197144,
      "reward_std": 0.20320609211921692,
      "rewards/cross_solution_unittest_reward/mean": 0.3774036541581154,
      "rewards/cross_solution_unittest_reward/std": 0.21862108260393143,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 63
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 842.0,
      "completions/mean_length": 497.3567810058594,
      "completions/mean_terminated_length": 489.29766845703125,
      "completions/min_length": 259.0,
      "completions/min_terminated_length": 259.0,
      "entropy": 0.09716796875,
      "epoch": 0.5839749073282008,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.04765147343277931,
      "kl": 0.00534820556640625,
      "learning_rate": 7.735032325742355e-06,
      "loss": -0.0042,
      "num_tokens": 26145418.0,
      "reward": 1.0832295417785645,
      "reward_std": 0.20443449914455414,
      "rewards/cross_solution_unittest_reward/mean": 0.5416147857904434,
      "rewards/cross_solution_unittest_reward/std": 0.1956881359219551,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 64
    },
    {
      "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": -5.0,
      "completions/max_length": 1014.0,
      "completions/max_terminated_length": 1014.0,
      "completions/mean_length": 469.390625,
      "completions/mean_terminated_length": 469.390625,
      "completions/min_length": 178.0,
      "completions/min_terminated_length": 178.0,
      "entropy": 0.110504150390625,
      "epoch": 0.5930995152552039,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.033948320895433426,
      "kl": 0.004923343658447266,
      "learning_rate": 7.4578166580651335e-06,
      "loss": -0.0063,
      "num_tokens": 26531240.0,
      "reward": 0.9129461050033569,
      "reward_std": 0.2018149346113205,
      "rewards/cross_solution_unittest_reward/mean": 0.4564730077981949,
      "rewards/cross_solution_unittest_reward/std": 0.2923680394887924,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 65
    },
    {
      "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": -4.953125,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1764.0,
      "completions/mean_length": 536.3333740234375,
      "completions/mean_terminated_length": 512.3359375,
      "completions/min_length": 253.0,
      "completions/min_terminated_length": 253.0,
      "entropy": 0.0997772216796875,
      "epoch": 0.602224123182207,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030758032575249672,
      "kl": 0.004119873046875,
      "learning_rate": 7.182674431585703e-06,
      "loss": 0.0112,
      "num_tokens": 27009184.0,
      "reward": 0.9593736529350281,
      "reward_std": 0.22100883722305298,
      "rewards/cross_solution_unittest_reward/mean": 0.47968684136867523,
      "rewards/cross_solution_unittest_reward/std": 0.2801002562046051,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 66
    },
    {
      "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": -5.0,
      "completions/max_length": 792.0,
      "completions/max_terminated_length": 792.0,
      "completions/mean_length": 486.00262451171875,
      "completions/mean_terminated_length": 486.00262451171875,
      "completions/min_length": 243.0,
      "completions/min_terminated_length": 243.0,
      "entropy": 0.0943145751953125,
      "epoch": 0.6113487311092102,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03071185201406479,
      "kl": 0.004786491394042969,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.0004,
      "num_tokens": 27460937.0,
      "reward": 1.0903098583221436,
      "reward_std": 0.25236761569976807,
      "rewards/cross_solution_unittest_reward/mean": 0.5451549142599106,
      "rewards/cross_solution_unittest_reward/std": 0.24181817471981049,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 67
    },
    {
      "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": -5.0,
      "completions/max_length": 1091.0,
      "completions/max_terminated_length": 1091.0,
      "completions/mean_length": 514.515625,
      "completions/mean_terminated_length": 514.515625,
      "completions/min_length": 257.0,
      "completions/min_terminated_length": 257.0,
      "entropy": 0.0963897705078125,
      "epoch": 0.6204733390362133,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030294137075543404,
      "kl": 0.0035681724548339844,
      "learning_rate": 6.639506067845698e-06,
      "loss": 0.0073,
      "num_tokens": 27860791.0,
      "reward": 0.9012066721916199,
      "reward_std": 0.18023371696472168,
      "rewards/cross_solution_unittest_reward/mean": 0.45060331374406815,
      "rewards/cross_solution_unittest_reward/std": 0.19457216560840607,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 68
    },
    {
      "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": -5.0,
      "completions/max_length": 2792.0,
      "completions/max_terminated_length": 2792.0,
      "completions/mean_length": 488.8046875,
      "completions/mean_terminated_length": 488.8046875,
      "completions/min_length": 251.0,
      "completions/min_terminated_length": 251.0,
      "entropy": 0.100738525390625,
      "epoch": 0.6295979469632165,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.029943598434329033,
      "kl": 0.004576683044433594,
      "learning_rate": 6.3719229464935915e-06,
