{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9578544061302682,
  "eval_steps": 500,
  "global_step": 2000,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 251.94140625,
      "completions/mean_terminated_length": 251.94140625,
      "completions/min_length": 165.0,
      "completions/min_terminated_length": 165.0,
      "entropy": 1.8460286458333333,
      "epoch": 0.0004789272030651341,
      "fork/depth_mean": 0.715498149394989,
      "fork/depth_std": 0.24286088347434998,
      "fork_head/loss": 2.829284906387329,
      "fork_head/num_positive": 6.0,
      "fork_head/pred_top1_mass": 0.006579761393368244,
      "frac_reward_zero_std": 0.59375,
      "grad_norm": 5.722615718841553,
      "learning_rate": 3e-06,
      "loss": 0.7092,
      "num_tokens": 220466.0,
      "reward": 0.4437796473503113,
      "reward_std": 0.4658518433570862,
      "rewards/code_format_reward/mean": 0.22265625,
      "rewards/code_format_reward/std": 0.4020913541316986,
      "rewards/code_reward/mean": 0.22112339735031128,
      "rewards/code_reward/std": 0.30220580101013184,
      "step": 1,
      "step_time": 455.6584514901042
    },
    {
      "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,
      "entropy": 1.4259259237183466,
      "epoch": 0.0009578544061302681,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14364464581012726,
      "learning_rate": 3e-06,
      "loss": -0.0042,
      "step": 2,
      "step_time": 9.726059943437576
    },
    {
      "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,
      "entropy": 1.8463541666666667,
      "epoch": 0.0014367816091954023,
      "fork_head/loss": 2.6864945888519287,
      "fork_head/num_positive": 6.0,
      "fork_head/pred_top1_mass": 0.020550303161144257,
      "grad_norm": 5.33871603012085,
      "learning_rate": 3e-06,
      "loss": 0.675,
      "step": 3,
      "step_time": 6.980820283293724
    },
    {
      "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,
      "entropy": 1.4289279513888888,
      "epoch": 0.0019157088122605363,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13988333940505981,
      "learning_rate": 3e-06,
      "loss": -0.0054,
      "step": 4,
      "step_time": 7.808648154139519
    },
    {
      "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,
      "entropy": 1.8598090277777777,
      "epoch": 0.0023946360153256703,
      "fork_head/loss": 2.330185890197754,
      "fork_head/num_positive": 6.0,
      "fork_head/pred_top1_mass": 0.34331420063972473,
      "grad_norm": 4.0790276527404785,
      "learning_rate": 3e-06,
      "loss": 0.5924,
      "step": 5,
      "step_time": 8.139731802046299
    },
    {
      "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,
      "entropy": 1.423828125,
      "epoch": 0.0028735632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.156802237033844,
      "learning_rate": 3e-06,
      "loss": -0.0055,
      "step": 6,
      "step_time": 7.0433134958148
    },
    {
      "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,
      "entropy": 1.8321397569444444,
      "epoch": 0.003352490421455939,
      "fork_head/loss": 1.8361632823944092,
      "fork_head/num_positive": 6.0,
      "fork_head/pred_top1_mass": 0.34430134296417236,
      "grad_norm": 1.45439875125885,
      "learning_rate": 3e-06,
      "loss": 0.4861,
      "step": 7,
      "step_time": 8.519813820719719
    },
    {
      "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,
      "entropy": 1.4141348401705425,
      "epoch": 0.0038314176245210726,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12923823297023773,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 8,
      "step_time": 7.293802157044411
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 248.427734375,
      "completions/mean_terminated_length": 248.4129180908203,
      "completions/min_length": 94.0,
      "completions/min_terminated_length": 94.0,
      "entropy": 1.2770898302396139,
      "epoch": 0.004310344827586207,
      "fork/depth_mean": 0.44066473841667175,
      "fork/depth_std": 0.2581905424594879,
      "fork_head/loss": 2.753633737564087,
      "fork_head/num_positive": 8.366666793823242,
      "fork_head/pred_top1_mass": 0.07240662723779678,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 1.4421236515045166,
      "learning_rate": 3e-06,
      "loss": 0.6937,
      "num_tokens": 446333.0,
      "reward": 0.42859503626823425,
      "reward_std": 0.5597556233406067,
      "rewards/code_format_reward/mean": 0.2841796875,
      "rewards/code_format_reward/std": 0.4343424141407013,
      "rewards/code_reward/mean": 0.14441536366939545,
      "rewards/code_reward/std": 0.29635319113731384,
      "step": 9,
      "step_time": 710.4428443983197
    },
    {
      "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,
      "entropy": 1.742008239030838,
      "epoch": 0.004789272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.26099786162376404,
      "learning_rate": 3e-06,
      "loss": -0.0155,
      "step": 10,
      "step_time": 9.12552060931921
    },
    {
      "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,
      "entropy": 1.2774522423744201,
      "epoch": 0.005268199233716475,
      "fork_head/loss": 2.712696075439453,
      "fork_head/num_positive": 8.366666793823242,
      "fork_head/pred_top1_mass": 0.07952509075403214,
      "grad_norm": 1.3114644289016724,
      "learning_rate": 3e-06,
      "loss": 0.685,
      "step": 11,
      "step_time": 9.270118854939938
    },
    {
      "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,
      "entropy": 1.7354290187358856,
      "epoch": 0.005747126436781609,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.23080012202262878,
      "learning_rate": 3e-06,
      "loss": -0.0157,
      "step": 12,
      "step_time": 8.311069034039974
    },
    {
      "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,
      "entropy": 1.2852387269337973,
      "epoch": 0.006226053639846743,
      "fork_head/loss": 2.6701886653900146,
      "fork_head/num_positive": 8.366666793823242,
      "fork_head/pred_top1_mass": 0.09189368784427643,
      "grad_norm": 1.1719558238983154,
      "learning_rate": 3e-06,
      "loss": 0.674,
      "step": 13,
      "step_time": 8.67317271977663
    },
    {
      "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,
      "entropy": 1.7420381605625153,
      "epoch": 0.006704980842911878,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18211869895458221,
      "learning_rate": 3e-06,
      "loss": -0.0157,
      "step": 14,
      "step_time": 8.545758605003357
    },
    {
      "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,
      "entropy": 1.2812564969062805,
      "epoch": 0.007183908045977011,
      "fork_head/loss": 2.6094284057617188,
      "fork_head/num_positive": 8.366666793823242,
      "fork_head/pred_top1_mass": 0.15050630271434784,
      "grad_norm": 1.00422203540802,
      "learning_rate": 3e-06,
      "loss": 0.6579,
      "step": 15,
      "step_time": 8.295676939189434
    },
    {
      "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,
      "entropy": 1.743624433875084,
      "epoch": 0.007662835249042145,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19789303839206696,
      "learning_rate": 3e-06,
      "loss": -0.0162,
      "step": 16,
      "step_time": 8.634957537055016
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 250.794921875,
      "completions/mean_terminated_length": 250.76425170898438,
      "completions/min_length": 152.0,
      "completions/min_terminated_length": 152.0,
      "entropy": 1.7393534295260906,
      "epoch": 0.008141762452107279,
      "fork/depth_mean": 0.2582264840602875,
      "fork/depth_std": 0.17530721426010132,
      "fork_head/loss": 2.718217134475708,
      "fork_head/num_positive": 10.640000343322754,
      "fork_head/pred_top1_mass": 0.07392557710409164,
      "frac_reward_zero_std": 0.21875,
      "grad_norm": 0.8959291577339172,
      "learning_rate": 3e-06,
      "loss": 0.6891,
      "num_tokens": 666244.0,
      "reward": 0.32312896847724915,
      "reward_std": 0.44159135222435,
      "rewards/code_format_reward/mean": 0.0849609375,
      "rewards/code_format_reward/std": 0.26928257942199707,
      "rewards/code_reward/mean": 0.23816804587841034,
      "rewards/code_reward/std": 0.3781704008579254,
      "step": 17,
      "step_time": 793.6200416907668
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.011111111442248026,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.011111111442248026,
      "entropy": 1.5538650274276733,
      "epoch": 0.008620689655172414,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09454979002475739,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 18,
      "step_time": 8.78907810151577
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.015625,
      "entropy": 1.7502332888543606,
      "epoch": 0.009099616858237548,
      "fork_head/loss": 2.692525863647461,
      "fork_head/num_positive": 10.640000343322754,
      "fork_head/pred_top1_mass": 0.07988409698009491,
      "grad_norm": 0.8463331460952759,
      "learning_rate": 3e-06,
      "loss": 0.6833,
      "step": 19,
      "step_time": 9.429565943777561
    },
    {
      "clip_ratio/high_max": 0.022222222884496053,
      "clip_ratio/high_mean": 0.005555555721124013,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.005555555721124013,
      "entropy": 1.5574225942293802,
      "epoch": 0.009578544061302681,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11352519690990448,
      "learning_rate": 3e-06,
      "loss": -0.004,
      "step": 20,
      "step_time": 8.314368918538094
    },
    {
      "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,
      "entropy": 1.742492001503706,
      "epoch": 0.010057471264367816,
      "fork_head/loss": 2.659151077270508,
      "fork_head/num_positive": 10.640000343322754,
      "fork_head/pred_top1_mass": 0.09344788640737534,
      "grad_norm": 0.7682421803474426,
      "learning_rate": 3e-06,
      "loss": 0.674,
      "step": 21,
      "step_time": 8.567935965955257
    },
    {
      "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,
      "entropy": 1.545882531007131,
      "epoch": 0.01053639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09155088663101196,
      "learning_rate": 3e-06,
      "loss": -0.0044,
      "step": 22,
      "step_time": 8.534242175519466
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.015625,
      "entropy": 1.7478237636387348,
      "epoch": 0.011015325670498084,
      "fork_head/loss": 2.6274337768554688,
      "fork_head/num_positive": 10.640000343322754,
      "fork_head/pred_top1_mass": 0.12934909760951996,
      "grad_norm": 0.6521260738372803,
      "learning_rate": 3e-06,
      "loss": 0.6661,
      "step": 23,
      "step_time": 8.154608376324177
    },
    {
      "clip_ratio/high_max": 0.022222222884496053,
      "clip_ratio/high_mean": 0.005555555721124013,
      "clip_ratio/low_mean": 0.011111111442248026,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.01666666716337204,
      "entropy": 1.5564453125,
      "epoch": 0.011494252873563218,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09396857023239136,
      "learning_rate": 3e-06,
      "loss": -0.0041,
      "step": 24,
      "step_time": 8.83865650743246
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 248.677734375,
      "completions/mean_terminated_length": 248.64903259277344,
      "completions/min_length": 93.0,
      "completions/min_terminated_length": 93.0,
      "entropy": 1.7377224445343018,
      "epoch": 0.011973180076628353,
      "fork/depth_mean": 0.2559109628200531,
      "fork/depth_std": 0.2323300689458847,
      "fork_head/loss": 2.7329134941101074,
      "fork_head/num_positive": 10.535714149475098,
      "fork_head/pred_top1_mass": 0.06558293849229813,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 1.0918742418289185,
      "learning_rate": 3e-06,
      "loss": 0.685,
      "num_tokens": 891887.0,
      "reward": 0.27595603466033936,
      "reward_std": 0.4314592480659485,
      "rewards/code_format_reward/mean": 0.08203125,
      "rewards/code_format_reward/std": 0.254334956407547,
      "rewards/code_reward/mean": 0.19392478466033936,
      "rewards/code_reward/std": 0.35153821110725403,
      "step": 25,
      "step_time": 1105.6121295690536
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.2284078853470939,
      "epoch": 0.012452107279693486,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09742733091115952,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 26,
      "step_time": 8.341794639825821
    },
    {
      "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,
      "entropy": 1.7366048177083333,
      "epoch": 0.01293103448275862,
      "fork_head/loss": 2.709704875946045,
      "fork_head/num_positive": 10.535714149475098,
      "fork_head/pred_top1_mass": 0.09324157238006592,
      "grad_norm": 0.943398118019104,
      "learning_rate": 3e-06,
      "loss": 0.6783,
      "step": 27,
      "step_time": 9.125020310282707
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.232131234237126,
      "epoch": 0.013409961685823755,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0971648320555687,
      "learning_rate": 3e-06,
      "loss": -0.0018,
      "step": 28,
      "step_time": 7.5903173834085464
    },
    {
      "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,
      "entropy": 1.7322721401850383,
      "epoch": 0.013888888888888888,
      "fork_head/loss": 2.677337408065796,
      "fork_head/num_positive": 10.535714149475098,
      "fork_head/pred_top1_mass": 0.11638545244932175,
      "grad_norm": 0.7593768835067749,
      "learning_rate": 3e-06,
      "loss": 0.6694,
      "step": 29,
      "step_time": 8.25031067430973
    },
    {
      "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,
      "entropy": 1.239742981536048,
      "epoch": 0.014367816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10931398719549179,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 30,
      "step_time": 7.576368257403374
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.016666666666666666,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.016666666666666666,
      "entropy": 1.7413650274276733,
      "epoch": 0.014846743295019157,
      "fork_head/loss": 2.6471290588378906,
      "fork_head/num_positive": 10.535714149475098,
      "fork_head/pred_top1_mass": 0.11972916126251221,
      "grad_norm": 0.6792694926261902,
      "learning_rate": 3e-06,
      "loss": 0.66,
      "step": 31,
      "step_time": 7.997422412037849
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.2369597852230072,
      "epoch": 0.01532567049808429,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09656476229429245,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 32,
      "step_time": 8.30939307063818
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.017578125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 251.314453125,
      "completions/mean_terminated_length": 251.23060607910156,
      "completions/min_length": 145.0,
      "completions/min_terminated_length": 145.0,
      "entropy": 1.7289887229601542,
      "epoch": 0.015804597701149427,
      "fork/depth_mean": 0.24849361181259155,
      "fork/depth_std": 0.20971310138702393,
      "fork_head/loss": 2.750793218612671,
      "fork_head/num_positive": 10.392857551574707,
      "fork_head/pred_top1_mass": 0.05677121505141258,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.9009348154067993,
      "learning_rate": 3e-06,
      "loss": 0.6858,
      "num_tokens": 1114896.0,
      "reward": 0.3341137766838074,
      "reward_std": 0.45625680685043335,
      "rewards/code_format_reward/mean": 0.166015625,
      "rewards/code_format_reward/std": 0.3466040790081024,
      "rewards/code_reward/mean": 0.16809815168380737,
      "rewards/code_reward/std": 0.326517790555954,
      "step": 33,
      "step_time": 753.8249189481139
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.03125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "entropy": 1.5283087976276875,
      "epoch": 0.016283524904214558,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2436455488204956,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 34,
      "step_time": 8.427571676671505
    },
    {
      "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,
      "entropy": 1.7374486406644185,
      "epoch": 0.016762452107279693,
      "fork_head/loss": 2.7354328632354736,
      "fork_head/num_positive": 10.392857551574707,
      "fork_head/pred_top1_mass": 0.07823926955461502,
      "grad_norm": 0.90395188331604,
      "learning_rate": 3e-06,
      "loss": 0.6828,
      "step": 35,
      "step_time": 9.168097496032715
    },
    {
      "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,
      "entropy": 1.5437405072152615,
      "epoch": 0.017241379310344827,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.24566903710365295,
      "learning_rate": 3e-06,
      "loss": -0.0007,
      "step": 36,
      "step_time": 7.513980835676193
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.716335368156433,
      "epoch": 0.017720306513409962,
      "fork_head/loss": 2.7076592445373535,
      "fork_head/num_positive": 10.392857551574707,
      "fork_head/pred_top1_mass": 0.07823926955461502,
      "grad_norm": 0.839942991733551,
      "learning_rate": 3e-06,
      "loss": 0.6733,
      "step": 37,
      "step_time": 8.510633528232574
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.03125,
      "clip_ratio/low_mean": 0.028125000186264515,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.059375000186264515,
      "entropy": 1.52667236328125,
      "epoch": 0.018199233716475097,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19772188365459442,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 38,
      "step_time": 8.190028190612793
    },
    {
      "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,
      "entropy": 1.7277878999710083,
      "epoch": 0.01867816091954023,
      "fork_head/loss": 2.6737723350524902,
      "fork_head/num_positive": 10.392857551574707,
      "fork_head/pred_top1_mass": 0.10476752370595932,
      "grad_norm": 0.7709957361221313,
      "learning_rate": 3e-06,
      "loss": 0.6646,
      "step": 39,
      "step_time": 8.10255840420723
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.012500000186264515,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.012500000186264515,
      "entropy": 1.546929258853197,
      "epoch": 0.019157088122605363,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20185846090316772,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 40,
      "step_time": 7.5891823545098305
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 252.029296875,
      "completions/mean_terminated_length": 252.01373291015625,
      "completions/min_length": 137.0,
      "completions/min_terminated_length": 137.0,
      "entropy": 1.7633087188005447,
      "epoch": 0.019636015325670497,
      "fork/depth_mean": 0.2388589233160019,
      "fork/depth_std": 0.1947963982820511,
      "fork_head/loss": 2.748295783996582,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.06624330580234528,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.8751810789108276,
      "learning_rate": 3e-06,
      "loss": 0.6825,
      "num_tokens": 1334207.0,
      "reward": 0.31063133478164673,
      "reward_std": 0.394540935754776,
      "rewards/code_format_reward/mean": 0.11328125,
      "rewards/code_format_reward/std": 0.29486700892448425,
      "rewards/code_reward/mean": 0.19735011458396912,
      "rewards/code_reward/std": 0.31829386949539185,
      "step": 41,
      "step_time": 782.2157406955957
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.012500000186264515,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.012500000186264515,
      "entropy": 1.6175781413912773,
      "epoch": 0.020114942528735632,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19079811871051788,
      "learning_rate": 3e-06,
      "loss": 0.0036,
      "step": 42,
      "step_time": 8.717117220163345
    },
    {
      "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,
      "entropy": 1.7655273377895355,
      "epoch": 0.020593869731800767,
      "fork_head/loss": 2.7313876152038574,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.08982071280479431,
      "grad_norm": 0.8152995109558105,
      "learning_rate": 3e-06,
      "loss": 0.6785,
      "step": 43,
      "step_time": 9.632876090705395
    },
    {
      "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,
      "entropy": 1.5997585766017437,
      "epoch": 0.0210727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.24043777585029602,
      "learning_rate": 3e-06,
      "loss": 0.004,
      "step": 44,
      "step_time": 8.82364346086979
    },
    {
      "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,
      "entropy": 1.7628424018621445,
      "epoch": 0.021551724137931036,
      "fork_head/loss": 2.706272840499878,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.07735563814640045,
      "grad_norm": 0.748870849609375,
      "learning_rate": 3e-06,
      "loss": 0.6721,
      "step": 45,
      "step_time": 8.82179507613182
    },
    {
      "clip_ratio/high_max": 0.03125,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.010416666977107525,
      "entropy": 1.6217841170728207,
      "epoch": 0.022030651340996167,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19158822298049927,
      "learning_rate": 3e-06,
      "loss": 0.0036,
      "step": 46,
      "step_time": 9.283285796642303
    },
    {
      "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,
      "entropy": 1.7575988844037056,
      "epoch": 0.022509578544061302,
      "fork_head/loss": 2.676161527633667,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.09021718055009842,
      "grad_norm": 0.7127240896224976,
      "learning_rate": 3e-06,
      "loss": 0.6637,
      "step": 47,
      "step_time": 8.52501928806305
    },
    {
      "clip_ratio/high_max": 0.03125,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.010416666977107525,
      "entropy": 1.6177751384675503,
      "epoch": 0.022988505747126436,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17444540560245514,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 48,
      "step_time": 9.350535936653614
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 249.806640625,
      "completions/mean_terminated_length": 249.7701416015625,
      "completions/min_length": 141.0,
      "completions/min_terminated_length": 141.0,
      "entropy": 1.645231120288372,
      "epoch": 0.02346743295019157,
      "fork/depth_mean": 0.20813676714897156,
      "fork/depth_std": 0.2023322731256485,
      "fork_head/loss": 2.7251060009002686,
      "fork_head/num_positive": 11.071428298950195,
      "fork_head/pred_top1_mass": 0.07330572605133057,
      "frac_reward_zero_std": 0.125,
      "grad_norm": 0.7912424802780151,
      "learning_rate": 3e-06,
      "loss": 0.668,
      "num_tokens": 1560380.0,
      "reward": 0.4000995457172394,
      "reward_std": 0.5264614224433899,
      "rewards/code_format_reward/mean": 0.1884765625,
      "rewards/code_format_reward/std": 0.380703866481781,
      "rewards/code_reward/mean": 0.21162298321723938,
      "rewards/code_reward/std": 0.3662030100822449,
      "step": 49,
      "step_time": 891.1354386433959
    },
    {
      "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,
      "entropy": 1.6503204504648845,
      "epoch": 0.023946360153256706,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.33536002039909363,
      "learning_rate": 3e-06,
      "loss": 0.0157,
      "step": 50,
      "step_time": 8.483312383294106
    },
    {
      "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,
      "entropy": 1.6503580585122108,
      "epoch": 0.02442528735632184,
      "fork_head/loss": 2.7123520374298096,
      "fork_head/num_positive": 11.071428298950195,
      "fork_head/pred_top1_mass": 0.07811455428600311,
      "grad_norm": 0.7666037678718567,
      "learning_rate": 3e-06,
      "loss": 0.6627,
      "step": 51,
      "step_time": 8.935653232038021
    },
    {
      "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,
      "entropy": 1.6521585583686829,
      "epoch": 0.02490421455938697,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.3115832209587097,
      "learning_rate": 3e-06,
      "loss": 0.0144,
      "step": 52,
      "step_time": 8.003259278833866
    },
    {
      "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,
      "entropy": 1.6578003093600273,
      "epoch": 0.025383141762452106,
      "fork_head/loss": 2.6937143802642822,
      "fork_head/num_positive": 11.071428298950195,
      "fork_head/pred_top1_mass": 0.09371833503246307,
      "grad_norm": 0.6712632775306702,
      "learning_rate": 3e-06,
      "loss": 0.6555,
      "step": 53,
      "step_time": 7.743285425007343
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.6701960484186809,
      "epoch": 0.02586206896551724,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2647401988506317,
      "learning_rate": 3e-06,
      "loss": 0.0113,
      "step": 54,
      "step_time": 8.436087384819984
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.05208333395421505,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05208333395421505,
      "entropy": 1.6500091552734375,
      "epoch": 0.026340996168582376,
      "fork_head/loss": 2.673043727874756,
      "fork_head/num_positive": 11.071428298950195,
      "fork_head/pred_top1_mass": 0.09872584044933319,
      "grad_norm": 0.5988104939460754,
      "learning_rate": 3e-06,
      "loss": 0.6455,
      "step": 55,
      "step_time": 7.842362232506275
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.6658297260602315,
      "epoch": 0.02681992337164751,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2772415578365326,
      "learning_rate": 3e-06,
      "loss": 0.0108,
      "step": 56,
      "step_time": 7.979304105043411
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 250.4765625,
      "completions/mean_terminated_length": 250.4765625,
      "completions/min_length": 121.0,
      "completions/min_terminated_length": 121.0,
      "entropy": 1.4555562287569046,
      "epoch": 0.027298850574712645,
      "fork/depth_mean": 0.222345769405365,
      "fork/depth_std": 0.2207973152399063,
      "fork_head/loss": 2.7522084712982178,
      "fork_head/num_positive": 11.354838371276855,
      "fork_head/pred_top1_mass": 0.05551491305232048,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7934682369232178,
      "learning_rate": 3e-06,
      "loss": 0.67,
      "num_tokens": 1784864.0,
      "reward": 0.38203611969947815,
      "reward_std": 0.45684391260147095,
      "rewards/code_format_reward/mean": 0.1611328125,
      "rewards/code_format_reward/std": 0.3425372838973999,
      "rewards/code_reward/mean": 0.22090329229831696,
      "rewards/code_reward/std": 0.35053515434265137,
      "step": 57,
      "step_time": 786.2950336560607
    },
    {
      "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,
      "entropy": 1.4506076256434122,
      "epoch": 0.027777777777777776,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1891869455575943,
      "learning_rate": 3e-06,
      "loss": 0.0211,
      "step": 58,
      "step_time": 8.184207864105701
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.4461568146944046,
      "epoch": 0.02825670498084291,
      "fork_head/loss": 2.737870693206787,
      "fork_head/num_positive": 11.354838371276855,
      "fork_head/pred_top1_mass": 0.058239273726940155,
      "grad_norm": 0.7308641672134399,
      "learning_rate": 3e-06,
      "loss": 0.6661,
      "step": 59,
      "step_time": 7.851589113473892
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.4394965291023254,
      "epoch": 0.028735632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19798265397548676,
      "learning_rate": 3e-06,
      "loss": 0.0214,
      "step": 60,
      "step_time": 8.355403795838356
    },
    {
      "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,
      "entropy": 1.473968505859375,
      "epoch": 0.02921455938697318,
      "fork_head/loss": 2.716456174850464,
      "fork_head/num_positive": 11.354838371276855,
      "fork_head/pred_top1_mass": 0.058768127113580704,
      "grad_norm": 0.6539332270622253,
      "learning_rate": 3e-06,
      "loss": 0.6608,
      "step": 61,
      "step_time": 7.524000406265259
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.4532118002573648,
      "epoch": 0.029693486590038315,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1937580704689026,
      "learning_rate": 3e-06,
      "loss": 0.0203,
      "step": 62,
      "step_time": 8.41039239615202
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.4522705078125,
      "epoch": 0.03017241379310345,
      "fork_head/loss": 2.6924941539764404,
      "fork_head/num_positive": 11.354838371276855,
      "fork_head/pred_top1_mass": 0.07813070714473724,
      "grad_norm": 0.6078172922134399,
      "learning_rate": 3e-06,
      "loss": 0.6539,
      "step": 63,
      "step_time": 7.931280635297298
    },
    {
      "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,
      "entropy": 1.4613498330116272,
      "epoch": 0.03065134099616858,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18468615412712097,
      "learning_rate": 3e-06,
      "loss": 0.0187,
      "step": 64,
      "step_time": 7.987094931304455
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.016666666666666666,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.016666666666666666,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 250.888671875,
      "completions/mean_terminated_length": 250.85855102539062,
      "completions/min_length": 110.0,
      "completions/min_terminated_length": 110.0,
      "entropy": 1.6608045895894368,
      "epoch": 0.031130268199233715,
      "fork/depth_mean": 0.1971900761127472,
      "fork/depth_std": 0.16759760677814484,
      "fork_head/loss": 2.7267963886260986,
      "fork_head/num_positive": 10.870967864990234,
      "fork_head/pred_top1_mass": 0.07507643848657608,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7493215799331665,
      "learning_rate": 3e-06,
      "loss": 0.6673,
      "num_tokens": 2021607.0,
      "reward": 0.43861904740333557,
      "reward_std": 0.5203312635421753,
      "rewards/code_format_reward/mean": 0.1875,
      "rewards/code_format_reward/std": 0.3647909164428711,
      "rewards/code_reward/mean": 0.25111907720565796,
      "rewards/code_reward/std": 0.38289710879325867,
      "step": 65,
      "step_time": 732.4875741899014
    },
    {
      "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,
      "entropy": 1.6049330870310465,
      "epoch": 0.031609195402298854,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2641911506652832,
      "learning_rate": 3e-06,
      "loss": 0.0124,
      "step": 66,
      "step_time": 9.425761252641678
    },
    {
      "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,
      "entropy": 1.6371961832046509,
      "epoch": 0.032088122605363985,
      "fork_head/loss": 2.7112691402435303,
      "fork_head/num_positive": 10.870967864990234,
      "fork_head/pred_top1_mass": 0.07623767852783203,
      "grad_norm": 0.7106398344039917,
      "learning_rate": 3e-06,
      "loss": 0.6633,
      "step": 67,
      "step_time": 8.422638155519962
    },
    {
      "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,
      "entropy": 1.6135459979375204,
      "epoch": 0.032567049808429116,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.260549932718277,
      "learning_rate": 3e-06,
      "loss": 0.0123,
      "step": 68,
      "step_time": 8.237969636917114
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.020833333333333332,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.020833333333333332,
      "entropy": 1.6614691734313964,
      "epoch": 0.033045977011494254,
      "fork_head/loss": 2.6891586780548096,
      "fork_head/num_positive": 10.870967864990234,
      "fork_head/pred_top1_mass": 0.08659953624010086,
      "grad_norm": 0.6861870884895325,
      "learning_rate": 3e-06,
      "loss": 0.6545,
      "step": 69,
      "step_time": 8.314754910767078
    },
    {
      "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,
      "entropy": 1.617364724477132,
      "epoch": 0.033524904214559385,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2835683226585388,
      "learning_rate": 3e-06,
      "loss": 0.0097,
      "step": 70,
      "step_time": 8.035150483250618
    },
    {
      "clip_ratio/high_max": 0.03333333333333333,
      "clip_ratio/high_mean": 0.008333333333333333,
      "clip_ratio/low_mean": 0.05416666766007741,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06250000099341074,
      "entropy": 1.663880745569865,
      "epoch": 0.034003831417624523,
      "fork_head/loss": 2.6651015281677246,
      "fork_head/num_positive": 10.870967864990234,
      "fork_head/pred_top1_mass": 0.08448668569326401,
      "grad_norm": 0.5829343795776367,
      "learning_rate": 3e-06,
      "loss": 0.6479,
      "step": 71,
      "step_time": 8.726863972842693
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023333333432674408,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023333333432674408,
      "entropy": 1.6176114002863566,
      "epoch": 0.034482758620689655,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.27477702498435974,
      "learning_rate": 3e-06,
      "loss": 0.0093,
      "step": 72,
      "step_time": 8.470994397997856
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 248.330078125,
      "completions/mean_terminated_length": 248.28488159179688,
      "completions/min_length": 89.0,
      "completions/min_terminated_length": 89.0,
      "entropy": 1.5874742306768894,
      "epoch": 0.034961685823754786,
      "fork/depth_mean": 0.2345808893442154,
      "fork/depth_std": 0.22355885803699493,
      "fork_head/loss": 2.7349355220794678,
      "fork_head/num_positive": 10.4375,
      "fork_head/pred_top1_mass": 0.07046256959438324,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8004422783851624,
      "learning_rate": 3e-06,
      "loss": 0.6914,
      "num_tokens": 2241584.0,
      "reward": 0.41280221939086914,
      "reward_std": 0.49552085995674133,
      "rewards/code_format_reward/mean": 0.1796875,
      "rewards/code_format_reward/std": 0.3687094748020172,
      "rewards/code_reward/mean": 0.23311470448970795,
      "rewards/code_reward/std": 0.37320205569267273,
      "step": 73,
      "step_time": 824.2227686494589
    },
    {
      "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,
      "entropy": 1.450536028544108,
      "epoch": 0.035440613026819924,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14568890631198883,
      "learning_rate": 3e-06,
      "loss": -0.001,
      "step": 74,
      "step_time": 8.624849006533623
    },
    {
      "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,
      "entropy": 1.5952541679143906,
      "epoch": 0.035919540229885055,
      "fork_head/loss": 2.719402551651001,
      "fork_head/num_positive": 10.4375,
      "fork_head/pred_top1_mass": 0.07775117456912994,
      "grad_norm": 0.7479220628738403,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "step": 75,
      "step_time": 9.059083499014378
    },
    {
      "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,
      "entropy": 1.4489116748174031,
      "epoch": 0.03639846743295019,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13552092015743256,
      "learning_rate": 3e-06,
      "loss": -0.0015,
      "step": 76,
      "step_time": 7.298877276480198
    },
    {
      "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,
      "entropy": 1.6073900014162064,
      "epoch": 0.036877394636015325,
      "fork_head/loss": 2.695291519165039,
      "fork_head/num_positive": 10.4375,
      "fork_head/pred_top1_mass": 0.10276663303375244,
      "grad_norm": 0.6762702465057373,
      "learning_rate": 3e-06,
      "loss": 0.6792,
      "step": 77,
      "step_time": 8.404901273548603
    },
    {
      "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,
      "entropy": 1.455418841044108,
      "epoch": 0.03735632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13262280821800232,
      "learning_rate": 3e-06,
      "loss": -0.0023,
      "step": 78,
      "step_time": 7.432597532868385
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03125,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "entropy": 1.6003662087023258,
      "epoch": 0.037835249042145594,
      "fork_head/loss": 2.668043851852417,
      "fork_head/num_positive": 10.4375,
      "fork_head/pred_top1_mass": 0.11752656102180481,
      "grad_norm": 0.6513739228248596,
      "learning_rate": 3e-06,
      "loss": 0.6719,
      "step": 79,
      "step_time": 8.00691781193018
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.016666666666666666,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.016666666666666666,
      "entropy": 1.4543858448664346,
      "epoch": 0.038314176245210725,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12417446076869965,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 80,
      "step_time": 7.964618980884552
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 249.083984375,
      "completions/mean_terminated_length": 249.056884765625,
      "completions/min_length": 116.0,
      "completions/min_terminated_length": 116.0,
      "entropy": 1.559768684208393,
      "epoch": 0.03879310344827586,
      "fork/depth_mean": 0.26469525694847107,
      "fork/depth_std": 0.25598645210266113,
      "fork_head/loss": 2.7449216842651367,
      "fork_head/num_positive": 10.100000381469727,
      "fork_head/pred_top1_mass": 0.07169187068939209,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 1.097815990447998,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "num_tokens": 2466411.0,
      "reward": 0.30931127071380615,
      "reward_std": 0.4227736294269562,
      "rewards/code_format_reward/mean": 0.1826171875,
      "rewards/code_format_reward/std": 0.3530002534389496,
      "rewards/code_reward/mean": 0.12669408321380615,
      "rewards/code_reward/std": 0.2702362537384033,
      "step": 81,
      "step_time": 730.799570158124
    },
    {
      "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,
      "entropy": 1.6437723711133003,
      "epoch": 0.039272030651340994,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15933610498905182,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 82,
      "step_time": 8.349621079862118
    },
    {
      "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,
      "entropy": 1.552098587155342,
      "epoch": 0.03975095785440613,
      "fork_head/loss": 2.7245538234710693,
      "fork_head/num_positive": 10.100000381469727,
      "fork_head/pred_top1_mass": 0.07267247885465622,
      "grad_norm": 0.9375616908073425,
      "learning_rate": 3e-06,
      "loss": 0.6718,
      "step": 83,
      "step_time": 8.67477621883154
    },
    {
      "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,
      "entropy": 1.647070299834013,
      "epoch": 0.040229885057471264,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14117228984832764,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 84,
      "step_time": 8.112283036112785
    },
    {
      "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,
      "entropy": 1.551152542233467,
      "epoch": 0.040708812260536395,
      "fork_head/loss": 2.694902181625366,
      "fork_head/num_positive": 10.100000381469727,
      "fork_head/pred_top1_mass": 0.07221952080726624,
      "grad_norm": 0.7618127465248108,
      "learning_rate": 3e-06,
      "loss": 0.6629,
      "step": 85,
      "step_time": 8.310085572302341
    },
    {
      "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,
      "entropy": 1.66017659381032,
      "epoch": 0.04118773946360153,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1514984369277954,
      "learning_rate": 3e-06,
      "loss": 0.0023,
      "step": 86,
      "step_time": 8.739887230098248
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.015625,
      "entropy": 1.5509236603975296,
      "epoch": 0.041666666666666664,
      "fork_head/loss": 2.6618919372558594,
      "fork_head/num_positive": 10.100000381469727,
      "fork_head/pred_top1_mass": 0.0804135799407959,
      "grad_norm": 0.7262336015701294,
      "learning_rate": 3e-06,
      "loss": 0.6528,
      "step": 87,
      "step_time": 8.063064388930798
    },
    {
      "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,
      "entropy": 1.657354399561882,
      "epoch": 0.0421455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15493851900100708,
      "learning_rate": 3e-06,
      "loss": 0.0011,
      "step": 88,
      "step_time": 7.90926743298769
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.009765625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 247.62890625,
      "completions/mean_terminated_length": 247.54635620117188,
      "completions/min_length": 106.0,
      "completions/min_terminated_length": 106.0,
      "entropy": 1.31738892942667,
      "epoch": 0.042624521072796934,
      "fork/depth_mean": 0.23038941621780396,
      "fork/depth_std": 0.2390764057636261,
      "fork_head/loss": 2.757599353790283,
      "fork_head/num_positive": 10.633333206176758,
      "fork_head/pred_top1_mass": 0.07190661132335663,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.8058872222900391,
      "learning_rate": 3e-06,
      "loss": 0.6966,
      "num_tokens": 2686493.0,
      "reward": 0.37274467945098877,
      "reward_std": 0.45990586280822754,
      "rewards/code_format_reward/mean": 0.171875,
      "rewards/code_format_reward/std": 0.35768061876296997,
      "rewards/code_reward/mean": 0.20086967945098877,
      "rewards/code_reward/std": 0.3394486904144287,
      "step": 89,
      "step_time": 855.620008431375
    },
    {
      "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,
      "entropy": 1.6228190104166667,
      "epoch": 0.04310344827586207,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2176692932844162,
      "learning_rate": 3e-06,
      "loss": -0.004,
      "step": 90,
      "step_time": 7.985640503466129
    },
    {
      "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,
      "entropy": 1.3261535465717316,
      "epoch": 0.0435823754789272,
      "fork_head/loss": 2.742584466934204,
      "fork_head/num_positive": 10.633333206176758,
      "fork_head/pred_top1_mass": 0.07757739722728729,
      "grad_norm": 0.7229804396629333,
      "learning_rate": 3e-06,
      "loss": 0.6914,
      "step": 91,
      "step_time": 7.999469868838787
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.616308589776357,
      "epoch": 0.044061302681992334,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.22253946959972382,
      "learning_rate": 3e-06,
      "loss": -0.0042,
      "step": 92,
      "step_time": 7.795591339468956
    },
    {
      "clip_ratio/high_max": 0.02083333395421505,
      "clip_ratio/high_mean": 0.0052083334885537624,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.06770833348855376,
      "entropy": 1.3282338306307793,
      "epoch": 0.04454022988505747,
      "fork_head/loss": 2.7208757400512695,
      "fork_head/num_positive": 10.633333206176758,
      "fork_head/pred_top1_mass": 0.07378590852022171,
      "grad_norm": 0.6257479190826416,
      "learning_rate": 3e-06,
      "loss": 0.6858,
      "step": 93,
      "step_time": 7.349123306572437
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.055555556217829385,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.055555556217829385,
      "entropy": 1.6041775226593018,
      "epoch": 0.045019157088122604,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20265519618988037,
      "learning_rate": 3e-06,
      "loss": -0.0069,
      "step": 94,
      "step_time": 8.148776911199093
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.328769937157631,
      "epoch": 0.04549808429118774,
      "fork_head/loss": 2.6977930068969727,
      "fork_head/num_positive": 10.633333206176758,
      "fork_head/pred_top1_mass": 0.08510536700487137,
      "grad_norm": 0.5605509281158447,
      "learning_rate": 3e-06,
      "loss": 0.6795,
      "step": 95,
      "step_time": 7.561759188771248
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.06388888955116272,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09722222288449606,
      "entropy": 1.6047200481096904,
      "epoch": 0.04597701149425287,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20598573982715607,
      "learning_rate": 3e-06,
      "loss": -0.0068,
      "step": 96,
      "step_time": 7.779385514557362
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 244.814453125,
      "completions/mean_terminated_length": 244.7925567626953,
      "completions/min_length": 21.0,
      "completions/min_terminated_length": 21.0,
      "entropy": 1.4519769906997682,
      "epoch": 0.04645593869731801,
      "fork/depth_mean": 0.21678853034973145,
      "fork/depth_std": 0.24635162949562073,
      "fork_head/loss": 2.7168383598327637,
      "fork_head/num_positive": 10.466666221618652,
      "fork_head/pred_top1_mass": 0.07683847844600677,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7983229160308838,
      "learning_rate": 3e-06,
      "loss": 0.6715,
      "num_tokens": 2912606.0,
      "reward": 0.40840286016464233,
      "reward_std": 0.5119037628173828,
      "rewards/code_format_reward/mean": 0.2001953125,
      "rewards/code_format_reward/std": 0.37335410714149475,
      "rewards/code_reward/mean": 0.20820754766464233,
      "rewards/code_reward/std": 0.37530139088630676,
      "step": 97,
      "step_time": 762.6952515020967
    },
    {
      "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,
      "entropy": 1.519580416381359,
      "epoch": 0.04693486590038314,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17991243302822113,
      "learning_rate": 3e-06,
      "loss": 0.004,
      "step": 98,
      "step_time": 8.256162628531456
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.4602213621139526,
      "epoch": 0.04741379310344827,
      "fork_head/loss": 2.6994082927703857,
      "fork_head/num_positive": 10.466666221618652,
      "fork_head/pred_top1_mass": 0.080260269343853,
      "grad_norm": 0.7587929368019104,
      "learning_rate": 3e-06,
      "loss": 0.6671,
      "step": 99,
      "step_time": 7.947205059230328
    },
    {
      "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,
      "entropy": 1.511566162109375,
      "epoch": 0.04789272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17705956101417542,
      "learning_rate": 3e-06,
      "loss": 0.0048,
      "step": 100,
      "step_time": 7.891145341098309
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.022222222884496053,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.4686979134877522,
      "epoch": 0.04837164750957854,
      "fork_head/loss": 2.674748420715332,
      "fork_head/num_positive": 10.466666221618652,
      "fork_head/pred_top1_mass": 0.08725646138191223,
      "grad_norm": 0.6847788691520691,
      "learning_rate": 3e-06,
      "loss": 0.6583,
      "step": 101,
      "step_time": 7.9497526958584785
    },
    {
      "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,
      "entropy": 1.5015089213848114,
      "epoch": 0.04885057471264368,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1853141486644745,
      "learning_rate": 3e-06,
      "loss": 0.0032,
      "step": 102,
      "step_time": 7.646180666983128
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07222222288449605,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07222222288449605,
      "entropy": 1.4805251677831015,
      "epoch": 0.04932950191570881,
      "fork_head/loss": 2.64709734916687,
      "fork_head/num_positive": 10.466666221618652,
      "fork_head/pred_top1_mass": 0.11039599776268005,
      "grad_norm": 0.6315827369689941,
      "learning_rate": 3e-06,
      "loss": 0.6512,
      "step": 103,
      "step_time": 7.918535895645618
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0234375,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0234375,
      "entropy": 1.5065697506070137,
      "epoch": 0.04980842911877394,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19786830246448517,
      "learning_rate": 3e-06,
      "loss": 0.0017,
      "step": 104,
      "step_time": 7.896614596247673
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 249.9921875,
      "completions/mean_terminated_length": 249.956787109375,
      "completions/min_length": 79.0,
      "completions/min_terminated_length": 79.0,
      "entropy": 1.47593180462718,
      "epoch": 0.05028735632183908,
      "fork/depth_mean": 0.1669139713048935,
      "fork/depth_std": 0.1804734617471695,
      "fork_head/loss": 2.7481741905212402,
      "fork_head/num_positive": 12.166666984558105,
      "fork_head/pred_top1_mass": 0.09418733417987823,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.8199674487113953,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "num_tokens": 3137482.0,
      "reward": 0.4544333815574646,
      "reward_std": 0.5426991581916809,
      "rewards/code_format_reward/mean": 0.2392578125,
      "rewards/code_format_reward/std": 0.4077068567276001,
      "rewards/code_reward/mean": 0.215175598859787,
      "rewards/code_reward/std": 0.3590669631958008,
      "step": 105,
      "step_time": 765.0128696113825
    },
    {
      "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,
      "entropy": 1.752621129155159,
      "epoch": 0.05076628352490421,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2980705499649048,
      "learning_rate": 3e-06,
      "loss": 0.0067,
      "step": 106,
      "step_time": 8.35207711905241
    },
    {
      "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,
      "entropy": 1.4722574912011623,
      "epoch": 0.05124521072796935,
      "fork_head/loss": 2.735971689224243,
      "fork_head/num_positive": 12.166666984558105,
      "fork_head/pred_top1_mass": 0.09418733417987823,
      "grad_norm": 0.6629585027694702,
      "learning_rate": 3e-06,
      "loss": 0.6762,
      "step": 107,
      "step_time": 8.71271512657404
    },
    {
      "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,
      "entropy": 1.7691582515835762,
      "epoch": 0.05172413793103448,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2720845937728882,
      "learning_rate": 3e-06,
      "loss": 0.0067,
      "step": 108,
      "step_time": 7.770799390971661
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0069444444961845875,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0069444444961845875,
      "entropy": 1.4722093306481838,
      "epoch": 0.05220306513409962,
      "fork_head/loss": 2.7188539505004883,
      "fork_head/num_positive": 12.166666984558105,
      "fork_head/pred_top1_mass": 0.09214868396520615,
      "grad_norm": 0.62071293592453,
      "learning_rate": 3e-06,
      "loss": 0.6708,
      "step": 109,
      "step_time": 7.834910482168198
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.012500000186264515,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.012500000186264515,
      "entropy": 1.7770561873912811,
      "epoch": 0.05268199233716475,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21562834084033966,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 110,
      "step_time": 7.9708628207445145
    },
    {
      "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,
      "entropy": 1.4713066928088665,
      "epoch": 0.05316091954022988,
      "fork_head/loss": 2.701151132583618,
      "fork_head/num_positive": 12.166666984558105,
      "fork_head/pred_top1_mass": 0.0978783667087555,
      "grad_norm": 0.6057758927345276,
      "learning_rate": 3e-06,
      "loss": 0.6656,
      "step": 111,
      "step_time": 8.061193272471428
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.059375000186264515,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.12187500018626451,
      "entropy": 1.7712015770375729,
      "epoch": 0.05363984674329502,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15135392546653748,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 112,
      "step_time": 7.559063613414764
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.015625,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "completions/clipped_ratio": 0.01171875,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 245.376953125,
      "completions/mean_terminated_length": 245.25099182128906,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.4920152500271797,
      "epoch": 0.05411877394636015,
      "fork/depth_mean": 0.17415578663349152,
      "fork/depth_std": 0.17631393671035767,
      "fork_head/loss": 2.7193546295166016,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.08936985582113266,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8133645057678223,
      "learning_rate": 3e-06,
      "loss": 0.7031,
      "num_tokens": 3350779.0,
      "reward": 0.4019225835800171,
      "reward_std": 0.5229238271713257,
      "rewards/code_format_reward/mean": 0.212890625,
      "rewards/code_format_reward/std": 0.388924241065979,
      "rewards/code_reward/mean": 0.1890319585800171,
      "rewards/code_reward/std": 0.3418823778629303,
      "step": 113,
      "step_time": 688.1201862543821
    },
    {
      "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,
      "entropy": 1.7076171785593033,
      "epoch": 0.05459770114942529,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2604975700378418,
      "learning_rate": 3e-06,
      "loss": -0.0228,
      "step": 114,
      "step_time": 8.982920333743095
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.015625,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "entropy": 1.4921847954392433,
      "epoch": 0.05507662835249042,
      "fork_head/loss": 2.7038049697875977,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.08528460562229156,
      "grad_norm": 0.75169438123703,
      "learning_rate": 3e-06,
      "loss": 0.6987,
      "step": 115,
      "step_time": 7.700765982270241
    },
    {
      "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,
      "entropy": 1.7073011621832848,
      "epoch": 0.05555555555555555,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.24773140251636505,
      "learning_rate": 3e-06,
      "loss": -0.0242,
      "step": 116,
      "step_time": 8.621254093945026
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.015625,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "entropy": 1.498114675283432,
      "epoch": 0.05603448275862069,
      "fork_head/loss": 2.6811749935150146,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.10424870997667313,
      "grad_norm": 0.7148356437683105,
      "learning_rate": 3e-06,
      "loss": 0.6916,
      "step": 117,
      "step_time": 8.034182123839855
    },
    {
      "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,
      "entropy": 1.6980563625693321,
      "epoch": 0.05651340996168582,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.24711641669273376,
      "learning_rate": 3e-06,
      "loss": -0.0276,
      "step": 118,
      "step_time": 8.516617462038994
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.015625,
      "clip_ratio/low_mean": 0.10937500186264515,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.12500000186264515,
      "entropy": 1.5118815153837204,
      "epoch": 0.05699233716475096,
      "fork_head/loss": 2.656270980834961,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.11753301322460175,
      "grad_norm": 0.6442331075668335,
      "learning_rate": 3e-06,
      "loss": 0.6834,
      "step": 119,
      "step_time": 7.528894029557705
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.012500000186264515,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.012500000186264515,
      "entropy": 1.6785725951194763,
      "epoch": 0.05747126436781609,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2663419246673584,
      "learning_rate": 3e-06,
      "loss": -0.0298,
      "step": 120,
      "step_time": 8.709872044622898
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01171875,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 246.404296875,
      "completions/mean_terminated_length": 246.29052734375,
      "completions/min_length": 19.0,
      "completions/min_terminated_length": 19.0,
      "entropy": 1.6513129313786825,
      "epoch": 0.05795019157088123,
      "fork/depth_mean": 0.18335111439228058,
      "fork/depth_std": 0.16543802618980408,
      "fork_head/loss": 2.742692708969116,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.0672348290681839,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.650956928730011,
      "learning_rate": 3e-06,
      "loss": 0.6896,
      "num_tokens": 3564666.0,
      "reward": 0.4751059412956238,
      "reward_std": 0.5559310913085938,
      "rewards/code_format_reward/mean": 0.2294921875,
      "rewards/code_format_reward/std": 0.39265432953834534,
      "rewards/code_reward/mean": 0.24561375379562378,
      "rewards/code_reward/std": 0.3939843773841858,
      "step": 121,
      "step_time": 620.7411005049944
    },
    {
      "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,
      "entropy": 1.5256551131606102,
      "epoch": 0.05842911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1559724509716034,
      "learning_rate": 3e-06,
      "loss": 0.0048,
      "step": 122,
      "step_time": 7.6525548323988914
    },
    {
      "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,
      "entropy": 1.6626519123713175,
      "epoch": 0.05890804597701149,
      "fork_head/loss": 2.7318155765533447,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.07430391013622284,
      "grad_norm": 0.6097334623336792,
      "learning_rate": 3e-06,
      "loss": 0.6863,
      "step": 123,
      "step_time": 8.069602072238922
    },
    {
      "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,
      "entropy": 1.5277811661362648,
      "epoch": 0.05938697318007663,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1540878713130951,
      "learning_rate": 3e-06,
      "loss": 0.0045,
      "step": 124,
      "step_time": 7.565707050263882
    },
    {
      "clip_ratio/high_max": 0.03333333333333333,
      "clip_ratio/high_mean": 0.008333333333333333,
      "clip_ratio/low_mean": 0.016666666666666666,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.025,
      "entropy": 1.6811143636703492,
      "epoch": 0.05986590038314176,
      "fork_head/loss": 2.716808795928955,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.09356627613306046,
      "grad_norm": 0.5638452172279358,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 125,
      "step_time": 7.289768189191818
    },
    {
      "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,
      "entropy": 1.5451812744140625,
      "epoch": 0.0603448275862069,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17142528295516968,
      "learning_rate": 3e-06,
      "loss": 0.0029,
      "step": 126,
      "step_time": 7.7650845646858215
    },
    {
      "clip_ratio/high_max": 0.03333333333333333,
      "clip_ratio/high_mean": 0.008333333333333333,
      "clip_ratio/low_mean": 0.016666666666666666,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.025,
      "entropy": 1.692323136329651,
      "epoch": 0.06082375478927203,
      "fork_head/loss": 2.6986382007598877,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.09694520384073257,
      "grad_norm": 0.5413647294044495,
      "learning_rate": 3e-06,
      "loss": 0.6733,
      "step": 127,
      "step_time": 7.977322041988373
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.5524139404296875,
      "epoch": 0.06130268199233716,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17359082400798798,
      "learning_rate": 3e-06,
      "loss": 0.0017,
      "step": 128,
      "step_time": 7.64722241461277
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0078125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 241.67578125,
      "completions/mean_terminated_length": 241.56298828125,
      "completions/min_length": 81.0,
      "completions/min_terminated_length": 81.0,
      "entropy": 1.5333940903345744,
      "epoch": 0.0617816091954023,
      "fork/depth_mean": 0.19481468200683594,
      "fork/depth_std": 0.19901010394096375,
      "fork_head/loss": 2.7429027557373047,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.06686317920684814,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.9973364472389221,
      "learning_rate": 3e-06,
      "loss": 0.6657,
      "num_tokens": 3775972.0,
      "reward": 0.589870274066925,
      "reward_std": 0.5826863646507263,
      "rewards/code_format_reward/mean": 0.341796875,
      "rewards/code_format_reward/std": 0.432725727558136,
      "rewards/code_reward/mean": 0.24807342886924744,
      "rewards/code_reward/std": 0.36807844042778015,
      "step": 129,
      "step_time": 662.5190129354596
    },
    {
      "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,
      "entropy": 1.5261569619178772,
      "epoch": 0.06226053639846743,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1597854495048523,
      "learning_rate": 3e-06,
      "loss": 0.0273,
      "step": 130,
      "step_time": 8.299535103142262
    },
    {
      "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,
      "entropy": 1.5387369950612386,
      "epoch": 0.06273946360153257,
      "fork_head/loss": 2.7231228351593018,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.07426995784044266,
      "grad_norm": 0.8710717558860779,
      "learning_rate": 3e-06,
      "loss": 0.6578,
      "step": 131,
      "step_time": 9.10016879439354
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.043750000186264515,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.043750000186264515,
      "entropy": 1.535566546022892,
      "epoch": 0.06321839080459771,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13138404488563538,
      "learning_rate": 3e-06,
      "loss": 0.0271,
      "step": 132,
      "step_time": 8.455771706998348
    },
    {
      "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,
      "entropy": 1.5412869095802306,
      "epoch": 0.06369731800766283,
      "fork_head/loss": 2.695385456085205,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.10335095226764679,
      "grad_norm": 0.691987931728363,
      "learning_rate": 3e-06,
      "loss": 0.6505,
      "step": 133,
      "step_time": 8.164950102567673
    },
    {
      "clip_ratio/high_max": 0.03125,
      "clip_ratio/high_mean": 0.0062500000931322575,
      "clip_ratio/low_mean": 0.0506944446824491,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05694444524124265,
      "entropy": 1.5389051586389542,
      "epoch": 0.06417624521072797,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14229726791381836,
      "learning_rate": 3e-06,
      "loss": 0.0251,
      "step": 134,
      "step_time": 8.5243551582098
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.02444444497426351,
      "clip_ratio/low_mean": 0.07777777910232545,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.10222222407658894,
      "entropy": 1.5314778606096904,
      "epoch": 0.06465517241379311,
      "fork_head/loss": 2.668039321899414,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.1060040146112442,
      "grad_norm": 0.5746515393257141,
      "learning_rate": 3e-06,
      "loss": 0.6416,
      "step": 135,
      "step_time": 8.602493546903133
    },
    {
      "clip_ratio/high_max": 0.03125,
      "clip_ratio/high_mean": 0.0062500000931322575,
      "clip_ratio/low_mean": 0.05694444524124265,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06319444486871362,
      "entropy": 1.5506937578320503,
      "epoch": 0.06513409961685823,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14701895415782928,
      "learning_rate": 3e-06,
      "loss": 0.0246,
      "step": 136,
      "step_time": 8.51498706638813
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 243.26953125,
      "completions/mean_terminated_length": 243.26953125,
      "completions/min_length": 42.0,
      "completions/min_terminated_length": 42.0,
      "entropy": 1.67138671875,
      "epoch": 0.06561302681992337,
      "fork/depth_mean": 0.1404622346162796,
      "fork/depth_std": 0.14472581446170807,
      "fork_head/loss": 2.745065927505493,
      "fork_head/num_positive": 11.129032135009766,
      "fork_head/pred_top1_mass": 0.06981092691421509,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7696008682250977,
      "learning_rate": 3e-06,
      "loss": 0.6942,
      "num_tokens": 3992014.0,
      "reward": 0.4883415102958679,
      "reward_std": 0.5433435440063477,
      "rewards/code_format_reward/mean": 0.2919921875,
      "rewards/code_format_reward/std": 0.4256850481033325,
      "rewards/code_reward/mean": 0.19634932279586792,
      "rewards/code_reward/std": 0.3386116921901703,
      "step": 137,
      "step_time": 715.4003605768085
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.038461538461538464,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.6534830698600183,
      "epoch": 0.06609195402298851,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1402927041053772,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 138,
      "step_time": 8.486693792045116
    },
    {
      "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,
      "entropy": 1.6653994577271598,
      "epoch": 0.06657088122605365,
      "fork_head/loss": 2.731395959854126,
      "fork_head/num_positive": 11.129032135009766,
      "fork_head/pred_top1_mass": 0.07229851186275482,
      "grad_norm": 0.7309831380844116,
      "learning_rate": 3e-06,
      "loss": 0.6892,
      "step": 139,
      "step_time": 7.556033246219158
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.038461538461538464,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.6812024208215566,
      "epoch": 0.06704980842911877,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13954007625579834,
      "learning_rate": 3e-06,
      "loss": -0.0081,
      "step": 140,
      "step_time": 7.54819480329752
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.6655796595982142,
      "epoch": 0.06752873563218391,
      "fork_head/loss": 2.710542678833008,
      "fork_head/num_positive": 11.129032135009766,
      "fork_head/pred_top1_mass": 0.07580172270536423,
      "grad_norm": 0.6505802869796753,
      "learning_rate": 3e-06,
      "loss": 0.6813,
      "step": 141,
      "step_time": 7.2553441524505615
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.09615384615384616,
      "clip_ratio/low_min": 0.038461538461538464,
      "clip_ratio/region_mean": 0.1346153846153846,
      "entropy": 1.6964643460053663,
      "epoch": 0.06800766283524905,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11344940960407257,
      "learning_rate": 3e-06,
      "loss": -0.0093,
      "step": 142,
      "step_time": 7.646098352968693
    },
    {
      "clip_ratio/high_max": 0.20238095309053147,
      "clip_ratio/high_mean": 0.13095238112977572,
      "clip_ratio/low_mean": 0.032738095947674344,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1636904776096344,
      "entropy": 1.6567440969603402,
      "epoch": 0.06848659003831417,
      "fork_head/loss": 2.687056541442871,
      "fork_head/num_positive": 11.129032135009766,
      "fork_head/pred_top1_mass": 0.08856675028800964,
      "grad_norm": 0.5696991086006165,
      "learning_rate": 3e-06,
      "loss": 0.6728,
      "step": 143,
      "step_time": 7.984478339552879
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.019230769230769232,
      "clip_ratio/low_mean": 0.11538461538461539,
      "clip_ratio/low_min": 0.038461538461538464,
      "clip_ratio/region_mean": 0.1346153846153846,
      "entropy": 1.7166591515907874,
      "epoch": 0.06896551724137931,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10447943210601807,
      "learning_rate": 3e-06,
      "loss": -0.0094,
      "step": 144,
      "step_time": 7.334467053413391
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 245.109375,
      "completions/mean_terminated_length": 245.04519653320312,
      "completions/min_length": 73.0,
      "completions/min_terminated_length": 73.0,
      "entropy": 1.6098768413066864,
      "epoch": 0.06944444444444445,
      "fork/depth_mean": 0.2096957266330719,
      "fork/depth_std": 0.2233714610338211,
      "fork_head/loss": 2.7424802780151367,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.07131518423557281,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5682779550552368,
      "learning_rate": 3e-06,
      "loss": 0.6901,
      "num_tokens": 4214598.0,
      "reward": 0.6046602725982666,
      "reward_std": 0.5945462584495544,
      "rewards/code_format_reward/mean": 0.3583984375,
      "rewards/code_format_reward/std": 0.4478342831134796,
      "rewards/code_reward/mean": 0.2462618052959442,
      "rewards/code_reward/std": 0.3718891143798828,
      "step": 145,
      "step_time": 645.3035795241594
    },
    {
      "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,
      "entropy": 1.6104767597638643,
      "epoch": 0.06992337164750957,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19691331684589386,
      "learning_rate": 3e-06,
      "loss": 0.0051,
      "step": 146,
      "step_time": 7.794346787035465
    },
    {
      "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,
      "entropy": 1.6149359742800395,
      "epoch": 0.07040229885057471,
      "fork_head/loss": 2.733577251434326,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.07655691355466843,
      "grad_norm": 0.5124426484107971,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "step": 147,
      "step_time": 8.550788134336472
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.6342835976527288,
      "epoch": 0.07088122605363985,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.176438570022583,
      "learning_rate": 3e-06,
      "loss": 0.0027,
      "step": 148,
      "step_time": 7.433328032493591
    },
    {
      "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,
      "entropy": 1.6244167784849803,
      "epoch": 0.07136015325670499,
      "fork_head/loss": 2.721083641052246,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.07811205834150314,
      "grad_norm": 0.4306269586086273,
      "learning_rate": 3e-06,
      "loss": 0.6838,
      "step": 149,
      "step_time": 8.15010266751051
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.17948717967822,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.25641025660129696,
      "entropy": 1.6407251633130586,
      "epoch": 0.07183908045977011,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11097190529108047,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 150,
      "step_time": 7.377088889479637
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.662584086259206,
      "epoch": 0.07231800766283525,
      "fork_head/loss": 2.708937406539917,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.07536598294973373,
      "grad_norm": 0.3809207081794739,
      "learning_rate": 3e-06,
      "loss": 0.68,
      "step": 151,
      "step_time": 8.54037855565548
    },
    {
      "clip_ratio/high_max": 0.10256410332826468,
      "clip_ratio/high_mean": 0.08333333352437386,
      "clip_ratio/low_mean": 0.14102564121668154,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.22435897474105543,
      "entropy": 1.6554048978365385,
      "epoch": 0.07279693486590039,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09905069321393967,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 152,
      "step_time": 7.872317537665367
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 243.955078125,
      "completions/mean_terminated_length": 243.955078125,
      "completions/min_length": 36.0,
      "completions/min_terminated_length": 36.0,
      "entropy": 1.8831380208333333,
      "epoch": 0.07327586206896551,
      "fork/depth_mean": 0.1984960287809372,
      "fork/depth_std": 0.1907033920288086,
      "fork_head/loss": 2.712616205215454,
      "fork_head/num_positive": 10.903225898742676,
      "fork_head/pred_top1_mass": 0.0905723050236702,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6141097545623779,
      "learning_rate": 3e-06,
      "loss": 0.6743,
      "num_tokens": 4433519.0,
      "reward": 0.6129482984542847,
      "reward_std": 0.5823231339454651,
      "rewards/code_format_reward/mean": 0.39453125,
      "rewards/code_format_reward/std": 0.45395562052726746,
      "rewards/code_reward/mean": 0.21841703355312347,
      "rewards/code_reward/std": 0.3468570113182068,
      "step": 153,
      "step_time": 567.5708920881152
    },
    {
      "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,
      "entropy": 1.514453125,
      "epoch": 0.07375478927203065,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06872160732746124,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 154,
      "step_time": 7.524426117539406
    },
    {
      "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,
      "entropy": 1.8556857638888888,
      "epoch": 0.07423371647509579,
      "fork_head/loss": 2.7053427696228027,
      "fork_head/num_positive": 10.903225898742676,
      "fork_head/pred_top1_mass": 0.07930921763181686,
      "grad_norm": 0.5948209762573242,
      "learning_rate": 3e-06,
      "loss": 0.6722,
      "step": 155,
      "step_time": 8.089858129620552
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.513671875,
      "epoch": 0.07471264367816093,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06089162826538086,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 156,
      "step_time": 7.574751935899258
    },
    {
      "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,
      "entropy": 1.8832465277777777,
      "epoch": 0.07519157088122605,
      "fork_head/loss": 2.691455125808716,
      "fork_head/num_positive": 10.903225898742676,
      "fork_head/pred_top1_mass": 0.07609158009290695,
      "grad_norm": 0.5284668207168579,
      "learning_rate": 3e-06,
      "loss": 0.6687,
      "step": 157,
      "step_time": 7.542235851287842
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.532421875,
      "epoch": 0.07567049808429119,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06936169415712357,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 158,
      "step_time": 7.35250336676836
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.05555555555555555,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.8542751736111112,
      "epoch": 0.07614942528735633,
      "fork_head/loss": 2.6753127574920654,
      "fork_head/num_positive": 10.903225898742676,
      "fork_head/pred_top1_mass": 0.09585502743721008,
      "grad_norm": 0.4292150139808655,
      "learning_rate": 3e-06,
      "loss": 0.6634,
      "step": 159,
      "step_time": 7.045839138329029
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.5375325560569764,
      "epoch": 0.07662835249042145,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06974364817142487,
      "learning_rate": 3e-06,
      "loss": 0.0019,
      "step": 160,
      "step_time": 7.82091299444437
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 244.005859375,
      "completions/mean_terminated_length": 243.95883178710938,
      "completions/min_length": 105.0,
      "completions/min_terminated_length": 105.0,
      "entropy": 1.5716145932674408,
      "epoch": 0.07710727969348659,
      "fork/depth_mean": 0.23156358301639557,
      "fork/depth_std": 0.23513688147068024,
      "fork_head/loss": 2.714611530303955,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.07713735103607178,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6225348711013794,
      "learning_rate": 3e-06,
      "loss": 0.6695,
      "num_tokens": 4651810.0,
      "reward": 0.5540305972099304,
      "reward_std": 0.5587325692176819,
      "rewards/code_format_reward/mean": 0.3759765625,
      "rewards/code_format_reward/std": 0.4594613313674927,
      "rewards/code_reward/mean": 0.17805403470993042,
      "rewards/code_reward/std": 0.31976619362831116,
      "step": 161,
      "step_time": 625.5853372961283
    },
    {
      "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,
      "entropy": 1.5272435821019685,
      "epoch": 0.07758620689655173,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.119603231549263,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 162,
      "step_time": 8.242325350642204
    },
    {
      "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,
      "entropy": 1.5933431088924408,
      "epoch": 0.07806513409961686,
      "fork_head/loss": 2.7053208351135254,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.08038050681352615,
      "grad_norm": 0.5701454281806946,
      "learning_rate": 3e-06,
      "loss": 0.6675,
      "step": 163,
      "step_time": 7.5133800730109215
    },
    {
      "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,
      "entropy": 1.5216095814338098,
      "epoch": 0.07854406130268199,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12150757014751434,
      "learning_rate": 3e-06,
      "loss": 0.0048,
      "step": 164,
      "step_time": 7.8381712436676025
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.1875,
      "entropy": 1.573486328125,
      "epoch": 0.07902298850574713,
      "fork_head/loss": 2.6915338039398193,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.08421708643436432,
      "grad_norm": 0.5253466367721558,
      "learning_rate": 3e-06,
      "loss": 0.662,
      "step": 165,
      "step_time": 7.604714885354042
    },
    {
      "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,
      "entropy": 1.517102369895348,
      "epoch": 0.07950191570881227,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11271782219409943,
      "learning_rate": 3e-06,
      "loss": 0.0041,
      "step": 166,
      "step_time": 7.746861778199673
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.5452474057674408,
      "epoch": 0.07998084291187739,
      "fork_head/loss": 2.6747817993164062,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.10424143075942993,
      "grad_norm": 0.5135942697525024,
      "learning_rate": 3e-06,
      "loss": 0.6565,
      "step": 167,
      "step_time": 7.660115323960781
    },
    {
      "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,
      "entropy": 1.5014022497030406,
      "epoch": 0.08045977011494253,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1222093254327774,
      "learning_rate": 3e-06,
      "loss": 0.0035,
      "step": 168,
      "step_time": 7.599498435854912
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01171875,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 237.666015625,
      "completions/mean_terminated_length": 237.44862365722656,
      "completions/min_length": 44.0,
      "completions/min_terminated_length": 44.0,
      "entropy": 1.714248664677143,
      "epoch": 0.08093869731800767,
      "fork/depth_mean": 0.25201019644737244,
      "fork/depth_std": 0.2337212860584259,
      "fork_head/loss": 2.7301719188690186,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.07927811145782471,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6389692425727844,
      "learning_rate": 3e-06,
      "loss": 0.6813,
      "num_tokens": 4869111.0,
      "reward": 0.5935791730880737,
      "reward_std": 0.5516173839569092,
      "rewards/code_format_reward/mean": 0.3857421875,
      "rewards/code_format_reward/std": 0.46304935216903687,
      "rewards/code_reward/mean": 0.20783698558807373,
      "rewards/code_reward/std": 0.3743649423122406,
      "step": 169,
      "step_time": 617.9267398715019
    },
    {
      "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,
      "entropy": 1.801883028103755,
      "epoch": 0.08141762452107279,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1529417783021927,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 170,
      "step_time": 8.271762922406197
    },
    {
      "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,
      "entropy": 1.720438651740551,
      "epoch": 0.08189655172413793,
      "fork_head/loss": 2.71901273727417,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.07833171635866165,
      "grad_norm": 0.6214330196380615,
      "learning_rate": 3e-06,
      "loss": 0.6769,
      "step": 171,
      "step_time": 8.511127404868603
    },
    {
      "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,
      "entropy": 1.790777747447674,
      "epoch": 0.08237547892720307,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15440599620342255,
      "learning_rate": 3e-06,
      "loss": -0.003,
      "step": 172,
      "step_time": 8.508676864206791
    },
    {
      "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,
      "entropy": 1.7284952849149704,
      "epoch": 0.0828544061302682,
      "fork_head/loss": 2.7011897563934326,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.08370465785264969,
      "grad_norm": 0.5588541626930237,
      "learning_rate": 3e-06,
      "loss": 0.671,
      "step": 173,
      "step_time": 7.743534319102764
    },
    {
      "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,
      "entropy": 1.8029346832862267,
      "epoch": 0.08333333333333333,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1495984047651291,
      "learning_rate": 3e-06,
      "loss": -0.0038,
      "step": 174,
      "step_time": 8.310873813927174
    },
    {
      "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,
      "entropy": 1.7331543043255806,
      "epoch": 0.08381226053639847,
      "fork_head/loss": 2.680063247680664,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.09458339959383011,
      "grad_norm": 0.50445955991745,
      "learning_rate": 3e-06,
      "loss": 0.6648,
      "step": 175,
      "step_time": 7.841958485543728
    },
    {
      "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,
      "entropy": 1.8064152644230769,
      "epoch": 0.0842911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1386292725801468,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 176,
      "step_time": 8.462002284824848
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 239.283203125,
      "completions/mean_terminated_length": 239.18467712402344,
      "completions/min_length": 68.0,
      "completions/min_terminated_length": 68.0,
      "entropy": 2.1216145833333333,
      "epoch": 0.08477011494252873,
      "fork/depth_mean": 0.2242422103881836,
      "fork/depth_std": 0.23020599782466888,
      "fork_head/loss": 2.730151891708374,
      "fork_head/num_positive": 11.516129493713379,
      "fork_head/pred_top1_mass": 0.07286382466554642,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7838069796562195,
      "learning_rate": 3e-06,
      "loss": 0.6704,
      "num_tokens": 5081000.0,
      "reward": 0.6930122375488281,
      "reward_std": 0.6424105167388916,
      "rewards/code_format_reward/mean": 0.4384765625,
      "rewards/code_format_reward/std": 0.46145352721214294,
      "rewards/code_reward/mean": 0.2545356750488281,
      "rewards/code_reward/std": 0.3904031217098236,
      "step": 177,
      "step_time": 535.6928007304668
    },
    {
      "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,
      "entropy": 1.6666066497564316,
      "epoch": 0.08524904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2149481475353241,
      "learning_rate": 3e-06,
      "loss": 0.0075,
      "step": 178,
      "step_time": 8.226006776094437
    },
    {
      "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,
      "entropy": 2.1215972185134886,
      "epoch": 0.085727969348659,
      "fork_head/loss": 2.714024305343628,
      "fork_head/num_positive": 11.516129493713379,
      "fork_head/pred_top1_mass": 0.06955792754888535,
      "grad_norm": 0.6839104294776917,
      "learning_rate": 3e-06,
      "loss": 0.6657,
      "step": 179,
      "step_time": 7.704845756292343
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.010416666977107525,
      "entropy": 1.672006219625473,
      "epoch": 0.08620689655172414,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20281462371349335,
      "learning_rate": 3e-06,
      "loss": 0.0059,
      "step": 180,
      "step_time": 8.198175013065338
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.016666666666666666,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.016666666666666666,
      "entropy": 2.1118858496348065,
      "epoch": 0.08668582375478927,
      "fork_head/loss": 2.693378210067749,
      "fork_head/num_positive": 11.516129493713379,
      "fork_head/pred_top1_mass": 0.08335172384977341,
      "grad_norm": 0.5798757672309875,
      "learning_rate": 3e-06,
      "loss": 0.6598,
      "step": 181,
      "step_time": 7.2186983078718185
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.010416666977107525,
      "entropy": 1.6639434844255447,
      "epoch": 0.0871647509578544,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1944732964038849,
      "learning_rate": 3e-06,
      "loss": 0.005,
      "step": 182,
      "step_time": 7.851053439080715
    },
    {
      "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,
      "entropy": 2.116927083333333,
      "epoch": 0.08764367816091954,
      "fork_head/loss": 2.6730525493621826,
      "fork_head/num_positive": 11.516129493713379,
      "fork_head/pred_top1_mass": 0.09906712919473648,
      "grad_norm": 0.554607629776001,
      "learning_rate": 3e-06,
      "loss": 0.6542,
      "step": 183,
      "step_time": 7.815625697374344
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.0078125,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.018229166977107525,
      "entropy": 1.675650030374527,
      "epoch": 0.08812260536398467,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16592462360858917,
      "learning_rate": 3e-06,
      "loss": 0.0044,
      "step": 184,
      "step_time": 7.620532542467117
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 241.0546875,
      "completions/mean_terminated_length": 241.0254364013672,
      "completions/min_length": 46.0,
      "completions/min_terminated_length": 46.0,
      "entropy": 1.450047349387949,
      "epoch": 0.0886015325670498,
      "fork/depth_mean": 0.20260776579380035,
      "fork/depth_std": 0.2203216552734375,
      "fork_head/loss": 2.7371954917907715,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.07043854892253876,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8351646065711975,
      "learning_rate": 3e-06,
      "loss": 0.6783,
      "num_tokens": 5299412.0,
      "reward": 0.5542190074920654,
      "reward_std": 0.5813146233558655,
      "rewards/code_format_reward/mean": 0.3466796875,
      "rewards/code_format_reward/std": 0.4461497664451599,
      "rewards/code_reward/mean": 0.20753934979438782,
      "rewards/code_reward/std": 0.3553330600261688,
      "step": 185,
      "step_time": 510.0203702002764
    },
    {
      "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,
      "entropy": 2.323046875,
      "epoch": 0.08908045977011494,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09422620385885239,
      "learning_rate": 3e-06,
      "loss": 0.0029,
      "step": 186,
      "step_time": 8.028488427400589
    },
    {
      "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,
      "entropy": 1.4468809149482034,
      "epoch": 0.08955938697318008,
      "fork_head/loss": 2.7243523597717285,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.07225506752729416,
      "grad_norm": 0.6280652284622192,
      "learning_rate": 3e-06,
      "loss": 0.6747,
      "step": 187,
      "step_time": 8.42279277741909
    },
    {
      "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,
      "entropy": 2.3266927123069765,
      "epoch": 0.09003831417624521,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09607056528329849,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 188,
      "step_time": 7.383212983608246
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.4028172276236794,
      "epoch": 0.09051724137931035,
      "fork_head/loss": 2.7109620571136475,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.07480332255363464,
      "grad_norm": 0.5414120554924011,
      "learning_rate": 3e-06,
      "loss": 0.6689,
      "step": 189,
      "step_time": 8.273227967321873
    },
    {
      "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,
      "entropy": 2.364322912693024,
      "epoch": 0.09099616858237548,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08938618749380112,
      "learning_rate": 3e-06,
      "loss": 0.0027,
      "step": 190,
      "step_time": 7.383739709854126
    },
    {
      "clip_ratio/high_max": 0.2727272727272727,
      "clip_ratio/high_mean": 0.22727272727272727,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.22727272727272727,
      "entropy": 1.3563565340909092,
      "epoch": 0.09147509578544061,
      "fork_head/loss": 2.6971659660339355,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.08390404284000397,
      "grad_norm": 0.5741972923278809,
      "learning_rate": 3e-06,
      "loss": 0.6642,
      "step": 191,
      "step_time": 7.426052436232567
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.2,
      "clip_ratio/low_min": 0.2,
      "clip_ratio/region_mean": 0.2,
      "entropy": 2.3990885376930238,
      "epoch": 0.09195402298850575,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11912108957767487,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 192,
      "step_time": 7.970463417470455
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 238.474609375,
      "completions/mean_terminated_length": 238.37132263183594,
      "completions/min_length": 102.0,
      "completions/min_terminated_length": 102.0,
      "entropy": 1.9085286498069762,
      "epoch": 0.09243295019157088,
      "fork/depth_mean": 0.1920716017484665,
      "fork/depth_std": 0.20616963505744934,
      "fork_head/loss": 2.7380127906799316,
      "fork_head/num_positive": 11.967741966247559,
      "fork_head/pred_top1_mass": 0.07046186178922653,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5399562120437622,
      "learning_rate": 3e-06,
      "loss": 0.6897,
      "num_tokens": 5508743.0,
      "reward": 0.6123549938201904,
      "reward_std": 0.6037150621414185,
      "rewards/code_format_reward/mean": 0.396484375,
      "rewards/code_format_reward/std": 0.45440584421157837,
      "rewards/code_reward/mean": 0.21587061882019043,
      "rewards/code_reward/std": 0.3649078905582428,
      "step": 193,
      "step_time": 612.7553169950843
    },
    {
      "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,
      "entropy": 1.7281087239583333,
      "epoch": 0.09291187739463602,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1429772973060608,
      "learning_rate": 3e-06,
      "loss": -0.0057,
      "step": 194,
      "step_time": 7.7963990941643715
    },
    {
      "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,
      "entropy": 1.9042806029319763,
      "epoch": 0.09339080459770115,
      "fork_head/loss": 2.7282867431640625,
      "fork_head/num_positive": 11.967741966247559,
      "fork_head/pred_top1_mass": 0.07152995467185974,
      "grad_norm": 0.487428218126297,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "step": 195,
      "step_time": 7.78619509190321
    },
    {
      "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,
      "entropy": 1.7164713541666667,
      "epoch": 0.09386973180076628,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18241949379444122,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 196,
      "step_time": 7.601539857685566
    },
    {
      "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,
      "entropy": 1.8741373658180236,
      "epoch": 0.09434865900383142,
      "fork_head/loss": 2.7163333892822266,
      "fork_head/num_positive": 11.967741966247559,
      "fork_head/pred_top1_mass": 0.07583353668451309,
      "grad_norm": 0.4260818660259247,
      "learning_rate": 3e-06,
      "loss": 0.6833,
      "step": 197,
      "step_time": 7.76293408870697
    },
    {
      "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,
      "entropy": 1.7362467447916667,
      "epoch": 0.09482758620689655,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17134106159210205,
      "learning_rate": 3e-06,
      "loss": -0.0072,
      "step": 198,
      "step_time": 7.651630595326424
    },
    {
      "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,
      "entropy": 1.880859375,
      "epoch": 0.09530651340996169,
      "fork_head/loss": 2.7037291526794434,
      "fork_head/num_positive": 11.967741966247559,
      "fork_head/pred_top1_mass": 0.07428161799907684,
      "grad_norm": 0.3756798803806305,
      "learning_rate": 3e-06,
      "loss": 0.6798,
      "step": 199,
      "step_time": 7.45420041680336
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.7525227864583333,
      "epoch": 0.09578544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0983741283416748,
      "learning_rate": 3e-06,
      "loss": -0.0081,
      "step": 200,
      "step_time": 7.430532492697239
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 233.384765625,
      "completions/mean_terminated_length": 233.384765625,
      "completions/min_length": 64.0,
      "completions/min_terminated_length": 64.0,
      "entropy": 1.82265625,
      "epoch": 0.09626436781609195,
      "fork/depth_mean": 0.239730104804039,
      "fork/depth_std": 0.2562384605407715,
      "fork_head/loss": 2.7515902519226074,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.04756607115268707,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.1247532367706299,
      "learning_rate": 3e-06,
      "loss": 0.6897,
      "num_tokens": 5715180.0,
      "reward": 0.6424655914306641,
      "reward_std": 0.6414240598678589,
      "rewards/code_format_reward/mean": 0.48046875,
      "rewards/code_format_reward/std": 0.4571773409843445,
      "rewards/code_reward/mean": 0.16199688613414764,
      "rewards/code_reward/std": 0.31838804483413696,
      "step": 201,
      "step_time": 493.96854554116726
    },
    {
      "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,
      "entropy": 1.46826171875,
      "epoch": 0.09674329501915709,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14791351556777954,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 202,
      "step_time": 7.928949072957039
    },
    {
      "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,
      "entropy": 1.818359375,
      "epoch": 0.09722222222222222,
      "fork_head/loss": 2.7272002696990967,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.08624599874019623,
      "grad_norm": 0.9927728176116943,
      "learning_rate": 3e-06,
      "loss": 0.683,
      "step": 203,
      "step_time": 7.839807920157909
    },
    {
      "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,
      "entropy": 1.4737630248069764,
      "epoch": 0.09770114942528736,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13414588570594788,
      "learning_rate": 3e-06,
      "loss": -0.0032,
      "step": 204,
      "step_time": 7.503169745206833
    },
    {
      "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,
      "entropy": 1.81923828125,
      "epoch": 0.09818007662835249,
      "fork_head/loss": 2.6887102127075195,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.10255374014377594,
      "grad_norm": 0.8374541401863098,
      "learning_rate": 3e-06,
      "loss": 0.6736,
      "step": 205,
      "step_time": 7.5643575116992
    },
    {
      "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,
      "entropy": 1.4627278685569762,
      "epoch": 0.09865900383141762,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15319670736789703,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 206,
      "step_time": 7.47791800647974
    },
    {
      "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,
      "entropy": 1.8060546875,
      "epoch": 0.09913793103448276,
      "fork_head/loss": 2.6471190452575684,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.1423809975385666,
      "grad_norm": 0.8020092248916626,
      "learning_rate": 3e-06,
      "loss": 0.6623,
      "step": 207,
      "step_time": 7.47884114831686
    },
    {
      "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,
      "entropy": 1.4637044310569762,
      "epoch": 0.09961685823754789,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17806211113929749,
      "learning_rate": 3e-06,
      "loss": -0.0036,
      "step": 208,
      "step_time": 7.518914639949799
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 236.21484375,
      "completions/mean_terminated_length": 236.13726806640625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.7323404923081398,
      "epoch": 0.10009578544061302,
      "fork/depth_mean": 0.28791749477386475,
      "fork/depth_std": 0.2931916117668152,
      "fork_head/loss": 2.710245132446289,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.07798101007938385,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7206663489341736,
      "learning_rate": 3e-06,
      "loss": 0.6675,
      "num_tokens": 5925850.0,
      "reward": 0.6479203701019287,
      "reward_std": 0.6177699565887451,
      "rewards/code_format_reward/mean": 0.4990234375,
      "rewards/code_format_reward/std": 0.44402867555618286,
      "rewards/code_reward/mean": 0.1488969326019287,
      "rewards/code_reward/std": 0.3165914714336395,
      "step": 209,
      "step_time": 512.9408536404371
    },
    {
      "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,
      "entropy": 1.9116336107254028,
      "epoch": 0.10057471264367816,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2472791075706482,
      "learning_rate": 3e-06,
      "loss": 0.0069,
      "step": 210,
      "step_time": 7.584155306220055
    },
    {
      "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,
      "entropy": 1.7280273512005806,
      "epoch": 0.1010536398467433,
      "fork_head/loss": 2.7016959190368652,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.08344785124063492,
      "grad_norm": 0.6120437979698181,
      "learning_rate": 3e-06,
      "loss": 0.6641,
      "step": 211,
      "step_time": 8.331801004707813
    },
    {
      "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,
      "entropy": 1.929487182543828,
      "epoch": 0.10153256704980843,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21625347435474396,
      "learning_rate": 3e-06,
      "loss": 0.0063,
      "step": 212,
      "step_time": 7.558121360838413
    },
    {
      "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,
      "entropy": 1.7302144318819046,
      "epoch": 0.10201149425287356,
      "fork_head/loss": 2.689422369003296,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.10213067382574081,
      "grad_norm": 0.595379114151001,
      "learning_rate": 3e-06,
      "loss": 0.6605,
      "step": 213,
      "step_time": 7.693633444607258
    },
    {
      "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,
      "entropy": 1.9324669471153846,
      "epoch": 0.1024904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.25431516766548157,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 214,
      "step_time": 7.604380585253239
    },
    {
      "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,
      "entropy": 1.7444864884018898,
      "epoch": 0.10296934865900383,
      "fork_head/loss": 2.672088861465454,
      "fork_head/num_positive": 11.09375,
      "fork_head/pred_top1_mass": 0.09878737479448318,
      "grad_norm": 0.6365586519241333,
      "learning_rate": 3e-06,
      "loss": 0.6554,
      "step": 215,
      "step_time": 7.475438162684441
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.9320663030330951,
      "epoch": 0.10344827586206896,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11829949170351028,
      "learning_rate": 3e-06,
      "loss": 0.0047,
      "step": 216,
      "step_time": 7.584404073655605
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.019230769230769232,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.019230769230769232,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 224.630859375,
      "completions/mean_terminated_length": 224.56947326660156,
      "completions/min_length": 86.0,
      "completions/min_terminated_length": 86.0,
      "entropy": 1.8906876032169049,
      "epoch": 0.1039272030651341,
      "fork/depth_mean": 0.2737963795661926,
      "fork/depth_std": 0.27723702788352966,
      "fork_head/loss": 2.727414608001709,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.08521943539381027,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5890733003616333,
      "learning_rate": 3e-06,
      "loss": 0.6904,
      "num_tokens": 6135373.0,
      "reward": 0.5819035768508911,
      "reward_std": 0.5468857884407043,
      "rewards/code_format_reward/mean": 0.4423828125,
      "rewards/code_format_reward/std": 0.443589448928833,
      "rewards/code_reward/mean": 0.1395207643508911,
      "rewards/code_reward/std": 0.3031037449836731,
      "step": 217,
      "step_time": 604.0309075415134
    },
    {
      "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,
      "entropy": 1.6483930945396423,
      "epoch": 0.10440613026819924,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08433660119771957,
      "learning_rate": 3e-06,
      "loss": -0.0106,
      "step": 218,
      "step_time": 8.223316729068756
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.019230769230769232,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.019230769230769232,
      "entropy": 1.8860677113899817,
      "epoch": 0.10488505747126436,
      "fork_head/loss": 2.7189338207244873,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.083920918405056,
      "grad_norm": 0.5456047654151917,
      "learning_rate": 3e-06,
      "loss": 0.6876,
      "step": 219,
      "step_time": 7.620181478559971
    },
    {
      "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,
      "entropy": 1.664698757908561,
      "epoch": 0.1053639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08342716842889786,
      "learning_rate": 3e-06,
      "loss": -0.0105,
      "step": 220,
      "step_time": 7.571695372462273
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.019230769230769232,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.019230769230769232,
      "entropy": 1.900415668120751,
      "epoch": 0.10584291187739464,
      "fork_head/loss": 2.705334186553955,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.08247815072536469,
      "grad_norm": 0.49129363894462585,
      "learning_rate": 3e-06,
      "loss": 0.6835,
      "step": 221,
      "step_time": 7.693280093371868
    },
    {
      "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,
      "entropy": 1.6602006554603577,
      "epoch": 0.10632183908045977,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08111627399921417,
      "learning_rate": 3e-06,
      "loss": -0.0107,
      "step": 222,
      "step_time": 7.647177815437317
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.019230769230769232,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.019230769230769232,
      "entropy": 1.906012122447674,
      "epoch": 0.1068007662835249,
      "fork_head/loss": 2.6890008449554443,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.08677852898836136,
      "grad_norm": 0.45553848147392273,
      "learning_rate": 3e-06,
      "loss": 0.6782,
      "step": 223,
      "step_time": 8.003170922398567
    },
    {
      "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,
      "entropy": 1.663293096152219,
      "epoch": 0.10727969348659004,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09834904223680496,
      "learning_rate": 3e-06,
      "loss": -0.0111,
      "step": 224,
      "step_time": 7.577278308570385
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777778605620067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777778605620067,
      "completions/clipped_ratio": 0.0078125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 228.658203125,
      "completions/mean_terminated_length": 228.4429168701172,
      "completions/min_length": 35.0,
      "completions/min_terminated_length": 35.0,
      "entropy": 1.7790473103523254,
      "epoch": 0.10775862068965517,
      "fork/depth_mean": 0.25506407022476196,
      "fork/depth_std": 0.2593793272972107,
      "fork_head/loss": 2.718899726867676,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07911711931228638,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.995102047920227,
      "learning_rate": 3e-06,
      "loss": 0.6927,
      "num_tokens": 6340142.0,
      "reward": 0.6579333543777466,
      "reward_std": 0.6016430258750916,
      "rewards/code_format_reward/mean": 0.4873046875,
      "rewards/code_format_reward/std": 0.4493253827095032,
      "rewards/code_reward/mean": 0.1706286370754242,
      "rewards/code_reward/std": 0.34252408146858215,
      "step": 225,
      "step_time": 577.0438985303044
    },
    {
      "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,
      "entropy": 1.4147431308572942,
      "epoch": 0.1082375478927203,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11739934980869293,
      "learning_rate": 3e-06,
      "loss": -0.0104,
      "step": 226,
      "step_time": 7.689344167709351
    },
    {
      "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,
      "entropy": 1.7604275147120159,
      "epoch": 0.10871647509578544,
      "fork_head/loss": 2.702239513397217,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.09457971900701523,
      "grad_norm": 0.5722391605377197,
      "learning_rate": 3e-06,
      "loss": 0.6881,
      "step": 227,
      "step_time": 8.265742339193821
    },
    {
      "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,
      "entropy": 1.39404296875,
      "epoch": 0.10919540229885058,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12469405680894852,
      "learning_rate": 3e-06,
      "loss": -0.0112,
      "step": 228,
      "step_time": 7.351524293422699
    },
    {
      "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,
      "entropy": 1.7936415076255798,
      "epoch": 0.1096743295019157,
      "fork_head/loss": 2.689850091934204,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.1071953997015953,
      "grad_norm": 0.5558217763900757,
      "learning_rate": 3e-06,
      "loss": 0.6845,
      "step": 229,
      "step_time": 7.354934066534042
    },
    {
      "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,
      "entropy": 1.4070933948863635,
      "epoch": 0.11015325670498084,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1248968318104744,
      "learning_rate": 3e-06,
      "loss": -0.012,
      "step": 230,
      "step_time": 7.625793866813183
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777778605620067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777778605620067,
      "entropy": 1.800271252791087,
      "epoch": 0.11063218390804598,
      "fork_head/loss": 2.680142402648926,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.10545417666435242,
      "grad_norm": 0.7282523512840271,
      "learning_rate": 3e-06,
      "loss": 0.681,
      "step": 231,
      "step_time": 7.608129933476448
    },
    {
      "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,
      "entropy": 1.397002252665433,
      "epoch": 0.1111111111111111,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11283720284700394,
      "learning_rate": 3e-06,
      "loss": -0.0123,
      "step": 232,
      "step_time": 7.387894950807095
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 235.560546875,
      "completions/mean_terminated_length": 235.5205535888672,
      "completions/min_length": 103.0,
      "completions/min_terminated_length": 103.0,
      "entropy": 1.3676313920454546,
      "epoch": 0.11159003831417624,
      "fork/depth_mean": 0.23116089403629303,
      "fork/depth_std": 0.22877784073352814,
      "fork_head/loss": 2.722573757171631,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.07666009664535522,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6375036239624023,
      "learning_rate": 3e-06,
      "loss": 0.673,
      "num_tokens": 6562109.0,
      "reward": 0.562158465385437,
      "reward_std": 0.6102018356323242,
      "rewards/code_format_reward/mean": 0.4501953125,
      "rewards/code_format_reward/std": 0.46707648038864136,
      "rewards/code_reward/mean": 0.11196315288543701,
      "rewards/code_reward/std": 0.252400279045105,
      "step": 233,
      "step_time": 516.2511024624109
    },
    {
      "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,
      "entropy": 1.9831321022727273,
      "epoch": 0.11206896551724138,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08962038904428482,
      "learning_rate": 3e-06,
      "loss": 0.0063,
      "step": 234,
      "step_time": 8.320552565157413
    },
    {
      "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,
      "entropy": 1.3813920454545454,
      "epoch": 0.11254789272030652,
      "fork_head/loss": 2.710026741027832,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.10020364820957184,
      "grad_norm": 0.5912087559700012,
      "learning_rate": 3e-06,
      "loss": 0.6697,
      "step": 235,
      "step_time": 7.797637104988098
    },
    {
      "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,
      "entropy": 1.9790926846590908,
      "epoch": 0.11302681992337164,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09222979843616486,
      "learning_rate": 3e-06,
      "loss": 0.0056,
      "step": 236,
      "step_time": 7.655877836048603
    },
    {
      "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,
      "entropy": 1.4002130681818181,
      "epoch": 0.11350574712643678,
      "fork_head/loss": 2.6924681663513184,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.11215775460004807,
      "grad_norm": 0.5356510877609253,
      "learning_rate": 3e-06,
      "loss": 0.6655,
      "step": 237,
      "step_time": 7.78903640806675
    },
    {
      "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,
      "entropy": 1.9814009232954546,
      "epoch": 0.11398467432950192,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09264083951711655,
      "learning_rate": 3e-06,
      "loss": 0.0058,
      "step": 238,
      "step_time": 7.075154945254326
    },
    {
      "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,
      "entropy": 1.3818359375,
      "epoch": 0.11446360153256704,
      "fork_head/loss": 2.6724414825439453,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.11220884323120117,
      "grad_norm": 0.5152155756950378,
      "learning_rate": 3e-06,
      "loss": 0.6597,
      "step": 239,
      "step_time": 7.461415611207485
    },
    {
      "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,
      "entropy": 1.9779385653409092,
      "epoch": 0.11494252873563218,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10192609578371048,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 240,
      "step_time": 7.772572904825211
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 229.4921875,
      "completions/mean_terminated_length": 229.4921875,
      "completions/min_length": 90.0,
      "completions/min_terminated_length": 90.0,
      "entropy": 1.7548828125,
      "epoch": 0.11542145593869732,
      "fork/depth_mean": 0.2449028193950653,
      "fork/depth_std": 0.2177785187959671,
      "fork_head/loss": 2.728468894958496,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.07905665785074234,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6836417317390442,
      "learning_rate": 3e-06,
      "loss": 0.6873,
      "num_tokens": 6777257.0,
      "reward": 0.642899751663208,
      "reward_std": 0.6749046444892883,
      "rewards/code_format_reward/mean": 0.453125,
      "rewards/code_format_reward/std": 0.4679032564163208,
      "rewards/code_reward/mean": 0.1897747814655304,
      "rewards/code_reward/std": 0.3602587878704071,
      "step": 241,
      "step_time": 563.6823398843408
    },
    {
      "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,
      "entropy": 1.7010967548076923,
      "epoch": 0.11590038314176246,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1871815174818039,
      "learning_rate": 3e-06,
      "loss": -0.0056,
      "step": 242,
      "step_time": 7.812990620732307
    },
    {
      "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,
      "entropy": 1.7453776121139526,
      "epoch": 0.11637931034482758,
      "fork_head/loss": 2.7163190841674805,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.08257326483726501,
      "grad_norm": 0.5917626619338989,
      "learning_rate": 3e-06,
      "loss": 0.6839,
      "step": 243,
      "step_time": 7.81565161049366
    },
    {
      "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,
      "entropy": 1.7021734714508057,
      "epoch": 0.11685823754789272,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19630169868469238,
      "learning_rate": 3e-06,
      "loss": -0.0071,
      "step": 244,
      "step_time": 7.97839467972517
    },
    {
      "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,
      "entropy": 1.7345377564430238,
      "epoch": 0.11733716475095786,
      "fork_head/loss": 2.6985843181610107,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.08927018195390701,
      "grad_norm": 0.520252525806427,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "step": 245,
      "step_time": 7.251650720834732
    },
    {
      "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,
      "entropy": 1.683293260060824,
      "epoch": 0.11781609195402298,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19527283310890198,
      "learning_rate": 3e-06,
      "loss": -0.007,
      "step": 246,
      "step_time": 8.340748384594917
    },
    {
      "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,
      "entropy": 1.7483072876930237,
      "epoch": 0.11829501915708812,
      "fork_head/loss": 2.678921699523926,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.0908241793513298,
      "grad_norm": 0.5049571990966797,
      "learning_rate": 3e-06,
      "loss": 0.6734,
      "step": 247,
      "step_time": 7.271320305764675
    },
    {
      "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,
      "entropy": 1.6978165002969594,
      "epoch": 0.11877394636015326,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19809016585350037,
      "learning_rate": 3e-06,
      "loss": -0.0096,
      "step": 248,
      "step_time": 8.256942182779312
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.009765625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 230.697265625,
      "completions/mean_terminated_length": 230.44773864746094,
      "completions/min_length": 86.0,
      "completions/min_terminated_length": 86.0,
      "entropy": 1.7994791666666667,
      "epoch": 0.11925287356321838,
      "fork/depth_mean": 0.26379621028900146,
      "fork/depth_std": 0.23345805704593658,
      "fork_head/loss": 2.7383248805999756,
      "fork_head/num_positive": 9.967741966247559,
      "fork_head/pred_top1_mass": 0.07315102964639664,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.9770631194114685,
      "learning_rate": 3e-06,
      "loss": 0.6931,
      "num_tokens": 6997086.0,
      "reward": 0.7261761426925659,
      "reward_std": 0.6488388180732727,
      "rewards/code_format_reward/mean": 0.5703125,
      "rewards/code_format_reward/std": 0.45752835273742676,
      "rewards/code_reward/mean": 0.15586361289024353,
      "rewards/code_reward/std": 0.32364901900291443,
      "step": 249,
      "step_time": 465.3884741514921
    },
    {
      "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,
      "entropy": 1.1563895089285714,
      "epoch": 0.11973180076628352,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04186144843697548,
      "learning_rate": 3e-06,
      "loss": -0.0067,
      "step": 250,
      "step_time": 7.692061066627502
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.7934389511744182,
      "epoch": 0.12021072796934866,
      "fork_head/loss": 2.714620351791382,
      "fork_head/num_positive": 9.967741966247559,
      "fork_head/pred_top1_mass": 0.07891633361577988,
      "grad_norm": 0.8132622241973877,
      "learning_rate": 3e-06,
      "loss": 0.6859,
      "step": 251,
      "step_time": 7.572969034314156
    },
    {
      "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,
      "entropy": 1.1645275354385376,
      "epoch": 0.1206896551724138,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.042020395398139954,
      "learning_rate": 3e-06,
      "loss": -0.0065,
      "step": 252,
      "step_time": 7.365267664194107
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.7678313122855291,
      "epoch": 0.12116858237547892,
      "fork_head/loss": 2.6828701496124268,