      "loss": 0.0022,
      "num_tokens": 28294524.0,
      "reward": 0.9089535474777222,
      "reward_std": 0.21831755340099335,
      "rewards/cross_solution_unittest_reward/mean": 0.4544767662882805,
      "rewards/cross_solution_unittest_reward/std": 0.20777993649244308,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 69
    },
    {
      "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": -5.0,
      "completions/max_length": 1007.0,
      "completions/max_terminated_length": 1007.0,
      "completions/mean_length": 517.0234375,
      "completions/mean_terminated_length": 517.0234375,
      "completions/min_length": 252.0,
      "completions/min_terminated_length": 252.0,
      "entropy": 0.09405517578125,
      "epoch": 0.6387225548902196,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.032788679003715515,
      "kl": 0.00408935546875,
      "learning_rate": 6.107298936826086e-06,
      "loss": 0.003,
      "num_tokens": 28738997.0,
      "reward": 0.9689019918441772,
      "reward_std": 0.2762044668197632,
      "rewards/cross_solution_unittest_reward/mean": 0.4844510108232498,
      "rewards/cross_solution_unittest_reward/std": 0.21708913147449493,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 70
    },
    {
      "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": -5.0,
      "completions/max_length": 898.0,
      "completions/max_terminated_length": 898.0,
      "completions/mean_length": 492.09637451171875,
      "completions/mean_terminated_length": 492.09637451171875,
      "completions/min_length": 280.0,
      "completions/min_terminated_length": 280.0,
      "entropy": 0.10223388671875,
      "epoch": 0.6478471628172227,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03405945748090744,
      "kl": 0.004323005676269531,
      "learning_rate": 5.845849869981137e-06,
      "loss": -0.0087,
      "num_tokens": 29140394.0,
      "reward": 0.6452049016952515,
      "reward_std": 0.18944227695465088,
      "rewards/cross_solution_unittest_reward/mean": 0.32260240986943245,
      "rewards/cross_solution_unittest_reward/std": 0.19827812910079956,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 71
    },
    {
      "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": -5.0,
      "completions/max_length": 1050.0,
      "completions/max_terminated_length": 1050.0,
      "completions/mean_length": 511.7578125,
      "completions/mean_terminated_length": 511.7578125,
      "completions/min_length": 275.0,
      "completions/min_terminated_length": 275.0,
      "entropy": 0.0999755859375,
      "epoch": 0.6569717707442259,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.04706510528922081,
      "kl": 0.005205631256103516,
      "learning_rate": 5.587788987567785e-06,
      "loss": 0.0065,
      "num_tokens": 29547821.0,
      "reward": 0.9646930694580078,
      "reward_std": 0.1975850760936737,
      "rewards/cross_solution_unittest_reward/mean": 0.4823464900255203,
      "rewards/cross_solution_unittest_reward/std": 0.26742689311504364,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 72
    },
    {
      "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": -5.0,
      "completions/max_length": 960.0,
      "completions/max_terminated_length": 960.0,
      "completions/mean_length": 469.0859375,
      "completions/mean_terminated_length": 469.0859375,
      "completions/min_length": 207.0,
      "completions/min_terminated_length": 207.0,
      "entropy": 0.103759765625,
      "epoch": 0.666096378671229,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.062288798391819,
      "kl": 0.007381916046142578,
      "learning_rate": 5.333326767743263e-06,
      "loss": -0.0033,
      "num_tokens": 29945526.0,
      "reward": 1.0015891790390015,
      "reward_std": 0.24343864619731903,
      "rewards/cross_solution_unittest_reward/mean": 0.5007946044206619,
      "rewards/cross_solution_unittest_reward/std": 0.24663101881742477,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 73
    },
    {
      "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": -5.0,
      "completions/max_length": 970.0,
      "completions/max_terminated_length": 970.0,
      "completions/mean_length": 514.5989990234375,
      "completions/mean_terminated_length": 514.5989990234375,
      "completions/min_length": 197.0,
      "completions/min_terminated_length": 197.0,
      "entropy": 0.0988922119140625,
      "epoch": 0.6752209865982322,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.034228503704071045,
      "kl": 0.004460811614990234,
      "learning_rate": 5.082670753543961e-06,
      "loss": 0.0019,
      "num_tokens": 30349572.0,
      "reward": 0.8880095481872559,
      "reward_std": 0.2291492521762848,
      "rewards/cross_solution_unittest_reward/mean": 0.44400477409362793,
      "rewards/cross_solution_unittest_reward/std": 0.24787399172782898,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 74
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1092.0,
      "completions/mean_length": 510.9739685058594,
      "completions/mean_terminated_length": 502.9504089355469,
      "completions/min_length": 277.0,
      "completions/min_terminated_length": 277.0,
      "entropy": 0.106231689453125,
      "epoch": 0.6843455945252352,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030502676963806152,