      "fork_head/num_positive": 9.967741966247559,
      "fork_head/pred_top1_mass": 0.09240839630365372,
      "grad_norm": 0.6885731816291809,
      "learning_rate": 3e-06,
      "loss": 0.6772,
      "step": 253,
      "step_time": 7.674375757575035
    },
    {
      "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,
      "entropy": 1.1662946428571428,
      "epoch": 0.12164750957854406,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06069132313132286,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 254,
      "step_time": 7.751405782997608
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.05555555555555555,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.16666666666666666,
      "entropy": 1.7607421875,
      "epoch": 0.1221264367816092,
      "fork_head/loss": 2.6488096714019775,
      "fork_head/num_positive": 9.967741966247559,
      "fork_head/pred_top1_mass": 0.1271485537290573,
      "grad_norm": 0.6799867153167725,
      "learning_rate": 3e-06,
      "loss": 0.6684,
      "step": 255,
      "step_time": 7.356069914996624
    },
    {
      "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,
      "entropy": 1.1710844550813948,
      "epoch": 0.12260536398467432,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0409386046230793,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 256,
      "step_time": 7.899344503879547
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 224.90625,
      "completions/mean_terminated_length": 224.78433227539062,
      "completions/min_length": 52.0,
      "completions/min_terminated_length": 52.0,
      "entropy": 1.2464409748713174,
      "epoch": 0.12308429118773946,
      "fork/depth_mean": 0.2294158935546875,
      "fork/depth_std": 0.23998697102069855,
      "fork_head/loss": 2.7571089267730713,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.07743460685014725,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0530741214752197,
      "learning_rate": 3e-06,
      "loss": 0.7023,
      "num_tokens": 7210238.0,
      "reward": 0.6249740719795227,
      "reward_std": 0.6209197044372559,
      "rewards/code_format_reward/mean": 0.4951171875,
      "rewards/code_format_reward/std": 0.45273205637931824,
      "rewards/code_reward/mean": 0.1298568844795227,
      "rewards/code_reward/std": 0.2922990620136261,
      "step": 257,
      "step_time": 493.29864779114723
    },
    {
      "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,
      "entropy": 1.605712890625,
      "epoch": 0.1235632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18225838243961334,
      "learning_rate": 3e-06,
      "loss": -0.0101,
      "step": 258,
      "step_time": 8.946067608892918
    },
    {
      "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,
      "entropy": 1.2413628538449606,
      "epoch": 0.12404214559386974,
      "fork_head/loss": 2.737577199935913,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.07638527452945709,
      "grad_norm": 0.7264775633811951,
      "learning_rate": 3e-06,
      "loss": 0.6972,
      "step": 259,
      "step_time": 7.587924115359783
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.03125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "entropy": 1.6015574149787426,
      "epoch": 0.12452107279693486,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1531110256910324,
      "learning_rate": 3e-06,
      "loss": -0.0105,
      "step": 260,
      "step_time": 8.171219266951084
    },
    {
      "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,
      "entropy": 1.2489583412806193,
      "epoch": 0.125,
      "fork_head/loss": 2.718639612197876,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.07954675704240799,
      "grad_norm": 0.46732866764068604,
      "learning_rate": 3e-06,
      "loss": 0.6916,
      "step": 261,
      "step_time": 7.798991501331329
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.03125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "entropy": 1.5924326591193676,
      "epoch": 0.12547892720306514,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15545779466629028,
      "learning_rate": 3e-06,
      "loss": -0.0111,
      "step": 262,
      "step_time": 8.340430103242397
    },
    {
      "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,
      "entropy": 1.2571614662806192,
      "epoch": 0.12595785440613028,
      "fork_head/loss": 2.708346366882324,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.07603175193071365,
      "grad_norm": 0.4903040826320648,
      "learning_rate": 3e-06,
      "loss": 0.6886,
      "step": 263,
      "step_time": 7.460551828145981
    },
    {
      "clip_ratio/high_max": 0.1875,
      "clip_ratio/high_mean": 0.09375,
      "clip_ratio/low_mean": 0.03125,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.5821990966796875,
      "epoch": 0.12643678160919541,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10816199332475662,
      "learning_rate": 3e-06,
      "loss": -0.013,
      "step": 264,
      "step_time": 8.253050543367863
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 224.37890625,
      "completions/mean_terminated_length": 224.37890625,
      "completions/min_length": 39.0,
      "completions/min_terminated_length": 39.0,
      "entropy": 1.9966023862361908,
      "epoch": 0.12691570881226052,
      "fork/depth_mean": 0.2305891364812851,
      "fork/depth_std": 0.21103104948997498,
      "fork_head/loss": 2.7302613258361816,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.06918355822563171,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6670970320701599,
      "learning_rate": 3e-06,
      "loss": 0.6907,
      "num_tokens": 7417312.0,
      "reward": 0.7929556965827942,
      "reward_std": 0.6302420496940613,
      "rewards/code_format_reward/mean": 0.591796875,
      "rewards/code_format_reward/std": 0.45153695344924927,
      "rewards/code_reward/mean": 0.2011588215827942,
      "rewards/code_reward/std": 0.34236058592796326,
      "step": 265,
      "step_time": 472.4491008371115
    },
    {
      "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,
      "entropy": 1.27734375,
      "epoch": 0.12739463601532566,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14387798309326172,
      "learning_rate": 3e-06,
      "loss": -0.0079,
      "step": 266,
      "step_time": 7.278028771281242
    },
    {
      "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,
      "entropy": 2.0050048902630806,
      "epoch": 0.1278735632183908,
      "fork_head/loss": 2.721966028213501,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.0746309831738472,
      "grad_norm": 0.6037601828575134,
      "learning_rate": 3e-06,
      "loss": 0.6877,
      "step": 267,
      "step_time": 8.322195202112198
    },
    {
      "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,
      "entropy": 1.2671223878860474,
      "epoch": 0.12835249042145594,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15378960967063904,
      "learning_rate": 3e-06,
      "loss": -0.008,
      "step": 268,
      "step_time": 7.8955328688025475
    },
    {
      "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,
      "entropy": 2.00880941003561,
      "epoch": 0.12883141762452108,
      "fork_head/loss": 2.7065956592559814,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.0755472332239151,
      "grad_norm": 0.4952162504196167,
      "learning_rate": 3e-06,
      "loss": 0.683,
      "step": 269,
      "step_time": 7.310508951544762
    },
    {
      "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,
      "entropy": 1.2582899371782938,
      "epoch": 0.12931034482758622,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13788093626499176,
      "learning_rate": 3e-06,
      "loss": -0.0086,
      "step": 270,
      "step_time": 7.558171756565571
    },
    {
      "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,
      "entropy": 2.0026652067899704,
      "epoch": 0.12978927203065135,
      "fork_head/loss": 2.690725326538086,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.08108112961053848,
      "grad_norm": 0.47597581148147583,
      "learning_rate": 3e-06,
      "loss": 0.6778,
      "step": 271,
      "step_time": 7.242053762078285
    },
    {
      "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,
      "entropy": 1.2396918376286825,
      "epoch": 0.13026819923371646,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14428779482841492,
      "learning_rate": 3e-06,
      "loss": -0.0099,
      "step": 272,
      "step_time": 7.802125506103039
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 225.46875,
      "completions/mean_terminated_length": 225.34902954101562,
      "completions/min_length": 21.0,
      "completions/min_terminated_length": 21.0,
      "entropy": 1.868958330154419,
      "epoch": 0.1307471264367816,
      "fork/depth_mean": 0.20556823909282684,
      "fork/depth_std": 0.20480017364025116,
      "fork_head/loss": 2.738886833190918,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.07706328481435776,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8656755685806274,
      "learning_rate": 3e-06,
      "loss": 0.6864,
      "num_tokens": 7627360.0,
      "reward": 0.7843291759490967,
      "reward_std": 0.5576251745223999,
      "rewards/code_format_reward/mean": 0.6025390625,
      "rewards/code_format_reward/std": 0.4158472716808319,
      "rewards/code_reward/mean": 0.1817900538444519,
      "rewards/code_reward/std": 0.324050635099411,
      "step": 273,
      "step_time": 525.4006648883224
    },
    {
      "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,
      "entropy": 1.4696875015894573,
      "epoch": 0.13122605363984674,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19173258543014526,
      "learning_rate": 3e-06,
      "loss": 0.0019,
      "step": 274,
      "step_time": 7.976119205355644
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.8669270833333333,
      "epoch": 0.13170498084291188,
      "fork_head/loss": 2.7239203453063965,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.07851837575435638,
      "grad_norm": 0.59714275598526,
      "learning_rate": 3e-06,
      "loss": 0.6816,
      "step": 275,
      "step_time": 7.700700677931309
    },
    {
      "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,
      "entropy": 1.484791668256124,
      "epoch": 0.13218390804597702,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18978048861026764,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 276,
      "step_time": 7.865974098443985
    },
    {
      "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,
      "entropy": 1.8655121485392252,
      "epoch": 0.13266283524904215,
      "fork_head/loss": 2.7086782455444336,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.08310332894325256,
      "grad_norm": 0.43948686122894287,
      "learning_rate": 3e-06,
      "loss": 0.6774,
      "step": 277,
      "step_time": 7.550845220685005
    },
    {
      "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,
      "entropy": 1.4758854150772094,
      "epoch": 0.1331417624521073,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20337019860744476,
      "learning_rate": 3e-06,
      "loss": -0.0002,
      "step": 278,
      "step_time": 7.649259619414806
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.046666666865348816,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.046666666865348816,
      "entropy": 1.8919618050257365,
      "epoch": 0.1336206896551724,
      "fork_head/loss": 2.695821523666382,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.10961947590112686,
      "grad_norm": 0.4325365722179413,
      "learning_rate": 3e-06,
      "loss": 0.6736,
      "step": 279,
      "step_time": 7.868260324001312
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.016666666666666666,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.11666666666666667,
      "entropy": 1.4943359375,
      "epoch": 0.13409961685823754,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16562646627426147,
      "learning_rate": 3e-06,
      "loss": -0.0021,
      "step": 280,
      "step_time": 7.194701738655567
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 231.39453125,
      "completions/mean_terminated_length": 231.39453125,
      "completions/min_length": 80.0,
      "completions/min_terminated_length": 80.0,
      "entropy": 2.0400327994273257,
      "epoch": 0.13457854406130268,
      "fork/depth_mean": 0.2757253050804138,
      "fork/depth_std": 0.23659169673919678,
      "fork_head/loss": 2.7148430347442627,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.07288803905248642,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6614143252372742,
      "learning_rate": 3e-06,
      "loss": 0.6697,
      "num_tokens": 7842426.0,
      "reward": 0.8892805576324463,
      "reward_std": 0.6298314332962036,
      "rewards/code_format_reward/mean": 0.6640625,
      "rewards/code_format_reward/std": 0.41313737630844116,
      "rewards/code_reward/mean": 0.2252180576324463,
      "rewards/code_reward/std": 0.35270050168037415,
      "step": 281,
      "step_time": 489.2958885356784
    },
    {
      "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,
      "entropy": 1.736914054552714,
      "epoch": 0.13505747126436782,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13231585919857025,
      "learning_rate": 3e-06,
      "loss": 0.0093,
      "step": 282,
      "step_time": 7.609129838645458
    },
    {
      "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,
      "entropy": 2.0150490815822897,
      "epoch": 0.13553639846743296,
      "fork_head/loss": 2.7049262523651123,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.08237145841121674,
      "grad_norm": 0.6364495158195496,
      "learning_rate": 3e-06,
      "loss": 0.6668,
      "step": 283,
      "step_time": 8.363687053322792
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.751573348045349,
      "epoch": 0.1360153256704981,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12934423983097076,
      "learning_rate": 3e-06,
      "loss": 0.0085,
      "step": 284,
      "step_time": 7.736310683190823
    },
    {
      "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,
      "entropy": 2.040784001350403,
      "epoch": 0.13649425287356323,
      "fork_head/loss": 2.687063455581665,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.09609653800725937,
      "grad_norm": 0.5475329160690308,
      "learning_rate": 3e-06,
      "loss": 0.6623,
      "step": 285,
      "step_time": 7.749227970838547
    },
    {
      "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,
      "entropy": 1.7502170085906983,
      "epoch": 0.13697318007662834,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13839083909988403,
      "learning_rate": 3e-06,
      "loss": 0.0084,
      "step": 286,
      "step_time": 8.523310117423534
    },
    {
      "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,
      "entropy": 2.025622248649597,
      "epoch": 0.13745210727969348,
      "fork_head/loss": 2.6665151119232178,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.10305389761924744,
      "grad_norm": 0.46813535690307617,
      "learning_rate": 3e-06,
      "loss": 0.6566,
      "step": 287,
      "step_time": 7.5221890062093735
    },
    {
      "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,
      "entropy": 1.7503689289093018,
      "epoch": 0.13793103448275862,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14164356887340546,
      "learning_rate": 3e-06,
      "loss": 0.0078,
      "step": 288,
      "step_time": 8.594746761023998
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 222.328125,
      "completions/mean_terminated_length": 222.1960906982422,
      "completions/min_length": 42.0,
      "completions/min_terminated_length": 42.0,
      "entropy": 1.5981517632802327,
      "epoch": 0.13840996168582376,
      "fork/depth_mean": 0.3032437264919281,
      "fork/depth_std": 0.2595197558403015,
      "fork_head/loss": 2.718827962875366,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06673189252614975,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6160004138946533,
      "learning_rate": 3e-06,
      "loss": 0.6598,
      "num_tokens": 8052338.0,
      "reward": 0.8467532396316528,
      "reward_std": 0.6201764345169067,
      "rewards/code_format_reward/mean": 0.6689453125,
      "rewards/code_format_reward/std": 0.41530463099479675,
      "rewards/code_reward/mean": 0.17780792713165283,
      "rewards/code_reward/std": 0.3193133771419525,
      "step": 289,
      "step_time": 468.6339668110013
    },
    {
      "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,
      "entropy": 1.8732856001172746,
      "epoch": 0.1388888888888889,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14500382542610168,
      "learning_rate": 3e-06,
      "loss": 0.0238,
      "step": 290,
      "step_time": 9.31861724704504
    },
    {
      "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,
      "entropy": 1.594180409113566,
      "epoch": 0.13936781609195403,
      "fork_head/loss": 2.714256525039673,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07665688544511795,
      "grad_norm": 0.625364363193512,
      "learning_rate": 3e-06,
      "loss": 0.6575,
      "step": 291,
      "step_time": 7.22262217104435
    },
    {
      "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,
      "entropy": 1.873181734766279,
      "epoch": 0.13984674329501914,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.159972682595253,
      "learning_rate": 3e-06,
      "loss": 0.0235,
      "step": 292,
      "step_time": 8.097512155771255
    },
    {
      "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,
      "entropy": 1.596679695447286,
      "epoch": 0.14032567049808428,
      "fork_head/loss": 2.701524257659912,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07677275687456131,
      "grad_norm": 0.5711675882339478,
      "learning_rate": 3e-06,
      "loss": 0.6544,
      "step": 293,
      "step_time": 7.69357642531395
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.883847951889038,
      "epoch": 0.14080459770114942,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15803487598896027,
      "learning_rate": 3e-06,
      "loss": 0.0228,
      "step": 294,
      "step_time": 7.609806522727013
    },
    {
      "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,
      "entropy": 1.5899956544240317,
      "epoch": 0.14128352490421456,
      "fork_head/loss": 2.684748411178589,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07194025069475174,
      "grad_norm": 0.4854421019554138,
      "learning_rate": 3e-06,
      "loss": 0.6482,
      "step": 295,
      "step_time": 8.16280397027731
    },
    {
      "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,
      "entropy": 1.8851500494139535,
      "epoch": 0.1417624521072797,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16020043194293976,
      "learning_rate": 3e-06,
      "loss": 0.0208,
      "step": 296,
      "step_time": 7.598455987870693
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 225.9609375,
      "completions/mean_terminated_length": 225.9021453857422,
      "completions/min_length": 53.0,
      "completions/min_terminated_length": 53.0,
      "entropy": 1.805164921283722,
      "epoch": 0.14224137931034483,
      "fork/depth_mean": 0.3240509331226349,
      "fork/depth_std": 0.2666846215724945,
      "fork_head/loss": 2.7309110164642334,
      "fork_head/num_positive": 12.375,
      "fork_head/pred_top1_mass": 0.0673985481262207,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5487269163131714,
      "learning_rate": 3e-06,
      "loss": 0.6903,
      "num_tokens": 8261438.0,
      "reward": 0.8627381920814514,
      "reward_std": 0.6098452210426331,
      "rewards/code_format_reward/mean": 0.673828125,
      "rewards/code_format_reward/std": 0.42552903294563293,
      "rewards/code_reward/mean": 0.1889100819826126,
      "rewards/code_reward/std": 0.33303242921829224,
      "step": 297,
      "step_time": 515.2239556014538
    },
    {
      "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,
      "entropy": 1.7956387911524092,
      "epoch": 0.14272030651340997,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21802006661891937,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 298,
      "step_time": 7.679245755076408
    },
    {
      "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,
      "entropy": 1.7971137086550395,
      "epoch": 0.14319923371647508,
      "fork_head/loss": 2.722698926925659,
      "fork_head/num_positive": 12.375,
      "fork_head/pred_top1_mass": 0.06366441398859024,
      "grad_norm": 0.5192992091178894,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "step": 299,
      "step_time": 9.014138452708721
    },
    {
      "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,
      "entropy": 1.7864350761686052,
      "epoch": 0.14367816091954022,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19632799923419952,
      "learning_rate": 3e-06,
      "loss": -0.006,
      "step": 300,
      "step_time": 7.565100498497486
    },
    {
      "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,
      "entropy": 1.8001844525337218,
      "epoch": 0.14415708812260536,
      "fork_head/loss": 2.7101430892944336,
      "fork_head/num_positive": 12.375,
      "fork_head/pred_top1_mass": 0.06726670265197754,
      "grad_norm": 0.4819996654987335,
      "learning_rate": 3e-06,
      "loss": 0.684,
      "step": 301,
      "step_time": 8.149663634598255
    },
    {
      "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,
      "entropy": 1.7892950092043196,
      "epoch": 0.1446360153256705,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19784896075725555,
      "learning_rate": 3e-06,
      "loss": -0.0074,
      "step": 302,
      "step_time": 8.066067703068256
    },
    {
      "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,
      "entropy": 1.7811197916666666,
      "epoch": 0.14511494252873564,
      "fork_head/loss": 2.695065975189209,
      "fork_head/num_positive": 12.375,
      "fork_head/pred_top1_mass": 0.07567436993122101,
      "grad_norm": 0.4492378830909729,
      "learning_rate": 3e-06,
      "loss": 0.6786,
      "step": 303,
      "step_time": 7.812012031674385
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.790690107004983,
      "epoch": 0.14559386973180077,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14894026517868042,
      "learning_rate": 3e-06,
      "loss": -0.0073,
      "step": 304,
      "step_time": 7.76327895373106
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 226.87109375,
      "completions/mean_terminated_length": 226.7568817138672,
      "completions/min_length": 57.0,
      "completions/min_terminated_length": 57.0,
      "entropy": 1.834560284247765,
      "epoch": 0.1460727969348659,
      "fork/depth_mean": 0.2533308267593384,
      "fork/depth_std": 0.21087850630283356,
      "fork_head/loss": 2.722168207168579,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.0741736888885498,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6303244233131409,
      "learning_rate": 3e-06,
      "loss": 0.6749,
      "num_tokens": 8473660.0,
      "reward": 0.9087932109832764,
      "reward_std": 0.667566180229187,
      "rewards/code_format_reward/mean": 0.6728515625,
      "rewards/code_format_reward/std": 0.43782204389572144,
      "rewards/code_reward/mean": 0.23594161868095398,
      "rewards/code_reward/std": 0.37771841883659363,
      "step": 305,
      "step_time": 474.6643754541874
    },
    {
      "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,
      "entropy": 2.02685546875,
      "epoch": 0.14655172413793102,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18074527382850647,
      "learning_rate": 3e-06,
      "loss": 0.0077,
      "step": 306,
      "step_time": 7.875670425593853
    },
    {
      "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,
      "entropy": 1.831380193050091,
      "epoch": 0.14703065134099616,
      "fork_head/loss": 2.710516929626465,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07261935621500015,
      "grad_norm": 0.4966951906681061,
      "learning_rate": 3e-06,
      "loss": 0.6715,
      "step": 307,
      "step_time": 7.709654197096825
    },
    {
      "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,
      "entropy": 2.0229899187882743,
      "epoch": 0.1475095785440613,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16931463778018951,
      "learning_rate": 3e-06,
      "loss": 0.0069,
      "step": 308,
      "step_time": 8.06404872238636
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.838291278252235,
      "epoch": 0.14798850574712644,
      "fork_head/loss": 2.698734998703003,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.0741865336894989,
      "grad_norm": 0.4475520849227905,
      "learning_rate": 3e-06,
      "loss": 0.6676,
      "step": 309,
      "step_time": 7.560082510113716
    },
    {
      "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,
      "entropy": 2.0276014506816864,
      "epoch": 0.14846743295019157,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18091410398483276,
      "learning_rate": 3e-06,
      "loss": 0.0064,
      "step": 310,
      "step_time": 7.677307166159153
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.836713745043828,
      "epoch": 0.1489463601532567,
      "fork_head/loss": 2.685004234313965,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.09126292169094086,
      "grad_norm": 0.5085337162017822,
      "learning_rate": 3e-06,
      "loss": 0.663,
      "step": 311,
      "step_time": 8.08677277714014
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.020833333333333332,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.020833333333333332,
      "entropy": 2.01791720589002,
      "epoch": 0.14942528735632185,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10575191676616669,
      "learning_rate": 3e-06,
      "loss": 0.0055,
      "step": 312,
      "step_time": 7.50250393897295
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 231.755859375,
      "completions/mean_terminated_length": 231.61297607421875,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.8962131076388888,
      "epoch": 0.14990421455938696,
      "fork/depth_mean": 0.2053501456975937,
      "fork/depth_std": 0.17884625494480133,
      "fork_head/loss": 2.7246205806732178,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.06758923083543777,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.588361382484436,
      "learning_rate": 3e-06,
      "loss": 0.6773,
      "num_tokens": 8681935.0,
      "reward": 0.8185000419616699,
      "reward_std": 0.6489876508712769,
      "rewards/code_format_reward/mean": 0.6015625,
      "rewards/code_format_reward/std": 0.4612557291984558,
      "rewards/code_reward/mean": 0.2169375866651535,
      "rewards/code_reward/std": 0.3394949436187744,
      "step": 313,
      "step_time": 475.8037943392992
    },
    {
      "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,
      "entropy": 1.636865234375,
      "epoch": 0.1503831417624521,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15945681929588318,
      "learning_rate": 3e-06,
      "loss": 0.0036,
      "step": 314,
      "step_time": 7.919688858091831
    },
    {
      "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,
      "entropy": 1.8888165553410847,
      "epoch": 0.15086206896551724,
      "fork_head/loss": 2.7158384323120117,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.06651289761066437,
      "grad_norm": 0.44445955753326416,
      "learning_rate": 3e-06,
      "loss": 0.6753,
      "step": 315,
      "step_time": 7.980330750346184
    },
    {
      "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,
      "entropy": 1.6525390625,
      "epoch": 0.15134099616858238,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15954940021038055,
      "learning_rate": 3e-06,
      "loss": 0.0032,
      "step": 316,
      "step_time": 7.052637055516243
    },
    {
      "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,
      "entropy": 1.8740415175755818,
      "epoch": 0.1518199233716475,
      "fork_head/loss": 2.7060165405273438,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.07756398618221283,
      "grad_norm": 0.3701857924461365,
      "learning_rate": 3e-06,
      "loss": 0.6713,
      "step": 317,
      "step_time": 7.743414923548698
    },
    {
      "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,
      "entropy": 1.6430989503860474,
      "epoch": 0.15229885057471265,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14625714719295502,
      "learning_rate": 3e-06,
      "loss": 0.002,
      "step": 318,
      "step_time": 7.501337364315987
    },
    {
      "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,
      "entropy": 1.9065031740400527,
      "epoch": 0.1527777777777778,
      "fork_head/loss": 2.696397542953491,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.08242988586425781,
      "grad_norm": 0.37546592950820923,
      "learning_rate": 3e-06,
      "loss": 0.6698,
      "step": 319,
      "step_time": 7.369455359876156
    },
    {
      "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,
      "entropy": 1.64619140625,
      "epoch": 0.1532567049808429,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16809457540512085,
      "learning_rate": 3e-06,
      "loss": 0.0018,
      "step": 320,
      "step_time": 7.748534597456455
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.009765625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 226.13671875,
      "completions/mean_terminated_length": 225.8422088623047,
      "completions/min_length": 80.0,
      "completions/min_terminated_length": 80.0,
      "entropy": 1.2740885416666667,
      "epoch": 0.15373563218390804,
      "fork/depth_mean": 0.23606997728347778,
      "fork/depth_std": 0.17695918679237366,
      "fork_head/loss": 2.7167868614196777,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.07795342057943344,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5594058036804199,
      "learning_rate": 3e-06,
      "loss": 0.6839,
      "num_tokens": 8888533.0,
      "reward": 1.0328788757324219,
      "reward_std": 0.7043209671974182,
      "rewards/code_format_reward/mean": 0.7080078125,
      "rewards/code_format_reward/std": 0.4198992848396301,
      "rewards/code_reward/mean": 0.32487109303474426,
      "rewards/code_reward/std": 0.4271993935108185,
      "step": 321,
      "step_time": 495.3187390640378
    },
    {
      "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,
      "entropy": 2.041015625,
      "epoch": 0.15421455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15556712448596954,
      "learning_rate": 3e-06,
      "loss": -0.007,
      "step": 322,
      "step_time": 8.159950129687786
    },
    {
      "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,
      "entropy": 1.3079427083333333,
      "epoch": 0.15469348659003831,
      "fork_head/loss": 2.706897020339966,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.07924072444438934,
      "grad_norm": 0.5310034155845642,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "step": 323,
      "step_time": 7.125004969537258
    },
    {
      "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,
      "entropy": 2.0640190972222223,
      "epoch": 0.15517241379310345,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15256129205226898,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 324,
      "step_time": 7.587127186357975
    },
    {
      "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,
      "entropy": 1.2981770833333333,
      "epoch": 0.1556513409961686,
      "fork_head/loss": 2.6908507347106934,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.10683846473693848,
      "grad_norm": 0.4837616980075836,
      "learning_rate": 3e-06,
      "loss": 0.6762,
      "step": 325,
      "step_time": 7.897359162569046
    },
    {
      "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,
      "entropy": 2.0520833333333335,
      "epoch": 0.15613026819923373,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14852143824100494,
      "learning_rate": 3e-06,
      "loss": -0.0074,
      "step": 326,
      "step_time": 7.6116795018315315
    },
    {
      "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,
      "entropy": 1.3098958333333333,
      "epoch": 0.15660919540229884,
      "fork_head/loss": 2.6717166900634766,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.10518841445446014,
      "grad_norm": 0.43955501914024353,
      "learning_rate": 3e-06,
      "loss": 0.6708,
      "step": 327,
      "step_time": 7.250779040157795
    },
    {
      "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,
      "entropy": 2.0506727430555554,
      "epoch": 0.15708812260536398,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16515782475471497,
      "learning_rate": 3e-06,
      "loss": -0.0082,
      "step": 328,
      "step_time": 7.5050159096717834
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 230.921875,
      "completions/mean_terminated_length": 230.921875,
      "completions/min_length": 94.0,
      "completions/min_terminated_length": 94.0,
      "entropy": 1.8610025942325592,
      "epoch": 0.15756704980842912,
      "fork/depth_mean": 0.19538667798042297,
      "fork/depth_std": 0.1568288952112198,
      "fork_head/loss": 2.744011640548706,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.06430058926343918,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6901955008506775,
      "learning_rate": 3e-06,
      "loss": 0.6907,
      "num_tokens": 9102157.0,
      "reward": 0.8648869395256042,
      "reward_std": 0.6476251482963562,
      "rewards/code_format_reward/mean": 0.634765625,
      "rewards/code_format_reward/std": 0.44281870126724243,
      "rewards/code_reward/mean": 0.23012134432792664,
      "rewards/code_reward/std": 0.38986003398895264,
      "step": 329,
      "step_time": 773.1669930145144
    },
    {
      "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,
      "entropy": 1.1823381696428572,
      "epoch": 0.15804597701149425,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09526235610246658,
      "learning_rate": 3e-06,
      "loss": -0.0065,
      "step": 330,
      "step_time": 6.9338622614741325
    },
    {
      "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,
      "entropy": 1.8512369692325592,
      "epoch": 0.1585249042145594,
      "fork_head/loss": 2.7313027381896973,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.07710318267345428,
      "grad_norm": 0.5774946212768555,
      "learning_rate": 3e-06,
      "loss": 0.6879,
      "step": 331,
      "step_time": 7.964941620826721
    },
    {
      "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,
      "entropy": 1.1697823660714286,
      "epoch": 0.15900383141762453,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0875936970114708,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 332,
      "step_time": 7.561043463647366
    },
    {
      "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,
      "entropy": 1.8694661557674408,
      "epoch": 0.15948275862068967,
      "fork_head/loss": 2.713989734649658,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.08555468916893005,
      "grad_norm": 0.4762364327907562,
      "learning_rate": 3e-06,
      "loss": 0.6832,
      "step": 333,
      "step_time": 7.555005118250847
    },
    {
      "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,
      "entropy": 1.1533203125,
      "epoch": 0.15996168582375478,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09257755428552628,
      "learning_rate": 3e-06,
      "loss": -0.0071,
      "step": 334,
      "step_time": 7.614955618977547
    },
    {
      "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,
      "entropy": 1.8854166567325592,
      "epoch": 0.16044061302681992,
      "fork_head/loss": 2.697458028793335,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.09355515986680984,
      "grad_norm": 0.5098318457603455,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 335,
      "step_time": 6.812476225197315
    },
    {
      "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,
      "entropy": 1.1527622767857142,
      "epoch": 0.16091954022988506,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09169309586286545,
      "learning_rate": 3e-06,
      "loss": -0.0076,
      "step": 336,
      "step_time": 7.813331633806229
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 224.224609375,
      "completions/mean_terminated_length": 224.1624298095703,
      "completions/min_length": 19.0,
      "completions/min_terminated_length": 19.0,
      "entropy": 2.2453243407336148,
      "epoch": 0.1613984674329502,
      "fork/depth_mean": 0.17006319761276245,
      "fork/depth_std": 0.16005894541740417,
      "fork_head/loss": 2.7267355918884277,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06774920970201492,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7142240405082703,
      "learning_rate": 3e-06,
      "loss": 0.6856,
      "num_tokens": 9303056.0,
      "reward": 0.7943264245986938,
      "reward_std": 0.6604226231575012,
      "rewards/code_format_reward/mean": 0.60546875,
      "rewards/code_format_reward/std": 0.464607834815979,
      "rewards/code_reward/mean": 0.18885764479637146,
      "rewards/code_reward/std": 0.3158889710903168,
      "step": 337,
      "step_time": 459.6888966858387
    },
    {
      "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,
      "entropy": 1.0711263020833333,
      "epoch": 0.16187739463601533,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.050140395760536194,
      "learning_rate": 3e-06,
      "loss": -0.0058,
      "step": 338,
      "step_time": 7.75412829965353
    },
    {
      "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,
      "entropy": 2.228633999824524,
      "epoch": 0.16235632183908047,
      "fork_head/loss": 2.7135977745056152,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06960640102624893,
      "grad_norm": 0.6439325213432312,
      "learning_rate": 3e-06,
      "loss": 0.6821,
      "step": 339,
      "step_time": 7.710120886564255
    },
    {
      "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,
      "entropy": 1.0730794270833333,
      "epoch": 0.16283524904214558,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04879250004887581,
      "learning_rate": 3e-06,
      "loss": -0.0055,
      "step": 340,
      "step_time": 7.548122890293598
    },
    {
      "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,
      "entropy": 2.2348780740391123,
      "epoch": 0.16331417624521072,
      "fork_head/loss": 2.692533016204834,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.09705841541290283,
      "grad_norm": 0.5327420830726624,
      "learning_rate": 3e-06,
      "loss": 0.6771,
      "step": 341,
      "step_time": 7.60172551125288
    },
    {
      "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,
      "entropy": 1.0724283854166667,
      "epoch": 0.16379310344827586,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0478668175637722,
      "learning_rate": 3e-06,
      "loss": -0.0059,
      "step": 342,
      "step_time": 6.903803624212742
    },
    {
      "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,
      "entropy": 2.2333392555063423,
      "epoch": 0.164272030651341,
      "fork_head/loss": 2.6727659702301025,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.09723392128944397,
      "grad_norm": 0.46078088879585266,
      "learning_rate": 3e-06,
      "loss": 0.6715,
      "step": 343,
      "step_time": 8.198415771126747
    },
    {
      "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,
      "entropy": 1.0641276041666667,
      "epoch": 0.16475095785440613,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0486062616109848,
      "learning_rate": 3e-06,
      "loss": -0.0058,
      "step": 344,
      "step_time": 7.343709081411362
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.009765625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 232.943359375,
      "completions/mean_terminated_length": 232.71597290039062,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.4326171875,
      "epoch": 0.16522988505747127,
      "fork/depth_mean": 0.1992252916097641,
      "fork/depth_std": 0.16573749482631683,
      "fork_head/loss": 2.7277889251708984,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07212618738412857,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6914125084877014,
      "learning_rate": 3e-06,
      "loss": 0.6804,
      "num_tokens": 9522435.0,
      "reward": 0.8534036874771118,
      "reward_std": 0.6563976407051086,
      "rewards/code_format_reward/mean": 0.6494140625,
      "rewards/code_format_reward/std": 0.43302926421165466,
      "rewards/code_reward/mean": 0.2039896547794342,
      "rewards/code_reward/std": 0.35727784037590027,
      "step": 345,
      "step_time": 462.82494056224823
    },
    {
      "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,
      "entropy": 2.0260416666666665,
      "epoch": 0.1657088122605364,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07372087240219116,
      "learning_rate": 3e-06,
      "loss": 0.0014,
      "step": 346,
      "step_time": 6.825615592300892
    },
    {
      "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,
      "entropy": 1.4453125,
      "epoch": 0.16618773946360152,
      "fork_head/loss": 2.719149589538574,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07333660870790482,
      "grad_norm": 0.5956210494041443,
      "learning_rate": 3e-06,
      "loss": 0.6783,
      "step": 347,
      "step_time": 7.825912185013294
    },
    {
      "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,
      "entropy": 2.0260416666666665,
      "epoch": 0.16666666666666666,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06984347850084305,
      "learning_rate": 3e-06,
      "loss": 0.0013,
      "step": 348,
      "step_time": 7.46814251691103
    },
    {
      "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,
      "entropy": 1.41796875,
      "epoch": 0.1671455938697318,
      "fork_head/loss": 2.7045464515686035,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.0836276113986969,
      "grad_norm": 0.5083640217781067,
      "learning_rate": 3e-06,
      "loss": 0.6745,
      "step": 349,
      "step_time": 7.039910949766636
    },
    {
      "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,
      "entropy": 2.046875,
      "epoch": 0.16762452107279693,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07273561507463455,
      "learning_rate": 3e-06,
      "loss": 0.0013,
      "step": 350,
      "step_time": 8.023671835660934
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.416015625,
      "epoch": 0.16810344827586207,
      "fork_head/loss": 2.6903793811798096,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08344666659832001,
      "grad_norm": 0.47382402420043945,
      "learning_rate": 3e-06,
      "loss": 0.6708,
      "step": 351,
      "step_time": 7.0672149658203125
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.3333333333333333,
      "clip_ratio/low_min": 0.3333333333333333,
      "clip_ratio/region_mean": 0.3333333333333333,
      "entropy": 2.125,
      "epoch": 0.1685823754789272,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.004320937674492598,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 352,
      "step_time": 7.330066934227943
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 220.37890625,
      "completions/mean_terminated_length": 220.30918884277344,
      "completions/min_length": 13.0,
      "completions/min_terminated_length": 13.0,
      "entropy": 1.3266059027777777,
      "epoch": 0.16906130268199235,
      "fork/depth_mean": 0.19590343534946442,
      "fork/depth_std": 0.17250585556030273,
      "fork_head/loss": 2.734978675842285,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.06229795515537262,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5546173453330994,
      "learning_rate": 3e-06,
      "loss": 0.6862,
      "num_tokens": 9719621.0,
      "reward": 0.9659653902053833,
      "reward_std": 0.7112915515899658,
      "rewards/code_format_reward/mean": 0.6611328125,
      "rewards/code_format_reward/std": 0.4287983179092407,
      "rewards/code_reward/mean": 0.3048325777053833,
      "rewards/code_reward/std": 0.39620596170425415,
      "step": 353,
      "step_time": 477.3187549635768
    },
    {
      "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,
      "entropy": 1.3746744751930238,
      "epoch": 0.16954022988505746,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13503913581371307,
      "learning_rate": 3e-06,
      "loss": -0.002,
      "step": 354,
      "step_time": 8.144000373780727
    },
    {
      "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,
      "entropy": 1.3161892361111112,
      "epoch": 0.1700191570881226,
      "fork_head/loss": 2.728870391845703,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.06367155909538269,
      "grad_norm": 0.5309804677963257,
      "learning_rate": 3e-06,
      "loss": 0.6851,
      "step": 355,
      "step_time": 7.580307677388191
    },
    {
      "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,
      "entropy": 1.3638671875,
      "epoch": 0.17049808429118773,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1403999626636505,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 356,
      "step_time": 7.344587720930576
    },
    {
      "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,
      "entropy": 1.3416883680555556,
      "epoch": 0.17097701149425287,
      "fork_head/loss": 2.7172534465789795,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.06832877546548843,
      "grad_norm": 0.4963780641555786,
      "learning_rate": 3e-06,
      "loss": 0.6815,
      "step": 357,
      "step_time": 7.550817847251892
    },
    {
      "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,
      "entropy": 1.3481119751930237,
      "epoch": 0.171455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17131009697914124,
      "learning_rate": 3e-06,
      "loss": -0.0036,
      "step": 358,
      "step_time": 7.107489742338657
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.3509114583333333,
      "epoch": 0.17193486590038315,
      "fork_head/loss": 2.703092098236084,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.0745689645409584,
      "grad_norm": 0.4556436836719513,
      "learning_rate": 3e-06,
      "loss": 0.6768,
      "step": 359,
      "step_time": 7.612797923386097
    },
    {
      "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,
      "entropy": 1.3394205689430236,
      "epoch": 0.1724137931034483,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13849741220474243,
      "learning_rate": 3e-06,
      "loss": -0.0044,
      "step": 360,
      "step_time": 7.685968495905399
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 222.46484375,
      "completions/mean_terminated_length": 222.3992156982422,
      "completions/min_length": 63.0,
      "completions/min_terminated_length": 63.0,
      "entropy": 1.4493272701899211,
      "epoch": 0.1728927203065134,
      "fork/depth_mean": 0.2276354283094406,
      "fork/depth_std": 0.20271845161914825,
      "fork_head/loss": 2.731901168823242,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.05998148396611214,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.48739323019981384,
      "learning_rate": 3e-06,
      "loss": 0.683,
      "num_tokens": 9930691.0,
      "reward": 0.8779771327972412,
      "reward_std": 0.5966188311576843,
      "rewards/code_format_reward/mean": 0.693359375,
      "rewards/code_format_reward/std": 0.41228070855140686,
      "rewards/code_reward/mean": 0.18461772799491882,
      "rewards/code_reward/std": 0.33190494775772095,
      "step": 361,
      "step_time": 475.10859228670597
    },
    {
      "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,
      "entropy": 1.3788519998391469,
      "epoch": 0.17337164750957854,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14171338081359863,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 362,
      "step_time": 7.312357343733311
    },
    {
      "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,
      "entropy": 1.443359375,
      "epoch": 0.17385057471264367,
      "fork_head/loss": 2.7234044075012207,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.056889019906520844,
      "grad_norm": 0.43908095359802246,
      "learning_rate": 3e-06,
      "loss": 0.6808,
      "step": 363,
      "step_time": 7.453264392912388
    },
    {
      "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,
      "entropy": 1.3713650206724803,
      "epoch": 0.1743295019157088,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1335553228855133,
      "learning_rate": 3e-06,
      "loss": -0.0007,
      "step": 364,
      "step_time": 7.820786029100418
    },
    {
      "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,
      "entropy": 1.4468315839767456,
      "epoch": 0.17480842911877395,
      "fork_head/loss": 2.7109827995300293,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.05439000576734543,
      "grad_norm": 0.42579594254493713,
      "learning_rate": 3e-06,
      "loss": 0.6777,
      "step": 365,
      "step_time": 7.0367905125021935
    },
    {
      "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,
      "entropy": 1.3776041666666667,
      "epoch": 0.1752873563218391,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13674353063106537,
      "learning_rate": 3e-06,
      "loss": -0.0015,
      "step": 366,
      "step_time": 7.991743981838226
    },
    {
      "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,
      "entropy": 1.4393446048100789,
      "epoch": 0.17576628352490423,
      "fork_head/loss": 2.6961309909820557,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.06505925953388214,
      "grad_norm": 0.4307600259780884,
      "learning_rate": 3e-06,
      "loss": 0.674,
      "step": 367,
      "step_time": 7.424127236008644
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.4011501769224803,
      "epoch": 0.17624521072796934,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18701954185962677,
      "learning_rate": 3e-06,
      "loss": -0.0024,
      "step": 368,
      "step_time": 7.321131214499474
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 226.716796875,
      "completions/mean_terminated_length": 226.54421997070312,
      "completions/min_length": 70.0,
      "completions/min_terminated_length": 70.0,
      "entropy": 1.463671875,
      "epoch": 0.17672413793103448,
      "fork/depth_mean": 0.27403753995895386,
      "fork/depth_std": 0.22919192910194397,
      "fork_head/loss": 2.7268192768096924,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.06232769042253494,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7538728713989258,
      "learning_rate": 3e-06,
      "loss": 0.6725,
      "num_tokens": 10132754.0,
      "reward": 0.9623229503631592,
      "reward_std": 0.6435428857803345,
      "rewards/code_format_reward/mean": 0.73046875,
      "rewards/code_format_reward/std": 0.4025473892688751,
      "rewards/code_reward/mean": 0.2318541556596756,
      "rewards/code_reward/std": 0.3619474172592163,
      "step": 369,
      "step_time": 475.27627197653055
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.459009418120751,
      "epoch": 0.1772030651340996,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1227140724658966,
      "learning_rate": 3e-06,
      "loss": 0.007,
      "step": 370,
      "step_time": 8.390628933906555
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.4569010416666666,
      "epoch": 0.17768199233716475,
      "fork_head/loss": 2.7159786224365234,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.06526605784893036,
      "grad_norm": 0.5612732172012329,
      "learning_rate": 3e-06,
      "loss": 0.67,
      "step": 371,
      "step_time": 7.859503053128719
    },
    {
      "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,
      "entropy": 1.4520733173076923,
      "epoch": 0.1781609195402299,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1405252069234848,
      "learning_rate": 3e-06,
      "loss": 0.0074,
      "step": 372,
      "step_time": 7.249375529587269
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.4623697916666667,
      "epoch": 0.17863984674329503,
      "fork_head/loss": 2.701735258102417,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.07649379223585129,
      "grad_norm": 0.49562421441078186,
      "learning_rate": 3e-06,
      "loss": 0.6656,
      "step": 373,
      "step_time": 7.837835997343063
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.4702774469669049,
      "epoch": 0.17911877394636017,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12890249490737915,
      "learning_rate": 3e-06,
      "loss": 0.007,
      "step": 374,
      "step_time": 7.688862465322018
    },
    {
      "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,
      "entropy": 1.4689670085906983,
      "epoch": 0.17959770114942528,
      "fork_head/loss": 2.687063217163086,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.0861346423625946,
      "grad_norm": 0.47790220379829407,
      "learning_rate": 3e-06,
      "loss": 0.661,
      "step": 375,
      "step_time": 7.645905800163746
    },
    {
      "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,
      "entropy": 1.4800180288461537,
      "epoch": 0.18007662835249041,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15732069313526154,
      "learning_rate": 3e-06,
      "loss": 0.0065,
      "step": 376,
      "step_time": 7.594266541302204
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 214.708984375,
      "completions/mean_terminated_length": 214.628173828125,
      "completions/min_length": 73.0,
      "completions/min_terminated_length": 73.0,
      "entropy": 1.5066105769230769,
      "epoch": 0.18055555555555555,
      "fork/depth_mean": 0.2988888919353485,
      "fork/depth_std": 0.23316486179828644,
      "fork_head/loss": 2.7319717407226562,
      "fork_head/num_positive": 10.387096405029297,
      "fork_head/pred_top1_mass": 0.05767272412776947,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6969423890113831,
      "learning_rate": 3e-06,
      "loss": 0.6712,
      "num_tokens": 10327181.0,
      "reward": 1.2015845775604248,
      "reward_std": 0.6087460517883301,
      "rewards/code_format_reward/mean": 0.8251953125,
      "rewards/code_format_reward/std": 0.3238859474658966,
      "rewards/code_reward/mean": 0.37638920545578003,
      "rewards/code_reward/std": 0.42657214403152466,
      "step": 377,
      "step_time": 464.4706045985222
    },
    {
      "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,
      "entropy": 1.5018554709174416,
      "epoch": 0.1810344827586207,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15868814289569855,
      "learning_rate": 3e-06,
      "loss": 0.0131,
      "step": 378,
      "step_time": 7.769110061228275
    },
    {
      "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,
      "entropy": 1.4868289232254028,
      "epoch": 0.18151340996168583,
      "fork_head/loss": 2.719095230102539,
      "fork_head/num_positive": 10.387096405029297,
      "fork_head/pred_top1_mass": 0.05958256125450134,
      "grad_norm": 0.6784685254096985,
      "learning_rate": 3e-06,
      "loss": 0.6668,
      "step": 379,
      "step_time": 7.937903553247452
    },
    {
      "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,
      "entropy": 1.5061789751052856,
      "epoch": 0.18199233716475097,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17826510965824127,
      "learning_rate": 3e-06,
      "loss": 0.0123,
      "step": 380,
      "step_time": 7.365585975348949
    },
    {
      "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,
      "entropy": 1.4793920241869414,
      "epoch": 0.1824712643678161,
      "fork_head/loss": 2.6953299045562744,
      "fork_head/num_positive": 10.387096405029297,
      "fork_head/pred_top1_mass": 0.06797120720148087,
      "grad_norm": 0.6110901832580566,
      "learning_rate": 3e-06,
      "loss": 0.6602,
      "step": 381,
      "step_time": 7.700469255447388
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.4820342118089849,
      "epoch": 0.18295019157088122,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11204194277524948,
      "learning_rate": 3e-06,
      "loss": 0.0114,
      "step": 382,
      "step_time": 7.656743302941322
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.4797926682692308,
      "epoch": 0.18342911877394635,
      "fork_head/loss": 2.666926145553589,
      "fork_head/num_positive": 10.387096405029297,
      "fork_head/pred_top1_mass": 0.08805125951766968,
      "grad_norm": 0.5239434838294983,
      "learning_rate": 3e-06,
      "loss": 0.653,
      "step": 383,
      "step_time": 7.3815402165055275
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.12121212211522189,
      "entropy": 1.4718454046682878,
      "epoch": 0.1839080459770115,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07553054392337799,
      "learning_rate": 3e-06,
      "loss": 0.0103,
      "step": 384,
      "step_time": 7.590367533266544
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 221.55859375,
      "completions/mean_terminated_length": 221.4235382080078,
      "completions/min_length": 86.0,
      "completions/min_terminated_length": 86.0,
      "entropy": 1.4725399911403656,
      "epoch": 0.18438697318007663,
      "fork/depth_mean": 0.2674293518066406,
      "fork/depth_std": 0.23201170563697815,
      "fork_head/loss": 2.7328081130981445,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.05840256065130234,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7757052183151245,
      "learning_rate": 3e-06,
      "loss": 0.6909,
      "num_tokens": 10529851.0,
      "reward": 0.9785274267196655,
      "reward_std": 0.6422846913337708,
      "rewards/code_format_reward/mean": 0.7099609375,
      "rewards/code_format_reward/std": 0.40587466955184937,
      "rewards/code_reward/mean": 0.26856642961502075,
      "rewards/code_reward/std": 0.37353989481925964,
      "step": 385,
      "step_time": 492.7525214403868
    },
    {
      "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,
      "entropy": 1.5324928868900647,
      "epoch": 0.18486590038314177,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12648221850395203,
      "learning_rate": 3e-06,
      "loss": -0.0072,
      "step": 386,
      "step_time": 8.685692586004734
    },
    {
      "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,
      "entropy": 1.4771437857832228,
      "epoch": 0.1853448275862069,
      "fork_head/loss": 2.718945264816284,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.0642550140619278,
      "grad_norm": 0.5283551216125488,
      "learning_rate": 3e-06,
      "loss": 0.6878,
      "step": 387,
      "step_time": 7.2713910937309265
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.5483842166987332,
      "epoch": 0.18582375478927204,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14258944988250732,
      "learning_rate": 3e-06,
      "loss": -0.0085,
      "step": 388,
      "step_time": 7.798112839460373
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.4769228994846344,
      "epoch": 0.18630268199233715,
      "fork_head/loss": 2.7064290046691895,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.0816936269402504,
      "grad_norm": 0.5032984018325806,
      "learning_rate": 3e-06,
      "loss": 0.6827,
      "step": 389,
      "step_time": 7.551597647368908
    },
    {
      "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,
      "entropy": 1.5453953688794917,
      "epoch": 0.1867816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13913506269454956,
      "learning_rate": 3e-06,
      "loss": -0.0092,
      "step": 390,
      "step_time": 7.959310933947563
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.4790852793625422,
      "epoch": 0.18726053639846743,
      "fork_head/loss": 2.6951956748962402,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.08479553461074829,
      "grad_norm": 0.6149808764457703,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "step": 391,
      "step_time": 7.905872620642185
    },
    {
      "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,
      "entropy": 1.5374348868023267,
      "epoch": 0.18773946360153257,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13884536921977997,
      "learning_rate": 3e-06,
      "loss": -0.0097,
      "step": 392,
      "step_time": 7.8724474385380745
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 233.287109375,
      "completions/mean_terminated_length": 233.24266052246094,
      "completions/min_length": 80.0,
      "completions/min_terminated_length": 80.0,
      "entropy": 0.9912109375,
      "epoch": 0.1882183908045977,
      "fork/depth_mean": 0.20886406302452087,
      "fork/depth_std": 0.19892726838588715,
      "fork_head/loss": 2.738424777984619,
      "fork_head/num_positive": 12.033333778381348,
      "fork_head/pred_top1_mass": 0.0638677179813385,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.8019521236419678,
      "learning_rate": 3e-06,
      "loss": 0.6831,
      "num_tokens": 10744030.0,
      "reward": 0.8285945057868958,
      "reward_std": 0.6138575673103333,
      "rewards/code_format_reward/mean": 0.6328125,
      "rewards/code_format_reward/std": 0.44119733572006226,
      "rewards/code_reward/mean": 0.19578202068805695,
      "rewards/code_reward/std": 0.3544686734676361,
      "step": 393,
      "step_time": 508.3670546039939
    },
    {
      "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,
      "entropy": 1.72451171875,
      "epoch": 0.18869731800766285,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0788620337843895,
      "learning_rate": 3e-06,
      "loss": 0.0016,
      "step": 394,
      "step_time": 7.6804643496870995
    },
    {
      "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,
      "entropy": 0.9710693359375,
      "epoch": 0.18917624521072796,
      "fork_head/loss": 2.724578857421875,
      "fork_head/num_positive": 12.033333778381348,
      "fork_head/pred_top1_mass": 0.05687286704778671,
      "grad_norm": 0.502154529094696,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "step": 395,
      "step_time": 8.093957401812077
    },
    {
      "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,
      "entropy": 1.71474609375,
      "epoch": 0.1896551724137931,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0810118168592453,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 396,
      "step_time": 7.625940270721912
    },
    {
      "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,
      "entropy": 0.9669189453125,
      "epoch": 0.19013409961685823,
      "fork_head/loss": 2.7128348350524902,
      "fork_head/num_positive": 12.033333778381348,
      "fork_head/pred_top1_mass": 0.06854002922773361,
      "grad_norm": 0.48798730969429016,
      "learning_rate": 3e-06,
      "loss": 0.6745,
      "step": 397,
      "step_time": 7.320172756910324
    },
    {
      "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,
      "entropy": 1.730859375,
      "epoch": 0.19061302681992337,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0841798186302185,
      "learning_rate": 3e-06,
      "loss": 0.0013,
      "step": 398,
      "step_time": 7.801449902355671
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 0.9654541015625,
      "epoch": 0.1910919540229885,
      "fork_head/loss": 2.7001395225524902,
      "fork_head/num_positive": 12.033333778381348,
      "fork_head/pred_top1_mass": 0.073609858751297,
      "grad_norm": 0.5392469763755798,
      "learning_rate": 3e-06,
      "loss": 0.6706,
      "step": 399,
      "step_time": 7.359844096004963
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.75869140625,
      "epoch": 0.19157088122605365,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07117857784032822,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 400,
      "step_time": 7.26955758780241
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 218.236328125,
      "completions/mean_terminated_length": 218.236328125,
      "completions/min_length": 13.0,
      "completions/min_terminated_length": 13.0,
      "entropy": 1.4213285914489202,
      "epoch": 0.19204980842911878,
      "fork/depth_mean": 0.21679092943668365,
      "fork/depth_std": 0.19149020314216614,
      "fork_head/loss": 2.7311887741088867,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.08557455241680145,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6850104331970215,
      "learning_rate": 3e-06,
      "loss": 0.6832,
      "num_tokens": 10944791.0,
      "reward": 0.9868971705436707,
      "reward_std": 0.5802099108695984,
      "rewards/code_format_reward/mean": 0.7744140625,
      "rewards/code_format_reward/std": 0.3653131127357483,
      "rewards/code_reward/mean": 0.21248313784599304,
      "rewards/code_reward/std": 0.34193649888038635,
      "step": 401,
      "step_time": 486.52844855189323
    },
    {
      "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,
      "entropy": 1.6207217233521598,
      "epoch": 0.1925287356321839,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1968306303024292,
      "learning_rate": 3e-06,
      "loss": 0.0008,
      "step": 402,
      "step_time": 7.930353283882141
    },
    {
      "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,
      "entropy": 1.409249437706811,
      "epoch": 0.19300766283524903,
      "fork_head/loss": 2.719025135040283,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.0850558802485466,
      "grad_norm": 0.6434490084648132,
      "learning_rate": 3e-06,
      "loss": 0.6796,
      "step": 403,
      "step_time": 7.5386383682489395
    },
    {
      "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,
      "entropy": 1.6190011160714286,
      "epoch": 0.19348659003831417,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19246844947338104,
      "learning_rate": 3e-06,
      "loss": -0.0004,
      "step": 404,
      "step_time": 7.344245232641697
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05952381023338863,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05952381023338863,
      "entropy": 1.4210844423089708,
      "epoch": 0.1939655172413793,
      "fork_head/loss": 2.6980011463165283,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.09586390852928162,
      "grad_norm": 0.5789645314216614,
      "learning_rate": 3e-06,
      "loss": 0.6726,
      "step": 405,
      "step_time": 7.869456417858601
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.6366489955357142,
      "epoch": 0.19444444444444445,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13041599094867706,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 406,
      "step_time": 7.419455394148827
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.4161435110228402,
      "epoch": 0.19492337164750959,
      "fork_head/loss": 2.6731033325195312,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.11375922709703445,
      "grad_norm": 0.5144776701927185,
      "learning_rate": 3e-06,
      "loss": 0.6654,
      "step": 407,
      "step_time": 8.137549541890621
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.13095238166196005,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13095238166196005,
      "entropy": 1.6448800223214286,
      "epoch": 0.19540229885057472,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11221117526292801,
      "learning_rate": 3e-06,
      "loss": -0.0026,
      "step": 408,
      "step_time": 7.580831967294216
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.13671875,
      "completions/mean_terminated_length": 218.99217224121094,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.8037109375,
      "epoch": 0.19588122605363983,
      "fork/depth_mean": 0.221403107047081,
      "fork/depth_std": 0.19368097186088562,
      "fork_head/loss": 2.7190985679626465,
      "fork_head/num_positive": 10.580645561218262,
      "fork_head/pred_top1_mass": 0.057538244873285294,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7307291626930237,
      "learning_rate": 3e-06,
      "loss": 0.671,
      "num_tokens": 11152989.0,
      "reward": 0.9557915329933167,
      "reward_std": 0.614459753036499,
      "rewards/code_format_reward/mean": 0.7314453125,
      "rewards/code_format_reward/std": 0.37750786542892456,
      "rewards/code_reward/mean": 0.22434625029563904,
      "rewards/code_reward/std": 0.3562663495540619,
      "step": 409,
      "step_time": 507.9058329164982
    },
    {
      "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,
      "entropy": 1.6647786438465118,
      "epoch": 0.19636015325670497,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.269741415977478,
      "learning_rate": 3e-06,
      "loss": 0.0074,
      "step": 410,
      "step_time": 7.6510466784238815
    },
    {
      "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,
      "entropy": 1.7979600694444444,
      "epoch": 0.1968390804597701,
      "fork_head/loss": 2.70540189743042,
      "fork_head/num_positive": 10.580645561218262,
      "fork_head/pred_top1_mass": 0.07781857252120972,
      "grad_norm": 0.6249101161956787,
      "learning_rate": 3e-06,
      "loss": 0.6677,
      "step": 411,
      "step_time": 7.577318273484707
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.6345052063465118,
      "epoch": 0.19731800766283525,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1804642677307129,
      "learning_rate": 3e-06,
      "loss": 0.0073,
      "step": 412,
      "step_time": 7.788366004824638
    },
    {
      "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,
      "entropy": 1.8077256944444444,
      "epoch": 0.1977969348659004,
      "fork_head/loss": 2.686223030090332,
      "fork_head/num_positive": 10.580645561218262,
      "fork_head/pred_top1_mass": 0.11145155876874924,
      "grad_norm": 0.5420905351638794,
      "learning_rate": 3e-06,
      "loss": 0.6629,
      "step": 413,
      "step_time": 7.53588230907917
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.6863932311534882,
      "epoch": 0.19827586206896552,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15033195912837982,
      "learning_rate": 3e-06,
      "loss": 0.0072,
      "step": 414,
      "step_time": 7.7888331189751625
    },
    {
      "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,
      "entropy": 1.8021918402777777,
      "epoch": 0.19875478927203066,
      "fork_head/loss": 2.6646225452423096,
      "fork_head/num_positive": 10.580645561218262,
      "fork_head/pred_top1_mass": 0.11994908004999161,
      "grad_norm": 0.5222252607345581,
      "learning_rate": 3e-06,
      "loss": 0.6569,
      "step": 415,
      "step_time": 7.48451741039753
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.6296875,
      "epoch": 0.19923371647509577,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18141140043735504,
      "learning_rate": 3e-06,
      "loss": 0.0063,
      "step": 416,
      "step_time": 7.316910855472088
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 229.85546875,
      "completions/mean_terminated_length": 229.75296020507812,
      "completions/min_length": 50.0,
      "completions/min_terminated_length": 50.0,
      "entropy": 1.7128634999195735,
      "epoch": 0.1997126436781609,
      "fork/depth_mean": 0.22732074558734894,
      "fork/depth_std": 0.21676751971244812,
      "fork_head/loss": 2.7485361099243164,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.07258612662553787,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8212964534759521,
      "learning_rate": 3e-06,
      "loss": 0.6848,
      "num_tokens": 11369843.0,
      "reward": 0.7772555351257324,
      "reward_std": 0.5737707614898682,
      "rewards/code_format_reward/mean": 0.640625,
      "rewards/code_format_reward/std": 0.43315389752388,
      "rewards/code_reward/mean": 0.136630579829216,
      "rewards/code_reward/std": 0.29768267273902893,
      "step": 417,
      "step_time": 475.83855947107077
    },
    {
      "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,
      "entropy": 1.3649576902389526,
      "epoch": 0.20019157088122605,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12664219737052917,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 418,
      "step_time": 7.558632120490074
    },
    {
      "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,
      "entropy": 1.71034071346124,
      "epoch": 0.2006704980842912,
      "fork_head/loss": 2.735710620880127,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.0714874416589737,
      "grad_norm": 0.7346537709236145,
      "learning_rate": 3e-06,
      "loss": 0.6814,
      "step": 419,
      "step_time": 7.48791791498661
    },
    {
      "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,
      "entropy": 1.3613606691360474,
      "epoch": 0.20114942528735633,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11243920028209686,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 420,
      "step_time": 7.537673212587833
    },
    {
      "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,
      "entropy": 1.7052408854166667,
      "epoch": 0.20162835249042146,
      "fork_head/loss": 2.7133216857910156,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.07545682787895203,
      "grad_norm": 0.5650261044502258,
      "learning_rate": 3e-06,
      "loss": 0.676,
      "step": 421,
      "step_time": 7.688019506633282
    },
    {
      "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,
      "entropy": 1.3557454347610474,
      "epoch": 0.2021072796934866,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12633274495601654,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 422,
      "step_time": 7.498547732830048
    },
    {
      "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,
      "entropy": 1.69688584903876,
      "epoch": 0.2025862068965517,
      "fork_head/loss": 2.69024920463562,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.1009012758731842,
      "grad_norm": 0.4450347125530243,
      "learning_rate": 3e-06,
      "loss": 0.6697,
      "step": 423,
      "step_time": 7.2101161032915115
    },
    {
      "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,
      "entropy": 1.3581705808639526,
      "epoch": 0.20306513409961685,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1256064921617508,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 424,
      "step_time": 7.285280928015709
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 223.20703125,
      "completions/mean_terminated_length": 223.14285278320312,
      "completions/min_length": 24.0,
      "completions/min_terminated_length": 24.0,
      "entropy": 1.0431315079331398,
      "epoch": 0.203544061302682,
      "fork/depth_mean": 0.1978253722190857,
      "fork/depth_std": 0.17083394527435303,
      "fork_head/loss": 2.7312071323394775,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07333047688007355,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7544835805892944,
      "learning_rate": 3e-06,
      "loss": 0.6759,
      "num_tokens": 11579805.0,
      "reward": 0.9203928709030151,
      "reward_std": 0.6547951698303223,
      "rewards/code_format_reward/mean": 0.701171875,
      "rewards/code_format_reward/std": 0.4226449429988861,
      "rewards/code_reward/mean": 0.2192210704088211,
      "rewards/code_reward/std": 0.3697650134563446,
      "step": 425,
      "step_time": 506.967924810946
    },
    {
      "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,
      "entropy": 1.457275390625,
      "epoch": 0.20402298850574713,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03641282767057419,
      "learning_rate": 3e-06,
      "loss": 0.0072,
      "step": 426,
      "step_time": 7.934421695768833
    },
    {
      "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,
      "entropy": 1.0524088516831398,
      "epoch": 0.20450191570881227,
      "fork_head/loss": 2.7208871841430664,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07036708295345306,
      "grad_norm": 0.7489346861839294,
      "learning_rate": 3e-06,
      "loss": 0.6734,
      "step": 427,
      "step_time": 7.4398568123579025
    },
    {
      "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,
      "entropy": 1.4677734375,
      "epoch": 0.2049808429118774,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0368666835129261,
      "learning_rate": 3e-06,
      "loss": 0.0071,
      "step": 428,
      "step_time": 7.304845839738846
    },
    {
      "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,
      "entropy": 1.0313313826918602,
      "epoch": 0.20545977011494254,
      "fork_head/loss": 2.69983172416687,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.10184267163276672,
      "grad_norm": 0.6144914031028748,
      "learning_rate": 3e-06,
      "loss": 0.6673,
      "step": 429,
      "step_time": 7.885627306997776
    },
    {
      "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,
      "entropy": 1.473388671875,
      "epoch": 0.20593869731800765,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03969532996416092,
      "learning_rate": 3e-06,
      "loss": 0.0069,
      "step": 430,
      "step_time": 7.573764875531197
    },
    {
      "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,
      "entropy": 1.0403645858168602,
      "epoch": 0.2064176245210728,
      "fork_head/loss": 2.675870418548584,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.10782823711633682,
      "grad_norm": 0.5442111492156982,
      "learning_rate": 3e-06,
      "loss": 0.6609,
      "step": 431,
      "step_time": 7.484921209514141
    },
    {
      "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,
      "entropy": 1.46923828125,
      "epoch": 0.20689655172413793,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0396454855799675,
      "learning_rate": 3e-06,
      "loss": 0.0066,
      "step": 432,
      "step_time": 7.677101299166679
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.939453125,
      "completions/mean_terminated_length": 219.7980499267578,
      "completions/min_length": 48.0,
      "completions/min_terminated_length": 48.0,
      "entropy": 1.5791903409090908,
      "epoch": 0.20737547892720307,
      "fork/depth_mean": 0.22032450139522552,
      "fork/depth_std": 0.19145099818706512,
      "fork_head/loss": 2.743943214416504,
      "fork_head/num_positive": 12.59375,
      "fork_head/pred_top1_mass": 0.07802894711494446,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8661714196205139,
      "learning_rate": 3e-06,
      "loss": 0.6902,
      "num_tokens": 11785390.0,
      "reward": 0.8559596538543701,
      "reward_std": 0.6829107999801636,
      "rewards/code_format_reward/mean": 0.6435546875,
      "rewards/code_format_reward/std": 0.4383716881275177,
      "rewards/code_reward/mean": 0.2124050110578537,
      "rewards/code_reward/std": 0.3499707877635956,
      "step": 433,
      "step_time": 508.47207295149565
    },
    {
      "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,
      "entropy": 1.468017578125,
      "epoch": 0.2078544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10448956489562988,
      "learning_rate": 3e-06,
      "loss": -0.0056,
      "step": 434,
      "step_time": 7.569570191204548
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.6081321022727273,
      "epoch": 0.20833333333333334,
      "fork_head/loss": 2.729660749435425,
      "fork_head/num_positive": 12.59375,
      "fork_head/pred_top1_mass": 0.0757344663143158,
      "grad_norm": 0.5405313968658447,
      "learning_rate": 3e-06,
      "loss": 0.686,
      "step": 435,
      "step_time": 7.920401498675346
    },
    {
      "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,
      "entropy": 1.45599365234375,
      "epoch": 0.20881226053639848,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07673072069883347,
      "learning_rate": 3e-06,
      "loss": -0.0056,
      "step": 436,
      "step_time": 7.056364595890045
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.6560724431818181,
      "epoch": 0.2092911877394636,
      "fork_head/loss": 2.717726945877075,
      "fork_head/num_positive": 12.59375,
      "fork_head/pred_top1_mass": 0.07389146089553833,
      "grad_norm": 0.41838538646698,
      "learning_rate": 3e-06,
      "loss": 0.683,
      "step": 437,
      "step_time": 7.721019230782986
    },
    {
      "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,
      "entropy": 1.4708251953125,
      "epoch": 0.20977011494252873,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07011871039867401,
      "learning_rate": 3e-06,
      "loss": -0.0062,
      "step": 438,
      "step_time": 7.549565263092518
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.7143110795454546,
      "epoch": 0.21024904214559387,
      "fork_head/loss": 2.7078535556793213,
      "fork_head/num_positive": 12.59375,
      "fork_head/pred_top1_mass": 0.08609329164028168,
      "grad_norm": 0.4951418936252594,
      "learning_rate": 3e-06,
      "loss": 0.6802,
      "step": 439,
      "step_time": 7.3077511340379715
    },
    {
      "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,
      "entropy": 1.46441650390625,
      "epoch": 0.210727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06736687570810318,
      "learning_rate": 3e-06,
      "loss": -0.0064,
      "step": 440,
      "step_time": 7.9837101846933365
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 218.76953125,
      "completions/mean_terminated_length": 218.62353515625,
      "completions/min_length": 98.0,
      "completions/min_terminated_length": 98.0,
      "entropy": 1.7307477678571428,
      "epoch": 0.21120689655172414,
      "fork/depth_mean": 0.21820005774497986,
      "fork/depth_std": 0.19155558943748474,
      "fork_head/loss": 2.724839925765991,
      "fork_head/num_positive": 10.84375,
      "fork_head/pred_top1_mass": 0.06802421063184738,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6340774297714233,
      "learning_rate": 3e-06,
      "loss": 0.6757,
      "num_tokens": 11988360.0,
      "reward": 0.9282667636871338,
      "reward_std": 0.6926395297050476,
      "rewards/code_format_reward/mean": 0.654296875,
      "rewards/code_format_reward/std": 0.4318769574165344,
      "rewards/code_reward/mean": 0.2739698886871338,
      "rewards/code_reward/std": 0.4169972836971283,
      "step": 441,
      "step_time": 458.3893976509571
    },
    {
      "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,
      "entropy": 1.68671875,
      "epoch": 0.21168582375478928,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08276502043008804,
      "learning_rate": 3e-06,
      "loss": 0.0051,
      "step": 442,
      "step_time": 7.794153906404972
    },
    {
      "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,
      "entropy": 1.7052176339285714,
      "epoch": 0.21216475095785442,
      "fork_head/loss": 2.7125754356384277,
      "fork_head/num_positive": 10.84375,
      "fork_head/pred_top1_mass": 0.07191567122936249,
      "grad_norm": 0.6005361676216125,
      "learning_rate": 3e-06,
      "loss": 0.672,
      "step": 443,
      "step_time": 7.072033114731312
    },
    {
      "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,
      "entropy": 1.70390625,
      "epoch": 0.21264367816091953,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06047524884343147,
      "learning_rate": 3e-06,
      "loss": 0.0049,
      "step": 444,
      "step_time": 7.927282959222794
    },
    {
      "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,
      "entropy": 1.7021484375,
      "epoch": 0.21312260536398467,
      "fork_head/loss": 2.6907877922058105,
      "fork_head/num_positive": 10.84375,
      "fork_head/pred_top1_mass": 0.10029703378677368,
      "grad_norm": 0.5545859336853027,
      "learning_rate": 3e-06,
      "loss": 0.6662,
      "step": 445,
      "step_time": 7.038097865879536
    },
    {
      "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,
      "entropy": 1.70859375,
      "epoch": 0.2136015325670498,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0639958307147026,
      "learning_rate": 3e-06,
      "loss": 0.0049,
      "step": 446,
      "step_time": 6.996253289282322
    },
    {
      "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,
      "entropy": 1.7073102678571428,
      "epoch": 0.21408045977011494,
      "fork_head/loss": 2.6659934520721436,
      "fork_head/num_positive": 10.84375,
      "fork_head/pred_top1_mass": 0.11183373630046844,
      "grad_norm": 0.510964572429657,
      "learning_rate": 3e-06,
      "loss": 0.6594,
      "step": 447,
      "step_time": 7.9876499474048615
    },
    {
      "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,
      "entropy": 1.72109375,
      "epoch": 0.21455938697318008,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05898105353116989,
      "learning_rate": 3e-06,
      "loss": 0.0047,
      "step": 448,
      "step_time": 6.988285504281521
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 233.19140625,
      "completions/mean_terminated_length": 233.1467742919922,
      "completions/min_length": 66.0,
      "completions/min_terminated_length": 66.0,
      "entropy": 1.7516741071428572,
      "epoch": 0.21503831417624522,
      "fork/depth_mean": 0.20314298570156097,
      "fork/depth_std": 0.1760406494140625,
      "fork_head/loss": 2.7327919006347656,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07539404183626175,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4988114833831787,
      "learning_rate": 3e-06,
      "loss": 0.6847,
      "num_tokens": 12205674.0,
      "reward": 0.8338286280632019,
      "reward_std": 0.5917543172836304,
      "rewards/code_format_reward/mean": 0.6435546875,
      "rewards/code_format_reward/std": 0.42706555128097534,
      "rewards/code_reward/mean": 0.1902739256620407,
      "rewards/code_reward/std": 0.35319164395332336,
      "step": 449,
      "step_time": 525.2545355707407
    },
    {
      "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,
      "entropy": 1.336187903697674,
      "epoch": 0.21551724137931033,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12179934978485107,
      "learning_rate": 3e-06,
      "loss": -0.0031,
      "step": 450,
      "step_time": 7.287766367197037
    },
    {
      "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,
      "entropy": 1.7336774553571428,
      "epoch": 0.21599616858237547,
      "fork_head/loss": 2.727095603942871,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07944023609161377,
      "grad_norm": 0.4910355806350708,
      "learning_rate": 3e-06,
      "loss": 0.6831,
      "step": 451,
      "step_time": 8.827800840139389
    },
    {
      "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,
      "entropy": 1.3180088171592126,
      "epoch": 0.2164750957854406,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11276113986968994,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 452,
      "step_time": 7.044648453593254
    },
    {
      "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,
      "entropy": 1.7566964285714286,
      "epoch": 0.21695402298850575,
      "fork_head/loss": 2.7160143852233887,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.08377410471439362,
      "grad_norm": 0.45412537455558777,
      "learning_rate": 3e-06,
      "loss": 0.6801,
      "step": 453,
      "step_time": 7.340071223676205
    },
    {
      "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,
      "entropy": 1.3113481539946337,
      "epoch": 0.21743295019157088,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11583419144153595,
      "learning_rate": 3e-06,
      "loss": -0.0035,
      "step": 454,
      "step_time": 8.083771251142025
    },
    {
      "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,
      "entropy": 1.7596261160714286,
      "epoch": 0.21791187739463602,
      "fork_head/loss": 2.7018165588378906,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.08323933184146881,
      "grad_norm": 0.41121289134025574,
      "learning_rate": 3e-06,
      "loss": 0.6758,
      "step": 455,
      "step_time": 7.047412559390068
    },
    {
      "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,
      "entropy": 1.3076923076923077,
      "epoch": 0.21839080459770116,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11721662431955338,
      "learning_rate": 3e-06,
      "loss": -0.0048,
      "step": 456,
      "step_time": 7.5007089748978615
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 217.625,
      "completions/mean_terminated_length": 217.54989624023438,
      "completions/min_length": 51.0,
      "completions/min_terminated_length": 51.0,
      "entropy": 1.8935275574525197,
      "epoch": 0.21886973180076627,
      "fork/depth_mean": 0.18884053826332092,
      "fork/depth_std": 0.17753426730632782,
      "fork_head/loss": 2.744840621948242,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.06135697662830353,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.49683621525764465,
      "learning_rate": 3e-06,
      "loss": 0.682,
      "num_tokens": 12403002.0,
      "reward": 0.9766608476638794,
      "reward_std": 0.6194941401481628,
      "rewards/code_format_reward/mean": 0.720703125,
      "rewards/code_format_reward/std": 0.40316882729530334,
      "rewards/code_reward/mean": 0.2559577524662018,
      "rewards/code_reward/std": 0.3793812394142151,
      "step": 457,
      "step_time": 525.1238992214203
    },
    {
      "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,
      "entropy": 1.5457356691360473,
      "epoch": 0.2193486590038314,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12510329484939575,
      "learning_rate": 3e-06,
      "loss": 0.0037,
      "step": 458,
      "step_time": 7.645316906273365
    },
    {
      "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,
      "entropy": 1.8917914529641469,
      "epoch": 0.21982758620689655,
      "fork_head/loss": 2.738121509552002,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.06710817664861679,
      "grad_norm": 0.472100168466568,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "step": 459,
      "step_time": 7.618349693715572
    },
    {
      "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,
      "entropy": 1.5379720091819764,
      "epoch": 0.22030651340996169,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16268949210643768,
      "learning_rate": 3e-06,
      "loss": 0.0032,
      "step": 460,
      "step_time": 7.8872488513588905
    },
    {
      "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,
      "entropy": 1.9150933126608531,
      "epoch": 0.22078544061302682,
      "fork_head/loss": 2.727113723754883,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.06886545568704605,
      "grad_norm": 0.4675130844116211,
      "learning_rate": 3e-06,
      "loss": 0.6768,
      "step": 461,
      "step_time": 7.111623659729958
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.5290364623069763,
      "epoch": 0.22126436781609196,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.029502147808670998,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 462,
      "step_time": 7.669439189136028
    },
    {
      "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,
      "entropy": 1.9073621928691864,
      "epoch": 0.2217432950191571,
      "fork_head/loss": 2.713550090789795,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.07664624601602554,
      "grad_norm": 0.4286182224750519,
      "learning_rate": 3e-06,
      "loss": 0.6732,
      "step": 463,
      "step_time": 8.01371992379427
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.5193196654319763,
      "epoch": 0.2222222222222222,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.029946204274892807,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 464,
      "step_time": 7.430034354329109
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 228.333984375,
      "completions/mean_terminated_length": 228.333984375,
      "completions/min_length": 79.0,
      "completions/min_terminated_length": 79.0,
      "entropy": 1.5408709512816534,
      "epoch": 0.22270114942528735,
      "fork/depth_mean": 0.18569150567054749,
      "fork/depth_std": 0.17038945853710175,
      "fork_head/loss": 2.7230687141418457,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.07381927222013474,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.508620023727417,
      "learning_rate": 3e-06,
      "loss": 0.6891,
      "num_tokens": 12619525.0,
      "reward": 0.8355199694633484,
      "reward_std": 0.6311467885971069,
      "rewards/code_format_reward/mean": 0.646484375,
      "rewards/code_format_reward/std": 0.44018691778182983,
      "rewards/code_reward/mean": 0.1890355944633484,
      "rewards/code_reward/std": 0.3433837592601776,
      "step": 465,
      "step_time": 542.3775353878736
    },
    {
      "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,
      "entropy": 2.082430752840909,
      "epoch": 0.2231800766283525,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16779355704784393,
      "learning_rate": 3e-06,
      "loss": -0.0134,
      "step": 466,
      "step_time": 7.714014567434788
    },
    {
      "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,
      "entropy": 1.5534939236111112,
      "epoch": 0.22365900383141762,
      "fork_head/loss": 2.714327573776245,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.07388342171907425,
      "grad_norm": 0.4801587462425232,
      "learning_rate": 3e-06,
      "loss": 0.6863,
      "step": 467,
      "step_time": 7.997204661369324
    },
    {
      "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,
      "entropy": 2.086159446022727,
      "epoch": 0.22413793103448276,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11829755455255508,
      "learning_rate": 3e-06,
      "loss": -0.014,
      "step": 468,
      "step_time": 7.494575336575508
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.5858289930555556,
      "epoch": 0.2246168582375479,
      "fork_head/loss": 2.700683832168579,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.07641252875328064,
      "grad_norm": 0.4731031060218811,
      "learning_rate": 3e-06,
      "loss": 0.682,
      "step": 469,
      "step_time": 7.44773331284523
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 2.0366654829545454,
      "epoch": 0.22509578544061304,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09733033925294876,
      "learning_rate": 3e-06,
      "loss": -0.0151,
      "step": 470,
      "step_time": 8.145076356828213
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.5804398126072354,
      "epoch": 0.22557471264367815,
      "fork_head/loss": 2.683286190032959,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.07889707386493683,
      "grad_norm": 0.43979427218437195,
      "learning_rate": 3e-06,
      "loss": 0.6778,
      "step": 471,
      "step_time": 7.418669246137142
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.18181818181818182,
      "entropy": 2.0442560369318183,
      "epoch": 0.2260536398467433,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09337176382541656,
      "learning_rate": 3e-06,
      "loss": -0.0153,
      "step": 472,
      "step_time": 7.774471543729305
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0078125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 227.369140625,
      "completions/mean_terminated_length": 227.14370727539062,
      "completions/min_length": 42.0,
      "completions/min_terminated_length": 42.0,
      "entropy": 1.4333433554722712,
      "epoch": 0.22653256704980843,
      "fork/depth_mean": 0.19809512794017792,
      "fork/depth_std": 0.16936635971069336,
      "fork_head/loss": 2.7166013717651367,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.0760771781206131,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.586919367313385,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "num_tokens": 12831602.0,
      "reward": 0.91984623670578,
      "reward_std": 0.6765034198760986,
      "rewards/code_format_reward/mean": 0.6806640625,
      "rewards/code_format_reward/std": 0.4138941466808319,
      "rewards/code_reward/mean": 0.23918214440345764,
      "rewards/code_reward/std": 0.3890198767185211,
      "step": 473,
      "step_time": 523.4231154099107
    },
    {
      "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,
      "entropy": 1.99072265625,
      "epoch": 0.22701149425287356,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20526711642742157,
      "learning_rate": 3e-06,
      "loss": 0.005,
      "step": 474,
      "step_time": 7.748822286725044
    },
    {
      "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,
      "entropy": 1.428961341197674,
      "epoch": 0.2274904214559387,
      "fork_head/loss": 2.708242416381836,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.07939338684082031,
      "grad_norm": 0.5475062727928162,
      "learning_rate": 3e-06,
      "loss": 0.6752,
      "step": 475,
      "step_time": 7.590849705040455
    },
    {
      "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,
      "entropy": 2.001220703125,
      "epoch": 0.22796934865900384,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.245212122797966,
      "learning_rate": 3e-06,
      "loss": 0.0032,
      "step": 476,
      "step_time": 7.934845432639122
    },
    {
      "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,
      "entropy": 1.4246294039946337,
      "epoch": 0.22844827586206898,
      "fork_head/loss": 2.6939749717712402,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.0930638238787651,
      "grad_norm": 0.4689394235610962,
      "learning_rate": 3e-06,
      "loss": 0.6714,
      "step": 477,
      "step_time": 7.254658855497837
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.25,
      "entropy": 2.0384114583333335,
      "epoch": 0.2289272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07241401076316833,
      "learning_rate": 3e-06,
      "loss": 0.0006,
      "step": 478,
      "step_time": 7.969769820570946
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.025641026405187752,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.025641026405187752,
      "entropy": 1.4310897405330951,
      "epoch": 0.22940613026819923,
      "fork_head/loss": 2.6779584884643555,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.09175387769937515,
      "grad_norm": 0.3991106450557709,
      "learning_rate": 3e-06,
      "loss": 0.6662,
      "step": 479,
      "step_time": 7.531509190797806
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.25,
      "entropy": 2.0563151041666665,
      "epoch": 0.22988505747126436,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07596747577190399,
      "learning_rate": 3e-06,
      "loss": 0.0006,
      "step": 480,
      "step_time": 7.16516151279211
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 223.900390625,
      "completions/mean_terminated_length": 223.77452087402344,
      "completions/min_length": 88.0,
      "completions/min_terminated_length": 88.0,
      "entropy": 1.54296875,
      "epoch": 0.2303639846743295,
      "fork/depth_mean": 0.22068437933921814,
      "fork/depth_std": 0.20080235600471497,
      "fork_head/loss": 2.739276885986328,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.057747066020965576,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6043835282325745,
      "learning_rate": 3e-06,
      "loss": 0.6852,
      "num_tokens": 13041295.0,
      "reward": 0.8645498752593994,
      "reward_std": 0.6230220198631287,
      "rewards/code_format_reward/mean": 0.6787109375,
      "rewards/code_format_reward/std": 0.4003373086452484,
      "rewards/code_reward/mean": 0.1858389526605606,
      "rewards/code_reward/std": 0.34449058771133423,
      "step": 481,
      "step_time": 502.48632980138063
    },
    {
      "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,
      "entropy": 2.2193080357142856,
      "epoch": 0.23084291187739464,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.059459902346134186,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 482,
      "step_time": 7.909461043775082
    },
    {
      "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,
      "entropy": 1.5347222222222223,
      "epoch": 0.23132183908045978,
      "fork_head/loss": 2.7294015884399414,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.058631639927625656,
      "grad_norm": 0.5076257586479187,
      "learning_rate": 3e-06,
      "loss": 0.6826,
      "step": 483,
      "step_time": 7.971340261399746
    },
    {
      "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,
      "entropy": 2.229771205357143,
      "epoch": 0.23180076628352492,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05812695622444153,
      "learning_rate": 3e-06,
      "loss": 0.0006,
      "step": 484,
      "step_time": 7.053235739469528
    },
    {
      "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,
      "entropy": 1.5564236111111112,
      "epoch": 0.23227969348659003,
      "fork_head/loss": 2.7148995399475098,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.0821918249130249,
      "grad_norm": 0.4405119717121124,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "step": 485,
      "step_time": 7.789397172629833
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 2.2352120535714284,
      "epoch": 0.23275862068965517,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05581849440932274,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 486,
      "step_time": 7.54198557138443
    },
    {
      "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,
      "entropy": 1.5407986111111112,
      "epoch": 0.2332375478927203,
      "fork_head/loss": 2.6979947090148926,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.0846923217177391,
      "grad_norm": 0.4446457028388977,
      "learning_rate": 3e-06,
      "loss": 0.6745,
      "step": 487,
      "step_time": 7.338490128517151
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 2.2426060267857144,
      "epoch": 0.23371647509578544,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.061108075082302094,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 488,
      "step_time": 7.901680365204811
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 215.65625,
      "completions/mean_terminated_length": 215.57730102539062,
      "completions/min_length": 61.0,
      "completions/min_terminated_length": 61.0,
      "entropy": 1.8760653300718828,
      "epoch": 0.23419540229885058,
      "fork/depth_mean": 0.22604762017726898,
      "fork/depth_std": 0.20985688269138336,
      "fork_head/loss": 2.737182140350342,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.051120318472385406,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7701677680015564,
      "learning_rate": 3e-06,
      "loss": 0.686,
      "num_tokens": 13246415.0,
      "reward": 0.8342366218566895,
      "reward_std": 0.5967313051223755,
      "rewards/code_format_reward/mean": 0.666015625,
      "rewards/code_format_reward/std": 0.40153491497039795,
      "rewards/code_reward/mean": 0.16822099685668945,
      "rewards/code_reward/std": 0.31438905000686646,
      "step": 489,
      "step_time": 534.2943422719836
    },
    {
      "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,
      "entropy": 1.3727213541666667,
      "epoch": 0.23467432950191572,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16557268798351288,
      "learning_rate": 3e-06,
      "loss": -0.0032,
      "step": 490,
      "step_time": 7.7145813927054405
    },
    {
      "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,
      "entropy": 1.8851207494735718,
      "epoch": 0.23515325670498086,
      "fork_head/loss": 2.724857807159424,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.05608823895454407,
      "grad_norm": 0.5276307463645935,
      "learning_rate": 3e-06,
      "loss": 0.6828,
      "step": 491,
      "step_time": 7.274155490100384
    },
    {
      "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,
      "entropy": 1.3811848958333333,
      "epoch": 0.23563218390804597,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15287582576274872,
      "learning_rate": 3e-06,
      "loss": -0.0031,
      "step": 492,
      "step_time": 8.63173707574606
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.8827237324281172,
      "epoch": 0.2361111111111111,
      "fork_head/loss": 2.71225643157959,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.05687762051820755,
      "grad_norm": 0.4373615086078644,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "step": 493,
      "step_time": 7.3158760741353035
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05555555555555555,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.3967013888888888,
      "epoch": 0.23659003831417624,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08023694902658463,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 494,
      "step_time": 7.524633973836899
    },
    {
      "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,
      "entropy": 1.8806522326035933,
      "epoch": 0.23706896551724138,
      "fork_head/loss": 2.7000370025634766,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.0712122693657875,
      "grad_norm": 0.4666943848133087,
      "learning_rate": 3e-06,
      "loss": 0.6752,
      "step": 495,
      "step_time": 7.858401633799076
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05555555555555555,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.3971354166666667,
      "epoch": 0.23754789272030652,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07718116044998169,
      "learning_rate": 3e-06,
      "loss": -0.0055,
      "step": 496,
      "step_time": 6.919654689729214
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 232.0234375,
      "completions/mean_terminated_length": 231.92942810058594,
      "completions/min_length": 78.0,
      "completions/min_terminated_length": 78.0,
      "entropy": 1.8937322443181819,
      "epoch": 0.23802681992337166,
      "fork/depth_mean": 0.3106261193752289,
      "fork/depth_std": 0.23912958800792694,
      "fork_head/loss": 2.722501516342163,
      "fork_head/num_positive": 10.6875,
      "fork_head/pred_top1_mass": 0.05420929566025734,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.666443943977356,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "num_tokens": 13460443.0,
      "reward": 1.0360456705093384,
      "reward_std": 0.590480625629425,
      "rewards/code_format_reward/mean": 0.7861328125,
      "rewards/code_format_reward/std": 0.31070762872695923,
      "rewards/code_reward/mean": 0.24991290271282196,
      "rewards/code_reward/std": 0.3808183968067169,
      "step": 497,
      "step_time": 428.2988211661577
    },
    {
      "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,
      "entropy": 1.545735674245017,
      "epoch": 0.23850574712643677,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1433788239955902,
      "learning_rate": 3e-06,
      "loss": 0.0006,
      "step": 498,
      "step_time": 7.7402573972940445
    },
    {
      "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,
      "entropy": 1.9193892045454546,
      "epoch": 0.2389846743295019,
      "fork_head/loss": 2.7095894813537598,
      "fork_head/num_positive": 10.6875,
      "fork_head/pred_top1_mass": 0.06716816127300262,
      "grad_norm": 0.6226699352264404,
      "learning_rate": 3e-06,
      "loss": 0.6746,
      "step": 499,
      "step_time": 7.863756112754345
    },
    {
      "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,
      "entropy": 1.5516183035714286,
      "epoch": 0.23946360153256704,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1857476830482483,
      "learning_rate": 3e-06,
      "loss": -0.0007,
      "step": 500,
      "step_time": 7.562820732593536
    },
    {
      "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,
      "entropy": 1.9001242897727273,
      "epoch": 0.23994252873563218,
      "fork_head/loss": 2.688950777053833,
      "fork_head/num_positive": 10.6875,
      "fork_head/pred_top1_mass": 0.106669582426548,
      "grad_norm": 0.573951005935669,
      "learning_rate": 3e-06,
      "loss": 0.6698,
      "step": 501,
      "step_time": 7.381594829261303
    },
    {
      "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,
      "entropy": 1.5386439732142858,
      "epoch": 0.24042145593869732,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12371935695409775,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 502,
      "step_time": 8.081864766776562
    },
    {
      "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,
      "entropy": 1.9128196022727273,
      "epoch": 0.24090038314176246,
      "fork_head/loss": 2.662698745727539,
      "fork_head/num_positive": 10.6875,
      "fork_head/pred_top1_mass": 0.1106816753745079,
      "grad_norm": 0.5288758277893066,
      "learning_rate": 3e-06,
      "loss": 0.6628,
      "step": 503,
      "step_time": 7.1914307326078415
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05952381023338863,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05952381023338863,
      "entropy": 1.5431470104626246,
      "epoch": 0.2413793103448276,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09664784371852875,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 504,
      "step_time": 8.11844266206026
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.203125,
      "completions/mean_terminated_length": 219.058837890625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.9982346754807692,
      "epoch": 0.2418582375478927,
      "fork/depth_mean": 0.30262380838394165,
      "fork/depth_std": 0.2393195778131485,
      "fork_head/loss": 2.7154154777526855,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.06809810549020767,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5871378779411316,
      "learning_rate": 3e-06,
      "loss": 0.6739,
      "num_tokens": 13660755.0,
      "reward": 0.9704515933990479,
      "reward_std": 0.5366705656051636,
      "rewards/code_format_reward/mean": 0.76953125,
      "rewards/code_format_reward/std": 0.331941694021225,
      "rewards/code_reward/mean": 0.20092038810253143,
      "rewards/code_reward/std": 0.360167533159256,
      "step": 505,
      "step_time": 445.1597457230091
    },
    {
      "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,
      "entropy": 1.4049479166666667,
      "epoch": 0.24233716475095785,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16616207361221313,
      "learning_rate": 3e-06,
      "loss": 0.0072,
      "step": 506,
      "step_time": 8.115596659481525
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 2.020294978068425,
      "epoch": 0.24281609195402298,
      "fork_head/loss": 2.7055325508117676,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.07935597747564316,
      "grad_norm": 0.5388314127922058,
      "learning_rate": 3e-06,
      "loss": 0.6712,
      "step": 507,
      "step_time": 7.541972905397415
    },
    {
      "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,
      "entropy": 1.4221643606821697,
      "epoch": 0.24329501915708812,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1607627272605896,
      "learning_rate": 3e-06,
      "loss": 0.0055,
      "step": 508,
      "step_time": 7.560894161462784
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.11538461538461539,
      "entropy": 2.0087640285491943,
      "epoch": 0.24377394636015326,
      "fork_head/loss": 2.6904752254486084,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08014081418514252,
      "grad_norm": 0.4856332838535309,
      "learning_rate": 3e-06,
      "loss": 0.6652,
      "step": 509,
      "step_time": 7.631102658808231
    },
    {
      "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,
      "entropy": 1.415653944015503,
      "epoch": 0.2442528735632184,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1625078022480011,
      "learning_rate": 3e-06,
      "loss": 0.0047,
      "step": 510,
      "step_time": 7.427282840013504
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.11538461538461539,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.11538461538461539,
      "entropy": 2.0039438100961537,
      "epoch": 0.24473180076628354,
      "fork_head/loss": 2.6720101833343506,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08893817663192749,
      "grad_norm": 0.4518047571182251,
      "learning_rate": 3e-06,
      "loss": 0.6598,
      "step": 511,
      "step_time": 8.011291600763798
    },
    {
      "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,
      "entropy": 1.4255642361111112,
      "epoch": 0.24521072796934865,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1483761966228485,
      "learning_rate": 3e-06,
      "loss": 0.0036,
      "step": 512,
      "step_time": 6.996781289577484
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.017578125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 225.359375,
      "completions/mean_terminated_length": 224.81112670898438,
      "completions/min_length": 65.0,
      "completions/min_terminated_length": 65.0,
      "entropy": 1.4951171875,
      "epoch": 0.24568965517241378,
      "fork/depth_mean": 0.28064337372779846,
      "fork/depth_std": 0.22832654416561127,
      "fork_head/loss": 2.7205753326416016,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.06493460386991501,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6162589192390442,
      "learning_rate": 3e-06,
      "loss": 0.6785,
      "num_tokens": 13871291.0,
      "reward": 0.9579378962516785,
      "reward_std": 0.621293306350708,
      "rewards/code_format_reward/mean": 0.7314453125,
      "rewards/code_format_reward/std": 0.3813759386539459,
      "rewards/code_reward/mean": 0.22649258375167847,
      "rewards/code_reward/std": 0.3597337007522583,
      "step": 513,
      "step_time": 502.24782808870077
    },
    {
      "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,
      "entropy": 1.4782986111111112,
      "epoch": 0.24616858237547892,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07208183407783508,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 514,
      "step_time": 6.837856277823448
    },
    {
      "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,
      "entropy": 1.4954427083333333,
      "epoch": 0.24664750957854406,
      "fork_head/loss": 2.705958366394043,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.06524641811847687,
      "grad_norm": 0.5879212617874146,
      "learning_rate": 3e-06,
      "loss": 0.6747,
      "step": 515,
      "step_time": 7.969695255160332
    },
    {
      "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,
      "entropy": 1.4737413194444444,
      "epoch": 0.2471264367816092,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07506668567657471,
      "learning_rate": 3e-06,
      "loss": 0.0002,
      "step": 516,
      "step_time": 7.499054983258247
    },
    {
      "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,
      "entropy": 1.5110677083333333,
      "epoch": 0.24760536398467434,
      "fork_head/loss": 2.684135913848877,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.106992207467556,
      "grad_norm": 0.4927445650100708,
      "learning_rate": 3e-06,
      "loss": 0.6692,
      "step": 517,
      "step_time": 7.285921066999435
    },
    {
      "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,
      "entropy": 1.44921875,
      "epoch": 0.24808429118773948,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0677558109164238,
      "learning_rate": 3e-06,
      "loss": -0.0001,
      "step": 518,
      "step_time": 8.22396495193243
    },
    {
      "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,
      "entropy": 1.5045572916666667,
      "epoch": 0.24856321839080459,
      "fork_head/loss": 2.6630778312683105,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.1042696163058281,
      "grad_norm": 0.43468207120895386,
      "learning_rate": 3e-06,
      "loss": 0.6639,
      "step": 519,
      "step_time": 6.9037302657961845
    },
    {
      "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,
      "entropy": 1.4357638888888888,
      "epoch": 0.24904214559386972,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0710841566324234,
      "learning_rate": 3e-06,
      "loss": -0.0002,
      "step": 520,
      "step_time": 7.5371701791882515
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 209.25390625,
      "completions/mean_terminated_length": 209.25390625,
      "completions/min_length": 83.0,