      "kl": 0.003562450408935547,
      "learning_rate": 4.836025383610382e-06,
      "loss": -0.0003,
      "num_tokens": 30727274.0,
      "reward": 0.9304172396659851,
      "reward_std": 0.20447766780853271,
      "rewards/cross_solution_unittest_reward/mean": 0.46520860493183136,
      "rewards/cross_solution_unittest_reward/std": 0.23417390882968903,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 75
    },
    {
      "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": -5.0,
      "completions/max_length": 888.0,
      "completions/max_terminated_length": 888.0,
      "completions/mean_length": 464.9453125,
      "completions/mean_terminated_length": 464.9453125,
      "completions/min_length": 217.0,
      "completions/min_terminated_length": 217.0,
      "entropy": 0.1075897216796875,
      "epoch": 0.6934702024522383,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03586795926094055,
      "kl": 0.004437446594238281,
      "learning_rate": 4.593591825444028e-06,
      "loss": 0.0051,
      "num_tokens": 31071701.0,
      "reward": 0.7625653147697449,
      "reward_std": 0.1792157143354416,
      "rewards/cross_solution_unittest_reward/mean": 0.38128262013196945,
      "rewards/cross_solution_unittest_reward/std": 0.22236895561218262,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 76
    },
    {
      "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": -4.96875,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 890.0,
      "completions/mean_length": 497.3125,
      "completions/mean_terminated_length": 481.1518249511719,
      "completions/min_length": 241.0,
      "completions/min_terminated_length": 241.0,
      "entropy": 0.105224609375,
      "epoch": 0.7025948103792415,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03737795352935791,
      "kl": 0.003971576690673828,
      "learning_rate": 4.355567811332311e-06,
      "loss": -0.0003,
      "num_tokens": 31477149.0,
      "reward": 0.9584690928459167,
      "reward_std": 0.2223634421825409,
      "rewards/cross_solution_unittest_reward/mean": 0.47923456132411957,
      "rewards/cross_solution_unittest_reward/std": 0.20396708697080612,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 77
    },
    {
      "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": -5.0,
      "completions/max_length": 880.0,
      "completions/max_terminated_length": 880.0,
      "completions/mean_length": 495.7161560058594,
      "completions/mean_terminated_length": 495.7161560058594,
      "completions/min_length": 236.0,
      "completions/min_terminated_length": 236.0,
      "entropy": 0.097930908203125,
      "epoch": 0.7117194183062446,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.159642294049263,
      "kl": 0.006262063980102539,
      "learning_rate": 4.12214747707527e-06,
      "loss": -0.0001,
      "num_tokens": 31912712.0,
      "reward": 0.9255527257919312,
      "reward_std": 0.2571796178817749,
      "rewards/cross_solution_unittest_reward/mean": 0.46277637779712677,
      "rewards/cross_solution_unittest_reward/std": 0.25626206398010254,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 78
    },
    {
      "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": -4.9375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1027.0,
      "completions/mean_length": 512.7135620117188,
      "completions/mean_terminated_length": 480.38421630859375,
      "completions/min_length": 224.0,
      "completions/min_terminated_length": 224.0,
      "entropy": 0.1019744873046875,
      "epoch": 0.7208440262332478,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.032687410712242126,
      "kl": 0.004110813140869141,
      "learning_rate": 3.893521203645618e-06,
      "loss": -0.0029,
      "num_tokens": 32308370.0,
      "reward": 0.9185789823532104,
      "reward_std": 0.19305042922496796,
      "rewards/cross_solution_unittest_reward/mean": 0.45928946137428284,
      "rewards/cross_solution_unittest_reward/std": 0.23322047293186188,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 79
    },
    {
      "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": -5.0,
      "completions/max_length": 938.0,
      "completions/max_terminated_length": 938.0,
      "completions/mean_length": 471.16668701171875,
      "completions/mean_terminated_length": 471.16668701171875,
      "completions/min_length": 246.0,
      "completions/min_terminated_length": 246.0,
      "entropy": 0.0972900390625,
      "epoch": 0.7299686341602509,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.034349896013736725,
      "kl": 0.0041866302490234375,
      "learning_rate": 3.6698754619112974e-06,
      "loss": -0.0032,
      "num_tokens": 32667882.0,
      "reward": 0.8720422983169556,
      "reward_std": 0.18141961097717285,
      "rewards/cross_solution_unittest_reward/mean": 0.436021126806736,
      "rewards/cross_solution_unittest_reward/std": 0.23245983570814133,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 80
    },
    {
      "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": -5.0,
      "completions/max_length": 952.0,
      "completions/max_terminated_length": 952.0,
      "completions/mean_length": 484.87762451171875,
      "completions/mean_terminated_length": 484.87762451171875,