      "completions/min_terminated_length": 83.0,
      "entropy": 1.5815873579545454,
      "epoch": 0.24952107279693486,
      "fork/depth_mean": 0.25390687584877014,
      "fork/depth_std": 0.21391654014587402,
      "fork_head/loss": 2.743408203125,
      "fork_head/num_positive": 12.71875,
      "fork_head/pred_top1_mass": 0.07545756548643112,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6935438513755798,
      "learning_rate": 3e-06,
      "loss": 0.6815,
      "num_tokens": 14067757.0,
      "reward": 0.9104833006858826,
      "reward_std": 0.6139981150627136,
      "rewards/code_format_reward/mean": 0.705078125,
      "rewards/code_format_reward/std": 0.3968239724636078,
      "rewards/code_reward/mean": 0.20540517568588257,
      "rewards/code_reward/std": 0.3559519350528717,
      "step": 521,
      "step_time": 477.3065570294857
    },
    {
      "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,
      "entropy": 1.9936671473763206,
      "epoch": 0.25,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08482667058706284,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 522,
      "step_time": 8.39413870871067
    },
    {
      "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,
      "entropy": 1.5930397727272727,
      "epoch": 0.25047892720306514,
      "fork_head/loss": 2.7381606101989746,
      "fork_head/num_positive": 12.71875,
      "fork_head/pred_top1_mass": 0.07441359758377075,
      "grad_norm": 0.6072656512260437,
      "learning_rate": 3e-06,
      "loss": 0.6795,
      "step": 523,
      "step_time": 7.279388301074505
    },
    {
      "clip_ratio/high_max": 0.045454545454545456,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.9812381701035933,
      "epoch": 0.2509578544061303,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08207912743091583,
      "learning_rate": 3e-06,
      "loss": 0.0002,
      "step": 524,
      "step_time": 7.3226741552352905
    },
    {
      "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,
      "entropy": 1.5912642045454546,
      "epoch": 0.2514367816091954,
      "fork_head/loss": 2.7241358757019043,
      "fork_head/num_positive": 12.71875,
      "fork_head/pred_top1_mass": 0.0797693133354187,
      "grad_norm": 0.4755915105342865,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "step": 525,
      "step_time": 7.89828397333622
    },
    {
      "clip_ratio/high_max": 0.045454545454545456,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.9853515625,
      "epoch": 0.25191570881226055,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08390266448259354,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 526,
      "step_time": 7.186623349785805
    },
    {
      "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,
      "entropy": 1.5890447443181819,
      "epoch": 0.2523946360153257,
      "fork_head/loss": 2.708657741546631,
      "fork_head/num_positive": 12.71875,
      "fork_head/pred_top1_mass": 0.08401107788085938,
      "grad_norm": 0.44038519263267517,
      "learning_rate": 3e-06,
      "loss": 0.6702,
      "step": 527,
      "step_time": 7.386976130306721
    },
    {
      "clip_ratio/high_max": 0.045454545454545456,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.9790187098763206,
      "epoch": 0.25287356321839083,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0860094353556633,
      "learning_rate": 3e-06,
      "loss": -0.0003,
      "step": 528,
      "step_time": 8.115377426147461
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.07142857142857142,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 224.599609375,
      "completions/mean_terminated_length": 224.599609375,
      "completions/min_length": 93.0,
      "completions/min_terminated_length": 93.0,
      "entropy": 1.7213774153164454,
      "epoch": 0.2533524904214559,
      "fork/depth_mean": 0.24099315702915192,
      "fork/depth_std": 0.1871427744626999,
      "fork_head/loss": 2.723708391189575,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.06331875175237656,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.43400830030441284,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "num_tokens": 14270464.0,
      "reward": 1.010643482208252,
      "reward_std": 0.6248570084571838,
      "rewards/code_format_reward/mean": 0.763671875,
      "rewards/code_format_reward/std": 0.35224828124046326,
      "rewards/code_reward/mean": 0.2469717264175415,
      "rewards/code_reward/std": 0.3903096616268158,
      "step": 529,
      "step_time": 514.3924599438906
    },
    {
      "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,
      "entropy": 1.5936631917953492,
      "epoch": 0.25383141762452105,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12162766605615616,
      "learning_rate": 3e-06,
      "loss": 0.0014,
      "step": 530,
      "step_time": 7.958508417010307
    },
    {
      "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,
      "entropy": 1.7042643172400338,
      "epoch": 0.2543103448275862,
      "fork_head/loss": 2.718430280685425,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.0677669569849968,
      "grad_norm": 0.41673797369003296,
      "learning_rate": 3e-06,
      "loss": 0.6773,
      "step": 531,
      "step_time": 8.050729297101498
    },
    {
      "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,
      "entropy": 1.6027018229166667,
      "epoch": 0.2547892720306513,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14608050882816315,
      "learning_rate": 3e-06,
      "loss": 0.0012,
      "step": 532,
      "step_time": 7.318672388792038
    },
    {
      "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,
      "entropy": 1.720749625137874,
      "epoch": 0.25526819923371646,
      "fork_head/loss": 2.7102489471435547,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.06946807354688644,
      "grad_norm": 0.3853345513343811,
      "learning_rate": 3e-06,
      "loss": 0.6744,
      "step": 533,
      "step_time": 7.540452800691128
    },
    {
      "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,
      "entropy": 1.6132052977879843,
      "epoch": 0.2557471264367816,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11227232217788696,
      "learning_rate": 3e-06,
      "loss": 0.0008,
      "step": 534,
      "step_time": 7.6669433787465096
    },
    {
      "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,
      "entropy": 1.7319103394235884,
      "epoch": 0.25622605363984674,
      "fork_head/loss": 2.698131561279297,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.06803533434867859,
      "grad_norm": 0.3570137917995453,
      "learning_rate": 3e-06,
      "loss": 0.6701,
      "step": 535,
      "step_time": 7.4410795867443085
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.03333333333333333,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.6259982665379842,
      "epoch": 0.2567049808429119,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10485237836837769,
      "learning_rate": 3e-06,
      "loss": -0.0005,
      "step": 536,
      "step_time": 7.3624334037303925
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 221.287109375,
      "completions/mean_terminated_length": 221.1510009765625,
      "completions/min_length": 56.0,
      "completions/min_terminated_length": 56.0,
      "entropy": 1.4800180288461537,
      "epoch": 0.257183908045977,
      "fork/depth_mean": 0.3151340186595917,
      "fork/depth_std": 0.24626950919628143,
      "fork_head/loss": 2.7203924655914307,
      "fork_head/num_positive": 11.4375,
      "fork_head/pred_top1_mass": 0.06649376451969147,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5934690237045288,
      "learning_rate": 3e-06,
      "loss": 0.6773,
      "num_tokens": 14477731.0,
      "reward": 0.9491225481033325,
      "reward_std": 0.593697726726532,
      "rewards/code_format_reward/mean": 0.736328125,
      "rewards/code_format_reward/std": 0.36986351013183594,
      "rewards/code_reward/mean": 0.2127944529056549,
      "rewards/code_reward/std": 0.35137495398521423,
      "step": 537,
      "step_time": 412.9940000548959
    },
    {
      "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,
      "entropy": 1.4171424278846154,
      "epoch": 0.25766283524904215,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06323738396167755,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 538,
      "step_time": 8.026048131287098
    },
    {
      "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,
      "entropy": 1.4573818078407874,
      "epoch": 0.2581417624521073,
      "fork_head/loss": 2.70914363861084,
      "fork_head/num_positive": 11.4375,
      "fork_head/pred_top1_mass": 0.07268311828374863,
      "grad_norm": 0.5511836409568787,
      "learning_rate": 3e-06,
      "loss": 0.6736,
      "step": 539,
      "step_time": 7.601077035069466
    },
    {
      "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,
      "entropy": 1.403345355620751,
      "epoch": 0.25862068965517243,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0591726079583168,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 540,
      "step_time": 7.638015978038311
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.4566306059177105,
      "epoch": 0.25909961685823757,
      "fork_head/loss": 2.690561294555664,
      "fork_head/num_positive": 11.4375,
      "fork_head/pred_top1_mass": 0.07424134761095047,
      "grad_norm": 0.47509825229644775,
      "learning_rate": 3e-06,
      "loss": 0.6681,
      "step": 541,
      "step_time": 7.65003664791584
    },
    {
      "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,
      "entropy": 1.4189453125,
      "epoch": 0.2595785440613027,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.058000680059194565,
      "learning_rate": 3e-06,
      "loss": 0.002,
      "step": 542,
      "step_time": 7.4370725974440575
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.4486678655330951,
      "epoch": 0.2600574712643678,
      "fork_head/loss": 2.6708831787109375,
      "fork_head/num_positive": 11.4375,
      "fork_head/pred_top1_mass": 0.08417456597089767,
      "grad_norm": 0.41057366132736206,
      "learning_rate": 3e-06,
      "loss": 0.663,
      "step": 543,
      "step_time": 7.643835335969925
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.386969153697674,
      "epoch": 0.26053639846743293,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05560031533241272,
      "learning_rate": 3e-06,
      "loss": 0.0014,
      "step": 544,
      "step_time": 7.617923945188522
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 221.267578125,
      "completions/mean_terminated_length": 221.06288146972656,
      "completions/min_length": 56.0,
      "completions/min_terminated_length": 56.0,
      "entropy": 1.7206488847732544,
      "epoch": 0.26101532567049807,
      "fork/depth_mean": 0.22402195632457733,
      "fork/depth_std": 0.18422053754329681,
      "fork_head/loss": 2.7221896648406982,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.0734180361032486,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4337543547153473,
      "learning_rate": 3e-06,
      "loss": 0.6851,
      "num_tokens": 14684988.0,
      "reward": 0.9430587291717529,
      "reward_std": 0.6468262672424316,
      "rewards/code_format_reward/mean": 0.69921875,
      "rewards/code_format_reward/std": 0.4046686589717865,
      "rewards/code_reward/mean": 0.24383996427059174,
      "rewards/code_reward/std": 0.3768720030784607,
      "step": 545,
      "step_time": 475.7034825384617
    },
    {
      "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,
      "entropy": 1.7252459565798441,
      "epoch": 0.2614942528735632,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17355720698833466,
      "learning_rate": 3e-06,
      "loss": -0.0076,
      "step": 546,
      "step_time": 7.209214337170124
    },
    {
      "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,
      "entropy": 1.7237820078929265,
      "epoch": 0.26197318007662834,
      "fork_head/loss": 2.7179770469665527,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.07081353664398193,
      "grad_norm": 0.4544186294078827,
      "learning_rate": 3e-06,
      "loss": 0.6836,
      "step": 547,
      "step_time": 7.245548278093338
    },
    {
      "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,
      "entropy": 1.7149667223294576,
      "epoch": 0.2624521072796935,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18970191478729248,
      "learning_rate": 3e-06,
      "loss": -0.008,
      "step": 548,
      "step_time": 8.142438925802708
    },
    {
      "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,
      "entropy": 1.7375623732805252,
      "epoch": 0.2629310344827586,
      "fork_head/loss": 2.7099609375,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.077761210501194,
      "grad_norm": 0.4735352694988251,
      "learning_rate": 3e-06,
      "loss": 0.68,
      "step": 549,
      "step_time": 6.880097143352032
    },
    {
      "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,
      "entropy": 1.715878172715505,
      "epoch": 0.26340996168582376,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21870557963848114,
      "learning_rate": 3e-06,
      "loss": -0.0102,
      "step": 550,
      "step_time": 8.05680327117443
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06944444527228673,
      "clip_ratio/low_min": 0.041666666666666664,
      "clip_ratio/region_mean": 0.06944444527228673,
      "entropy": 1.7457818041245143,
      "epoch": 0.2638888888888889,
      "fork_head/loss": 2.6984755992889404,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.08217956125736237,
      "grad_norm": 0.40873241424560547,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "step": 551,
      "step_time": 7.622295409440994
    },
    {
      "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,
      "entropy": 1.7373119155565897,
      "epoch": 0.26436781609195403,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17018096148967743,
      "learning_rate": 3e-06,
      "loss": -0.0124,
      "step": 552,
      "step_time": 7.647312164306641
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 227.068359375,
      "completions/mean_terminated_length": 227.01173400878906,
      "completions/min_length": 98.0,
      "completions/min_terminated_length": 98.0,
      "entropy": 1.8525390625,
      "epoch": 0.26484674329501917,
      "fork/depth_mean": 0.23475702106952667,
      "fork/depth_std": 0.19307276606559753,
      "fork_head/loss": 2.7351796627044678,
      "fork_head/num_positive": 11.300000190734863,
      "fork_head/pred_top1_mass": 0.06612087786197662,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6947343945503235,
      "learning_rate": 3e-06,
      "loss": 0.6951,
      "num_tokens": 14892783.0,
      "reward": 0.8667208552360535,
      "reward_std": 0.6451014876365662,
      "rewards/code_format_reward/mean": 0.662109375,
      "rewards/code_format_reward/std": 0.4209413528442383,
      "rewards/code_reward/mean": 0.2046114206314087,
      "rewards/code_reward/std": 0.34639665484428406,
      "step": 553,
      "step_time": 510.45434333384037
    },
    {
      "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,
      "entropy": 1.2580450177192688,
      "epoch": 0.2653256704980843,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08113007247447968,
      "learning_rate": 3e-06,
      "loss": -0.011,
      "step": 554,
      "step_time": 8.403685733675957
    },
    {
      "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,
      "entropy": 1.8673270089285714,
      "epoch": 0.26580459770114945,
      "fork_head/loss": 2.7213335037231445,
      "fork_head/num_positive": 11.300000190734863,
      "fork_head/pred_top1_mass": 0.07429276406764984,
      "grad_norm": 0.5665345788002014,
      "learning_rate": 3e-06,
      "loss": 0.6903,
      "step": 555,
      "step_time": 8.031419858336449
    },
    {
      "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,
      "entropy": 1.260091142995017,
      "epoch": 0.2662835249042146,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08262232691049576,
      "learning_rate": 3e-06,
      "loss": -0.0109,
      "step": 556,
      "step_time": 7.607694692909718
    },
    {
      "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,
      "entropy": 1.8602120535714286,
      "epoch": 0.26676245210727967,
      "fork_head/loss": 2.7036550045013428,
      "fork_head/num_positive": 11.300000190734863,
      "fork_head/pred_top1_mass": 0.07602262496948242,
      "grad_norm": 0.5237346291542053,
      "learning_rate": 3e-06,
      "loss": 0.6855,
      "step": 557,
      "step_time": 7.523875705897808
    },
    {
      "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,
      "entropy": 1.2627883212906974,
      "epoch": 0.2672413793103448,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0781203955411911,
      "learning_rate": 3e-06,
      "loss": -0.0112,
      "step": 558,
      "step_time": 7.458724178373814
    },
    {
      "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,
      "entropy": 1.8823939732142858,
      "epoch": 0.26772030651340994,
      "fork_head/loss": 2.6841776371002197,
      "fork_head/num_positive": 11.300000190734863,
      "fork_head/pred_top1_mass": 0.08810552954673767,
      "grad_norm": 0.5058808326721191,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 559,
      "step_time": 7.35491294413805
    },
    {
      "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,
      "entropy": 1.2424200177192688,
      "epoch": 0.2681992337164751,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07560888677835464,
      "learning_rate": 3e-06,
      "loss": -0.0119,
      "step": 560,
      "step_time": 7.853423468768597
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 225.236328125,
      "completions/mean_terminated_length": 225.11569213867188,
      "completions/min_length": 77.0,
      "completions/min_terminated_length": 77.0,
      "entropy": 1.5312906901041667,
      "epoch": 0.2686781609195402,
      "fork/depth_mean": 0.21581554412841797,
      "fork/depth_std": 0.1770457923412323,
      "fork_head/loss": 2.720167875289917,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.06122070550918579,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5329287052154541,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "num_tokens": 15096792.0,
      "reward": 0.8790266513824463,
      "reward_std": 0.6189442276954651,
      "rewards/code_format_reward/mean": 0.6962890625,
      "rewards/code_format_reward/std": 0.39324766397476196,
      "rewards/code_reward/mean": 0.1827375292778015,
      "rewards/code_reward/std": 0.34424257278442383,
      "step": 561,
      "step_time": 515.9964241310954
    },
    {
      "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,
      "entropy": 1.9197265625,
      "epoch": 0.26915708812260536,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18288378417491913,
      "learning_rate": 3e-06,
      "loss": 0.0012,
      "step": 562,
      "step_time": 7.744786985218525
    },
    {
      "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,
      "entropy": 1.5466715494791667,
      "epoch": 0.2696360153256705,
      "fork_head/loss": 2.710299253463745,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.07941070199012756,
      "grad_norm": 0.4710041582584381,
      "learning_rate": 3e-06,
      "loss": 0.6742,
      "step": 563,
      "step_time": 8.143693692982197
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.9491861939430237,
      "epoch": 0.27011494252873564,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12945808470249176,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 564,
      "step_time": 7.509607560932636
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5633544921875,
      "epoch": 0.2705938697318008,
      "fork_head/loss": 2.6978344917297363,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.08915610611438751,
      "grad_norm": 0.4063486158847809,
      "learning_rate": 3e-06,
      "loss": 0.6685,
      "step": 565,
      "step_time": 7.175612576305866
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.03333333432674408,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333432674408,
      "entropy": 1.9204427123069763,
      "epoch": 0.2710727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1139797642827034,
      "learning_rate": 3e-06,
      "loss": -0.0003,
      "step": 566,
      "step_time": 8.12374659627676
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.020833333333333332,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.10416666666666667,
      "entropy": 1.5831298828125,
      "epoch": 0.27155172413793105,
      "fork_head/loss": 2.683760404586792,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.08515390753746033,
      "grad_norm": 0.3772883415222168,
      "learning_rate": 3e-06,
      "loss": 0.6655,
      "step": 567,
      "step_time": 7.496451154351234
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.03333333432674408,
      "clip_ratio/low_mean": 0.15,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1833333343267441,
      "entropy": 1.9927083373069763,
      "epoch": 0.2720306513409962,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06650248169898987,
      "learning_rate": 3e-06,
      "loss": -0.0013,
      "step": 568,
      "step_time": 8.040253795683384
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 210.74609375,
      "completions/mean_terminated_length": 210.74609375,
      "completions/min_length": 61.0,
      "completions/min_terminated_length": 61.0,
      "entropy": 1.724609375,
      "epoch": 0.2725095785440613,
      "fork/depth_mean": 0.18617990612983704,
      "fork/depth_std": 0.16442398726940155,
      "fork_head/loss": 2.731126070022583,
      "fork_head/num_positive": 10.77419376373291,
      "fork_head/pred_top1_mass": 0.0775202140212059,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5816218852996826,
      "learning_rate": 3e-06,
      "loss": 0.6906,
      "num_tokens": 15287542.0,
      "reward": 0.8210683465003967,
      "reward_std": 0.5960471630096436,
      "rewards/code_format_reward/mean": 0.6396484375,
      "rewards/code_format_reward/std": 0.40968015789985657,
      "rewards/code_reward/mean": 0.18141992390155792,
      "rewards/code_reward/std": 0.3335700035095215,
      "step": 569,
      "step_time": 521.3318581804633
    },
    {
      "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,
      "entropy": 1.5440538227558136,
      "epoch": 0.27298850574712646,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07936476916074753,
      "learning_rate": 3e-06,
      "loss": -0.0098,
      "step": 570,
      "step_time": 8.965047985315323
    },
    {
      "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,
      "entropy": 1.7137784090909092,
      "epoch": 0.27346743295019155,
      "fork_head/loss": 2.7197604179382324,
      "fork_head/num_positive": 10.77419376373291,
      "fork_head/pred_top1_mass": 0.07628713548183441,
      "grad_norm": 0.5266129970550537,
      "learning_rate": 3e-06,
      "loss": 0.6882,
      "step": 571,
      "step_time": 7.414936192333698
    },
    {
      "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,
      "entropy": 1.546875,
      "epoch": 0.2739463601532567,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08023258298635483,
      "learning_rate": 3e-06,
      "loss": -0.0105,
      "step": 572,
      "step_time": 7.110700488090515
    },
    {
      "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,
      "entropy": 1.7116477272727273,
      "epoch": 0.2744252873563218,
      "fork_head/loss": 2.7020785808563232,
      "fork_head/num_positive": 10.77419376373291,
      "fork_head/pred_top1_mass": 0.0799580067396164,
      "grad_norm": 0.503560483455658,
      "learning_rate": 3e-06,
      "loss": 0.6823,
      "step": 573,
      "step_time": 7.747186467051506
    },
    {
      "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,
      "entropy": 1.5449761251608531,
      "epoch": 0.27490421455938696,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08251477777957916,
      "learning_rate": 3e-06,
      "loss": -0.0104,
      "step": 574,
      "step_time": 7.055122934281826
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.7327769886363635,
      "epoch": 0.2753831417624521,
      "fork_head/loss": 2.679321527481079,
      "fork_head/num_positive": 10.77419376373291,
      "fork_head/pred_top1_mass": 0.11872972548007965,
      "grad_norm": 0.5094435214996338,
      "learning_rate": 3e-06,
      "loss": 0.6757,
      "step": 575,
      "step_time": 7.789535351097584
    },
    {
      "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,
      "entropy": 1.5477430522441864,
      "epoch": 0.27586206896551724,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07665704935789108,
      "learning_rate": 3e-06,
      "loss": -0.0112,
      "step": 576,
      "step_time": 7.531896412372589
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 212.10546875,
      "completions/mean_terminated_length": 211.84677124023438,
      "completions/min_length": 39.0,
      "completions/min_terminated_length": 39.0,
      "entropy": 2.037471055984497,
      "epoch": 0.2763409961685824,
      "fork/depth_mean": 0.18820075690746307,
      "fork/depth_std": 0.16953063011169434,
      "fork_head/loss": 2.721686840057373,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.0718887448310852,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6714509129524231,
      "learning_rate": 3e-06,
      "loss": 0.6757,
      "num_tokens": 15492204.0,
      "reward": 0.8651357293128967,
      "reward_std": 0.6973443031311035,
      "rewards/code_format_reward/mean": 0.634765625,
      "rewards/code_format_reward/std": 0.45048609375953674,
      "rewards/code_reward/mean": 0.23037010431289673,
      "rewards/code_reward/std": 0.3771602213382721,
      "step": 577,
      "step_time": 521.1457083076239
    },
    {
      "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,
      "entropy": 1.4398082386363635,
      "epoch": 0.2768199233716475,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0752640888094902,
      "learning_rate": 3e-06,
      "loss": 0.0056,
      "step": 578,
      "step_time": 7.7540847808122635
    },
    {
      "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,
      "entropy": 2.0252459579043918,
      "epoch": 0.27729885057471265,
      "fork_head/loss": 2.709702253341675,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.0733831375837326,
      "grad_norm": 0.5284579992294312,
      "learning_rate": 3e-06,
      "loss": 0.6727,
      "step": 579,
      "step_time": 7.430787973105907
    },
    {
      "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,
      "entropy": 1.4450905539772727,
      "epoch": 0.2777777777777778,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07450630515813828,
      "learning_rate": 3e-06,
      "loss": 0.0055,
      "step": 580,
      "step_time": 7.645877055823803
    },
    {
      "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,
      "entropy": 1.9933449162377253,
      "epoch": 0.27825670498084293,
      "fork_head/loss": 2.6931145191192627,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.0814858004450798,
      "grad_norm": 0.4656166136264801,
      "learning_rate": 3e-06,
      "loss": 0.6683,
      "step": 581,
      "step_time": 7.011630974709988
    },
    {
      "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,
      "entropy": 1.47314453125,
      "epoch": 0.27873563218390807,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07267221063375473,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 582,
      "step_time": 7.8594492599368095
    },
    {
      "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,
      "entropy": 1.9829282495710585,
      "epoch": 0.2792145593869732,
      "fork_head/loss": 2.6759896278381348,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.0979909598827362,
      "grad_norm": 0.4703916311264038,
      "learning_rate": 3e-06,
      "loss": 0.6635,
      "step": 583,
      "step_time": 7.761670842766762
    },
    {
      "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,
      "entropy": 1.4668856534090908,
      "epoch": 0.2796934865900383,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07694123685359955,
      "learning_rate": 3e-06,
      "loss": 0.0047,
      "step": 584,
      "step_time": 7.483692370355129
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.025641026405187752,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.025641026405187752,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 226.267578125,
      "completions/mean_terminated_length": 226.267578125,
      "completions/min_length": 35.0,
      "completions/min_terminated_length": 35.0,
      "entropy": 1.623948321892665,
      "epoch": 0.2801724137931034,
      "fork/depth_mean": 0.18445022404193878,
      "fork/depth_std": 0.157170370221138,
      "fork_head/loss": 2.7403247356414795,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.06901146471500397,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6365469098091125,
      "learning_rate": 3e-06,
      "loss": 0.6874,
      "num_tokens": 15698917.0,
      "reward": 0.8845361471176147,
      "reward_std": 0.5280846953392029,
      "rewards/code_format_reward/mean": 0.751953125,
      "rewards/code_format_reward/std": 0.3714309632778168,
      "rewards/code_reward/mean": 0.13258299231529236,
      "rewards/code_reward/std": 0.2748987674713135,
      "step": 585,
      "step_time": 540.9984596818686
    },
    {
      "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,
      "entropy": 1.6799859007199605,
      "epoch": 0.28065134099616856,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09843004494905472,
      "learning_rate": 3e-06,
      "loss": -0.001,
      "step": 586,
      "step_time": 8.028708331286907
    },
    {
      "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,
      "entropy": 1.62464943757424,
      "epoch": 0.2811302681992337,
      "fork_head/loss": 2.7286319732666016,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.06970737874507904,
      "grad_norm": 0.5802482962608337,
      "learning_rate": 3e-06,
      "loss": 0.6836,
      "step": 587,
      "step_time": 8.235529839992523
    },
    {
      "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,
      "entropy": 1.6593966980775197,
      "epoch": 0.28160919540229884,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0919867604970932,
      "learning_rate": 3e-06,
      "loss": -0.0007,
      "step": 588,
      "step_time": 7.594922088086605
    },
    {
      "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,
      "entropy": 1.631735778771914,
      "epoch": 0.282088122605364,
      "fork_head/loss": 2.709601402282715,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.07088892161846161,
      "grad_norm": 0.5262558460235596,
      "learning_rate": 3e-06,
      "loss": 0.6785,
      "step": 589,
      "step_time": 7.360910303890705
    },
    {
      "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,
      "entropy": 1.6754828592141469,
      "epoch": 0.2825670498084291,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08582975715398788,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 590,
      "step_time": 7.679121121764183
    },
    {
      "clip_ratio/high_max": 0.15384615384615385,
      "clip_ratio/high_mean": 0.06410256486672622,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06410256486672622,
      "entropy": 1.6136067693050091,
      "epoch": 0.28304597701149425,
      "fork_head/loss": 2.6873624324798584,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.10218571871519089,
      "grad_norm": 0.4604213833808899,
      "learning_rate": 3e-06,
      "loss": 0.6714,
      "step": 591,
      "step_time": 7.986861199140549
    },
    {
      "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,
      "entropy": 1.657090922196706,
      "epoch": 0.2835249042145594,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09002962708473206,
      "learning_rate": 3e-06,
      "loss": -0.0015,
      "step": 592,
      "step_time": 7.391556419432163
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 215.73046875,
      "completions/mean_terminated_length": 215.5725555419922,
      "completions/min_length": 16.0,
      "completions/min_terminated_length": 16.0,
      "entropy": 1.823828125,
      "epoch": 0.28400383141762453,
      "fork/depth_mean": 0.22839586436748505,
      "fork/depth_std": 0.20960934460163116,
      "fork_head/loss": 2.7262701988220215,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.07312370836734772,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6288830041885376,
      "learning_rate": 3e-06,
      "loss": 0.6767,
      "num_tokens": 15891003.0,
      "reward": 0.9384843111038208,
      "reward_std": 0.6319583654403687,
      "rewards/code_format_reward/mean": 0.7119140625,
      "rewards/code_format_reward/std": 0.3963071405887604,
      "rewards/code_reward/mean": 0.22657017409801483,
      "rewards/code_reward/std": 0.3678654134273529,
      "step": 593,
      "step_time": 463.2421717196703
    },
    {
      "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,
      "entropy": 1.3302734375,
      "epoch": 0.28448275862068967,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1239134818315506,
      "learning_rate": 3e-06,
      "loss": 0.0063,
      "step": 594,
      "step_time": 7.630407825112343
    },
    {
      "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,
      "entropy": 1.79931640625,
      "epoch": 0.2849616858237548,
      "fork_head/loss": 2.7128186225891113,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.07911041378974915,
      "grad_norm": 0.5869691967964172,
      "learning_rate": 3e-06,
      "loss": 0.6726,
      "step": 595,
      "step_time": 8.347485706210136
    },
    {
      "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,
      "entropy": 1.325,
      "epoch": 0.28544061302681994,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10973557084798813,
      "learning_rate": 3e-06,
      "loss": 0.0063,
      "step": 596,
      "step_time": 7.293516226112843
    },
    {
      "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,
      "entropy": 1.8001953125,
      "epoch": 0.2859195402298851,
      "fork_head/loss": 2.6907496452331543,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.08048442006111145,
      "grad_norm": 0.5184904932975769,
      "learning_rate": 3e-06,
      "loss": 0.6661,
      "step": 597,
      "step_time": 7.681838996708393
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.31494140625,
      "epoch": 0.28639846743295017,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07883136719465256,
      "learning_rate": 3e-06,
      "loss": 0.0056,
      "step": 598,
      "step_time": 7.362156890332699
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.814990234375,
      "epoch": 0.2868773946360153,
      "fork_head/loss": 2.666875123977661,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.0795058161020279,
      "grad_norm": 0.4603860676288605,
      "learning_rate": 3e-06,
      "loss": 0.6588,
      "step": 599,
      "step_time": 7.496118627488613
    },
    {
      "clip_ratio/high_max": 0.2,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.29990234375,
      "epoch": 0.28735632183908044,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07440328598022461,
      "learning_rate": 3e-06,
      "loss": 0.0042,
      "step": 600,
      "step_time": 7.792205795645714
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01171875,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 222.376953125,
      "completions/mean_terminated_length": 221.978271484375,
      "completions/min_length": 65.0,
      "completions/min_terminated_length": 65.0,
      "entropy": 1.8092122395833334,
      "epoch": 0.2878352490421456,
      "fork/depth_mean": 0.2080942541360855,
      "fork/depth_std": 0.2184683233499527,
      "fork_head/loss": 2.748006820678711,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06128740683197975,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7098738551139832,
      "learning_rate": 3e-06,
      "loss": 0.6844,
      "num_tokens": 16094204.0,
      "reward": 0.8016788959503174,
      "reward_std": 0.6349448561668396,
      "rewards/code_format_reward/mean": 0.6220703125,
      "rewards/code_format_reward/std": 0.435958594083786,
      "rewards/code_reward/mean": 0.17960861325263977,
      "rewards/code_reward/std": 0.3345843553543091,
      "step": 601,
      "step_time": 436.58914121985435
    },
    {
      "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,
      "entropy": 1.9296991314206804,
      "epoch": 0.2883141762452107,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15188314020633698,
      "learning_rate": 3e-06,
      "loss": 0.0069,
      "step": 602,
      "step_time": 8.55135378986597
    },
    {
      "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,
      "entropy": 1.8310980955759684,
      "epoch": 0.28879310344827586,
      "fork_head/loss": 2.736492156982422,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.061254847794771194,
      "grad_norm": 0.6193290948867798,
      "learning_rate": 3e-06,
      "loss": 0.6807,
      "step": 603,
      "step_time": 7.72374464571476
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.9395228709493364,
      "epoch": 0.289272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1151939183473587,
      "learning_rate": 3e-06,
      "loss": 0.0058,
      "step": 604,
      "step_time": 7.548311702907085
    },
    {
      "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,
      "entropy": 1.8285590330759685,
      "epoch": 0.28975095785440613,
      "fork_head/loss": 2.7199511528015137,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06686128675937653,
      "grad_norm": 0.5020304322242737,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "step": 605,
      "step_time": 8.076370432972908
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.9404180645942688,
      "epoch": 0.29022988505747127,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11707878112792969,
      "learning_rate": 3e-06,
      "loss": 0.006,
      "step": 606,
      "step_time": 8.115542933344841
    },
    {
      "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,
      "entropy": 1.8345594565073648,
      "epoch": 0.2907088122605364,
      "fork_head/loss": 2.7032291889190674,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06811530888080597,
      "grad_norm": 0.4512539207935333,
      "learning_rate": 3e-06,
      "loss": 0.6699,
      "step": 607,
      "step_time": 8.010513983666897
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.9505905849593026,
      "epoch": 0.29118773946360155,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1095135435461998,
      "learning_rate": 3e-06,
      "loss": 0.0047,
      "step": 608,
      "step_time": 7.5395612716674805
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 220.21484375,
      "completions/mean_terminated_length": 220.21484375,
      "completions/min_length": 74.0,
      "completions/min_terminated_length": 74.0,
      "entropy": 1.8912402391433716,
      "epoch": 0.2916666666666667,
      "fork/depth_mean": 0.18964122235774994,
      "fork/depth_std": 0.18570935726165771,
      "fork_head/loss": 2.732189893722534,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06855854392051697,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6678314805030823,
      "learning_rate": 3e-06,
      "loss": 0.6801,
      "num_tokens": 16296570.0,
      "reward": 0.9010780453681946,
      "reward_std": 0.6420041918754578,
      "rewards/code_format_reward/mean": 0.6982421875,
      "rewards/code_format_reward/std": 0.41997209191322327,
      "rewards/code_reward/mean": 0.20283585786819458,
      "rewards/code_reward/std": 0.3483645021915436,
      "step": 609,
      "step_time": 539.7851459756494
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.012820513202593876,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.012820513202593876,
      "entropy": 1.8072165525876558,
      "epoch": 0.2921455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12974148988723755,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 610,
      "step_time": 7.769888818264008
    },
    {
      "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,
      "entropy": 1.894752597808838,
      "epoch": 0.29262452107279696,
      "fork_head/loss": 2.7223715782165527,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06930200010538101,
      "grad_norm": 0.6558368802070618,
      "learning_rate": 3e-06,
      "loss": 0.6776,
      "step": 611,
      "step_time": 7.563888058066368
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.012820513202593876,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.012820513202593876,
      "entropy": 1.809770629956172,
      "epoch": 0.29310344827586204,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13005003333091736,
      "learning_rate": 3e-06,
      "loss": -0.0004,
      "step": 612,
      "step_time": 7.987238705158234
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.9106477856636048,
      "epoch": 0.2935823754789272,
      "fork_head/loss": 2.702803373336792,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.0820528194308281,
      "grad_norm": 0.5501610636711121,
      "learning_rate": 3e-06,
      "loss": 0.6711,
      "step": 613,
      "step_time": 7.27111853659153
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05128205166413234,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05128205166413234,
      "entropy": 1.8121744761100183,
      "epoch": 0.2940613026819923,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12154275178909302,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 614,
      "step_time": 8.189748018980026
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.020000000298023225,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.12000000029802323,
      "entropy": 1.9103287696838378,
      "epoch": 0.29454022988505746,
      "fork_head/loss": 2.6823065280914307,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.10408564656972885,
      "grad_norm": 0.4105882942676544,
      "learning_rate": 3e-06,
      "loss": 0.6651,
      "step": 615,
      "step_time": 7.562656119465828
    },
    {
      "clip_ratio/high_max": 0.038461538461538464,
      "clip_ratio/high_mean": 0.009615384615384616,
      "clip_ratio/low_mean": 0.0897435901256708,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.09935897474105541,
      "entropy": 1.8349609375,
      "epoch": 0.2950191570881226,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09387866407632828,
      "learning_rate": 3e-06,
      "loss": -0.0023,
      "step": 616,
      "step_time": 7.760994374752045
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 214.455078125,
      "completions/mean_terminated_length": 214.455078125,
      "completions/min_length": 42.0,
      "completions/min_terminated_length": 42.0,
      "entropy": 1.504150390625,
      "epoch": 0.29549808429118773,
      "fork/depth_mean": 0.23933938145637512,
      "fork/depth_std": 0.2441420704126358,
      "fork_head/loss": 2.7436931133270264,
      "fork_head/num_positive": 12.34375,
      "fork_head/pred_top1_mass": 0.07106556743383408,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7921316623687744,
      "learning_rate": 3e-06,
      "loss": 0.6897,
      "num_tokens": 16501187.0,
      "reward": 0.8316623568534851,
      "reward_std": 0.6038548350334167,
      "rewards/code_format_reward/mean": 0.650390625,
      "rewards/code_format_reward/std": 0.4264262914657593,
      "rewards/code_reward/mean": 0.1812717318534851,
      "rewards/code_reward/std": 0.3006955087184906,
      "step": 617,
      "step_time": 492.92415917664766
    },
    {
      "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,
      "entropy": 1.6594050480769231,
      "epoch": 0.2959770114942529,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13676893711090088,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 618,
      "step_time": 8.345895260572433
    },
    {
      "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,
      "entropy": 1.4964701384305954,
      "epoch": 0.296455938697318,
      "fork_head/loss": 2.735042095184326,
      "fork_head/num_positive": 12.34375,
      "fork_head/pred_top1_mass": 0.06972265988588333,
      "grad_norm": 0.7497224807739258,
      "learning_rate": 3e-06,
      "loss": 0.6848,
      "step": 619,
      "step_time": 7.879541821777821
    },
    {
      "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,
      "entropy": 1.6650766225961537,
      "epoch": 0.29693486590038315,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1935858279466629,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 620,
      "step_time": 7.39503276348114
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0416666679084301,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0416666679084301,
      "entropy": 1.4866689071059227,
      "epoch": 0.2974137931034483,
      "fork_head/loss": 2.7168922424316406,
      "fork_head/num_positive": 12.34375,
      "fork_head/pred_top1_mass": 0.07594449073076248,
      "grad_norm": 0.5336112380027771,
      "learning_rate": 3e-06,
      "loss": 0.6777,
      "step": 621,
      "step_time": 8.020542927086353
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1346153846153846,
      "clip_ratio/low_min": 0.11538461538461539,
      "clip_ratio/region_mean": 0.1346153846153846,
      "entropy": 1.6974909855769231,
      "epoch": 0.2978927203065134,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11668190360069275,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 622,
      "step_time": 7.555262863636017
    },
    {
      "clip_ratio/high_max": 0.15625,
      "clip_ratio/high_mean": 0.0494791679084301,
      "clip_ratio/low_mean": 0.0416666679084301,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09114583395421505,
      "entropy": 1.4755350723862648,
      "epoch": 0.29837164750957856,
      "fork_head/loss": 2.6991500854492188,
      "fork_head/num_positive": 12.34375,
      "fork_head/pred_top1_mass": 0.07968703657388687,
      "grad_norm": 0.4292620122432709,
      "learning_rate": 3e-06,
      "loss": 0.6722,
      "step": 623,
      "step_time": 8.10886064171791
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.1346153846153846,
      "clip_ratio/low_min": 0.11538461538461539,
      "clip_ratio/region_mean": 0.21153846153846154,
      "entropy": 1.7154071514423077,
      "epoch": 0.2988505747126437,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0966154932975769,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 624,
      "step_time": 7.359820418059826
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 223.798828125,
      "completions/mean_terminated_length": 223.60903930664062,
      "completions/min_length": 20.0,
      "completions/min_terminated_length": 20.0,
      "entropy": 1.5155924303191048,
      "epoch": 0.29932950191570884,
      "fork/depth_mean": 0.21236059069633484,
      "fork/depth_std": 0.1914059817790985,
      "fork_head/loss": 2.7185277938842773,
      "fork_head/num_positive": 11.9375,
      "fork_head/pred_top1_mass": 0.08187419176101685,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6173702478408813,
      "learning_rate": 3e-06,
      "loss": 0.6766,
      "num_tokens": 16710300.0,
      "reward": 0.8876798152923584,
      "reward_std": 0.5945818424224854,
      "rewards/code_format_reward/mean": 0.7060546875,
      "rewards/code_format_reward/std": 0.40061408281326294,
      "rewards/code_reward/mean": 0.1816251575946808,
      "rewards/code_reward/std": 0.33169838786125183,
      "step": 625,
      "step_time": 533.8292061761022
    },
    {
      "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,
      "entropy": 1.5363706946372986,
      "epoch": 0.2998084291187739,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1921708583831787,
      "learning_rate": 3e-06,
      "loss": 0.0014,
      "step": 626,
      "step_time": 8.153130851686
    },
    {
      "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,
      "entropy": 1.5362583781991686,
      "epoch": 0.30028735632183906,
      "fork_head/loss": 2.710524559020996,
      "fork_head/num_positive": 11.9375,
      "fork_head/pred_top1_mass": 0.08126959204673767,
      "grad_norm": 0.6001931428909302,
      "learning_rate": 3e-06,
      "loss": 0.6745,
      "step": 627,
      "step_time": 8.387876458466053
    },
    {
      "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,
      "entropy": 1.5356132250565748,
      "epoch": 0.3007662835249042,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1995445042848587,
      "learning_rate": 3e-06,
      "loss": 0.0002,
      "step": 628,
      "step_time": 7.223354555666447
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.5449125639029913,
      "epoch": 0.30124521072796934,
      "fork_head/loss": 2.694979667663574,
      "fork_head/num_positive": 11.9375,
      "fork_head/pred_top1_mass": 0.10201819241046906,
      "grad_norm": 0.49403297901153564,
      "learning_rate": 3e-06,
      "loss": 0.6678,
      "step": 629,
      "step_time": 7.509460888803005
    },
    {
      "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,
      "entropy": 1.5492287690822895,
      "epoch": 0.3017241379310345,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19895809888839722,
      "learning_rate": 3e-06,
      "loss": -0.0012,
      "step": 630,
      "step_time": 8.257183633744717
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.541387643132891,
      "epoch": 0.3022030651340996,
      "fork_head/loss": 2.6765549182891846,
      "fork_head/num_positive": 11.9375,
      "fork_head/pred_top1_mass": 0.11615848541259766,
      "grad_norm": 0.41078948974609375,
      "learning_rate": 3e-06,
      "loss": 0.6619,
      "step": 631,
      "step_time": 7.4038675501942635
    },
    {
      "clip_ratio/high_max": 0.15384615384615385,
      "clip_ratio/high_mean": 0.06410256486672622,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.14102564178980315,
      "entropy": 1.5253780896847065,
      "epoch": 0.30268199233716475,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17495541274547577,
      "learning_rate": 3e-06,
      "loss": -0.0032,
      "step": 632,
      "step_time": 7.798579998314381
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 222.583984375,
      "completions/mean_terminated_length": 222.583984375,
      "completions/min_length": 45.0,
      "completions/min_terminated_length": 45.0,
      "entropy": 1.475065106153488,
      "epoch": 0.3031609195402299,
      "fork/depth_mean": 0.21613526344299316,
      "fork/depth_std": 0.21873313188552856,
      "fork_head/loss": 2.7206902503967285,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.059968795627355576,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6261299848556519,
      "learning_rate": 3e-06,
      "loss": 0.6858,
      "num_tokens": 16915351.0,
      "reward": 0.9537851810455322,
      "reward_std": 0.6043094396591187,
      "rewards/code_format_reward/mean": 0.7177734375,
      "rewards/code_format_reward/std": 0.3855721354484558,
      "rewards/code_reward/mean": 0.23601175844669342,
      "rewards/code_reward/std": 0.3688889145851135,
      "step": 633,
      "step_time": 534.9244987219572
    },
    {
      "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,
      "entropy": 1.806884765625,
      "epoch": 0.303639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12101179361343384,
      "learning_rate": 3e-06,
      "loss": -0.0074,
      "step": 634,
      "step_time": 7.2032326981425285
    },
    {
      "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,
      "entropy": 1.4630208432674408,
      "epoch": 0.30411877394636017,
      "fork_head/loss": 2.7113306522369385,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.0637202262878418,
      "grad_norm": 0.7387445569038391,
      "learning_rate": 3e-06,
      "loss": 0.6834,
      "step": 635,
      "step_time": 8.006466440856457
    },
    {
      "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,
      "entropy": 1.790771484375,
      "epoch": 0.3045977011494253,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14345544576644897,
      "learning_rate": 3e-06,
      "loss": -0.0084,
      "step": 636,
      "step_time": 7.996973820030689
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.499609375,
      "epoch": 0.30507662835249044,
      "fork_head/loss": 2.6956028938293457,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.0751015692949295,
      "grad_norm": 0.49548715353012085,
      "learning_rate": 3e-06,
      "loss": 0.6791,
      "step": 637,
      "step_time": 7.044672206044197
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.78173828125,
      "epoch": 0.3055555555555556,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0689394548535347,
      "learning_rate": 3e-06,
      "loss": -0.0087,
      "step": 638,
      "step_time": 7.8598800748586655
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.5273437440395354,
      "epoch": 0.30603448275862066,
      "fork_head/loss": 2.6781766414642334,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.08540204912424088,
      "grad_norm": 0.4089917540550232,
      "learning_rate": 3e-06,
      "loss": 0.6738,
      "step": 639,
      "step_time": 7.538481131196022
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.751953125,
      "epoch": 0.3065134099616858,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06784500926733017,
      "learning_rate": 3e-06,
      "loss": -0.009,
      "step": 640,
      "step_time": 7.404009826481342
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 218.29296875,
      "completions/mean_terminated_length": 218.07073974609375,
      "completions/min_length": 51.0,
      "completions/min_terminated_length": 51.0,
      "entropy": 1.65625,
      "epoch": 0.30699233716475094,
      "fork/depth_mean": 0.21597009897232056,
      "fork/depth_std": 0.20204676687717438,
      "fork_head/loss": 2.7241525650024414,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.07420142740011215,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.49875277280807495,
      "learning_rate": 3e-06,
      "loss": 0.686,
      "num_tokens": 17116493.0,
      "reward": 0.856134295463562,
      "reward_std": 0.6487326622009277,
      "rewards/code_format_reward/mean": 0.642578125,
      "rewards/code_format_reward/std": 0.4222106337547302,
      "rewards/code_reward/mean": 0.2135562300682068,
      "rewards/code_reward/std": 0.36519330739974976,
      "step": 641,
      "step_time": 455.38885851949453
    },
    {
      "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,
      "entropy": 1.7205078125,
      "epoch": 0.3074712643678161,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18280288577079773,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 642,
      "step_time": 7.597839377820492
    },
    {
      "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,
      "entropy": 1.6543782552083333,
      "epoch": 0.3079501915708812,
      "fork_head/loss": 2.7141432762145996,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.07424861192703247,
      "grad_norm": 0.4652221202850342,
      "learning_rate": 3e-06,
      "loss": 0.6829,
      "step": 643,
      "step_time": 8.417336195707321
    },
    {
      "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,
      "entropy": 1.70537109375,
      "epoch": 0.30842911877394635,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2537893056869507,
      "learning_rate": 3e-06,
      "loss": -0.0054,
      "step": 644,
      "step_time": 7.612181209027767
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.6805826822916667,
      "epoch": 0.3089080459770115,
      "fork_head/loss": 2.7004759311676025,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.08055607229471207,
      "grad_norm": 0.41791832447052,
      "learning_rate": 3e-06,
      "loss": 0.6781,
      "step": 645,
      "step_time": 7.871202304959297
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.6845703125,
      "epoch": 0.30938697318007663,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0934046134352684,
      "learning_rate": 3e-06,
      "loss": -0.0057,
      "step": 646,
      "step_time": 7.702960282564163
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.7041829427083333,
      "epoch": 0.30986590038314177,
      "fork_head/loss": 2.684476375579834,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.09517206996679306,
      "grad_norm": 0.400000661611557,
      "learning_rate": 3e-06,
      "loss": 0.6739,
      "step": 647,
      "step_time": 6.896198615431786
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.676171875,
      "epoch": 0.3103448275862069,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10422783344984055,
      "learning_rate": 3e-06,
      "loss": -0.0063,
      "step": 648,
      "step_time": 7.864935398101807
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 215.91796875,
      "completions/mean_terminated_length": 215.91796875,
      "completions/min_length": 69.0,
      "completions/min_terminated_length": 69.0,
      "entropy": 1.3813747763633728,
      "epoch": 0.31082375478927204,
      "fork/depth_mean": 0.20456261932849884,
      "fork/depth_std": 0.18847538530826569,
      "fork_head/loss": 2.7262113094329834,
      "fork_head/num_positive": 11.066666603088379,
      "fork_head/pred_top1_mass": 0.06470367312431335,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7199698090553284,
      "learning_rate": 3e-06,
      "loss": 0.6839,
      "num_tokens": 17322739.0,
      "reward": 1.0090311765670776,
      "reward_std": 0.5848614573478699,
      "rewards/code_format_reward/mean": 0.775390625,
      "rewards/code_format_reward/std": 0.34171754121780396,
      "rewards/code_reward/mean": 0.23364046216011047,
      "rewards/code_reward/std": 0.36589619517326355,
      "step": 649,
      "step_time": 534.9201189056039
    },
    {
      "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,
      "entropy": 1.6793891042470932,
      "epoch": 0.3113026819923372,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1524110585451126,
      "learning_rate": 3e-06,
      "loss": 0.0011,
      "step": 650,
      "step_time": 8.155594818294048
    },
    {
      "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,
      "entropy": 1.3837212423483531,
      "epoch": 0.3117816091954023,
      "fork_head/loss": 2.7108428478240967,
      "fork_head/num_positive": 11.066666603088379,
      "fork_head/pred_top1_mass": 0.0850321501493454,
      "grad_norm": 0.571770429611206,
      "learning_rate": 3e-06,
      "loss": 0.6794,
      "step": 651,
      "step_time": 7.109252884984016
    },
    {
      "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,
      "entropy": 1.6980794320503871,
      "epoch": 0.31226053639846746,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15162867307662964,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 652,
      "step_time": 8.677077554166317
    },
    {
      "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,
      "entropy": 1.370713969071706,
      "epoch": 0.31273946360153254,
      "fork_head/loss": 2.692225933074951,
      "fork_head/num_positive": 11.066666603088379,
      "fork_head/pred_top1_mass": 0.0833837240934372,
      "grad_norm": 0.482656866312027,
      "learning_rate": 3e-06,
      "loss": 0.674,
      "step": 653,
      "step_time": 7.255523130297661
    },
    {
      "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,
      "entropy": 1.6806098024050395,
      "epoch": 0.3132183908045977,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15111157298088074,
      "learning_rate": 3e-06,
      "loss": -0.0021,
      "step": 654,
      "step_time": 7.509786084294319
    },
    {
      "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,
      "entropy": 1.3731825053691864,
      "epoch": 0.3136973180076628,
      "fork_head/loss": 2.6740713119506836,
      "fork_head/num_positive": 11.066666603088379,
      "fork_head/pred_top1_mass": 0.09403259307146072,
      "grad_norm": 0.45635083317756653,
      "learning_rate": 3e-06,
      "loss": 0.6688,
      "step": 655,
      "step_time": 8.008772447705269
    },
    {
      "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,
      "entropy": 1.6907009482383728,
      "epoch": 0.31417624521072796,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15294505655765533,
      "learning_rate": 3e-06,
      "loss": -0.0018,
      "step": 656,
      "step_time": 7.332108490169048
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.443359375,
      "completions/mean_terminated_length": 219.443359375,
      "completions/min_length": 57.0,
      "completions/min_terminated_length": 57.0,
      "entropy": 0.9811197916666666,
      "epoch": 0.3146551724137931,
      "fork/depth_mean": 0.19162006676197052,
      "fork/depth_std": 0.18475784361362457,
      "fork_head/loss": 2.7303831577301025,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.07861144095659256,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5186684131622314,
      "learning_rate": 3e-06,
      "loss": 0.6869,
      "num_tokens": 17526022.0,
      "reward": 0.7638766765594482,
      "reward_std": 0.6961973905563354,
      "rewards/code_format_reward/mean": 0.5732421875,
      "rewards/code_format_reward/std": 0.45867037773132324,
      "rewards/code_reward/mean": 0.19063451886177063,
      "rewards/code_reward/std": 0.3502853810787201,
      "step": 657,
      "step_time": 443.389454215765
    },
    {
      "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,
      "entropy": 1.822265625,
      "epoch": 0.31513409961685823,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10308587551116943,
      "learning_rate": 3e-06,
      "loss": -0.0044,
      "step": 658,
      "step_time": 7.209618397057056
    },
    {
      "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,
      "entropy": 0.9856770833333334,
      "epoch": 0.31561302681992337,
      "fork_head/loss": 2.723689079284668,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.0777750238776207,
      "grad_norm": 0.49085649847984314,
      "learning_rate": 3e-06,
      "loss": 0.6852,
      "step": 659,
      "step_time": 8.119602978229523
    },
    {
      "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,
      "entropy": 1.8400390625,
      "epoch": 0.3160919540229885,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1529390513896942,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 660,
      "step_time": 7.448671281337738
    },
    {
      "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,
      "entropy": 0.9742838541666666,
      "epoch": 0.31657088122605365,
      "fork_head/loss": 2.712096929550171,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.07842861860990524,
      "grad_norm": 0.4190363883972168,
      "learning_rate": 3e-06,
      "loss": 0.6822,
      "step": 661,
      "step_time": 7.556257277727127
    },
    {
      "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,
      "entropy": 1.81845703125,
      "epoch": 0.3170498084291188,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1045880913734436,
      "learning_rate": 3e-06,
      "loss": -0.0054,
      "step": 662,
      "step_time": 7.712523058056831
    },
    {
      "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,
      "entropy": 0.9674479166666666,
      "epoch": 0.3175287356321839,
      "fork_head/loss": 2.6975297927856445,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.08150549232959747,
      "grad_norm": 0.3705242872238159,
      "learning_rate": 3e-06,
      "loss": 0.678,
      "step": 663,
      "step_time": 7.276861183345318
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.815625,
      "epoch": 0.31800766283524906,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08969177305698395,
      "learning_rate": 3e-06,
      "loss": -0.0063,
      "step": 664,
      "step_time": 7.5478555560112
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 222.193359375,
      "completions/mean_terminated_length": 222.193359375,
      "completions/min_length": 35.0,
      "completions/min_terminated_length": 35.0,
      "entropy": 2.1327039930555554,
      "epoch": 0.3184865900383142,
      "fork/depth_mean": 0.20343045890331268,
      "fork/depth_std": 0.1897556185722351,
      "fork_head/loss": 2.7396552562713623,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.07507901638746262,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6701818704605103,
      "learning_rate": 3e-06,
      "loss": 0.6799,
      "num_tokens": 17736505.0,
      "reward": 0.8062930107116699,
      "reward_std": 0.6340053081512451,
      "rewards/code_format_reward/mean": 0.6376953125,
      "rewards/code_format_reward/std": 0.4267074167728424,
      "rewards/code_reward/mean": 0.16859763860702515,
      "rewards/code_reward/std": 0.3250091075897217,
      "step": 665,
      "step_time": 551.4275472983718
    },
    {
      "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,
      "entropy": 0.987548828125,
      "epoch": 0.31896551724137934,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04516947269439697,
      "learning_rate": 3e-06,
      "loss": 0.0028,
      "step": 666,
      "step_time": 7.646697327494621
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.05555555555555555,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 2.1064453125,
      "epoch": 0.3194444444444444,
      "fork_head/loss": 2.7272746562957764,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.07482146471738815,
      "grad_norm": 0.5969884395599365,
      "learning_rate": 3e-06,
      "loss": 0.6761,
      "step": 667,
      "step_time": 7.730435386300087
    },
    {
      "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,
      "entropy": 0.9951171875,
      "epoch": 0.31992337164750956,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04634403809905052,
      "learning_rate": 3e-06,
      "loss": 0.0028,
      "step": 668,
      "step_time": 7.596455290913582
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.05555555555555555,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 2.0718315972222223,
      "epoch": 0.3204022988505747,
      "fork_head/loss": 2.7069778442382812,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.08673342317342758,
      "grad_norm": 0.5352058410644531,
      "learning_rate": 3e-06,
      "loss": 0.6706,
      "step": 669,
      "step_time": 7.4087376818060875
    },
    {
      "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,
      "entropy": 0.98681640625,
      "epoch": 0.32088122605363983,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0467439703643322,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 670,
      "step_time": 7.473184496164322
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.05555555555555555,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 2.0698784722222223,
      "epoch": 0.321360153256705,
      "fork_head/loss": 2.683602809906006,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.09047335386276245,
      "grad_norm": 0.46537283062934875,
      "learning_rate": 3e-06,
      "loss": 0.6638,
      "step": 671,
      "step_time": 8.199319489300251
    },
    {
      "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,
      "entropy": 0.9873046875,
      "epoch": 0.3218390804597701,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.046790603548288345,
      "learning_rate": 3e-06,
      "loss": 0.0023,
      "step": 672,
      "step_time": 7.163857780396938
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 224.244140625,
      "completions/mean_terminated_length": 224.18199157714844,
      "completions/min_length": 19.0,
      "completions/min_terminated_length": 19.0,
      "entropy": 1.221571187178294,
      "epoch": 0.32231800766283525,
      "fork/depth_mean": 0.18855378031730652,
      "fork/depth_std": 0.1790890395641327,
      "fork_head/loss": 2.7297375202178955,
      "fork_head/num_positive": 12.75,
      "fork_head/pred_top1_mass": 0.07750516384840012,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5383930206298828,
      "learning_rate": 3e-06,
      "loss": 0.6845,
      "num_tokens": 17951462.0,
      "reward": 0.8927725553512573,
      "reward_std": 0.6383034586906433,
      "rewards/code_format_reward/mean": 0.6923828125,
      "rewards/code_format_reward/std": 0.42037233710289,
      "rewards/code_reward/mean": 0.20038969814777374,
      "rewards/code_reward/std": 0.34130048751831055,
      "step": 673,
      "step_time": 510.41448144614697
    },
    {
      "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,
      "entropy": 1.5961037599123442,
      "epoch": 0.3227969348659004,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09082793444395065,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 674,
      "step_time": 7.631755709648132
    },
    {
      "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,
      "entropy": 1.2225748697916667,
      "epoch": 0.3232758620689655,
      "fork_head/loss": 2.725512981414795,
      "fork_head/num_positive": 12.75,
      "fork_head/pred_top1_mass": 0.07844067364931107,
      "grad_norm": 0.5358536243438721,
      "learning_rate": 3e-06,
      "loss": 0.6831,
      "step": 675,
      "step_time": 7.987068571150303
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.6088241247030406,
      "epoch": 0.32375478927203066,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08834149688482285,
      "learning_rate": 3e-06,
      "loss": -0.0036,
      "step": 676,
      "step_time": 7.2638164311647415
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.027777778605620067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06944444527228673,
      "entropy": 1.226209859053294,
      "epoch": 0.3242337164750958,
      "fork_head/loss": 2.7156169414520264,
      "fork_head/num_positive": 12.75,
      "fork_head/pred_top1_mass": 0.07169769704341888,
      "grad_norm": 0.416379451751709,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "step": 677,
      "step_time": 7.421171054244041
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.11538461538461539,
      "entropy": 1.5969050480769231,
      "epoch": 0.32471264367816094,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.054728515446186066,
      "learning_rate": 3e-06,
      "loss": -0.0039,
      "step": 678,
      "step_time": 7.634840793907642
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0486111119389534,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111119389534,
      "entropy": 1.2154134114583333,
      "epoch": 0.3251915708812261,
      "fork_head/loss": 2.7050235271453857,
      "fork_head/num_positive": 12.75,
      "fork_head/pred_top1_mass": 0.06953050941228867,
      "grad_norm": 0.326135516166687,
      "learning_rate": 3e-06,
      "loss": 0.6749,
      "step": 679,
      "step_time": 7.07073050737381
    },
    {
      "clip_ratio/high_max": 0.15384615384615385,
      "clip_ratio/high_mean": 0.11538461538461539,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.19230769230769232,
      "entropy": 1.6094000339508057,
      "epoch": 0.32567049808429116,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03880935534834862,
      "learning_rate": 3e-06,
      "loss": -0.0042,
      "step": 680,
      "step_time": 7.996875569224358
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 227.126953125,
      "completions/mean_terminated_length": 227.01373291015625,
      "completions/min_length": 56.0,
      "completions/min_terminated_length": 56.0,
      "entropy": 1.2802734375,
      "epoch": 0.3261494252873563,
      "fork/depth_mean": 0.20994806289672852,
      "fork/depth_std": 0.18573883175849915,
      "fork_head/loss": 2.71833872795105,
      "fork_head/num_positive": 9.625,
      "fork_head/pred_top1_mass": 0.0768435001373291,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8513457775115967,
      "learning_rate": 3e-06,
      "loss": 0.675,
      "num_tokens": 18160503.0,
      "reward": 0.9713340997695923,
      "reward_std": 0.5498830080032349,
      "rewards/code_format_reward/mean": 0.7900390625,
      "rewards/code_format_reward/std": 0.34317272901535034,
      "rewards/code_reward/mean": 0.1812949925661087,
      "rewards/code_reward/std": 0.3318769037723541,
      "step": 681,
      "step_time": 475.1939110606909
    },
    {
      "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,
      "entropy": 1.8153211805555556,
      "epoch": 0.32662835249042144,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11672507971525192,
      "learning_rate": 3e-06,
      "loss": 0.0067,
      "step": 682,
      "step_time": 8.275830678641796
    },
    {
      "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,
      "entropy": 1.278076171875,
      "epoch": 0.3271072796934866,
      "fork_head/loss": 2.7036948204040527,
      "fork_head/num_positive": 9.625,
      "fork_head/pred_top1_mass": 0.07828715443611145,
      "grad_norm": 0.7386294603347778,
      "learning_rate": 3e-06,
      "loss": 0.6717,
      "step": 683,
      "step_time": 7.5180483385920525
    },
    {
      "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,
      "entropy": 1.8213975694444444,
      "epoch": 0.3275862068965517,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10631651431322098,
      "learning_rate": 3e-06,
      "loss": 0.0061,
      "step": 684,
      "step_time": 7.788953810930252
    },
    {
      "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,
      "entropy": 1.251708984375,
      "epoch": 0.32806513409961685,
      "fork_head/loss": 2.6810946464538574,
      "fork_head/num_positive": 9.625,
      "fork_head/pred_top1_mass": 0.09841211885213852,
      "grad_norm": 0.5590615272521973,
      "learning_rate": 3e-06,
      "loss": 0.6655,
      "step": 685,
      "step_time": 7.467975527048111
    },
    {
      "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,
      "entropy": 1.8115234375,
      "epoch": 0.328544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0969315841794014,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 686,
      "step_time": 7.4674952775239944
    },
    {
      "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,
      "entropy": 1.259765625,
      "epoch": 0.3290229885057471,
      "fork_head/loss": 2.6586368083953857,
      "fork_head/num_positive": 9.625,
      "fork_head/pred_top1_mass": 0.10931036621332169,
      "grad_norm": 0.46070224046707153,
      "learning_rate": 3e-06,
      "loss": 0.6589,
      "step": 687,
      "step_time": 7.869917958974838
    },
    {
      "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,
      "entropy": 1.8166232638888888,
      "epoch": 0.32950191570881227,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10297614336013794,
      "learning_rate": 3e-06,
      "loss": 0.0045,
      "step": 688,
      "step_time": 7.409209027886391
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 217.73828125,
      "completions/mean_terminated_length": 217.5127716064453,
      "completions/min_length": 100.0,
      "completions/min_terminated_length": 100.0,
      "entropy": 1.8271484375,
      "epoch": 0.3299808429118774,
      "fork/depth_mean": 0.1865006536245346,
      "fork/depth_std": 0.1837741583585739,
      "fork_head/loss": 2.7077279090881348,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.08376343548297882,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.616621196269989,
      "learning_rate": 3e-06,
      "loss": 0.6665,
      "num_tokens": 18367105.0,
      "reward": 0.9509727954864502,
      "reward_std": 0.6688423752784729,
      "rewards/code_format_reward/mean": 0.6875,
      "rewards/code_format_reward/std": 0.4030218720436096,
      "rewards/code_reward/mean": 0.2634727954864502,
      "rewards/code_reward/std": 0.38958191871643066,
      "step": 689,
      "step_time": 546.439685203135
    },
    {
      "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,
      "entropy": 1.7144639765222867,
      "epoch": 0.33045977011494254,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08991137146949768,
      "learning_rate": 3e-06,
      "loss": 0.0035,
      "step": 690,
      "step_time": 8.10956171900034
    },
    {
      "clip_ratio/high_max": 0.2,
      "clip_ratio/high_mean": 0.13333333432674407,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13333333432674407,
      "entropy": 1.80419921875,
      "epoch": 0.3309386973180077,
      "fork_head/loss": 2.7008485794067383,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.08425312489271164,
      "grad_norm": 0.48550206422805786,
      "learning_rate": 3e-06,
      "loss": 0.6609,
      "step": 691,
      "step_time": 8.342555098235607
    },
    {
      "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,
      "entropy": 1.6860080311695735,
      "epoch": 0.3314176245210728,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09550526738166809,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 692,
      "step_time": 7.485546194016933
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.03333333432674408,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333432674409,
      "entropy": 1.81943359375,
      "epoch": 0.33189655172413796,
      "fork_head/loss": 2.6892876625061035,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.09203357249498367,
      "grad_norm": 0.43742161989212036,
      "learning_rate": 3e-06,
      "loss": 0.6572,
      "step": 693,
      "step_time": 7.684609934687614
    },
    {
      "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,
      "entropy": 1.6902669270833333,
      "epoch": 0.33237547892720304,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09170864522457123,
      "learning_rate": 3e-06,
      "loss": 0.0023,
      "step": 694,
      "step_time": 7.752978950738907
    },
    {
      "clip_ratio/high_max": 0.4,
      "clip_ratio/high_mean": 0.28333333432674407,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.33333333432674406,
      "entropy": 1.7534505128860474,
      "epoch": 0.3328544061302682,
      "fork_head/loss": 2.675682306289673,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.09637767821550369,
      "grad_norm": 0.3991265892982483,
      "learning_rate": 3e-06,
      "loss": 0.6517,
      "step": 695,
      "step_time": 7.913252949714661
    },
    {
      "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,
      "entropy": 1.6782769113779068,
      "epoch": 0.3333333333333333,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09723030775785446,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 696,
      "step_time": 7.381743349134922
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 225.447265625,
      "completions/mean_terminated_length": 225.26719665527344,
      "completions/min_length": 26.0,
      "completions/min_terminated_length": 26.0,
      "entropy": 1.5160807251930237,
      "epoch": 0.33381226053639845,
      "fork/depth_mean": 0.2163148671388626,
      "fork/depth_std": 0.23229363560676575,
      "fork_head/loss": 2.7276713848114014,
      "fork_head/num_positive": 12.84375,
      "fork_head/pred_top1_mass": 0.07403130829334259,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5032326579093933,
      "learning_rate": 3e-06,
      "loss": 0.6944,
      "num_tokens": 18579670.0,
      "reward": 0.8899000883102417,
      "reward_std": 0.6009580492973328,
      "rewards/code_format_reward/mean": 0.6748046875,
      "rewards/code_format_reward/std": 0.41523101925849915,
      "rewards/code_reward/mean": 0.2150954306125641,
      "rewards/code_reward/std": 0.34519317746162415,
      "step": 697,
      "step_time": 636.0058447569609
    },
    {
      "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,
      "entropy": 1.8851218819618225,
      "epoch": 0.3342911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10647077113389969,
      "learning_rate": 3e-06,
      "loss": -0.0126,
      "step": 698,
      "step_time": 9.344668924808502
    },
    {
      "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,
      "entropy": 1.5111653685569764,
      "epoch": 0.33477011494252873,
      "fork_head/loss": 2.719756841659546,
      "fork_head/num_positive": 12.84375,
      "fork_head/pred_top1_mass": 0.07854773849248886,
      "grad_norm": 0.4755583107471466,
      "learning_rate": 3e-06,
      "loss": 0.692,
      "step": 699,
      "step_time": 7.801806338131428
    },
    {
      "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,
      "entropy": 1.8704607983430226,
      "epoch": 0.33524904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11279706656932831,
      "learning_rate": 3e-06,
      "loss": -0.0139,
      "step": 700,
      "step_time": 8.82988341897726
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.03333333432674408,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13333333432674407,
      "entropy": 1.5127604126930236,
      "epoch": 0.335727969348659,
      "fork_head/loss": 2.7062125205993652,
      "fork_head/num_positive": 12.84375,
      "fork_head/pred_top1_mass": 0.07946242392063141,
      "grad_norm": 0.4135279059410095,
      "learning_rate": 3e-06,
      "loss": 0.6875,
      "step": 701,
      "step_time": 7.390668369829655
    },
    {
      "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,
      "entropy": 1.8591489493846893,
      "epoch": 0.33620689655172414,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11901779472827911,
      "learning_rate": 3e-06,
      "loss": -0.0141,
      "step": 702,
      "step_time": 8.496706292033195
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.03333333432674408,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13333333432674407,
      "entropy": 1.5277994751930237,
      "epoch": 0.3366858237547893,
      "fork_head/loss": 2.689971923828125,
      "fork_head/num_positive": 12.84375,
      "fork_head/pred_top1_mass": 0.0964873880147934,
      "grad_norm": 0.3664446175098419,
      "learning_rate": 3e-06,
      "loss": 0.6833,
      "step": 703,
      "step_time": 7.888358347117901
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.8635091135899227,
      "epoch": 0.3371647509578544,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09574859589338303,
      "learning_rate": 3e-06,
      "loss": -0.0145,
      "step": 704,
      "step_time": 8.85973409563303
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 209.400390625,
      "completions/mean_terminated_length": 209.30918884277344,
      "completions/min_length": 55.0,
      "completions/min_terminated_length": 55.0,
      "entropy": 1.635541870043828,
      "epoch": 0.33764367816091956,
      "fork/depth_mean": 0.20754772424697876,
      "fork/depth_std": 0.24080538749694824,
      "fork_head/loss": 2.727405071258545,
      "fork_head/num_positive": 11.677419662475586,
      "fork_head/pred_top1_mass": 0.06672926247119904,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7044205069541931,
      "learning_rate": 3e-06,
      "loss": 0.6796,
      "num_tokens": 18778163.0,
      "reward": 1.0151164531707764,
      "reward_std": 0.6040914058685303,
      "rewards/code_format_reward/mean": 0.765625,
      "rewards/code_format_reward/std": 0.365795373916626,
      "rewards/code_reward/mean": 0.24949154257774353,
      "rewards/code_reward/std": 0.37464362382888794,
      "step": 705,
      "step_time": 676.6101103574038
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.03333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.4676323811213174,
      "epoch": 0.3381226053639847,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.36822789907455444,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 706,
      "step_time": 7.685910180211067
    },
    {
      "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,
      "entropy": 1.6233723988899817,
      "epoch": 0.33860153256704983,
      "fork_head/loss": 2.7132763862609863,
      "fork_head/num_positive": 11.677419662475586,
      "fork_head/pred_top1_mass": 0.07435332983732224,
      "grad_norm": 0.6278664469718933,
      "learning_rate": 3e-06,
      "loss": 0.6751,
      "step": 707,
      "step_time": 8.535944685339928
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.03333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.4585395018259684,
      "epoch": 0.3390804597701149,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16322576999664307,
      "learning_rate": 3e-06,
      "loss": 0.0008,
      "step": 708,
      "step_time": 7.884283624589443
    },
    {
      "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,
      "entropy": 1.625776245043828,
      "epoch": 0.33955938697318006,
      "fork_head/loss": 2.692636728286743,
      "fork_head/num_positive": 11.677419662475586,
      "fork_head/pred_top1_mass": 0.08982746303081512,
      "grad_norm": 0.5636675953865051,
      "learning_rate": 3e-06,
      "loss": 0.6686,
      "step": 709,
      "step_time": 7.372974865138531
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.03333333333333333,
      "clip_ratio/low_mean": 0.022222222884496053,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.055555556217829385,
      "entropy": 1.4724500934282938,
      "epoch": 0.3400383141762452,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15269599854946136,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 710,
      "step_time": 8.320651702582836
    },
    {
      "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,
      "entropy": 1.6319611347638643,
      "epoch": 0.34051724137931033,
      "fork_head/loss": 2.668609142303467,
      "fork_head/num_positive": 11.677419662475586,
      "fork_head/pred_top1_mass": 0.10090870410203934,
      "grad_norm": 0.5415878891944885,
      "learning_rate": 3e-06,
      "loss": 0.6622,
      "step": 711,
      "step_time": 7.2334348782896996
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.03333333333333333,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.03333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.4809136311213176,
      "epoch": 0.34099616858237547,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13519859313964844,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 712,
      "step_time": 8.202105812728405
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 202.615234375,
      "completions/mean_terminated_length": 202.615234375,
      "completions/min_length": 9.0,
      "completions/min_terminated_length": 9.0,
      "entropy": 1.5374038900647844,
      "epoch": 0.3414750957854406,
      "fork/depth_mean": 0.24635399878025055,
      "fork/depth_std": 0.2977292835712433,
      "fork_head/loss": 2.7416093349456787,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.06181345880031586,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.778921902179718,
      "learning_rate": 3e-06,
      "loss": 0.6832,
      "num_tokens": 18976158.0,
      "reward": 1.0272530317306519,
      "reward_std": 0.6584952473640442,
      "rewards/code_format_reward/mean": 0.732421875,
      "rewards/code_format_reward/std": 0.39154917001724243,
      "rewards/code_reward/mean": 0.29483118653297424,
      "rewards/code_reward/std": 0.39903807640075684,
      "step": 713,
      "step_time": 442.8453555777669
    },
    {
      "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,
      "entropy": 1.533253986388445,
      "epoch": 0.34195402298850575,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13894139230251312,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 714,
      "step_time": 8.764260172843933
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.5439530611038208,
      "epoch": 0.3424329501915709,
      "fork_head/loss": 2.7295944690704346,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.06637348234653473,
      "grad_norm": 0.5583345890045166,
      "learning_rate": 3e-06,
      "loss": 0.6781,
      "step": 715,
      "step_time": 7.305598184466362
    },
    {
      "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,
      "entropy": 1.536956787109375,
      "epoch": 0.342911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1266748011112213,
      "learning_rate": 3e-06,
      "loss": 0.0054,
      "step": 716,
      "step_time": 8.068137496709824
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05357142857142857,
      "entropy": 1.5292581234659468,
      "epoch": 0.34339080459770116,
      "fork_head/loss": 2.716549873352051,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.06738004833459854,
      "grad_norm": 0.5443685054779053,
      "learning_rate": 3e-06,
      "loss": 0.6728,
      "step": 717,
      "step_time": 7.840320594608784
    },
    {
      "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,
      "entropy": 1.53206380084157,
      "epoch": 0.3438697318007663,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12693127989768982,
      "learning_rate": 3e-06,
      "loss": 0.0037,
      "step": 718,
      "step_time": 7.50559376180172
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.07142857142857142,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08928571428571429,
      "entropy": 1.5286070789609636,
      "epoch": 0.34434865900383144,
      "fork_head/loss": 2.701568841934204,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.07853259146213531,
      "grad_norm": 0.5399172306060791,
      "learning_rate": 3e-06,
      "loss": 0.6676,
      "step": 719,
      "step_time": 7.995094113051891
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.046875,
      "clip_ratio/low_min": 0.03125,
      "clip_ratio/region_mean": 0.046875,
      "entropy": 1.532114665955305,
      "epoch": 0.3448275862068966,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10186656564474106,
      "learning_rate": 3e-06,
      "loss": 0.0027,
      "step": 720,
      "step_time": 8.118474565446377
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 216.630859375,
      "completions/mean_terminated_length": 216.630859375,
      "completions/min_length": 75.0,
      "completions/min_terminated_length": 75.0,
      "entropy": 1.6520905627144709,
      "epoch": 0.3453065134099617,
      "fork/depth_mean": 0.2058226615190506,
      "fork/depth_std": 0.2252112627029419,
      "fork_head/loss": 2.7309837341308594,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.07215196639299393,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.48545345664024353,
      "learning_rate": 3e-06,
      "loss": 0.6868,
      "num_tokens": 19178993.0,
      "reward": 0.9671177268028259,
      "reward_std": 0.664161205291748,
      "rewards/code_format_reward/mean": 0.6982421875,
      "rewards/code_format_reward/std": 0.3859783709049225,
      "rewards/code_reward/mean": 0.2688755691051483,
      "rewards/code_reward/std": 0.4002925455570221,
      "step": 721,
      "step_time": 510.83108699321747
    },
    {
      "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,
      "entropy": 1.577392578125,
      "epoch": 0.3457854406130268,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14968739449977875,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 722,
      "step_time": 7.264333039522171
    },
    {
      "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,
      "entropy": 1.6631944444444444,
      "epoch": 0.34626436781609193,
      "fork_head/loss": 2.7231578826904297,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.07450009882450104,
      "grad_norm": 0.4865049421787262,
      "learning_rate": 3e-06,
      "loss": 0.6846,
      "step": 723,
      "step_time": 8.538562022149563
    },
    {
      "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,
      "entropy": 1.5673828125,
      "epoch": 0.34674329501915707,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13440175354480743,
      "learning_rate": 3e-06,
      "loss": -0.0026,
      "step": 724,
      "step_time": 7.263335570693016
    },
    {
      "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,
      "entropy": 1.685836222436693,
      "epoch": 0.3472222222222222,
      "fork_head/loss": 2.7093048095703125,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.075654536485672,
      "grad_norm": 0.44958430528640747,
      "learning_rate": 3e-06,
      "loss": 0.6809,
      "step": 725,
      "step_time": 7.0841928869485855
    },
    {
      "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,
      "entropy": 1.5904541015625,
      "epoch": 0.34770114942528735,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15681231021881104,
      "learning_rate": 3e-06,
      "loss": -0.0034,
      "step": 726,
      "step_time": 8.00648682564497
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.6902850071589153,
      "epoch": 0.3481800766283525,
      "fork_head/loss": 2.6920933723449707,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.09506721794605255,
      "grad_norm": 0.3975068926811218,
      "learning_rate": 3e-06,
      "loss": 0.6766,
      "step": 727,
      "step_time": 7.155944608151913
    },
    {
      "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,
      "entropy": 1.6214599609375,
      "epoch": 0.3486590038314176,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10808105021715164,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 728,
      "step_time": 7.558407336473465
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.8125,
      "completions/mean_terminated_length": 219.8125,
      "completions/min_length": 43.0,
      "completions/min_terminated_length": 43.0,
      "entropy": 1.5127766927083333,
      "epoch": 0.34913793103448276,
      "fork/depth_mean": 0.20522412657737732,
      "fork/depth_std": 0.23015163838863373,
      "fork_head/loss": 2.717991352081299,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.06772784143686295,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5520479083061218,
      "learning_rate": 3e-06,
      "loss": 0.6809,
      "num_tokens": 19381809.0,
      "reward": 0.965728759765625,
      "reward_std": 0.6023224592208862,
      "rewards/code_format_reward/mean": 0.7470703125,
      "rewards/code_format_reward/std": 0.37010499835014343,
      "rewards/code_reward/mean": 0.2186584770679474,
      "rewards/code_reward/std": 0.3716493546962738,
      "step": 729,
      "step_time": 541.0664793252945
    },
    {
      "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,
      "entropy": 1.9462002840909092,
      "epoch": 0.3496168582375479,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12245267629623413,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 730,
      "step_time": 8.021084293723106
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5031467080116272,
      "epoch": 0.35009578544061304,
      "fork_head/loss": 2.707916021347046,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.07316039502620697,
      "grad_norm": 0.5374640226364136,
      "learning_rate": 3e-06,
      "loss": 0.6783,
      "step": 731,
      "step_time": 7.839607924222946
    },
    {
      "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,
      "entropy": 1.9480868252840908,
      "epoch": 0.3505747126436782,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1201181635260582,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 732,
      "step_time": 7.543438531458378
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5109592080116272,
      "epoch": 0.3510536398467433,
      "fork_head/loss": 2.690521717071533,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.09496341645717621,
      "grad_norm": 0.5009539723396301,
      "learning_rate": 3e-06,
      "loss": 0.6733,
      "step": 733,
      "step_time": 7.9085995852947235
    },
    {
      "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,
      "entropy": 1.9452903053977273,
      "epoch": 0.35153256704980845,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11611872911453247,
      "learning_rate": 3e-06,
      "loss": -0.0035,
      "step": 734,
      "step_time": 7.223447546362877
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5138617654641469,
      "epoch": 0.35201149425287354,
      "fork_head/loss": 2.6688342094421387,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.09763947874307632,
      "grad_norm": 0.4556938409805298,
      "learning_rate": 3e-06,
      "loss": 0.6679,
      "step": 735,
      "step_time": 7.999595284461975
    },
    {
      "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,
      "entropy": 1.95263671875,
      "epoch": 0.3524904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11600667983293533,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 736,
      "step_time": 7.83284916728735
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 216.78125,
      "completions/mean_terminated_length": 216.70449829101562,
      "completions/min_length": 59.0,
      "completions/min_terminated_length": 59.0,
      "entropy": 1.9818793402777777,
      "epoch": 0.3529693486590038,
      "fork/depth_mean": 0.19982801377773285,
      "fork/depth_std": 0.23787730932235718,
      "fork_head/loss": 2.732302665710449,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.04801912605762482,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6047839522361755,
      "learning_rate": 3e-06,
      "loss": 0.6862,
      "num_tokens": 19586465.0,
      "reward": 0.9308830499649048,
      "reward_std": 0.5457432270050049,
      "rewards/code_format_reward/mean": 0.7568359375,
      "rewards/code_format_reward/std": 0.368728905916214,
      "rewards/code_reward/mean": 0.17404711246490479,
      "rewards/code_reward/std": 0.30769509077072144,
      "step": 737,
      "step_time": 527.2322110012174
    },
    {
      "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,
      "entropy": 1.6331084424799138,
      "epoch": 0.35344827586206895,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20419102907180786,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 738,
      "step_time": 7.384117379784584
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.9818793402777777,
      "epoch": 0.3539272030651341,
      "fork_head/loss": 2.7214138507843018,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.059035107493400574,
      "grad_norm": 0.48087212443351746,
      "learning_rate": 3e-06,
      "loss": 0.6829,
      "step": 739,
      "step_time": 8.459095530211926
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.6246448863636365,
      "epoch": 0.3544061302681992,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16362956166267395,
      "learning_rate": 3e-06,
      "loss": -0.005,
      "step": 740,
      "step_time": 7.330384358763695
    },
    {
      "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,
      "entropy": 1.9852430555555556,
      "epoch": 0.35488505747126436,
      "fork_head/loss": 2.7072176933288574,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.06662015616893768,
      "grad_norm": 0.4442187547683716,
      "learning_rate": 3e-06,
      "loss": 0.6793,
      "step": 741,
      "step_time": 7.561429299414158
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.12121212211522189,
      "entropy": 1.6322798295454546,
      "epoch": 0.3553639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16714414954185486,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 742,
      "step_time": 7.851053610444069
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.9767795138888888,
      "epoch": 0.35584291187739464,
      "fork_head/loss": 2.6926729679107666,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.07951980084180832,
      "grad_norm": 0.4910830855369568,
      "learning_rate": 3e-06,
      "loss": 0.6757,
      "step": 743,
      "step_time": 6.863843210041523
    },
    {
      "clip_ratio/high_max": 0.18181818181818182,
      "clip_ratio/high_mean": 0.12121212211522189,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.15151515332135287,
      "entropy": 1.6305930397727273,
      "epoch": 0.3563218390804598,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11944794654846191,
      "learning_rate": 3e-06,
      "loss": -0.0064,
      "step": 744,
      "step_time": 7.5500867292284966
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 207.548828125,
      "completions/mean_terminated_length": 207.548828125,
      "completions/min_length": 56.0,
      "completions/min_terminated_length": 56.0,
      "entropy": 1.0755027532577515,
      "epoch": 0.3568007662835249,
      "fork/depth_mean": 0.21309520304203033,
      "fork/depth_std": 0.2253841608762741,
      "fork_head/loss": 2.7405498027801514,
      "fork_head/num_positive": 12.064516067504883,
      "fork_head/pred_top1_mass": 0.06707989424467087,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7100406885147095,
      "learning_rate": 3e-06,
      "loss": 0.6848,
      "num_tokens": 19784730.0,
      "reward": 1.1354396343231201,
      "reward_std": 0.6359319686889648,
      "rewards/code_format_reward/mean": 0.806640625,
      "rewards/code_format_reward/std": 0.3291375935077667,
      "rewards/code_reward/mean": 0.32879894971847534,
      "rewards/code_reward/std": 0.42371854186058044,
      "step": 745,
      "step_time": 581.6594801321626
    },
    {
      "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,
      "entropy": 1.8004557291666667,
      "epoch": 0.35727969348659006,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19390997290611267,
      "learning_rate": 3e-06,
      "loss": 0.0012,
      "step": 746,
      "step_time": 7.848995365202427
    },
    {
      "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,
      "entropy": 1.0692816840277777,
      "epoch": 0.3577586206896552,
      "fork_head/loss": 2.7269058227539062,
      "fork_head/num_positive": 12.064516067504883,
      "fork_head/pred_top1_mass": 0.08432915806770325,
      "grad_norm": 0.5557408332824707,
      "learning_rate": 3e-06,
      "loss": 0.6802,
      "step": 747,
      "step_time": 7.7453358098864555
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.7998046875,
      "epoch": 0.35823754789272033,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21778368949890137,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 748,
      "step_time": 7.978325359523296
    },
    {
      "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,
      "entropy": 1.0560981035232544,
      "epoch": 0.3587164750957854,
      "fork_head/loss": 2.7102458477020264,
      "fork_head/num_positive": 12.064516067504883,
      "fork_head/pred_top1_mass": 0.09370900690555573,
      "grad_norm": 0.4971499443054199,
      "learning_rate": 3e-06,
      "loss": 0.6748,
      "step": 749,
      "step_time": 7.75118924677372
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.7985026041666667,
      "epoch": 0.35919540229885055,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12622644007205963,
      "learning_rate": 3e-06,
      "loss": 0.0013,
      "step": 750,
      "step_time": 7.127865858376026
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777777777777776,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777777777777776,
      "entropy": 1.0676540798611112,
      "epoch": 0.3596743295019157,
      "fork_head/loss": 2.692354679107666,
      "fork_head/num_positive": 12.064516067504883,
      "fork_head/pred_top1_mass": 0.09470121562480927,
      "grad_norm": 0.430449515581131,
      "learning_rate": 3e-06,
      "loss": 0.6701,
      "step": 751,
      "step_time": 8.878038235008717
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.8263346354166667,
      "epoch": 0.36015325670498083,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1333412081003189,
      "learning_rate": 3e-06,
      "loss": 0.0,
      "step": 752,
      "step_time": 7.156874284148216
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 215.87109375,
      "completions/mean_terminated_length": 215.7137451171875,
      "completions/min_length": 28.0,
      "completions/min_terminated_length": 28.0,
      "entropy": 1.905810546875,
      "epoch": 0.36063218390804597,
      "fork/depth_mean": 0.21941779553890228,
      "fork/depth_std": 0.2024029642343521,
      "fork_head/loss": 2.727717876434326,
      "fork_head/num_positive": 10.193548202514648,
      "fork_head/pred_top1_mass": 0.09756942838430405,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.862207293510437,
      "learning_rate": 3e-06,
      "loss": 0.6832,
      "num_tokens": 19986680.0,
      "reward": 1.0043234825134277,
      "reward_std": 0.6665511727333069,
      "rewards/code_format_reward/mean": 0.728515625,
      "rewards/code_format_reward/std": 0.39135390520095825,
      "rewards/code_reward/mean": 0.27580779790878296,
      "rewards/code_reward/std": 0.4179062247276306,
      "step": 753,
      "step_time": 565.9920837059617
    },
    {
      "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,
      "entropy": 1.2352120535714286,
      "epoch": 0.3611111111111111,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2397601306438446,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 754,
      "step_time": 7.409824006259441
    },
    {
      "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,
      "entropy": 1.908203125,
      "epoch": 0.36159003831417624,
      "fork_head/loss": 2.707390069961548,
      "fork_head/num_positive": 10.193548202514648,
      "fork_head/pred_top1_mass": 0.09908689558506012,
      "grad_norm": 0.8142316937446594,
      "learning_rate": 3e-06,
      "loss": 0.6771,
      "step": 755,
      "step_time": 8.290784828364849
    },
    {
      "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,
      "entropy": 1.2418387276785714,
      "epoch": 0.3620689655172414,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2418060600757599,
      "learning_rate": 3e-06,
      "loss": 0.0002,
      "step": 756,
      "step_time": 7.723566889762878
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.900146484375,
      "epoch": 0.3625478927203065,
      "fork_head/loss": 2.6712453365325928,
      "fork_head/num_positive": 10.193548202514648,
      "fork_head/pred_top1_mass": 0.13652054965496063,
      "grad_norm": 0.7027667164802551,
      "learning_rate": 3e-06,
      "loss": 0.6674,
      "step": 757,
      "step_time": 7.3770225793123245
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08928571428571429,
      "entropy": 1.2637416294642858,
      "epoch": 0.36302681992337166,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07001189887523651,
      "learning_rate": 3e-06,
      "loss": -0.0013,
      "step": 758,
      "step_time": 8.280547395348549
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.886328125,
      "epoch": 0.3635057471264368,
      "fork_head/loss": 2.6316723823547363,
      "fork_head/num_positive": 10.193548202514648,
      "fork_head/pred_top1_mass": 0.15468518435955048,
      "grad_norm": 0.5829970240592957,
      "learning_rate": 3e-06,
      "loss": 0.6571,
      "step": 759,
      "step_time": 7.488341838121414
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08928571428571429,
      "entropy": 1.2576032366071428,
      "epoch": 0.36398467432950193,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07383731752634048,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 760,
      "step_time": 8.000662542879581
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 216.01171875,
      "completions/mean_terminated_length": 215.93345642089844,
      "completions/min_length": 81.0,
      "completions/min_terminated_length": 81.0,
      "entropy": 1.584449406181063,
      "epoch": 0.36446360153256707,
      "fork/depth_mean": 0.1962144523859024,
      "fork/depth_std": 0.19363239407539368,
      "fork_head/loss": 2.7439961433410645,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07305695116519928,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.616169810295105,
      "learning_rate": 3e-06,
      "loss": 0.6923,
      "num_tokens": 20199790.0,
      "reward": 0.914643406867981,
      "reward_std": 0.5821338891983032,
      "rewards/code_format_reward/mean": 0.7333984375,
      "rewards/code_format_reward/std": 0.3814661204814911,
      "rewards/code_reward/mean": 0.18124501407146454,
      "rewards/code_reward/std": 0.33785566687583923,
      "step": 761,
      "step_time": 590.5430392250419
    },
    {
      "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,
      "entropy": 1.5649341742197673,
      "epoch": 0.3649425287356322,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09062141925096512,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 762,
      "step_time": 8.31606774032116
    },
    {
      "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,
      "entropy": 1.5852632139410292,
      "epoch": 0.3654214559386973,
      "fork_head/loss": 2.7390780448913574,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07578494399785995,
      "grad_norm": 0.6309606432914734,
      "learning_rate": 3e-06,
      "loss": 0.6913,
      "step": 763,
      "step_time": 7.4144222214818
    },
    {
      "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,
      "entropy": 1.5530526717503865,
      "epoch": 0.36590038314176243,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11487968266010284,
      "learning_rate": 3e-06,
      "loss": -0.0073,
      "step": 764,
      "step_time": 7.8710146471858025
    },
    {
      "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,
      "entropy": 1.5774274596146174,
      "epoch": 0.36637931034482757,
      "fork_head/loss": 2.7247202396392822,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.0729348361492157,
      "grad_norm": 0.6144540309906006,
      "learning_rate": 3e-06,
      "loss": 0.6877,
      "step": 765,
      "step_time": 7.626275420188904
    },
    {
      "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,
      "entropy": 1.5531394799550375,
      "epoch": 0.3668582375478927,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10074780136346817,
      "learning_rate": 3e-06,
      "loss": -0.0075,
      "step": 766,
      "step_time": 7.8530654311180115
    },
    {
      "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,
      "entropy": 1.5845656565257482,
      "epoch": 0.36733716475095785,
      "fork_head/loss": 2.7050745487213135,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07690735161304474,
      "grad_norm": 0.5570534467697144,
      "learning_rate": 3e-06,
      "loss": 0.6822,
      "step": 767,
      "step_time": 7.953771933913231
    },
    {
      "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,
      "entropy": 1.5365270614624023,
      "epoch": 0.367816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0793386772274971,
      "learning_rate": 3e-06,
      "loss": -0.0081,
      "step": 768,
      "step_time": 7.36125697940588
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.517578125,
      "completions/mean_terminated_length": 205.3196258544922,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.6882884820302329,
      "epoch": 0.3682950191570881,
      "fork/depth_mean": 0.21479304134845734,
      "fork/depth_std": 0.22361113131046295,
      "fork_head/loss": 2.728188991546631,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.08605211973190308,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8079339861869812,
      "learning_rate": 3e-06,
      "loss": 0.6841,
      "num_tokens": 20405415.0,
      "reward": 0.9838517904281616,
      "reward_std": 0.6303704977035522,
      "rewards/code_format_reward/mean": 0.734375,
      "rewards/code_format_reward/std": 0.38278767466545105,
      "rewards/code_reward/mean": 0.24947677552700043,
      "rewards/code_reward/std": 0.39034411311149597,
      "step": 769,
      "step_time": 582.3078459575772
    },
    {
      "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,
      "entropy": 1.7773654500643412,
      "epoch": 0.36877394636015326,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1558549702167511,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 770,
      "step_time": 8.015472322702408
    },
    {
      "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,
      "entropy": 1.6734158039093017,
      "epoch": 0.3692528735632184,
      "fork_head/loss": 2.7115581035614014,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.09053447842597961,
      "grad_norm": 0.5755575895309448,
      "learning_rate": 3e-06,
      "loss": 0.6785,
      "step": 771,
      "step_time": 7.961096256971359
    },
    {
      "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,
      "entropy": 1.7714084188143413,
      "epoch": 0.36973180076628354,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1492033153772354,
      "learning_rate": 3e-06,
      "loss": -0.0022,
      "step": 772,
      "step_time": 7.182436607778072
    },
    {
      "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,
      "entropy": 1.677430550257365,
      "epoch": 0.3702107279693487,
      "fork_head/loss": 2.6935064792633057,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.09459517896175385,
      "grad_norm": 0.6030600666999817,
      "learning_rate": 3e-06,
      "loss": 0.673,
      "step": 773,
      "step_time": 7.938923165202141
    },
    {
      "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,
      "entropy": 1.7808810750643411,
      "epoch": 0.3706896551724138,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16572372615337372,
      "learning_rate": 3e-06,
      "loss": -0.0039,
      "step": 774,
      "step_time": 7.629017099738121
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.6822120984395346,
      "epoch": 0.37116858237547895,
      "fork_head/loss": 2.6724586486816406,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.10145974904298782,
      "grad_norm": 0.6474586129188538,
      "learning_rate": 3e-06,
      "loss": 0.6658,
      "step": 775,
      "step_time": 7.313154838979244
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.7710177938143412,
      "epoch": 0.3716475095785441,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09272966533899307,
      "learning_rate": 3e-06,
      "loss": -0.0058,
      "step": 776,
      "step_time": 7.205166421830654
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 221.197265625,
      "completions/mean_terminated_length": 221.129150390625,
      "completions/min_length": 22.0,
      "completions/min_terminated_length": 22.0,
      "entropy": 1.9934895833333333,
      "epoch": 0.37212643678160917,
      "fork/depth_mean": 0.20095166563987732,
      "fork/depth_std": 0.18103522062301636,
      "fork_head/loss": 2.7288155555725098,
      "fork_head/num_positive": 12.96875,
      "fork_head/pred_top1_mass": 0.0691717192530632,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5345715284347534,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "num_tokens": 20614428.0,
      "reward": 1.0164823532104492,
      "reward_std": 0.6332305073738098,
      "rewards/code_format_reward/mean": 0.7373046875,
      "rewards/code_format_reward/std": 0.3764534294605255,
      "rewards/code_reward/mean": 0.27917760610580444,
      "rewards/code_reward/std": 0.3699555993080139,
      "step": 777,
      "step_time": 586.5397960543633
    },
    {
      "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,
      "entropy": 1.6795247395833333,
      "epoch": 0.3726053639846743,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10597233474254608,
      "learning_rate": 3e-06,
      "loss": -0.0005,
      "step": 778,
      "step_time": 8.084084086120129
    },
    {
      "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,
      "entropy": 1.9826388888888888,
      "epoch": 0.37308429118773945,
      "fork_head/loss": 2.7198843955993652,
      "fork_head/num_positive": 12.96875,
      "fork_head/pred_top1_mass": 0.06652506440877914,
      "grad_norm": 0.35930585861206055,
      "learning_rate": 3e-06,
      "loss": 0.6841,
      "step": 779,
      "step_time": 7.656016372144222
    },
    {
      "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,
      "entropy": 1.6691080729166667,
      "epoch": 0.3735632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10876832902431488,
      "learning_rate": 3e-06,
      "loss": -0.0006,
      "step": 780,
      "step_time": 7.3285281881690025
    },
    {
      "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,
      "entropy": 1.9900173611111112,
      "epoch": 0.3740421455938697,
      "fork_head/loss": 2.7124176025390625,
      "fork_head/num_positive": 12.96875,
      "fork_head/pred_top1_mass": 0.06333920359611511,
      "grad_norm": 0.3700961470603943,
      "learning_rate": 3e-06,
      "loss": 0.6816,
      "step": 781,
      "step_time": 7.477032333612442
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.6650390625,
      "epoch": 0.37452107279693486,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07948971539735794,
      "learning_rate": 3e-06,
      "loss": -0.0011,
      "step": 782,
      "step_time": 7.656495682895184