      "completions/min_length": 194.0,
      "completions/min_terminated_length": 194.0,
      "entropy": 0.099090576171875,
      "epoch": 0.739093242087254,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03139191120862961,
      "kl": 0.0039272308349609375,
      "learning_rate": 3.4513926605471504e-06,
      "loss": 0.0016,
      "num_tokens": 33050987.0,
      "reward": 0.9457516074180603,
      "reward_std": 0.1805034577846527,
      "rewards/cross_solution_unittest_reward/mean": 0.47287583351135254,
      "rewards/cross_solution_unittest_reward/std": 0.19942139089107513,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 81
    },
    {
      "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": -5.0,
      "completions/max_length": 1118.0,
      "completions/max_terminated_length": 1118.0,
      "completions/mean_length": 475.60418701171875,
      "completions/mean_terminated_length": 475.60418701171875,
      "completions/min_length": 212.0,
      "completions/min_terminated_length": 212.0,
      "entropy": 0.1024169921875,
      "epoch": 0.7482178500142572,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030605316162109375,
      "kl": 0.004054546356201172,
      "learning_rate": 3.2382509972598087e-06,
      "loss": 0.0052,
      "num_tokens": 33439955.0,
      "reward": 0.8594505190849304,
      "reward_std": 0.20357158780097961,
      "rewards/cross_solution_unittest_reward/mean": 0.4297252967953682,
      "rewards/cross_solution_unittest_reward/std": 0.24937064200639725,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 82
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 808.0,
      "completions/mean_length": 483.18231201171875,
      "completions/mean_terminated_length": 475.086181640625,
      "completions/min_length": 264.0,
      "completions/min_terminated_length": 264.0,
      "entropy": 0.101776123046875,
      "epoch": 0.7573424579412603,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.034236159175634384,
      "kl": 0.004290103912353516,
      "learning_rate": 3.0306243134470668e-06,
      "loss": 0.0102,
      "num_tokens": 33834753.0,
      "reward": 0.8233698606491089,
      "reward_std": 0.25579404830932617,
      "rewards/cross_solution_unittest_reward/mean": 0.41168488562107086,
      "rewards/cross_solution_unittest_reward/std": 0.23754408955574036,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 83
    },
    {
      "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": -4.96875,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1117.0,
      "completions/mean_length": 483.5625,
      "completions/mean_terminated_length": 467.329833984375,
      "completions/min_length": 234.0,
      "completions/min_terminated_length": 234.0,
      "entropy": 0.110076904296875,
      "epoch": 0.7664670658682635,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03293149545788765,
      "kl": 0.0037369728088378906,
      "learning_rate": 2.8286819524103657e-06,
      "loss": -0.0012,
      "num_tokens": 34176409.0,
      "reward": 0.8954019546508789,
      "reward_std": 0.20441854000091553,
      "rewards/cross_solution_unittest_reward/mean": 0.44770099222660065,
      "rewards/cross_solution_unittest_reward/std": 0.21769069880247116,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 84
    },
    {
      "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": -5.0,
      "completions/max_length": 937.0,
      "completions/max_terminated_length": 937.0,
      "completions/mean_length": 485.5442810058594,
      "completions/mean_terminated_length": 485.5442810058594,
      "completions/min_length": 242.0,
      "completions/min_terminated_length": 242.0,
      "entropy": 0.0960693359375,
      "epoch": 0.7755916737952666,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.033041298389434814,
      "kl": 0.004325389862060547,
      "learning_rate": 2.6325886212359496e-06,
      "loss": 0.0025,
      "num_tokens": 34602482.0,
      "reward": 0.9020225405693054,
      "reward_std": 0.16266906261444092,
      "rewards/cross_solution_unittest_reward/mean": 0.4510112926363945,
      "rewards/cross_solution_unittest_reward/std": 0.19898167252540588,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 85
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 826.0,
      "completions/mean_length": 472.5625,
      "completions/mean_terminated_length": 464.43865966796875,
      "completions/min_length": 254.0,
      "completions/min_terminated_length": 254.0,
      "entropy": 0.106231689453125,
      "epoch": 0.7847162817222697,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03541736304759979,
      "kl": 0.005221366882324219,
      "learning_rate": 2.4425042564574186e-06,
      "loss": -0.0027,
      "num_tokens": 35012690.0,
      "reward": 0.8730189204216003,
      "reward_std": 0.19061098992824554,
      "rewards/cross_solution_unittest_reward/mean": 0.4365094378590584,
      "rewards/cross_solution_unittest_reward/std": 0.19948572665452957,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 86
    },
    {
      "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": -4.953125,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 932.0,
      "completions/mean_length": 523.3880615234375,
      "completions/mean_terminated_length": 499.2887268066406,