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05555555555555555,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.9973958333333333,
      "epoch": 0.375,
      "fork_head/loss": 2.7046542167663574,
      "fork_head/num_positive": 12.96875,
      "fork_head/pred_top1_mass": 0.06722826510667801,
      "grad_norm": 0.4440240263938904,
      "learning_rate": 3e-06,
      "loss": 0.679,
      "step": 783,
      "step_time": 7.383031666278839
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.67431640625,
      "epoch": 0.37547892720306514,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06229773908853531,
      "learning_rate": 3e-06,
      "loss": -0.0015,
      "step": 784,
      "step_time": 7.614772297441959
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.03030303120613098,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 212.55078125,
      "completions/mean_terminated_length": 212.55078125,
      "completions/min_length": 45.0,
      "completions/min_terminated_length": 45.0,
      "entropy": 1.1602228338068181,
      "epoch": 0.3759578544061303,
      "fork/depth_mean": 0.2233026623725891,
      "fork/depth_std": 0.19999560713768005,
      "fork_head/loss": 2.7313828468322754,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.06285910308361053,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6235535740852356,
      "learning_rate": 3e-06,
      "loss": 0.6845,
      "num_tokens": 20810822.0,
      "reward": 1.102421760559082,
      "reward_std": 0.6505752205848694,
      "rewards/code_format_reward/mean": 0.78515625,
      "rewards/code_format_reward/std": 0.3408496081829071,
      "rewards/code_reward/mean": 0.31726545095443726,
      "rewards/code_reward/std": 0.43852633237838745,
      "step": 785,
      "step_time": 496.5374985188246
    },
    {
      "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,
      "entropy": 1.3778921274038463,
      "epoch": 0.3764367816091954,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09807947278022766,
      "learning_rate": 3e-06,
      "loss": -0.0039,
      "step": 786,
      "step_time": 8.322511851787567
    },
    {
      "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,
      "entropy": 1.1593498533422297,
      "epoch": 0.37691570881226055,
      "fork_head/loss": 2.7197089195251465,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.06812302023172379,
      "grad_norm": 0.5562715530395508,
      "learning_rate": 3e-06,
      "loss": 0.6821,
      "step": 787,
      "step_time": 7.230255305767059
    },
    {
      "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,
      "entropy": 1.3814978966346154,
      "epoch": 0.3773946360153257,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09794700890779495,
      "learning_rate": 3e-06,
      "loss": -0.004,
      "step": 788,
      "step_time": 7.909790739417076
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.03030303120613098,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.147157614881342,
      "epoch": 0.37787356321839083,
      "fork_head/loss": 2.70320987701416,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.07153204828500748,
      "grad_norm": 0.512154221534729,
      "learning_rate": 3e-06,
      "loss": 0.6768,
      "step": 789,
      "step_time": 7.312155805528164
    },
    {
      "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,
      "entropy": 1.3741361177884615,
      "epoch": 0.3783524904214559,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08860759437084198,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 790,
      "step_time": 7.648989900946617
    },
    {
      "clip_ratio/high_max": 0.18181818181818182,
      "clip_ratio/high_mean": 0.06060606241226196,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06060606241226196,
      "entropy": 1.1538233973763206,
      "epoch": 0.37883141762452105,
      "fork_head/loss": 2.6847550868988037,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.09264470636844635,
      "grad_norm": 0.443055123090744,
      "learning_rate": 3e-06,
      "loss": 0.6707,
      "step": 791,
      "step_time": 7.358509749174118
    },
    {
      "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,
      "entropy": 1.3776667668269231,
      "epoch": 0.3793103448275862,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09169319272041321,
      "learning_rate": 3e-06,
      "loss": -0.0056,
      "step": 792,
      "step_time": 7.372601591050625
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.439453125,
      "completions/mean_terminated_length": 205.439453125,
      "completions/min_length": 23.0,
      "completions/min_terminated_length": 23.0,
      "entropy": 1.0991654829545454,
      "epoch": 0.3797892720306513,
      "fork/depth_mean": 0.19295844435691833,
      "fork/depth_std": 0.17778517305850983,
      "fork_head/loss": 2.7294840812683105,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.06459707766771317,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4761977791786194,
      "learning_rate": 3e-06,
      "loss": 0.6862,
      "num_tokens": 21002423.0,
      "reward": 1.0431039333343506,
      "reward_std": 0.6375479102134705,
      "rewards/code_format_reward/mean": 0.759765625,
      "rewards/code_format_reward/std": 0.35098734498023987,
      "rewards/code_reward/mean": 0.28333836793899536,
      "rewards/code_reward/std": 0.3949759602546692,
      "step": 793,
      "step_time": 483.916927754879
    },
    {
      "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,
      "entropy": 1.8249421252144709,
      "epoch": 0.38026819923371646,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11435260623693466,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 794,
      "step_time": 7.225110702216625
    },
    {
      "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,
      "entropy": 1.0782137784090908,
      "epoch": 0.3807471264367816,
      "fork_head/loss": 2.722360372543335,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07071362435817719,
      "grad_norm": 0.46653565764427185,
      "learning_rate": 3e-06,
      "loss": 0.6828,
      "step": 795,
      "step_time": 8.033819608390331
    },
    {
      "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,
      "entropy": 1.8352141247855291,
      "epoch": 0.38122605363984674,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13201694190502167,
      "learning_rate": 3e-06,
      "loss": -0.005,
      "step": 796,
      "step_time": 7.279679082334042
    },
    {
      "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,
      "entropy": 1.0684481534090908,
      "epoch": 0.3817049808429119,
      "fork_head/loss": 2.7099008560180664,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07285238057374954,
      "grad_norm": 0.42191892862319946,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 797,
      "step_time": 7.362039625644684
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.2222222222222222,
      "entropy": 1.8250868055555556,
      "epoch": 0.382183908045977,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.100193552672863,
      "learning_rate": 3e-06,
      "loss": -0.0053,
      "step": 798,
      "step_time": 7.492869898676872
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.0630326704545454,
      "epoch": 0.38266283524904215,
      "fork_head/loss": 2.6952056884765625,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.08028298616409302,
      "grad_norm": 0.38584035634994507,
      "learning_rate": 3e-06,
      "loss": 0.6741,
      "step": 799,
      "step_time": 7.29070757329464
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.8087384303410847,
      "epoch": 0.3831417624521073,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.114515520632267,
      "learning_rate": 3e-06,
      "loss": -0.0069,
      "step": 800,
      "step_time": 7.668492145836353
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 216.673828125,
      "completions/mean_terminated_length": 216.44204711914062,
      "completions/min_length": 77.0,
      "completions/min_terminated_length": 77.0,
      "entropy": 1.495600054661433,
      "epoch": 0.38362068965517243,
      "fork/depth_mean": 0.22177429497241974,
      "fork/depth_std": 0.19164343178272247,
      "fork_head/loss": 2.7157883644104004,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.08907759189605713,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4795312285423279,
      "learning_rate": 3e-06,
      "loss": 0.6826,
      "num_tokens": 21204000.0,
      "reward": 0.9752558469772339,
      "reward_std": 0.6160140037536621,
      "rewards/code_format_reward/mean": 0.7138671875,
      "rewards/code_format_reward/std": 0.3952544927597046,
      "rewards/code_reward/mean": 0.2613886594772339,
      "rewards/code_reward/std": 0.37938445806503296,
      "step": 801,
      "step_time": 497.7543801665306
    },
    {
      "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,
      "entropy": 1.2480912642045454,
      "epoch": 0.38409961685823757,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08684812486171722,
      "learning_rate": 3e-06,
      "loss": -0.0039,
      "step": 802,
      "step_time": 8.185460858047009
    },
    {
      "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,
      "entropy": 1.5012478331724803,
      "epoch": 0.3845785440613027,
      "fork_head/loss": 2.7068538665771484,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.09064877033233643,
      "grad_norm": 0.45305007696151733,
      "learning_rate": 3e-06,
      "loss": 0.6802,
      "step": 803,
      "step_time": 7.142401494085789
    },
    {
      "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,
      "entropy": 1.2411369518800215,
      "epoch": 0.3850574712643678,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0846947729587555,
      "learning_rate": 3e-06,
      "loss": -0.004,
      "step": 804,
      "step_time": 7.442957170307636
    },
    {
      "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,
      "entropy": 1.5028591652711232,
      "epoch": 0.38553639846743293,
      "fork_head/loss": 2.6929099559783936,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.09578577429056168,
      "grad_norm": 0.41928166151046753,
      "learning_rate": 3e-06,
      "loss": 0.6763,
      "step": 805,
      "step_time": 7.907518237829208
    },
    {
      "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,
      "entropy": 1.2377781759608875,
      "epoch": 0.38601532567049807,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08515917509794235,
      "learning_rate": 3e-06,
      "loss": -0.0046,
      "step": 806,
      "step_time": 7.3262916803359985
    },
    {
      "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,
      "entropy": 1.5033826927344005,
      "epoch": 0.3864942528735632,
      "fork_head/loss": 2.6772208213806152,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.11191599816083908,
      "grad_norm": 0.3882104158401489,
      "learning_rate": 3e-06,
      "loss": 0.6723,
      "step": 807,
      "step_time": 7.660202242434025
    },
    {
      "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,
      "entropy": 1.2327030030163852,
      "epoch": 0.38697318007662834,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08474381268024445,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 808,
      "step_time": 7.762344196438789
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.921875,
      "completions/mean_terminated_length": 219.8512725830078,
      "completions/min_length": 55.0,
      "completions/min_terminated_length": 55.0,
      "entropy": 1.3659446076913313,
      "epoch": 0.3874521072796935,
      "fork/depth_mean": 0.23137003183364868,
      "fork/depth_std": 0.23614810407161713,
      "fork_head/loss": 2.736767292022705,
      "fork_head/num_positive": 11.935483932495117,
      "fork_head/pred_top1_mass": 0.06521887332201004,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6703629493713379,
      "learning_rate": 3e-06,
      "loss": 0.6764,
      "num_tokens": 21401400.0,
      "reward": 1.026227355003357,
      "reward_std": 0.6224173903465271,
      "rewards/code_format_reward/mean": 0.76953125,
      "rewards/code_format_reward/std": 0.3629210889339447,
      "rewards/code_reward/mean": 0.25669610500335693,
      "rewards/code_reward/std": 0.3792003095149994,
      "step": 809,
      "step_time": 585.0133780762553
    },
    {
      "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,
      "entropy": 1.5307849730764116,
      "epoch": 0.3879310344827586,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.6387342810630798,
      "learning_rate": 3e-06,
      "loss": 0.0108,
      "step": 810,
      "step_time": 8.092165872454643
    },
    {
      "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,
      "entropy": 1.366181341084567,
      "epoch": 0.38840996168582376,
      "fork_head/loss": 2.72039794921875,
      "fork_head/num_positive": 11.935483932495117,
      "fork_head/pred_top1_mass": 0.06807070225477219,
      "grad_norm": 0.6230072975158691,
      "learning_rate": 3e-06,
      "loss": 0.6718,
      "step": 811,
      "step_time": 8.158597357571125
    },
    {
      "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,
      "entropy": 1.5697195870535714,
      "epoch": 0.3888888888888889,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.4868932068347931,
      "learning_rate": 3e-06,
      "loss": 0.0071,
      "step": 812,
      "step_time": 7.31486888974905
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.34289180690592,
      "epoch": 0.38936781609195403,
      "fork_head/loss": 2.6948013305664062,
      "fork_head/num_positive": 11.935483932495117,
      "fork_head/pred_top1_mass": 0.09938716888427734,
      "grad_norm": 0.555553674697876,
      "learning_rate": 3e-06,
      "loss": 0.6635,
      "step": 813,
      "step_time": 7.668551035225391
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.5987606985228402,
      "epoch": 0.38984674329501917,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15926599502563477,
      "learning_rate": 3e-06,
      "loss": 0.0059,
      "step": 814,
      "step_time": 7.655389226973057
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.3634736050258984,
      "epoch": 0.3903256704980843,
      "fork_head/loss": 2.6664175987243652,
      "fork_head/num_positive": 11.935483932495117,
      "fork_head/pred_top1_mass": 0.12197622656822205,
      "grad_norm": 0.5066251754760742,
      "learning_rate": 3e-06,
      "loss": 0.6564,
      "step": 815,
      "step_time": 7.345195218920708
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.6440313458442688,
      "epoch": 0.39080459770114945,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11836032569408417,
      "learning_rate": 3e-06,
      "loss": 0.005,
      "step": 816,
      "step_time": 7.925781950354576
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 215.001953125,
      "completions/mean_terminated_length": 215.001953125,
      "completions/min_length": 21.0,
      "completions/min_terminated_length": 21.0,
      "entropy": 1.0979817708333333,
      "epoch": 0.3912835249042146,
      "fork/depth_mean": 0.2335720658302307,
      "fork/depth_std": 0.21434886753559113,
      "fork_head/loss": 2.734794855117798,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.08334967494010925,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6975601315498352,
      "learning_rate": 3e-06,
      "loss": 0.682,
      "num_tokens": 21606105.0,
      "reward": 1.0150142908096313,
      "reward_std": 0.5804358124732971,
      "rewards/code_format_reward/mean": 0.7724609375,
      "rewards/code_format_reward/std": 0.3461889326572418,
      "rewards/code_reward/mean": 0.24255335330963135,
      "rewards/code_reward/std": 0.36016717553138733,
      "step": 817,
      "step_time": 502.3363419175148
    },
    {
      "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,
      "entropy": 1.5880533903837204,
      "epoch": 0.39176245210727967,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04345997795462608,
      "learning_rate": 3e-06,
      "loss": 0.0014,
      "step": 818,
      "step_time": 7.899125620722771
    },
    {
      "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,
      "entropy": 1.078125,
      "epoch": 0.3922413793103448,
      "fork_head/loss": 2.7189574241638184,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.08570631593465805,
      "grad_norm": 0.5604706406593323,
      "learning_rate": 3e-06,
      "loss": 0.6778,
      "step": 819,
      "step_time": 7.398917652666569
    },
    {
      "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,
      "entropy": 1.59130859375,
      "epoch": 0.39272030651340994,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.045036595314741135,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 820,
      "step_time": 7.349477373063564
    },
    {
      "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,
      "entropy": 1.0699869791666667,
      "epoch": 0.3931992337164751,
      "fork_head/loss": 2.699831485748291,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.09448228776454926,
      "grad_norm": 0.500640869140625,
      "learning_rate": 3e-06,
      "loss": 0.6728,
      "step": 821,
      "step_time": 7.12645248323679
    },
    {
      "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,
      "entropy": 1.5808919221162796,
      "epoch": 0.3936781609195402,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.045519351959228516,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 822,
      "step_time": 7.260529391467571
    },
    {
      "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,
      "entropy": 1.0696614583333333,
      "epoch": 0.39415708812260536,
      "fork_head/loss": 2.6779284477233887,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.1062241867184639,
      "grad_norm": 0.5018882751464844,
      "learning_rate": 3e-06,
      "loss": 0.6674,
      "step": 823,
      "step_time": 8.295032143592834
    },
    {
      "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,
      "entropy": 1.6118977814912796,
      "epoch": 0.3946360153256705,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.043360840529203415,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 824,
      "step_time": 7.280121557414532
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.055555557211240135,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.055555557211240135,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 218.90234375,
      "completions/mean_terminated_length": 218.90234375,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.661241312821706,
      "epoch": 0.39511494252873564,
      "fork/depth_mean": 0.22781458497047424,
      "fork/depth_std": 0.197799950838089,
      "fork_head/loss": 2.706392765045166,
      "fork_head/num_positive": 11.741935729980469,
      "fork_head/pred_top1_mass": 0.07639321684837341,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6806069612503052,
      "learning_rate": 3e-06,
      "loss": 0.6774,
      "num_tokens": 21817047.0,
      "reward": 0.933021068572998,
      "reward_std": 0.5522127747535706,
      "rewards/code_format_reward/mean": 0.7529296875,
      "rewards/code_format_reward/std": 0.35937434434890747,
      "rewards/code_reward/mean": 0.18009141087532043,
      "rewards/code_reward/std": 0.3158875107765198,
      "step": 825,
      "step_time": 534.6402083560824
    },
    {
      "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,
      "entropy": 1.638671875,
      "epoch": 0.3955938697318008,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.02796349488198757,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 826,
      "step_time": 7.3668771758675575
    },
    {
      "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,
      "entropy": 1.6722005208333333,
      "epoch": 0.3960727969348659,
      "fork_head/loss": 2.6943535804748535,
      "fork_head/num_positive": 11.741935729980469,
      "fork_head/pred_top1_mass": 0.08587805181741714,
      "grad_norm": 0.4877192974090576,
      "learning_rate": 3e-06,
      "loss": 0.6742,
      "step": 827,
      "step_time": 8.080149292945862
    },
    {
      "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,
      "entropy": 1.6201985776424408,
      "epoch": 0.39655172413793105,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.026742946356534958,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 828,
      "step_time": 7.426648497581482
    },
    {
      "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,
      "entropy": 1.7038845419883728,
      "epoch": 0.3970306513409962,
      "fork_head/loss": 2.6809651851654053,
      "fork_head/num_positive": 11.741935729980469,
      "fork_head/pred_top1_mass": 0.10146433115005493,
      "grad_norm": 0.47481346130371094,
      "learning_rate": 3e-06,
      "loss": 0.6703,
      "step": 829,
      "step_time": 7.293013751506805
    },
    {
      "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,
      "entropy": 1.5884602963924408,
      "epoch": 0.3975095785440613,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.028829781338572502,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 830,
      "step_time": 7.7029403150081635
    },
    {
      "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,
      "entropy": 1.6928168336550395,
      "epoch": 0.39798850574712646,
      "fork_head/loss": 2.6663949489593506,
      "fork_head/num_positive": 11.741935729980469,
      "fork_head/pred_top1_mass": 0.10995383560657501,
      "grad_norm": 0.514119565486908,
      "learning_rate": 3e-06,
      "loss": 0.6668,
      "step": 831,
      "step_time": 7.24723669141531
    },
    {
      "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,
      "entropy": 1.6099446713924408,
      "epoch": 0.39846743295019155,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.026035258546471596,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 832,
      "step_time": 8.072384312748909
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 218.537109375,
      "completions/mean_terminated_length": 218.537109375,
      "completions/min_length": 59.0,
      "completions/min_terminated_length": 59.0,
      "entropy": 1.600632450410298,
      "epoch": 0.3989463601532567,
      "fork/depth_mean": 0.227511465549469,
      "fork/depth_std": 0.20463356375694275,
      "fork_head/loss": 2.7376949787139893,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.07502467185258865,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 1.0859920978546143,
      "learning_rate": 3e-06,
      "loss": 0.6868,
      "num_tokens": 22017162.0,
      "reward": 1.005076289176941,
      "reward_std": 0.6041010022163391,
      "rewards/code_format_reward/mean": 0.76953125,
      "rewards/code_format_reward/std": 0.36426663398742676,
      "rewards/code_reward/mean": 0.2355450689792633,
      "rewards/code_reward/std": 0.3789195120334625,
      "step": 833,
      "step_time": 539.8341298475862
    },
    {
      "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,
      "entropy": 1.3021240234375,
      "epoch": 0.3994252873563218,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09701802581548691,
      "learning_rate": 3e-06,
      "loss": -0.0022,
      "step": 834,
      "step_time": 7.667466893792152
    },
    {
      "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,
      "entropy": 1.5936918726989202,
      "epoch": 0.39990421455938696,
      "fork_head/loss": 2.7126667499542236,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.08139089494943619,
      "grad_norm": 0.7876355051994324,
      "learning_rate": 3e-06,
      "loss": 0.6803,
      "step": 835,
      "step_time": 8.180159464478493
    },
    {
      "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,
      "entropy": 1.2955322265625,
      "epoch": 0.4003831417624521,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11001239717006683,
      "learning_rate": 3e-06,
      "loss": -0.0028,
      "step": 836,
      "step_time": 7.840687923133373
    },
    {
      "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,
      "entropy": 1.6078985333442688,
      "epoch": 0.40086206896551724,
      "fork_head/loss": 2.682901620864868,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.08155019581317902,
      "grad_norm": 0.667018473148346,
      "learning_rate": 3e-06,
      "loss": 0.6718,
      "step": 837,
      "step_time": 7.459107257425785
    },
    {
      "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,
      "entropy": 1.2772216697533925,
      "epoch": 0.4013409961685824,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1050325334072113,
      "learning_rate": 3e-06,
      "loss": -0.0034,
      "step": 838,
      "step_time": 7.500717349350452
    },
    {
      "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,
      "entropy": 1.6067708390099662,
      "epoch": 0.4018199233716475,
      "fork_head/loss": 2.6537420749664307,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.09724822640419006,
      "grad_norm": 0.750453770160675,
      "learning_rate": 3e-06,
      "loss": 0.6633,
      "step": 839,
      "step_time": 7.966572038829327
    },
    {
      "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,
      "entropy": 1.278700093428294,
      "epoch": 0.40229885057471265,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09935685247182846,
      "learning_rate": 3e-06,
      "loss": -0.0036,
      "step": 840,
      "step_time": 7.1078865602612495
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 216.974609375,
      "completions/mean_terminated_length": 216.974609375,
      "completions/min_length": 10.0,
      "completions/min_terminated_length": 10.0,
      "entropy": 1.583642578125,
      "epoch": 0.4027777777777778,
      "fork/depth_mean": 0.2253521978855133,
      "fork/depth_std": 0.2175724059343338,
      "fork_head/loss": 2.72943377494812,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.06824647635221481,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6088598966598511,
      "learning_rate": 3e-06,
      "loss": 0.6857,
      "num_tokens": 22232141.0,
      "reward": 0.9516414403915405,
      "reward_std": 0.5929017066955566,
      "rewards/code_format_reward/mean": 0.736328125,
      "rewards/code_format_reward/std": 0.3658737540245056,
      "rewards/code_reward/mean": 0.21531328558921814,
      "rewards/code_reward/std": 0.35599663853645325,
      "step": 841,
      "step_time": 575.541994355619
    },
    {
      "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,
      "entropy": 1.3717176616191864,
      "epoch": 0.40325670498084293,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09219923615455627,
      "learning_rate": 3e-06,
      "loss": -0.0081,
      "step": 842,
      "step_time": 7.829875476658344
    },
    {
      "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,
      "entropy": 1.587255859375,
      "epoch": 0.40373563218390807,
      "fork_head/loss": 2.7184698581695557,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.06932324916124344,
      "grad_norm": 0.4329098165035248,
      "learning_rate": 3e-06,
      "loss": 0.6835,
      "step": 843,
      "step_time": 8.565485998988152
    },
    {
      "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,
      "entropy": 1.367540140946706,
      "epoch": 0.4042145593869732,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09003652632236481,
      "learning_rate": 3e-06,
      "loss": -0.0082,
      "step": 844,
      "step_time": 7.253651402890682
    },
    {
      "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,
      "entropy": 1.59794921875,
      "epoch": 0.4046934865900383,
      "fork_head/loss": 2.7078683376312256,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.06846199184656143,
      "grad_norm": 0.44974687695503235,
      "learning_rate": 3e-06,
      "loss": 0.6802,
      "step": 845,
      "step_time": 8.044233553111553
    },
    {
      "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,
      "entropy": 1.3687608540058136,
      "epoch": 0.4051724137931034,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08839334547519684,
      "learning_rate": 3e-06,
      "loss": -0.0086,
      "step": 846,
      "step_time": 7.466761261224747
    },
    {
      "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,
      "entropy": 1.59677734375,
      "epoch": 0.40565134099616856,
      "fork_head/loss": 2.694694995880127,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.07886002212762833,
      "grad_norm": 0.4533076882362366,
      "learning_rate": 3e-06,
      "loss": 0.6767,
      "step": 847,
      "step_time": 7.26690585911274
    },
    {
      "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,
      "entropy": 1.3633355100949605,
      "epoch": 0.4061302681992337,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0883694589138031,
      "learning_rate": 3e-06,
      "loss": -0.0091,
      "step": 848,
      "step_time": 8.254733830690384
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 216.916015625,
      "completions/mean_terminated_length": 216.8395233154297,
      "completions/min_length": 79.0,
      "completions/min_terminated_length": 79.0,
      "entropy": 1.4343039772727273,
      "epoch": 0.40660919540229884,
      "fork/depth_mean": 0.2553071677684784,
      "fork/depth_std": 0.21470613777637482,
      "fork_head/loss": 2.7370269298553467,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.05667206645011902,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.528660237789154,
      "learning_rate": 3e-06,
      "loss": 0.682,
      "num_tokens": 22433138.0,
      "reward": 1.0622379779815674,
      "reward_std": 0.6032245755195618,
      "rewards/code_format_reward/mean": 0.7822265625,
      "rewards/code_format_reward/std": 0.34113961458206177,
      "rewards/code_reward/mean": 0.2800114154815674,
      "rewards/code_reward/std": 0.37479764223098755,
      "step": 849,
      "step_time": 686.2522453069687
    },
    {
      "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,
      "entropy": 1.229736328125,
      "epoch": 0.407088122605364,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10083392262458801,
      "learning_rate": 3e-06,
      "loss": 0.0019,
      "step": 850,
      "step_time": 7.331092961132526
    },
    {
      "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,
      "entropy": 1.4318773638118396,
      "epoch": 0.4075670498084291,
      "fork_head/loss": 2.72646164894104,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.05670548230409622,
      "grad_norm": 0.48066648840904236,
      "learning_rate": 3e-06,
      "loss": 0.679,
      "step": 851,
      "step_time": 7.288361959159374
    },
    {
      "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,
      "entropy": 1.2276340027650197,
      "epoch": 0.40804597701149425,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.102774977684021,
      "learning_rate": 3e-06,
      "loss": 0.0019,
      "step": 852,
      "step_time": 8.112967684864998
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.42791194265539,
      "epoch": 0.4085249042145594,
      "fork_head/loss": 2.7115836143493652,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.0640452653169632,
      "grad_norm": 0.4306630790233612,
      "learning_rate": 3e-06,
      "loss": 0.6744,
      "step": 853,
      "step_time": 6.918776705861092
    },
    {
      "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,
      "entropy": 1.2454833984375,
      "epoch": 0.40900383141762453,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10184211283922195,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 854,
      "step_time": 7.550267890095711
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.4400449774482034,
      "epoch": 0.40948275862068967,
      "fork_head/loss": 2.6954116821289062,
      "fork_head/num_positive": 12.0625,
      "fork_head/pred_top1_mass": 0.07143775373697281,
      "grad_norm": 0.39049461483955383,
      "learning_rate": 3e-06,
      "loss": 0.6702,
      "step": 855,
      "step_time": 7.584506064653397
    },
    {
      "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,
      "entropy": 1.2611490885416667,
      "epoch": 0.4099616858237548,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08579061925411224,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 856,
      "step_time": 6.960254341363907
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 214.685546875,
      "completions/mean_terminated_length": 214.6046905517578,
      "completions/min_length": 35.0,
      "completions/min_terminated_length": 35.0,
      "entropy": 1.3972981827599662,
      "epoch": 0.41044061302681994,
      "fork/depth_mean": 0.23677058517932892,
      "fork/depth_std": 0.20829232037067413,
      "fork_head/loss": 2.74336576461792,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06908975541591644,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.56536865234375,
      "learning_rate": 3e-06,
      "loss": 0.6761,
      "num_tokens": 22641297.0,
      "reward": 0.9779319763183594,
      "reward_std": 0.5991506576538086,
      "rewards/code_format_reward/mean": 0.7587890625,
      "rewards/code_format_reward/std": 0.3700120449066162,
      "rewards/code_reward/mean": 0.219142884016037,
      "rewards/code_reward/std": 0.362737238407135,
      "step": 857,
      "step_time": 518.915077701211
    },
    {
      "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,
      "entropy": 1.515640499336379,
      "epoch": 0.4109195402298851,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19292475283145905,
      "learning_rate": 3e-06,
      "loss": 0.0112,
      "step": 858,
      "step_time": 7.698102071881294
    },
    {
      "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,
      "entropy": 1.4097377232142858,
      "epoch": 0.41139846743295017,
      "fork_head/loss": 2.7338156700134277,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07016435265541077,
      "grad_norm": 0.5325244069099426,
      "learning_rate": 3e-06,
      "loss": 0.6725,
      "step": 859,
      "step_time": 8.716497421264648
    },
    {
      "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,
      "entropy": 1.5345071447747094,
      "epoch": 0.4118773946360153,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16556471586227417,
      "learning_rate": 3e-06,
      "loss": 0.0095,
      "step": 860,
      "step_time": 7.377664253115654
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.07142857142857142,
      "clip_ratio/low_mean": 0.08333333368812289,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.15476190511669433,
      "entropy": 1.4000185983521598,
      "epoch": 0.41235632183908044,
      "fork_head/loss": 2.7182559967041016,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.09113148599863052,
      "grad_norm": 0.47192060947418213,
      "learning_rate": 3e-06,
      "loss": 0.6666,
      "step": 861,
      "step_time": 7.894936881959438
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08928571428571429,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.08928571428571429,
      "entropy": 1.5726299030440194,
      "epoch": 0.4128352490421456,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13909564912319183,
      "learning_rate": 3e-06,
      "loss": 0.0078,
      "step": 862,
      "step_time": 7.511467427015305
    },
    {
      "clip_ratio/high_max": 0.21428571428571427,
      "clip_ratio/high_mean": 0.17857142857142858,
      "clip_ratio/low_mean": 0.08333333368812289,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.26190476225955145,
      "entropy": 1.38232421875,
      "epoch": 0.4133141762452107,
      "fork_head/loss": 2.698996067047119,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.09755697846412659,
      "grad_norm": 0.4218161702156067,
      "learning_rate": 3e-06,
      "loss": 0.6614,
      "step": 863,
      "step_time": 7.139984257519245
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.18452381023338862,
      "clip_ratio/low_min": 0.14285714285714285,
      "clip_ratio/region_mean": 0.18452381023338862,
      "entropy": 1.5918356329202652,
      "epoch": 0.41379310344827586,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14809979498386383,
      "learning_rate": 3e-06,
      "loss": 0.0069,
      "step": 864,
      "step_time": 8.151504628360271
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 210.884765625,
      "completions/mean_terminated_length": 210.79647827148438,
      "completions/min_length": 68.0,
      "completions/min_terminated_length": 68.0,
      "entropy": 1.6064453125,
      "epoch": 0.414272030651341,
      "fork/depth_mean": 0.2049277275800705,
      "fork/depth_std": 0.1837330311536789,
      "fork_head/loss": 2.730912685394287,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.06514078378677368,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5300213694572449,
      "learning_rate": 3e-06,
      "loss": 0.6814,
      "num_tokens": 22845238.0,
      "reward": 0.9887240529060364,
      "reward_std": 0.6763169169425964,
      "rewards/code_format_reward/mean": 0.7216796875,
      "rewards/code_format_reward/std": 0.39712604880332947,
      "rewards/code_reward/mean": 0.2670443654060364,
      "rewards/code_reward/std": 0.3890487253665924,
      "step": 865,
      "step_time": 549.2309124693274
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.3088009032336148,
      "epoch": 0.41475095785440613,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07669182866811752,
      "learning_rate": 3e-06,
      "loss": 0.0019,
      "step": 866,
      "step_time": 8.032601557672024
    },
    {
      "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,
      "entropy": 1.602783203125,
      "epoch": 0.41522988505747127,
      "fork_head/loss": 2.720888614654541,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.07542794942855835,
      "grad_norm": 0.47012951970100403,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "step": 867,
      "step_time": 7.445928677916527
    },
    {
      "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,
      "entropy": 1.3166134032336148,
      "epoch": 0.4157088122605364,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08537830412387848,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 868,
      "step_time": 7.856789171695709
    },
    {
      "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,
      "entropy": 1.58984375,
      "epoch": 0.41618773946360155,
      "fork_head/loss": 2.7072079181671143,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.07549908757209778,
      "grad_norm": 0.43056562542915344,
      "learning_rate": 3e-06,
      "loss": 0.6746,
      "step": 869,
      "step_time": 7.193701192736626
    },
    {
      "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,
      "entropy": 1.313446966084567,
      "epoch": 0.4166666666666667,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0864916443824768,
      "learning_rate": 3e-06,
      "loss": 0.0011,
      "step": 870,
      "step_time": 7.623782359063625
    },
    {
      "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,
      "entropy": 1.560546875,
      "epoch": 0.4171455938697318,
      "fork_head/loss": 2.6912388801574707,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.08114872872829437,
      "grad_norm": 0.42472219467163086,
      "learning_rate": 3e-06,
      "loss": 0.6698,
      "step": 871,
      "step_time": 7.5718784481287
    },
    {
      "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,
      "entropy": 1.3321940140290693,
      "epoch": 0.41762452107279696,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08324994146823883,
      "learning_rate": 3e-06,
      "loss": -0.0005,
      "step": 872,
      "step_time": 7.425164394080639
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 208.74609375,
      "completions/mean_terminated_length": 208.74609375,
      "completions/min_length": 42.0,
      "completions/min_terminated_length": 42.0,
      "entropy": 1.4240056818181819,
      "epoch": 0.41810344827586204,
      "fork/depth_mean": 0.2009989470243454,
      "fork/depth_std": 0.18707165122032166,
      "fork_head/loss": 2.720222234725952,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06374296545982361,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4585981070995331,
      "learning_rate": 3e-06,
      "loss": 0.6864,
      "num_tokens": 23051668.0,
      "reward": 0.966554582118988,
      "reward_std": 0.5718360543251038,
      "rewards/code_format_reward/mean": 0.7626953125,
      "rewards/code_format_reward/std": 0.3452713191509247,
      "rewards/code_reward/mean": 0.20385931432247162,
      "rewards/code_reward/std": 0.3522414267063141,
      "step": 873,
      "step_time": 840.3227040097117
    },
    {
      "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,
      "entropy": 1.4143629807692308,
      "epoch": 0.4185823754789272,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05950493365526199,
      "learning_rate": 3e-06,
      "loss": -0.0062,
      "step": 874,
      "step_time": 7.646500281989574
    },
    {
      "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,
      "entropy": 1.4275568181818181,
      "epoch": 0.4190613026819923,
      "fork_head/loss": 2.713116407394409,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06900843232870102,
      "grad_norm": 0.43479084968566895,
      "learning_rate": 3e-06,
      "loss": 0.6845,
      "step": 875,
      "step_time": 8.11783430725336
    },
    {
      "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,
      "entropy": 1.4148637789946337,
      "epoch": 0.41954022988505746,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.058215122669935226,
      "learning_rate": 3e-06,
      "loss": -0.0057,
      "step": 876,
      "step_time": 7.71786417812109
    },
    {
      "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,
      "entropy": 1.4361683238636365,
      "epoch": 0.4200191570881226,
      "fork_head/loss": 2.7011635303497314,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07185521721839905,
      "grad_norm": 0.39532312750816345,
      "learning_rate": 3e-06,
      "loss": 0.6808,
      "step": 877,
      "step_time": 7.277973063290119
    },
    {
      "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,
      "entropy": 1.4002904616869414,
      "epoch": 0.42049808429118773,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06135457381606102,
      "learning_rate": 3e-06,
      "loss": -0.0062,
      "step": 878,
      "step_time": 7.456381306052208
    },
    {
      "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,
      "entropy": 1.4250710227272727,
      "epoch": 0.4209770114942529,
      "fork_head/loss": 2.687469005584717,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07637306302785873,
      "grad_norm": 0.3541790544986725,
      "learning_rate": 3e-06,
      "loss": 0.6771,
      "step": 879,
      "step_time": 7.8526818081736565
    },
    {
      "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,
      "entropy": 1.409204730620751,
      "epoch": 0.421455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05658218637108803,
      "learning_rate": 3e-06,
      "loss": -0.0067,
      "step": 880,
      "step_time": 7.410332158207893
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 210.4375,
      "completions/mean_terminated_length": 210.4375,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.751171875,
      "epoch": 0.42193486590038315,
      "fork/depth_mean": 0.2133454829454422,
      "fork/depth_std": 0.1930825263261795,
      "fork_head/loss": 2.7267465591430664,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.07289819419384003,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5099272727966309,
      "learning_rate": 3e-06,
      "loss": 0.6903,
      "num_tokens": 23260484.0,
      "reward": 0.9592344760894775,
      "reward_std": 0.6667659878730774,
      "rewards/code_format_reward/mean": 0.69921875,
      "rewards/code_format_reward/std": 0.397348552942276,
      "rewards/code_reward/mean": 0.2600157558917999,
      "rewards/code_reward/std": 0.3669685125350952,
      "step": 881,
      "step_time": 494.737622641027
    },
    {
      "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,
      "entropy": 1.3561197916666667,
      "epoch": 0.4224137931034483,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05938082933425903,
      "learning_rate": 3e-06,
      "loss": -0.0091,
      "step": 882,
      "step_time": 7.579040862619877
    },
    {
      "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,
      "entropy": 1.7345703125,
      "epoch": 0.4228927203065134,
      "fork_head/loss": 2.7174088954925537,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.07473781704902649,
      "grad_norm": 0.46027672290802,
      "learning_rate": 3e-06,
      "loss": 0.6883,
      "step": 883,
      "step_time": 7.517536126077175
    },
    {
      "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,
      "entropy": 1.357421875,
      "epoch": 0.42337164750957856,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05611221119761467,
      "learning_rate": 3e-06,
      "loss": -0.0093,
      "step": 884,
      "step_time": 7.286620654165745
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.76953125,
      "epoch": 0.4238505747126437,
      "fork_head/loss": 2.70444655418396,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.08473888784646988,
      "grad_norm": 0.39447447657585144,
      "learning_rate": 3e-06,
      "loss": 0.6842,
      "step": 885,
      "step_time": 7.370326362550259
    },
    {
      "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,
      "entropy": 1.3509114583333333,
      "epoch": 0.42432950191570884,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06119067221879959,
      "learning_rate": 3e-06,
      "loss": -0.0093,
      "step": 886,
      "step_time": 7.6275840401649475
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.80234375,
      "epoch": 0.4248084291187739,
      "fork_head/loss": 2.690364360809326,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.0926831066608429,
      "grad_norm": 0.3652031123638153,
      "learning_rate": 3e-06,
      "loss": 0.6804,
      "step": 887,
      "step_time": 7.551897056400776
    },
    {
      "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,
      "entropy": 1.3374837239583333,
      "epoch": 0.42528735632183906,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06247921660542488,
      "learning_rate": 3e-06,
      "loss": -0.01,
      "step": 888,
      "step_time": 7.052240639925003
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 221.55859375,
      "completions/mean_terminated_length": 221.4235382080078,
      "completions/min_length": 53.0,
      "completions/min_terminated_length": 53.0,
      "entropy": 1.4580078125,
      "epoch": 0.4257662835249042,
      "fork/depth_mean": 0.2567891776561737,
      "fork/depth_std": 0.20270773768424988,
      "fork_head/loss": 2.7310338020324707,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.08011868596076965,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7464421987533569,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "num_tokens": 23466274.0,
      "reward": 1.011319875717163,
      "reward_std": 0.5807861089706421,
      "rewards/code_format_reward/mean": 0.7744140625,
      "rewards/code_format_reward/std": 0.3612730801105499,
      "rewards/code_reward/mean": 0.2369057536125183,
      "rewards/code_reward/std": 0.35081973671913147,
      "step": 889,
      "step_time": 520.7159244567156
    },
    {
      "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,
      "entropy": 1.1171875,
      "epoch": 0.42624521072796934,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.02064836025238037,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 890,
      "step_time": 7.563564985990524
    },
    {
      "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,
      "entropy": 1.4423828125,
      "epoch": 0.4267241379310345,
      "fork_head/loss": 2.71683406829834,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.08097118139266968,
      "grad_norm": 0.601705014705658,
      "learning_rate": 3e-06,
      "loss": 0.6771,
      "step": 891,
      "step_time": 8.282874636352062
    },
    {
      "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,
      "entropy": 1.10546875,
      "epoch": 0.4272030651340996,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.02015606127679348,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 892,
      "step_time": 7.287173978984356
    },
    {
      "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,
      "entropy": 1.453125,
      "epoch": 0.42768199233716475,
      "fork_head/loss": 2.6976118087768555,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.07959045469760895,
      "grad_norm": 0.41438183188438416,
      "learning_rate": 3e-06,
      "loss": 0.6722,
      "step": 893,
      "step_time": 7.256328374147415
    },
    {
      "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,
      "entropy": 1.1494140625,
      "epoch": 0.4281609195402299,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.019574927166104317,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 894,
      "step_time": 8.079481095075607
    },
    {
      "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,
      "entropy": 1.4765625,
      "epoch": 0.428639846743295,
      "fork_head/loss": 2.6828103065490723,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.07961405813694,
      "grad_norm": 0.3831006586551666,
      "learning_rate": 3e-06,
      "loss": 0.6685,
      "step": 895,
      "step_time": 7.305093303322792
    },
    {
      "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,
      "entropy": 1.13623046875,
      "epoch": 0.42911877394636017,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.019288931041955948,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 896,
      "step_time": 7.3265747502446175
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 214.72265625,
      "completions/mean_terminated_length": 214.72265625,
      "completions/min_length": 47.0,
      "completions/min_terminated_length": 47.0,
      "entropy": 0.685546875,
      "epoch": 0.4295977011494253,
      "fork/depth_mean": 0.20287586748600006,
      "fork/depth_std": 0.18530184030532837,
      "fork_head/loss": 2.7277939319610596,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.051262255758047104,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5721415877342224,
      "learning_rate": 3e-06,
      "loss": 0.6813,
      "num_tokens": 23664324.0,
      "reward": 0.9446200728416443,
      "reward_std": 0.5949184894561768,
      "rewards/code_format_reward/mean": 0.755859375,
      "rewards/code_format_reward/std": 0.3830621838569641,
      "rewards/code_reward/mean": 0.1887606680393219,
      "rewards/code_reward/std": 0.3391883075237274,
      "step": 897,
      "step_time": 542.6680956035852
    },
    {
      "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,
      "entropy": 1.265625,
      "epoch": 0.43007662835249044,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07600141316652298,
      "learning_rate": 3e-06,
      "loss": 0.0007,
      "step": 898,
      "step_time": 8.043963722884655
    },
    {
      "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,
      "entropy": 0.697265625,
      "epoch": 0.4305555555555556,
      "fork_head/loss": 2.7163147926330566,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.05955871194601059,
      "grad_norm": 0.5111100673675537,
      "learning_rate": 3e-06,
      "loss": 0.6786,
      "step": 899,
      "step_time": 7.119223415851593
    },
    {
      "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,
      "entropy": 1.2875,
      "epoch": 0.43103448275862066,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05437517166137695,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 900,
      "step_time": 7.542817369103432
    },
    {
      "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,
      "entropy": 0.689453125,
      "epoch": 0.4315134099616858,
      "fork_head/loss": 2.7006947994232178,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.07200596481561661,
      "grad_norm": 0.43386703729629517,
      "learning_rate": 3e-06,
      "loss": 0.6746,
      "step": 901,
      "step_time": 7.633872479200363
    },
    {
      "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,
      "entropy": 1.270703125,
      "epoch": 0.43199233716475094,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0629456639289856,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 902,
      "step_time": 7.323989287018776
    },
    {
      "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,
      "entropy": 0.6962890625,
      "epoch": 0.4324712643678161,
      "fork_head/loss": 2.6844310760498047,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.08294085413217545,
      "grad_norm": 0.38405582308769226,
      "learning_rate": 3e-06,
      "loss": 0.6705,
      "step": 903,
      "step_time": 7.79639694839716
    },
    {
      "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,
      "entropy": 1.276953125,
      "epoch": 0.4329501915708812,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.054111871868371964,
      "learning_rate": 3e-06,
      "loss": -0.0004,
      "step": 904,
      "step_time": 7.566027760505676
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0078125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 202.716796875,
      "completions/mean_terminated_length": 202.2972412109375,
      "completions/min_length": 53.0,
      "completions/min_terminated_length": 53.0,
      "entropy": 1.6860795454545454,
      "epoch": 0.43342911877394635,
      "fork/depth_mean": 0.24004942178726196,
      "fork/depth_std": 0.18974576890468597,
      "fork_head/loss": 2.7406163215637207,
      "fork_head/num_positive": 11.612903594970703,
      "fork_head/pred_top1_mass": 0.06908532977104187,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5148220062255859,
      "learning_rate": 3e-06,
      "loss": 0.6781,
      "num_tokens": 23861987.0,
      "reward": 1.0414643287658691,
      "reward_std": 0.6117954850196838,
      "rewards/code_format_reward/mean": 0.7646484375,
      "rewards/code_format_reward/std": 0.35635173320770264,
      "rewards/code_reward/mean": 0.27681589126586914,
      "rewards/code_reward/std": 0.40100762248039246,
      "step": 905,
      "step_time": 469.3864384666085
    },
    {
      "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,
      "entropy": 1.4037977457046509,
      "epoch": 0.4339080459770115,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21869385242462158,
      "learning_rate": 3e-06,
      "loss": 0.0083,
      "step": 906,
      "step_time": 8.08532240241766
    },
    {
      "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,
      "entropy": 1.6829723011363635,
      "epoch": 0.43438697318007663,
      "fork_head/loss": 2.733804702758789,
      "fork_head/num_positive": 11.612903594970703,
      "fork_head/pred_top1_mass": 0.06880516558885574,
      "grad_norm": 0.5038066506385803,
      "learning_rate": 3e-06,
      "loss": 0.6762,
      "step": 907,
      "step_time": 8.09018349647522
    },
    {
      "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,
      "entropy": 1.4017795165379843,
      "epoch": 0.43486590038314177,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1887073516845703,
      "learning_rate": 3e-06,
      "loss": 0.008,
      "step": 908,
      "step_time": 7.479303307831287
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.69921875,
      "epoch": 0.4353448275862069,
      "fork_head/loss": 2.720924139022827,
      "fork_head/num_positive": 11.612903594970703,
      "fork_head/pred_top1_mass": 0.06843646615743637,
      "grad_norm": 0.42365357279777527,
      "learning_rate": 3e-06,
      "loss": 0.6722,
      "step": 909,
      "step_time": 8.309599474072456
    },
    {
      "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,
      "entropy": 1.4086371501286825,
      "epoch": 0.43582375478927204,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21561133861541748,
      "learning_rate": 3e-06,
      "loss": 0.0068,
      "step": 910,
      "step_time": 7.3707460090518
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.7041015625,
      "epoch": 0.4363026819923372,
      "fork_head/loss": 2.705756664276123,
      "fork_head/num_positive": 11.612903594970703,
      "fork_head/pred_top1_mass": 0.071506068110466,
      "grad_norm": 0.37056583166122437,
      "learning_rate": 3e-06,
      "loss": 0.6683,
      "step": 911,
      "step_time": 7.445036336779594
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.4224609375,
      "epoch": 0.4367816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.28348052501678467,
      "learning_rate": 3e-06,
      "loss": 0.005,
      "step": 912,
      "step_time": 7.83700617402792
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 211.79296875,
      "completions/mean_terminated_length": 211.79296875,
      "completions/min_length": 9.0,
      "completions/min_terminated_length": 9.0,
      "entropy": 1.182567101258498,
      "epoch": 0.43726053639846746,
      "fork/depth_mean": 0.2532667815685272,
      "fork/depth_std": 0.1995331197977066,
      "fork_head/loss": 2.7214198112487793,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.07552462071180344,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7119666337966919,
      "learning_rate": 3e-06,
      "loss": 0.6793,
      "num_tokens": 24061865.0,
      "reward": 1.0568121671676636,
      "reward_std": 0.568919837474823,
      "rewards/code_format_reward/mean": 0.8095703125,
      "rewards/code_format_reward/std": 0.31956154108047485,
      "rewards/code_reward/mean": 0.24724188446998596,
      "rewards/code_reward/std": 0.361462265253067,
      "step": 913,
      "step_time": 491.7043771445751
    },
    {
      "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,
      "entropy": 1.5191169543699785,
      "epoch": 0.43773946360153254,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07493150979280472,
      "learning_rate": 3e-06,
      "loss": 0.0005,
      "step": 914,
      "step_time": 8.25861606746912
    },
    {
      "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,
      "entropy": 1.1833183031815748,
      "epoch": 0.4382183908045977,