      "completions/min_length": 246.0,
      "completions/min_terminated_length": 246.0,
      "entropy": 0.1020660400390625,
      "epoch": 0.7938408896492729,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.035699885338544846,
      "kl": 0.0038251876831054688,
      "learning_rate": 2.2585838936091753e-06,
      "loss": 0.0055,
      "num_tokens": 35446079.0,
      "reward": 0.9853703379631042,
      "reward_std": 0.1933068484067917,
      "rewards/cross_solution_unittest_reward/mean": 0.49268509447574615,
      "rewards/cross_solution_unittest_reward/std": 0.2293217033147812,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 87
    },
    {
      "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": -5.0,
      "completions/max_length": 949.0,
      "completions/max_terminated_length": 949.0,
      "completions/mean_length": 436.2578125,
      "completions/mean_terminated_length": 436.2578125,
      "completions/min_length": 236.0,
      "completions/min_terminated_length": 236.0,
      "entropy": 0.1071624755859375,
      "epoch": 0.802965497576276,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.12767620384693146,
      "kl": 0.010437488555908203,
      "learning_rate": 2.0809775407772505e-06,
      "loss": 0.0142,
      "num_tokens": 35801002.0,
      "reward": 0.9056326746940613,
      "reward_std": 0.22697775065898895,
      "rewards/cross_solution_unittest_reward/mean": 0.45281632244586945,
      "rewards/cross_solution_unittest_reward/std": 0.2281804159283638,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 88
    },
    {
      "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": -5.0,
      "completions/max_length": 821.0,
      "completions/max_terminated_length": 821.0,
      "completions/mean_length": 501.71356201171875,
      "completions/mean_terminated_length": 501.71356201171875,
      "completions/min_length": 316.0,
      "completions/min_terminated_length": 316.0,
      "entropy": 0.0912628173828125,
      "epoch": 0.8120901055032792,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030100375413894653,
      "kl": 0.004030466079711914,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.0095,
      "num_tokens": 36223060.0,
      "reward": 0.8723653554916382,
      "reward_std": 0.1536586731672287,
      "rewards/cross_solution_unittest_reward/mean": 0.4361826628446579,
      "rewards/cross_solution_unittest_reward/std": 0.24371575564146042,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 89
    },
    {
      "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": -5.0,
      "completions/max_length": 846.0,
      "completions/max_terminated_length": 846.0,
      "completions/mean_length": 482.5078125,
      "completions/mean_terminated_length": 482.5078125,
      "completions/min_length": 239.0,
      "completions/min_terminated_length": 239.0,
      "entropy": 0.096923828125,
      "epoch": 0.8212147134302823,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.040627796202898026,
      "kl": 0.007803916931152344,
      "learning_rate": 1.74528103037226e-06,
      "loss": -0.0038,
      "num_tokens": 36635343.0,
      "reward": 0.7952584028244019,
      "reward_std": 0.23465335369110107,
      "rewards/cross_solution_unittest_reward/mean": 0.39762915670871735,
      "rewards/cross_solution_unittest_reward/std": 0.2743668779730797,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 90
    },
    {
      "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": -5.0,
      "completions/max_length": 1028.0,
      "completions/max_terminated_length": 1028.0,
      "completions/mean_length": 482.26824951171875,
      "completions/mean_terminated_length": 482.26824951171875,
      "completions/min_length": 228.0,
      "completions/min_terminated_length": 228.0,
      "entropy": 0.1040496826171875,
      "epoch": 0.8303393213572854,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.036647334694862366,
      "kl": 0.0043468475341796875,
      "learning_rate": 1.587464671688187e-06,
      "loss": -0.005,
      "num_tokens": 37027198.0,
      "reward": 0.8686568140983582,
      "reward_std": 0.1460837721824646,
      "rewards/cross_solution_unittest_reward/mean": 0.4343283474445343,
      "rewards/cross_solution_unittest_reward/std": 0.21033113449811935,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 91
    },
    {
      "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": -4.96875,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 941.0,
      "completions/mean_length": 507.2317810058594,
      "completions/mean_terminated_length": 491.123046875,
      "completions/min_length": 206.0,
      "completions/min_terminated_length": 206.0,
      "entropy": 0.09820556640625,
      "epoch": 0.8394639292842886,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.033177897334098816,
      "kl": 0.004420280456542969,
      "learning_rate": 1.436509697484111e-06,
      "loss": -0.001,
      "num_tokens": 37446527.0,
      "reward": 0.8312811851501465,
      "reward_std": 0.22770294547080994,
      "rewards/cross_solution_unittest_reward/mean": 0.41564057022333145,
      "rewards/cross_solution_unittest_reward/std": 0.25115445256233215,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 92
    },
    {
      "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": -4.921875,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1317.0,
      "completions/mean_length": 578.09375,
      "completions/mean_terminated_length": 538.43798828125,
      "completions/min_length": 288.0,
      "completions/min_terminated_length": 288.0,
      "entropy": 0.086181640625,
      "epoch": 0.8485885372112917,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.0323200523853302,
      "kl": 0.004235267639160156,
      "learning_rate": 1.2925392288022299e-06,
      "loss": 0.0244,
      "num_tokens": 37887771.0,
      "reward": 0.7843211889266968,
      "reward_std": 0.22614744305610657,
      "rewards/cross_solution_unittest_reward/mean": 0.3921606168150902,
      "rewards/cross_solution_unittest_reward/std": 0.23859115689992905,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 93
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 961.0,
      "completions/mean_length": 527.734375,
      "completions/mean_terminated_length": 519.7545776367188,
      "completions/min_length": 171.0,
      "completions/min_terminated_length": 171.0,
      "entropy": 0.089080810546875,
      "epoch": 0.8577131451382949,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03161383792757988,
      "kl": 0.0037050247192382812,
      "learning_rate": 1.1556706900218572e-06,
      "loss": -0.0029,
      "num_tokens": 38306421.0,
      "reward": 0.8312766551971436,
      "reward_std": 0.25127798318862915,
      "rewards/cross_solution_unittest_reward/mean": 0.4156383126974106,
      "rewards/cross_solution_unittest_reward/std": 0.27913951128721237,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 94
    },
    {
      "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": -5.0,
      "completions/max_length": 762.0,
      "completions/max_terminated_length": 762.0,
      "completions/mean_length": 482.65106201171875,
      "completions/mean_terminated_length": 482.65106201171875,
      "completions/min_length": 243.0,
      "completions/min_terminated_length": 243.0,
      "entropy": 0.096649169921875,
      "epoch": 0.866837753065298,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030079808086156845,
      "kl": 0.004782915115356445,
      "learning_rate": 1.0260157130864178e-06,
      "loss": 0.0004,
      "num_tokens": 38748439.0,
      "reward": 0.8986494541168213,
      "reward_std": 0.20043903589248657,
      "rewards/cross_solution_unittest_reward/mean": 0.44932468235492706,
      "rewards/cross_solution_unittest_reward/std": 0.1927783116698265,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 95
    },
    {
      "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": -5.0,
      "completions/max_length": 878.0,
      "completions/max_terminated_length": 878.0,
      "completions/mean_length": 512.8932495117188,
      "completions/mean_terminated_length": 512.8932495117188,
      "completions/min_length": 225.0,
      "completions/min_terminated_length": 225.0,
      "entropy": 0.0985107421875,
      "epoch": 0.8759623609923011,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.035013340413570404,
      "kl": 0.004712104797363281,
      "learning_rate": 9.036800464548157e-07,
      "loss": 0.0024,
      "num_tokens": 39133230.0,
      "reward": 0.8117542266845703,
      "reward_std": 0.20024093985557556,
      "rewards/cross_solution_unittest_reward/mean": 0.40587710589170456,
      "rewards/cross_solution_unittest_reward/std": 0.2243197336792946,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 96
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 811.0,
      "completions/mean_length": 455.43231201171875,
      "completions/mean_terminated_length": 447.2637023925781,
      "completions/min_length": 228.0,
      "completions/min_terminated_length": 228.0,
      "entropy": 0.1064453125,
      "epoch": 0.8850869689193043,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03698151558637619,
      "kl": 0.004943370819091797,
      "learning_rate": 7.887634688515e-07,
      "loss": 0.0116,
      "num_tokens": 39453516.0,
      "reward": 1.0091822147369385,
      "reward_std": 0.17214202880859375,
      "rewards/cross_solution_unittest_reward/mean": 0.5045911073684692,
      "rewards/cross_solution_unittest_reward/std": 0.2308211475610733,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 97
    },
    {
      "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": -5.0,
      "completions/max_length": 905.0,
      "completions/max_terminated_length": 905.0,
      "completions/mean_length": 517.7213745117188,
      "completions/mean_terminated_length": 517.7213745117188,
      "completions/min_length": 332.0,
      "completions/min_terminated_length": 332.0,
      "entropy": 0.09765625,
      "epoch": 0.8942115768463074,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.0317588746547699,
      "kl": 0.004177570343017578,
      "learning_rate": 6.813597078854772e-07,
      "loss": -0.0053,
      "num_tokens": 39895593.0,
      "reward": 0.9603371620178223,
      "reward_std": 0.18226468563079834,
      "rewards/cross_solution_unittest_reward/mean": 0.48016858100891113,
      "rewards/cross_solution_unittest_reward/std": 0.2831304222345352,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 98
    },
    {
      "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": -5.0,
      "completions/max_length": 795.0,
      "completions/max_terminated_length": 795.0,
      "completions/mean_length": 476.8515625,
      "completions/mean_terminated_length": 476.8515625,
      "completions/min_length": 175.0,
      "completions/min_terminated_length": 175.0,
      "entropy": 0.1009063720703125,
      "epoch": 0.9033361847733106,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03179348632693291,
      "kl": 0.004183292388916016,
      "learning_rate": 5.815563636047539e-07,
      "loss": -0.0026,
      "num_tokens": 40260864.0,
      "reward": 0.9521991014480591,
      "reward_std": 0.2101556956768036,
      "rewards/cross_solution_unittest_reward/mean": 0.47609956562519073,
      "rewards/cross_solution_unittest_reward/std": 0.225713312625885,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 99
    },
    {
      "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": -5.0,
      "completions/max_length": 798.0,
      "completions/max_terminated_length": 798.0,
      "completions/mean_length": 457.1015625,
      "completions/mean_terminated_length": 457.1015625,
      "completions/min_length": 207.0,
      "completions/min_terminated_length": 207.0,
      "entropy": 0.1097564697265625,
      "epoch": 0.9124607927003137,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.03737381473183632,
      "kl": 0.0048885345458984375,
      "learning_rate": 4.894348370484648e-07,
      "loss": 0.0142,
      "num_tokens": 40678271.0,
      "reward": 0.8324875235557556,
      "reward_std": 0.20148925483226776,
      "rewards/cross_solution_unittest_reward/mean": 0.416243739426136,
      "rewards/cross_solution_unittest_reward/std": 0.25171586126089096,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 100
    },
    {
      "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": -4.984375,
      "completions/max_length": 3584.0,
      "completions/max_terminated_length": 1004.0,
      "completions/mean_length": 483.8359375,
      "completions/mean_terminated_length": 475.74151611328125,
      "completions/min_length": 254.0,
      "completions/min_terminated_length": 254.0,
      "entropy": 0.09332275390625,
      "epoch": 0.9215854006273168,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.030489500612020493,
      "kl": 0.004380226135253906,
      "learning_rate": 4.0507026385502747e-07,
      "loss": 0.0107,
      "num_tokens": 41066632.0,
      "reward": 0.9275814890861511,
      "reward_std": 0.1952568143606186,
      "rewards/cross_solution_unittest_reward/mean": 0.46379074454307556,
      "rewards/cross_solution_unittest_reward/std": 0.20886488258838654,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 101
    },
    {
      "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": -5.0,
      "completions/max_length": 876.0,
      "completions/max_terminated_length": 876.0,
      "completions/mean_length": 490.8802185058594,
      "completions/mean_terminated_length": 490.8802185058594,
      "completions/min_length": 230.0,
      "completions/min_terminated_length": 230.0,
      "entropy": 0.0925445556640625,
      "epoch": 0.93071000855432,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.02890767715871334,
      "kl": 0.004089832305908203,
      "learning_rate": 3.2853145298042954e-07,
      "loss": 0.0088,
      "num_tokens": 41490450.0,
      "reward": 1.0289883613586426,
      "reward_std": 0.23402146995067596,
      "rewards/cross_solution_unittest_reward/mean": 0.5144941955804825,
      "rewards/cross_solution_unittest_reward/std": 0.20954615622758865,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 102
    },
    {
      "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": -5.0,
      "completions/max_length": 1279.0,
      "completions/max_terminated_length": 1279.0,
      "completions/mean_length": 471.7421875,
      "completions/mean_terminated_length": 471.7421875,
      "completions/min_length": 234.0,
      "completions/min_terminated_length": 234.0,
      "entropy": 0.100799560546875,
      "epoch": 0.939834616481323,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.039594829082489014,
      "kl": 0.005157470703125,
      "learning_rate": 2.5988083057666534e-07,
      "loss": 0.0164,
      "num_tokens": 41862519.0,
      "reward": 1.0344469547271729,
      "reward_std": 0.2174767255783081,
      "rewards/cross_solution_unittest_reward/mean": 0.5172234624624252,
      "rewards/cross_solution_unittest_reward/std": 0.2513153925538063,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 103
    },
    {
      "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": -5.0,
      "completions/max_length": 833.0,
      "completions/max_terminated_length": 833.0,
      "completions/mean_length": 489.171875,
      "completions/mean_terminated_length": 489.171875,
      "completions/min_length": 246.0,
      "completions/min_terminated_length": 246.0,
      "entropy": 0.0948944091796875,
      "epoch": 0.9489592244083263,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.04726433753967285,
      "kl": 0.005943775177001953,
      "learning_rate": 1.9917438907606556e-07,
      "loss": 0.0073,
      "num_tokens": 42276585.0,
      "reward": 0.8814212679862976,
      "reward_std": 0.1902289092540741,
      "rewards/cross_solution_unittest_reward/mean": 0.4407106339931488,
      "rewards/cross_solution_unittest_reward/std": 0.1961411088705063,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 104
    },
    {
      "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": -5.0,
      "completions/max_length": 737.0,
      "completions/max_terminated_length": 737.0,
      "completions/mean_length": 453.78125,
      "completions/mean_terminated_length": 453.78125,
      "completions/min_length": 216.0,
      "completions/min_terminated_length": 216.0,
      "entropy": 0.10247802734375,
      "epoch": 0.9580838323353293,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.0373138003051281,
      "kl": 0.004960536956787109,
      "learning_rate": 1.464616415230702e-07,
      "loss": -0.0085,
      "num_tokens": 42634949.0,
      "reward": 0.7063878774642944,
      "reward_std": 0.22609077394008636,
      "rewards/cross_solution_unittest_reward/mean": 0.3531939312815666,
      "rewards/cross_solution_unittest_reward/std": 0.23600679636001587,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 105
    },
    {
      "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": -5.0,
      "completions/max_length": 814.0,
      "completions/max_terminated_length": 814.0,
      "completions/mean_length": 514.3932495117188,
      "completions/mean_terminated_length": 514.3932495117188,
      "completions/min_length": 289.0,
      "completions/min_terminated_length": 289.0,
      "entropy": 0.0894622802734375,
      "epoch": 0.9672084402623324,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.028397122398018837,
      "kl": 0.004255771636962891,
      "learning_rate": 1.0178558119067316e-07,
      "loss": -0.008,
      "num_tokens": 43057844.0,
      "reward": 0.8395287990570068,
      "reward_std": 0.23434078693389893,
      "rewards/cross_solution_unittest_reward/mean": 0.4197644293308258,
      "rewards/cross_solution_unittest_reward/std": 0.25368085503578186,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 106
    },
    {
      "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": -5.0,
      "completions/max_length": 2142.0,
      "completions/max_terminated_length": 2142.0,
      "completions/mean_length": 488.59375,
      "completions/mean_terminated_length": 488.59375,
      "completions/min_length": 283.0,
      "completions/min_terminated_length": 283.0,
      "entropy": 0.10888671875,
      "epoch": 0.9763330481893356,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.034389153122901917,
      "kl": 0.004794120788574219,
      "learning_rate": 6.51826465144978e-08,
      "loss": 0.003,
      "num_tokens": 43460008.0,
      "reward": 0.9071213603019714,
      "reward_std": 0.1905367076396942,
      "rewards/cross_solution_unittest_reward/mean": 0.4535607025027275,
      "rewards/cross_solution_unittest_reward/std": 0.2374211624264717,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 107
    },
    {
      "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": -5.0,
      "completions/max_length": 1084.0,
      "completions/max_terminated_length": 1084.0,
      "completions/mean_length": 535.3229370117188,
      "completions/mean_terminated_length": 535.3229370117188,
      "completions/min_length": 228.0,
      "completions/min_terminated_length": 228.0,
      "entropy": 0.0862884521484375,
      "epoch": 0.9854576561163387,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.029783420264720917,
      "kl": 0.0033812522888183594,
      "learning_rate": 3.668269137308666e-08,
      "loss": -0.0022,
      "num_tokens": 43903588.0,
      "reward": 1.0605829954147339,
      "reward_std": 0.17206411063671112,
      "rewards/cross_solution_unittest_reward/mean": 0.5302915126085281,
      "rewards/cross_solution_unittest_reward/std": 0.2089696228504181,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 108
    },
    {
      "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": -5.0,
      "completions/max_length": 940.0,
      "completions/max_terminated_length": 940.0,
      "completions/mean_length": 456.51824951171875,
      "completions/mean_terminated_length": 456.51824951171875,
      "completions/min_length": 188.0,
      "completions/min_terminated_length": 188.0,
      "entropy": 0.1035308837890625,
      "epoch": 0.9945822640433419,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.034084659069776535,
      "kl": 0.004221439361572266,
      "learning_rate": 1.630896073864352e-08,
      "loss": 0.0052,
      "num_tokens": 44271795.0,
      "reward": 1.0674467086791992,
      "reward_std": 0.20672927796840668,
      "rewards/cross_solution_unittest_reward/mean": 0.5337233692407608,
      "rewards/cross_solution_unittest_reward/std": 0.2225872203707695,
      "rewards/format_reward/mean": 0.0,
      "rewards/format_reward/std": 0.0,
      "step": 109
    },
    {
      "epoch": 0.9945822640433419,
      "step": 109,
      "total_flos": 0.0,
      "train_loss": 0.0018782940647274324,
      "train_runtime": 42443.7986,
      "train_samples_per_second": 0.031,
      "train_steps_per_second": 0.003
    }
  ],
  "logging_steps": 1,
  "max_steps": 110,
  "num_input_tokens_seen": 44271795,
  "num_train_epochs": 1,
  "save_steps": 25,
  "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": 2,
  "trial_name": null,
  "trial_params": null
}