      "fork_head/loss": 2.708681344985962,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.09699176251888275,
      "grad_norm": 0.5904754996299744,
      "learning_rate": 3e-06,
      "loss": 0.6753,
      "step": 915,
      "step_time": 7.673110380768776
    },
    {
      "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,
      "entropy": 1.521602749824524,
      "epoch": 0.4386973180076628,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07288707792758942,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 916,
      "step_time": 7.577087618410587
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.179387010060824,
      "epoch": 0.43917624521072796,
      "fork_head/loss": 2.692418098449707,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.09393495321273804,
      "grad_norm": 0.4171706736087799,
      "learning_rate": 3e-06,
      "loss": 0.6696,
      "step": 917,
      "step_time": 7.426110848784447
    },
    {
      "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,
      "entropy": 1.5218394886363635,
      "epoch": 0.4396551724137931,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07643019407987595,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 918,
      "step_time": 7.369853720068932
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.1787109375,
      "epoch": 0.44013409961685823,
      "fork_head/loss": 2.67793607711792,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.10499553382396698,
      "grad_norm": 0.42846906185150146,
      "learning_rate": 3e-06,
      "loss": 0.6658,
      "step": 919,
      "step_time": 7.931741468608379
    },
    {
      "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,
      "entropy": 1.526396783915433,
      "epoch": 0.44061302681992337,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07421018183231354,
      "learning_rate": 3e-06,
      "loss": 0.0,
      "step": 920,
      "step_time": 7.092398777604103
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 207.326171875,
      "completions/mean_terminated_length": 207.326171875,
      "completions/min_length": 65.0,
      "completions/min_terminated_length": 65.0,
      "entropy": 1.3062789373927646,
      "epoch": 0.4410919540229885,
      "fork/depth_mean": 0.21670089662075043,
      "fork/depth_std": 0.19596423208713531,
      "fork_head/loss": 2.737187623977661,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.059676989912986755,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4943302571773529,
      "learning_rate": 3e-06,
      "loss": 0.6862,
      "num_tokens": 24255856.0,
      "reward": 0.9937185049057007,
      "reward_std": 0.6204805970191956,
      "rewards/code_format_reward/mean": 0.7548828125,
      "rewards/code_format_reward/std": 0.3753349184989929,
      "rewards/code_reward/mean": 0.2388356626033783,
      "rewards/code_reward/std": 0.367684930562973,
      "step": 921,
      "step_time": 530.3014232069254
    },
    {
      "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,
      "entropy": 1.4778645833333333,
      "epoch": 0.44157088122605365,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0662485808134079,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 922,
      "step_time": 7.646176680922508
    },
    {
      "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,
      "entropy": 1.2953920695516798,
      "epoch": 0.4420498084291188,
      "fork_head/loss": 2.729830265045166,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06246616318821907,
      "grad_norm": 0.4508437216281891,
      "learning_rate": 3e-06,
      "loss": 0.6841,
      "step": 923,
      "step_time": 8.09556882083416
    },
    {
      "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,
      "entropy": 1.4520399305555556,
      "epoch": 0.4425287356321839,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06557616591453552,
      "learning_rate": 3e-06,
      "loss": -0.0045,
      "step": 924,
      "step_time": 6.944225199520588
    },
    {
      "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,
      "entropy": 1.295862270726098,
      "epoch": 0.44300766283524906,
      "fork_head/loss": 2.718475341796875,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.060962408781051636,
      "grad_norm": 0.4044680893421173,
      "learning_rate": 3e-06,
      "loss": 0.6804,
      "step": 925,
      "step_time": 7.622900404036045
    },
    {
      "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,
      "entropy": 1.4568142361111112,
      "epoch": 0.4434865900383142,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06916642934083939,
      "learning_rate": 3e-06,
      "loss": -0.0055,
      "step": 926,
      "step_time": 7.884590893983841
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.2862774862183466,
      "epoch": 0.44396551724137934,
      "fork_head/loss": 2.704763412475586,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06997178494930267,
      "grad_norm": 0.34500357508659363,
      "learning_rate": 3e-06,
      "loss": 0.6767,
      "step": 927,
      "step_time": 7.204565674066544
    },
    {
      "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,
      "entropy": 1.4661458333333333,
      "epoch": 0.4444444444444444,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06830139458179474,
      "learning_rate": 3e-06,
      "loss": -0.0054,
      "step": 928,
      "step_time": 7.487151235342026
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 211.66796875,
      "completions/mean_terminated_length": 211.58120727539062,
      "completions/min_length": 38.0,
      "completions/min_terminated_length": 38.0,
      "entropy": 1.687440818006342,
      "epoch": 0.44492337164750956,
      "fork/depth_mean": 0.2960663437843323,
      "fork/depth_std": 0.24083474278450012,
      "fork_head/loss": 2.723039388656616,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.07629352062940598,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4247783124446869,
      "learning_rate": 3e-06,
      "loss": 0.6853,
      "num_tokens": 24468726.0,
      "reward": 0.9870354533195496,
      "reward_std": 0.5715008974075317,
      "rewards/code_format_reward/mean": 0.775390625,
      "rewards/code_format_reward/std": 0.34456905722618103,
      "rewards/code_reward/mean": 0.21164482831954956,
      "rewards/code_reward/std": 0.36634257435798645,
      "step": 929,
      "step_time": 516.1777449175715
    },
    {
      "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,
      "entropy": 1.1513671875,
      "epoch": 0.4454022988505747,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05404955521225929,
      "learning_rate": 3e-06,
      "loss": -0.0059,
      "step": 930,
      "step_time": 8.200530469417572
    },
    {
      "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,
      "entropy": 1.6722892956300215,
      "epoch": 0.44588122605363983,
      "fork_head/loss": 2.7168819904327393,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.07882677763700485,
      "grad_norm": 0.4138026237487793,
      "learning_rate": 3e-06,
      "loss": 0.6838,
      "step": 931,
      "step_time": 8.120504520833492
    },
    {
      "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,
      "entropy": 1.1162109375,
      "epoch": 0.446360153256705,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05442662537097931,
      "learning_rate": 3e-06,
      "loss": -0.0064,
      "step": 932,
      "step_time": 7.369854629039764
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.6861387274482034,
      "epoch": 0.4468390804597701,
      "fork_head/loss": 2.7063708305358887,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.09192385524511337,
      "grad_norm": 0.3805992007255554,
      "learning_rate": 3e-06,
      "loss": 0.6795,
      "step": 933,
      "step_time": 7.392986990511417
    },
    {
      "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,
      "entropy": 1.13818359375,
      "epoch": 0.44731800766283525,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.053653839975595474,
      "learning_rate": 3e-06,
      "loss": -0.0067,
      "step": 934,
      "step_time": 7.742446854710579
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.707948630506342,
      "epoch": 0.4477969348659004,
      "fork_head/loss": 2.6932034492492676,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.09395979344844818,
      "grad_norm": 0.34379029273986816,
      "learning_rate": 3e-06,
      "loss": 0.6755,
      "step": 935,
      "step_time": 7.688577488064766
    },
    {
      "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,
      "entropy": 1.122802734375,
      "epoch": 0.4482758620689655,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05459655448794365,
      "learning_rate": 3e-06,
      "loss": -0.0071,
      "step": 936,
      "step_time": 7.816340655088425
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 200.94140625,
      "completions/mean_terminated_length": 200.94140625,
      "completions/min_length": 54.0,
      "completions/min_terminated_length": 54.0,
      "entropy": 1.8325892857142858,
      "epoch": 0.44875478927203066,
      "fork/depth_mean": 0.254923015832901,
      "fork/depth_std": 0.1993228793144226,
      "fork_head/loss": 2.7139270305633545,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07567459344863892,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4888107180595398,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "num_tokens": 24670632.0,
      "reward": 1.0830488204956055,
      "reward_std": 0.5441908836364746,
      "rewards/code_format_reward/mean": 0.8310546875,
      "rewards/code_format_reward/std": 0.2819981575012207,
      "rewards/code_reward/mean": 0.25199419260025024,
      "rewards/code_reward/std": 0.36532220244407654,
      "step": 937,
      "step_time": 464.32958559691906
    },
    {
      "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,
      "entropy": 1.5220424107142858,
      "epoch": 0.4492337164750958,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15757736563682556,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 938,
      "step_time": 7.670976050198078
    },
    {
      "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,
      "entropy": 1.8136160714285714,
      "epoch": 0.44971264367816094,
      "fork_head/loss": 2.7058169841766357,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.0807342529296875,
      "grad_norm": 0.45562613010406494,
      "learning_rate": 3e-06,
      "loss": 0.6769,
      "step": 939,
      "step_time": 8.407460115849972
    },
    {
      "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,
      "entropy": 1.5556640625,
      "epoch": 0.4501915708812261,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14045721292495728,
      "learning_rate": 3e-06,
      "loss": -0.0002,
      "step": 940,
      "step_time": 6.881553985178471
    },
    {
      "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,
      "entropy": 1.8119419642857142,
      "epoch": 0.45067049808429116,
      "fork_head/loss": 2.6926350593566895,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.09737242758274078,
      "grad_norm": 0.40481850504875183,
      "learning_rate": 3e-06,
      "loss": 0.6735,
      "step": 941,
      "step_time": 7.492333762347698
    },
    {
      "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,
      "entropy": 1.5537109375,
      "epoch": 0.4511494252873563,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08729696273803711,
      "learning_rate": 3e-06,
      "loss": -0.0002,
      "step": 942,
      "step_time": 7.643582992255688
    },
    {
      "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,
      "entropy": 1.8426339285714286,
      "epoch": 0.45162835249042144,
      "fork_head/loss": 2.676537275314331,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.10594099760055542,
      "grad_norm": 0.3716060221195221,
      "learning_rate": 3e-06,
      "loss": 0.669,
      "step": 943,
      "step_time": 7.360121816396713
    },
    {
      "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,
      "entropy": 1.5382254464285714,
      "epoch": 0.4521072796934866,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07771128416061401,
      "learning_rate": 3e-06,
      "loss": -0.0007,
      "step": 944,
      "step_time": 7.703730553388596
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 203.09765625,
      "completions/mean_terminated_length": 202.8902130126953,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.228162984053294,
      "epoch": 0.4525862068965517,
      "fork/depth_mean": 0.23106232285499573,
      "fork/depth_std": 0.19463308155536652,
      "fork_head/loss": 2.7382020950317383,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.052752405405044556,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6734388470649719,
      "learning_rate": 3e-06,
      "loss": 0.681,
      "num_tokens": 24866730.0,
      "reward": 1.0510777235031128,
      "reward_std": 0.6096309423446655,
      "rewards/code_format_reward/mean": 0.7841796875,
      "rewards/code_format_reward/std": 0.33806705474853516,
      "rewards/code_reward/mean": 0.2668980360031128,
      "rewards/code_reward/std": 0.37673744559288025,
      "step": 945,
      "step_time": 527.4406980648637
    },
    {
      "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,
      "entropy": 1.6510184151785714,
      "epoch": 0.45306513409961685,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15432041883468628,
      "learning_rate": 3e-06,
      "loss": 0.0065,
      "step": 946,
      "step_time": 8.589377790689468
    },
    {
      "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,
      "entropy": 1.2246364951133728,
      "epoch": 0.453544061302682,
      "fork_head/loss": 2.7208921909332275,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.05441433936357498,
      "grad_norm": 0.6379342079162598,
      "learning_rate": 3e-06,
      "loss": 0.6768,
      "step": 947,
      "step_time": 8.177442826330662
    },
    {
      "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,
      "entropy": 1.6658993703978402,
      "epoch": 0.4540229885057471,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15058431029319763,
      "learning_rate": 3e-06,
      "loss": 0.0056,
      "step": 948,
      "step_time": 7.856590926647186
    },
    {
      "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,
      "entropy": 1.2175021767616272,
      "epoch": 0.45450191570881227,
      "fork_head/loss": 2.6932103633880615,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.08803121745586395,
      "grad_norm": 0.574759840965271,
      "learning_rate": 3e-06,
      "loss": 0.6686,
      "step": 949,
      "step_time": 7.575967229902744
    },
    {
      "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,
      "entropy": 1.6949288504464286,
      "epoch": 0.4549808429118774,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14615289866924286,
      "learning_rate": 3e-06,
      "loss": 0.0051,
      "step": 950,
      "step_time": 7.389398939907551
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.2234157919883728,
      "epoch": 0.45545977011494254,
      "fork_head/loss": 2.6626646518707275,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.09016211330890656,
      "grad_norm": 0.497763454914093,
      "learning_rate": 3e-06,
      "loss": 0.6609,
      "step": 951,
      "step_time": 7.972046263515949
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.6965448260307312,
      "epoch": 0.4559386973180077,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13753600418567657,
      "learning_rate": 3e-06,
      "loss": 0.0027,
      "step": 952,
      "step_time": 7.9310629814863205
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 215.197265625,
      "completions/mean_terminated_length": 215.11741638183594,
      "completions/min_length": 67.0,
      "completions/min_terminated_length": 67.0,
      "entropy": 0.9064843773841857,
      "epoch": 0.4564176245210728,
      "fork/depth_mean": 0.2338983416557312,
      "fork/depth_std": 0.21725966036319733,
      "fork_head/loss": 2.7328248023986816,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07696494460105896,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6579711437225342,
      "learning_rate": 3e-06,
      "loss": 0.6858,
      "num_tokens": 25070127.0,
      "reward": 1.0808751583099365,
      "reward_std": 0.6616002917289734,
      "rewards/code_format_reward/mean": 0.7666015625,
      "rewards/code_format_reward/std": 0.35626593232154846,
      "rewards/code_reward/mean": 0.3142735958099365,
      "rewards/code_reward/std": 0.41956859827041626,
      "step": 953,
      "step_time": 532.336671359837
    },
    {
      "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,
      "entropy": 1.4329131137241016,
      "epoch": 0.45689655172413796,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11091909557580948,
      "learning_rate": 3e-06,
      "loss": -0.0038,
      "step": 954,
      "step_time": 8.008110024034977
    },
    {
      "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,
      "entropy": 0.9028515577316284,
      "epoch": 0.45737547892720304,
      "fork_head/loss": 2.7254185676574707,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07536175847053528,
      "grad_norm": 0.6253965497016907,
      "learning_rate": 3e-06,
      "loss": 0.6845,
      "step": 955,
      "step_time": 8.183514066040516
    },
    {
      "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,
      "entropy": 1.4193892099640586,
      "epoch": 0.4578544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11685952544212341,
      "learning_rate": 3e-06,
      "loss": -0.0049,
      "step": 956,
      "step_time": 7.180770561099052
    },
    {
      "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,
      "entropy": 0.9226757764816285,
      "epoch": 0.4583333333333333,
      "fork_head/loss": 2.7104244232177734,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.0853133499622345,
      "grad_norm": 0.5080447196960449,
      "learning_rate": 3e-06,
      "loss": 0.6797,
      "step": 957,
      "step_time": 7.697801232337952
    },
    {
      "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,
      "entropy": 1.4261363582177595,
      "epoch": 0.45881226053639845,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.128914475440979,
      "learning_rate": 3e-06,
      "loss": -0.0057,
      "step": 958,
      "step_time": 7.774745650589466
    },
    {
      "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,
      "entropy": 0.9158203125,
      "epoch": 0.4592911877394636,
      "fork_head/loss": 2.693122148513794,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.09201181679964066,
      "grad_norm": 0.43745285272598267,
      "learning_rate": 3e-06,
      "loss": 0.6743,
      "step": 959,
      "step_time": 7.374561697244644
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.4182054942304438,
      "epoch": 0.45977011494252873,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13051629066467285,
      "learning_rate": 3e-06,
      "loss": -0.0071,
      "step": 960,
      "step_time": 8.201466009020805
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 207.48046875,
      "completions/mean_terminated_length": 207.48046875,
      "completions/min_length": 51.0,
      "completions/min_terminated_length": 51.0,
      "entropy": 1.3896687775850296,
      "epoch": 0.46024904214559387,
      "fork/depth_mean": 0.27582329511642456,
      "fork/depth_std": 0.25038179755210876,
      "fork_head/loss": 2.7276217937469482,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07790513336658478,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5212891101837158,
      "learning_rate": 3e-06,
      "loss": 0.685,
      "num_tokens": 25270853.0,
      "reward": 1.0513066053390503,
      "reward_std": 0.6196501851081848,
      "rewards/code_format_reward/mean": 0.775390625,
      "rewards/code_format_reward/std": 0.34171754121780396,
      "rewards/code_reward/mean": 0.2759159803390503,
      "rewards/code_reward/std": 0.40767157077789307,
      "step": 961,
      "step_time": 511.30126052349806
    },
    {
      "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,
      "entropy": 1.6460611939430236,
      "epoch": 0.460727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09321466833353043,
      "learning_rate": 3e-06,
      "loss": -0.0005,
      "step": 962,
      "step_time": 8.4927674010396
    },
    {
      "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,
      "entropy": 1.39190673828125,
      "epoch": 0.46120689655172414,
      "fork_head/loss": 2.7180075645446777,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07439016550779343,
      "grad_norm": 0.4833277463912964,
      "learning_rate": 3e-06,
      "loss": 0.6815,
      "step": 963,
      "step_time": 7.600556068122387
    },
    {
      "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,
      "entropy": 1.6055989623069764,
      "epoch": 0.4616858237547893,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09547851234674454,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 964,
      "step_time": 7.586457945406437
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.4317423552274704,
      "epoch": 0.4621647509578544,
      "fork_head/loss": 2.703230857849121,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08151184022426605,
      "grad_norm": 0.42883947491645813,
      "learning_rate": 3e-06,
      "loss": 0.6766,
      "step": 965,
      "step_time": 7.995697654783726
    },
    {
      "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,
      "entropy": 1.6434895873069764,
      "epoch": 0.46264367816091956,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09399028867483139,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 966,
      "step_time": 7.6532449051737785
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.4443359375,
      "epoch": 0.4631226053639847,
      "fork_head/loss": 2.686455011367798,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08260917663574219,
      "grad_norm": 0.3986230194568634,
      "learning_rate": 3e-06,
      "loss": 0.6721,
      "step": 967,
      "step_time": 7.816210962831974
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.6452799439430237,
      "epoch": 0.46360153256704983,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04976845532655716,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 968,
      "step_time": 7.280328422784805
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 213.931640625,
      "completions/mean_terminated_length": 213.8493194580078,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.3087189518488371,
      "epoch": 0.4640804597701149,
      "fork/depth_mean": 0.314961701631546,
      "fork/depth_std": 0.2753729820251465,
      "fork_head/loss": 2.7319016456604004,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.05721404775977135,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.48384585976600647,
      "learning_rate": 3e-06,
      "loss": 0.699,
      "num_tokens": 25476690.0,
      "reward": 1.1333025693893433,
      "reward_std": 0.6371234059333801,
      "rewards/code_format_reward/mean": 0.8037109375,
      "rewards/code_format_reward/std": 0.3384173810482025,
      "rewards/code_reward/mean": 0.3295916020870209,
      "rewards/code_reward/std": 0.4083973467350006,
      "step": 969,
      "step_time": 509.4540775939822
    },
    {
      "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,
      "entropy": 1.5594292561213174,
      "epoch": 0.46455938697318006,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.26383134722709656,
      "learning_rate": 3e-06,
      "loss": -0.01,
      "step": 970,
      "step_time": 8.053634881973267
    },
    {
      "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,
      "entropy": 1.3074168654588552,
      "epoch": 0.4650383141762452,
      "fork_head/loss": 2.7245521545410156,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.06942223012447357,
      "grad_norm": 0.45350849628448486,
      "learning_rate": 3e-06,
      "loss": 0.6969,
      "step": 971,
      "step_time": 7.655743889510632
    },
    {
      "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,
      "entropy": 1.5476888020833333,
      "epoch": 0.46551724137931033,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.3351871967315674,
      "learning_rate": 3e-06,
      "loss": -0.0125,
      "step": 972,
      "step_time": 8.039575062692165
    },
    {
      "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,
      "entropy": 1.3113231154588552,
      "epoch": 0.46599616858237547,
      "fork_head/loss": 2.7126030921936035,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.07483865320682526,
      "grad_norm": 0.4039784073829651,
      "learning_rate": 3e-06,
      "loss": 0.6935,
      "step": 973,
      "step_time": 7.520588420331478
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.5544704834620158,
      "epoch": 0.4664750957854406,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.3314177393913269,
      "learning_rate": 3e-06,
      "loss": -0.0163,
      "step": 974,
      "step_time": 7.81552942097187
    },
    {
      "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,
      "entropy": 1.3151542452665477,
      "epoch": 0.46695402298850575,
      "fork_head/loss": 2.699105978012085,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.07834017276763916,
      "grad_norm": 0.3752405345439911,
      "learning_rate": 3e-06,
      "loss": 0.689,
      "step": 975,
      "step_time": 8.008563116192818
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.5482096354166666,
      "epoch": 0.4674329501915709,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.23664721846580505,
      "learning_rate": 3e-06,
      "loss": -0.0181,
      "step": 976,
      "step_time": 7.236974127590656
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 211.248046875,
      "completions/mean_terminated_length": 211.16046142578125,
      "completions/min_length": 67.0,
      "completions/min_terminated_length": 67.0,
      "entropy": 1.5824110209941864,
      "epoch": 0.467911877394636,
      "fork/depth_mean": 0.24778537452220917,
      "fork/depth_std": 0.2147752046585083,
      "fork_head/loss": 2.7313621044158936,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.0649673268198967,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 1.0214518308639526,
      "learning_rate": 3e-06,
      "loss": 0.6828,
      "num_tokens": 25675585.0,
      "reward": 1.1108922958374023,
      "reward_std": 0.600139319896698,
      "rewards/code_format_reward/mean": 0.8173828125,
      "rewards/code_format_reward/std": 0.33008119463920593,
      "rewards/code_reward/mean": 0.29350942373275757,
      "rewards/code_reward/std": 0.4064639210700989,
      "step": 977,
      "step_time": 516.5543987154961
    },
    {
      "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,
      "entropy": 1.3735351647649492,
      "epoch": 0.46839080459770116,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2011485993862152,
      "learning_rate": 3e-06,
      "loss": 0.0052,
      "step": 978,
      "step_time": 8.355601154267788
    },
    {
      "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,
      "entropy": 1.5925564269224803,
      "epoch": 0.4688697318007663,
      "fork_head/loss": 2.6973166465759277,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.10188795626163483,
      "grad_norm": 0.9365127682685852,
      "learning_rate": 3e-06,
      "loss": 0.6734,
      "step": 979,
      "step_time": 7.700878232717514
    },
    {
      "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,
      "entropy": 1.3771623883928572,
      "epoch": 0.46934865900383144,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19025321304798126,
      "learning_rate": 3e-06,
      "loss": 0.0042,
      "step": 980,
      "step_time": 7.395897448062897
    },
    {
      "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,
      "entropy": 1.5907118022441864,
      "epoch": 0.4698275862068966,
      "fork_head/loss": 2.651184320449829,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.12460047751665115,
      "grad_norm": 0.733594536781311,
      "learning_rate": 3e-06,
      "loss": 0.6613,
      "step": 981,
      "step_time": 7.607583165168762
    },
    {
      "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,
      "entropy": 1.4008556519235884,
      "epoch": 0.4703065134099617,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19067278504371643,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 982,
      "step_time": 7.967752166092396
    },
    {
      "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,
      "entropy": 1.5790473123391469,
      "epoch": 0.4707854406130268,
      "fork_head/loss": 2.6135339736938477,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.12301977723836899,
      "grad_norm": 0.5439470410346985,
      "learning_rate": 3e-06,
      "loss": 0.6512,
      "step": 983,
      "step_time": 7.584850184619427
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.397042419229235,
      "epoch": 0.47126436781609193,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13533389568328857,
      "learning_rate": 3e-06,
      "loss": 0.0011,
      "step": 984,
      "step_time": 7.457472570240498
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 217.677734375,
      "completions/mean_terminated_length": 217.677734375,
      "completions/min_length": 65.0,
      "completions/min_terminated_length": 65.0,
      "entropy": 1.69390869140625,
      "epoch": 0.47174329501915707,
      "fork/depth_mean": 0.27391698956489563,
      "fork/depth_std": 0.2324288785457611,
      "fork_head/loss": 2.7617428302764893,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.06679143756628036,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 1.115809679031372,
      "learning_rate": 3e-06,
      "loss": 0.6977,
      "num_tokens": 25892348.0,
      "reward": 0.8307856917381287,
      "reward_std": 0.6051014065742493,
      "rewards/code_format_reward/mean": 0.6669921875,
      "rewards/code_format_reward/std": 0.40173831582069397,
      "rewards/code_reward/mean": 0.16379351913928986,
      "rewards/code_reward/std": 0.320237398147583,
      "step": 985,
      "step_time": 567.7403046786785
    },
    {
      "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,
      "entropy": 1.2981732147080558,
      "epoch": 0.4722222222222222,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19932246208190918,
      "learning_rate": 3e-06,
      "loss": -0.0076,
      "step": 986,
      "step_time": 8.019039362668991
    },
    {
      "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,
      "entropy": 1.6978352814912796,
      "epoch": 0.47270114942528735,
      "fork_head/loss": 2.7451798915863037,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.07050874829292297,
      "grad_norm": 0.9272557497024536,
      "learning_rate": 3e-06,
      "loss": 0.6913,
      "step": 987,
      "step_time": 8.490571767091751
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.3192584301744188,
      "epoch": 0.4731800766283525,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21835991740226746,
      "learning_rate": 3e-06,
      "loss": -0.0085,
      "step": 988,
      "step_time": 7.909325107932091
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03125,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03125,
      "entropy": 1.72296142578125,
      "epoch": 0.4736590038314176,
      "fork_head/loss": 2.717808246612549,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.06348992139101028,
      "grad_norm": 0.61039799451828,
      "learning_rate": 3e-06,
      "loss": 0.6819,
      "step": 989,
      "step_time": 7.618289262056351
    },
    {
      "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,
      "entropy": 1.2963828444480896,
      "epoch": 0.47413793103448276,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1831832379102707,
      "learning_rate": 3e-06,
      "loss": -0.0104,
      "step": 990,
      "step_time": 8.502333611249924
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.7286478653550148,
      "epoch": 0.4746168582375479,
      "fork_head/loss": 2.697802782058716,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.07923592627048492,
      "grad_norm": 0.45344826579093933,
      "learning_rate": 3e-06,
      "loss": 0.6744,
      "step": 991,
      "step_time": 7.52256441116333
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.2850361210959298,
      "epoch": 0.47509578544061304,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10537654906511307,
      "learning_rate": 3e-06,
      "loss": -0.0127,
      "step": 992,
      "step_time": 8.112728118896484
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 213.208984375,
      "completions/mean_terminated_length": 213.208984375,
      "completions/min_length": 53.0,
      "completions/min_terminated_length": 53.0,
      "entropy": 1.0910866477272727,
      "epoch": 0.4755747126436782,
      "fork/depth_mean": 0.24417339265346527,
      "fork/depth_std": 0.2020234912633896,
      "fork_head/loss": 2.72629714012146,
      "fork_head/num_positive": 11.064516067504883,
      "fork_head/pred_top1_mass": 0.07843209058046341,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6835806965827942,
      "learning_rate": 3e-06,
      "loss": 0.6856,
      "num_tokens": 26096327.0,
      "reward": 1.059335470199585,
      "reward_std": 0.6150854229927063,
      "rewards/code_format_reward/mean": 0.7890625,
      "rewards/code_format_reward/std": 0.3475346565246582,
      "rewards/code_reward/mean": 0.27027297019958496,
      "rewards/code_reward/std": 0.3836914300918579,
      "step": 993,
      "step_time": 474.0674358308315
    },
    {
      "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,
      "entropy": 1.5272549715909092,
      "epoch": 0.4760536398467433,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14031368494033813,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 994,
      "step_time": 8.82107812911272
    },
    {
      "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,
      "entropy": 1.0797230113636365,
      "epoch": 0.47653256704980845,
      "fork_head/loss": 2.71779727935791,
      "fork_head/num_positive": 11.064516067504883,
      "fork_head/pred_top1_mass": 0.06760232150554657,
      "grad_norm": 0.6522120237350464,
      "learning_rate": 3e-06,
      "loss": 0.6836,
      "step": 995,
      "step_time": 7.5147126615047455
    },
    {
      "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,
      "entropy": 1.5382191051136365,
      "epoch": 0.47701149425287354,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12321728467941284,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 996,
      "step_time": 7.247686691582203
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.0836292613636365,
      "epoch": 0.4774904214559387,
      "fork_head/loss": 2.698885440826416,
      "fork_head/num_positive": 11.064516067504883,
      "fork_head/pred_top1_mass": 0.07394113391637802,
      "grad_norm": 0.5610222220420837,
      "learning_rate": 3e-06,
      "loss": 0.679,
      "step": 997,
      "step_time": 8.047448143362999
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.5599698153409092,
      "epoch": 0.4779693486590038,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09131287038326263,
      "learning_rate": 3e-06,
      "loss": -0.0004,
      "step": 998,
      "step_time": 7.237140528857708
    },
    {
      "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,
      "entropy": 1.0861150568181819,
      "epoch": 0.47844827586206895,
      "fork_head/loss": 2.6779468059539795,
      "fork_head/num_positive": 11.064516067504883,
      "fork_head/pred_top1_mass": 0.08891763538122177,
      "grad_norm": 0.49139854311943054,
      "learning_rate": 3e-06,
      "loss": 0.6735,
      "step": 999,
      "step_time": 7.7023592963814735
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.5721916502172297,
      "epoch": 0.4789272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09228658676147461,
      "learning_rate": 3e-06,
      "loss": -0.0007,
      "step": 1000,
      "step_time": 8.050060041248798
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 211.43359375,
      "completions/mean_terminated_length": 211.43359375,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.462920221415433,
      "epoch": 0.4794061302681992,
      "fork/depth_mean": 0.24573297798633575,
      "fork/depth_std": 0.20820119976997375,
      "fork_head/loss": 2.7301323413848877,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07149463146924973,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.48546576499938965,
      "learning_rate": 3e-06,
      "loss": 0.6731,
      "num_tokens": 26308917.0,
      "reward": 1.0166043043136597,
      "reward_std": 0.7025936245918274,
      "rewards/code_format_reward/mean": 0.71875,
      "rewards/code_format_reward/std": 0.3944328725337982,
      "rewards/code_reward/mean": 0.2978542745113373,
      "rewards/code_reward/std": 0.4267633557319641,
      "step": 1001,
      "step_time": 496.31192503124475
    },
    {
      "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,
      "entropy": 1.856734659936693,
      "epoch": 0.47988505747126436,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13206256926059723,
      "learning_rate": 3e-06,
      "loss": 0.014,
      "step": 1002,
      "step_time": 7.482280351221561
    },
    {
      "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,
      "entropy": 1.4634824774482034,
      "epoch": 0.4803639846743295,
      "fork_head/loss": 2.721578598022461,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07406995445489883,
      "grad_norm": 0.4695241153240204,
      "learning_rate": 3e-06,
      "loss": 0.6702,
      "step": 1003,
      "step_time": 7.635758571326733
    },
    {
      "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,
      "entropy": 1.8875144720077515,
      "epoch": 0.48084291187739464,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12192702293395996,
      "learning_rate": 3e-06,
      "loss": 0.0129,
      "step": 1004,
      "step_time": 7.74009794741869
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.4287997159090908,
      "epoch": 0.4813218390804598,
      "fork_head/loss": 2.710223913192749,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07673840969800949,
      "grad_norm": 0.43383288383483887,
      "learning_rate": 3e-06,
      "loss": 0.6656,
      "step": 1005,
      "step_time": 7.213490255177021
    },
    {
      "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,
      "entropy": 1.8806061877144709,
      "epoch": 0.4818007662835249,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12416373193264008,
      "learning_rate": 3e-06,
      "loss": 0.0124,
      "step": 1006,
      "step_time": 8.140226751565933
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.4402817270972512,
      "epoch": 0.48227969348659006,
      "fork_head/loss": 2.6952803134918213,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08341553062200546,
      "grad_norm": 0.42075005173683167,
      "learning_rate": 3e-06,
      "loss": 0.6607,
      "step": 1007,
      "step_time": 7.5880009308457375
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.037037038140826754,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.037037038140826754,
      "entropy": 1.9089265002144709,
      "epoch": 0.4827586206896552,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0829528197646141,
      "learning_rate": 3e-06,
      "loss": 0.0109,
      "step": 1008,
      "step_time": 7.200872540473938
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 202.2890625,
      "completions/mean_terminated_length": 202.2890625,
      "completions/min_length": 13.0,
      "completions/min_terminated_length": 13.0,
      "entropy": 1.3252676460478041,
      "epoch": 0.48323754789272033,
      "fork/depth_mean": 0.20487597584724426,
      "fork/depth_std": 0.20295871794223785,
      "fork_head/loss": 2.7237799167633057,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.06848414242267609,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5165417790412903,
      "learning_rate": 3e-06,
      "loss": 0.6697,
      "num_tokens": 26502089.0,
      "reward": 0.9382920265197754,
      "reward_std": 0.6841232776641846,
      "rewards/code_format_reward/mean": 0.685546875,
      "rewards/code_format_reward/std": 0.41468408703804016,
      "rewards/code_reward/mean": 0.2527451515197754,
      "rewards/code_reward/std": 0.38312283158302307,
      "step": 1009,
      "step_time": 481.7882182523608
    },
    {
      "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,
      "entropy": 1.6541489145972512,
      "epoch": 0.4837164750957854,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0919293761253357,
      "learning_rate": 3e-06,
      "loss": 0.0069,
      "step": 1010,
      "step_time": 8.822226703166962
    },
    {
      "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,
      "entropy": 1.324616617626614,
      "epoch": 0.48419540229885055,
      "fork_head/loss": 2.7136425971984863,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.07475621998310089,
      "grad_norm": 0.4758002460002899,
      "learning_rate": 3e-06,
      "loss": 0.6666,
      "step": 1011,
      "step_time": 7.663698308169842
    },
    {
      "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,
      "entropy": 1.650390625,
      "epoch": 0.4846743295019157,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09422582387924194,
      "learning_rate": 3e-06,
      "loss": 0.0062,
      "step": 1012,
      "step_time": 7.321164660155773
    },
    {
      "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,
      "entropy": 1.3028428819444444,
      "epoch": 0.48515325670498083,
      "fork_head/loss": 2.700329542160034,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.07566721737384796,
      "grad_norm": 0.43063125014305115,
      "learning_rate": 3e-06,
      "loss": 0.6623,
      "step": 1013,
      "step_time": 8.00186076760292
    },
    {
      "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,
      "entropy": 1.6598899256099353,
      "epoch": 0.48563218390804597,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08607188612222672,
      "learning_rate": 3e-06,
      "loss": 0.0059,
      "step": 1014,
      "step_time": 7.216205917298794
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.05555555555555555,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.3110170761744182,
      "epoch": 0.4861111111111111,
      "fork_head/loss": 2.6854937076568604,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.07639698684215546,
      "grad_norm": 0.40949514508247375,
      "learning_rate": 3e-06,
      "loss": 0.6586,
      "step": 1015,
      "step_time": 7.751359961926937
    },
    {
      "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,
      "entropy": 1.6555101871490479,
      "epoch": 0.48659003831417624,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08543746173381805,
      "learning_rate": 3e-06,
      "loss": 0.0053,
      "step": 1016,
      "step_time": 7.695951864123344
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 208.486328125,
      "completions/mean_terminated_length": 208.30001831054688,
      "completions/min_length": 57.0,
      "completions/min_terminated_length": 57.0,
      "entropy": 1.4809027777777777,
      "epoch": 0.4870689655172414,
      "fork/depth_mean": 0.24788527190685272,
      "fork/depth_std": 0.21636274456977844,
      "fork_head/loss": 2.728541851043701,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.07006144523620605,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5119579434394836,
      "learning_rate": 3e-06,
      "loss": 0.6764,
      "num_tokens": 26699570.0,
      "reward": 0.8648679256439209,
      "reward_std": 0.629080593585968,
      "rewards/code_format_reward/mean": 0.6806640625,
      "rewards/code_format_reward/std": 0.41271039843559265,
      "rewards/code_reward/mean": 0.1842038929462433,
      "rewards/code_reward/std": 0.3279275596141815,
      "step": 1017,
      "step_time": 449.1171400398016
    },
    {
      "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,
      "entropy": 1.7188223467932806,
      "epoch": 0.4875478927203065,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06606878340244293,
      "learning_rate": 3e-06,
      "loss": 0.0059,
      "step": 1018,
      "step_time": 7.029899872839451
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.513671875,
      "epoch": 0.48802681992337166,
      "fork_head/loss": 2.720686674118042,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.06670287251472473,
      "grad_norm": 0.454353928565979,
      "learning_rate": 3e-06,
      "loss": 0.6744,
      "step": 1019,
      "step_time": 7.724833518266678
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05555555555555555,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.7368706597222223,
      "epoch": 0.4885057471264368,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0642474815249443,
      "learning_rate": 3e-06,
      "loss": 0.0058,
      "step": 1020,
      "step_time": 7.9136470556259155
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.4672309027777777,
      "epoch": 0.48898467432950193,
      "fork_head/loss": 2.7079575061798096,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.06998372077941895,
      "grad_norm": 0.42056769132614136,
      "learning_rate": 3e-06,
      "loss": 0.6707,
      "step": 1021,
      "step_time": 7.064567156136036
    },
    {
      "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,
      "entropy": 1.7569082842932806,
      "epoch": 0.48946360153256707,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06530709564685822,
      "learning_rate": 3e-06,
      "loss": 0.0054,
      "step": 1022,
      "step_time": 8.3111362606287
    },
    {
      "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,
      "entropy": 1.4697627358966403,
      "epoch": 0.4899425287356322,
      "fork_head/loss": 2.6939504146575928,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.09327574819326401,
      "grad_norm": 0.393495112657547,
      "learning_rate": 3e-06,
      "loss": 0.6674,
      "step": 1023,
      "step_time": 8.199636869132519
    },
    {
      "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,
      "entropy": 1.7416811254289415,
      "epoch": 0.4904214559386973,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.058656007051467896,
      "learning_rate": 3e-06,
      "loss": 0.005,
      "step": 1024,
      "step_time": 7.087878458201885
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.8828125,
      "completions/mean_terminated_length": 219.8828125,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.5652669204605951,
      "epoch": 0.49090038314176243,
      "fork/depth_mean": 0.264182984828949,
      "fork/depth_std": 0.23198050260543823,
      "fork_head/loss": 2.7227399349212646,
      "fork_head/num_positive": 11.5625,
      "fork_head/pred_top1_mass": 0.06424079835414886,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.45905423164367676,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "num_tokens": 26903254.0,
      "reward": 0.9909409284591675,
      "reward_std": 0.5862190127372742,
      "rewards/code_format_reward/mean": 0.7685546875,
      "rewards/code_format_reward/std": 0.33784088492393494,
      "rewards/code_reward/mean": 0.22238624095916748,
      "rewards/code_reward/std": 0.35549163818359375,
      "step": 1025,
      "step_time": 494.49966610223055
    },
    {
      "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,
      "entropy": 1.7754313200712204,
      "epoch": 0.49137931034482757,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0939435288310051,
      "learning_rate": 3e-06,
      "loss": 0.0066,
      "step": 1026,
      "step_time": 8.22067704796791
    },
    {
      "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,
      "entropy": 1.5752857393688626,
      "epoch": 0.4918582375478927,
      "fork_head/loss": 2.715365409851074,
      "fork_head/num_positive": 11.5625,
      "fork_head/pred_top1_mass": 0.06758114695549011,
      "grad_norm": 0.4387786388397217,
      "learning_rate": 3e-06,
      "loss": 0.6762,
      "step": 1027,
      "step_time": 8.02699613571167
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.7985026091337204,
      "epoch": 0.49233716475095785,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05455535277724266,
      "learning_rate": 3e-06,
      "loss": 0.0061,
      "step": 1028,
      "step_time": 6.89534767717123
    },
    {
      "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,
      "entropy": 1.5772750311427646,
      "epoch": 0.492816091954023,
      "fork_head/loss": 2.7032618522644043,
      "fork_head/num_positive": 11.5625,
      "fork_head/pred_top1_mass": 0.07647062838077545,
      "grad_norm": 0.4306313395500183,
      "learning_rate": 3e-06,
      "loss": 0.6723,
      "step": 1029,
      "step_time": 7.836309544742107
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.811279296875,
      "epoch": 0.4932950191570881,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.24034926295280457,
      "learning_rate": 3e-06,
      "loss": 0.0055,
      "step": 1030,
      "step_time": 7.736632741987705
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777777777777776,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777777777777776,
      "entropy": 1.596987121635013,
      "epoch": 0.49377394636015326,
      "fork_head/loss": 2.688633680343628,
      "fork_head/num_positive": 11.5625,
      "fork_head/pred_top1_mass": 0.07322404533624649,
      "grad_norm": 0.6460742354393005,
      "learning_rate": 3e-06,
      "loss": 0.6681,
      "step": 1031,
      "step_time": 7.460740186274052
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.8145345002412796,
      "epoch": 0.4942528735632184,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.055641401559114456,
      "learning_rate": 3e-06,
      "loss": 0.0056,
      "step": 1032,
      "step_time": 7.85783239454031
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.1015625,
      "completions/mean_terminated_length": 205.001953125,
      "completions/min_length": 74.0,
      "completions/min_terminated_length": 74.0,
      "entropy": 1.4326923076923077,
      "epoch": 0.49473180076628354,
      "fork/depth_mean": 0.36572265625,
      "fork/depth_std": 0.2735910713672638,
      "fork_head/loss": 2.741021156311035,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.07105018943548203,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7501944303512573,
      "learning_rate": 3e-06,
      "loss": 0.6922,
      "num_tokens": 27102378.0,
      "reward": 0.9819098711013794,
      "reward_std": 0.5667937994003296,
      "rewards/code_format_reward/mean": 0.751953125,
      "rewards/code_format_reward/std": 0.35664793848991394,
      "rewards/code_reward/mean": 0.2299567610025406,
      "rewards/code_reward/std": 0.3612518310546875,
      "step": 1033,
      "step_time": 496.0557968094945
    },
    {
      "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,
      "entropy": 1.3978949685891469,
      "epoch": 0.4952107279693487,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07686935365200043,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 1034,
      "step_time": 7.861514545977116
    },
    {
      "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,
      "entropy": 1.4393028846153846,
      "epoch": 0.4956896551724138,
      "fork_head/loss": 2.723952054977417,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.06327656656503677,
      "grad_norm": 0.6370770931243896,
      "learning_rate": 3e-06,
      "loss": 0.6878,
      "step": 1035,
      "step_time": 7.508247151970863
    },
    {
      "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,
      "entropy": 1.3981119791666667,
      "epoch": 0.49616858237547895,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07595309615135193,
      "learning_rate": 3e-06,
      "loss": -0.0046,
      "step": 1036,
      "step_time": 7.9843229949474335
    },
    {
      "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,
      "entropy": 1.4329927884615385,
      "epoch": 0.4966475095785441,
      "fork_head/loss": 2.7009143829345703,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.0669068694114685,
      "grad_norm": 0.5069851875305176,
      "learning_rate": 3e-06,
      "loss": 0.6816,
      "step": 1037,
      "step_time": 7.3768022656440735
    },
    {
      "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,
      "entropy": 1.4031982421875,
      "epoch": 0.49712643678160917,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0711168721318245,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 1038,
      "step_time": 7.790051035583019
    },
    {
      "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,
      "entropy": 1.4477163461538463,
      "epoch": 0.4976053639846743,
      "fork_head/loss": 2.678901433944702,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.087595134973526,
      "grad_norm": 0.4544009864330292,
      "learning_rate": 3e-06,
      "loss": 0.6752,
      "step": 1039,
      "step_time": 7.66048976033926
    },
    {
      "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,
      "entropy": 1.40576171875,
      "epoch": 0.49808429118773945,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07054805755615234,
      "learning_rate": 3e-06,
      "loss": -0.005,
      "step": 1040,
      "step_time": 7.414915382862091
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 206.599609375,
      "completions/mean_terminated_length": 206.5029296875,
      "completions/min_length": 72.0,
      "completions/min_terminated_length": 72.0,
      "entropy": 1.2679733974593026,
      "epoch": 0.4985632183908046,
      "fork/depth_mean": 0.2465691715478897,
      "fork/depth_std": 0.22178083658218384,
      "fork_head/loss": 2.716606378555298,
      "fork_head/num_positive": 12.8125,
      "fork_head/pred_top1_mass": 0.07557542622089386,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5927292704582214,
      "learning_rate": 3e-06,
      "loss": 0.6689,
      "num_tokens": 27304077.0,
      "reward": 0.9761354327201843,
      "reward_std": 0.6017450094223022,
      "rewards/code_format_reward/mean": 0.7490234375,
      "rewards/code_format_reward/std": 0.3661282956600189,
      "rewards/code_reward/mean": 0.22711198031902313,
      "rewards/code_reward/std": 0.3498373329639435,
      "step": 1041,
      "step_time": 489.89209470897913
    },
    {
      "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,
      "entropy": 1.375390625,
      "epoch": 0.4990421455938697,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21957191824913025,
      "learning_rate": 3e-06,
      "loss": 0.0092,
      "step": 1042,
      "step_time": 8.362160548567772
    },
    {
      "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,
      "entropy": 1.272832955632891,
      "epoch": 0.49952107279693486,
      "fork_head/loss": 2.7094717025756836,
      "fork_head/num_positive": 12.8125,
      "fork_head/pred_top1_mass": 0.08008254319429398,
      "grad_norm": 0.51996248960495,
      "learning_rate": 3e-06,
      "loss": 0.6651,
      "step": 1043,
      "step_time": 8.084419026970863
    },
    {
      "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,
      "entropy": 1.3625,
      "epoch": 0.5,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21019621193408966,
      "learning_rate": 3e-06,
      "loss": 0.0073,
      "step": 1044,
      "step_time": 6.847898885607719
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.023809524519102915,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05952381023338863,
      "entropy": 1.2533947229385376,
      "epoch": 0.5004789272030651,
      "fork_head/loss": 2.699897527694702,
      "fork_head/num_positive": 12.8125,
      "fork_head/pred_top1_mass": 0.08198990672826767,
      "grad_norm": 0.522148847579956,
      "learning_rate": 3e-06,
      "loss": 0.6605,
      "step": 1045,
      "step_time": 8.037525065243244
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333432674408,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333432674408,
      "entropy": 1.3766276001930238,
      "epoch": 0.5009578544061303,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2114919275045395,
      "learning_rate": 3e-06,
      "loss": 0.0055,
      "step": 1046,
      "step_time": 7.334072694182396
    },
    {
      "clip_ratio/high_max": 0.21428571428571427,
      "clip_ratio/high_mean": 0.13095238166196005,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.20238095309053147,
      "entropy": 1.2388857858521598,
      "epoch": 0.5014367816091954,
      "fork_head/loss": 2.6896798610687256,
      "fork_head/num_positive": 12.8125,
      "fork_head/pred_top1_mass": 0.0872131809592247,
      "grad_norm": 0.351170152425766,
      "learning_rate": 3e-06,
      "loss": 0.6565,
      "step": 1047,
      "step_time": 7.488555125892162
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.13333333432674407,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1833333343267441,
      "entropy": 1.3880859375,
      "epoch": 0.5019157088122606,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07940579950809479,
      "learning_rate": 3e-06,
      "loss": 0.004,
      "step": 1048,
      "step_time": 7.9470261707901955
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 200.875,
      "completions/mean_terminated_length": 200.76712036132812,
      "completions/min_length": 42.0,
      "completions/min_terminated_length": 42.0,
      "entropy": 1.3075284090909092,
      "epoch": 0.5023946360153256,
      "fork/depth_mean": 0.289527028799057,
      "fork/depth_std": 0.24443870782852173,
      "fork_head/loss": 2.7357730865478516,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.05237080156803131,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6732966899871826,
      "learning_rate": 3e-06,
      "loss": 0.6857,
      "num_tokens": 27492989.0,
      "reward": 1.0498104095458984,
      "reward_std": 0.6444541215896606,
      "rewards/code_format_reward/mean": 0.7666015625,
      "rewards/code_format_reward/std": 0.34368467330932617,
      "rewards/code_reward/mean": 0.28320884704589844,
      "rewards/code_reward/std": 0.40928909182548523,
      "step": 1049,
      "step_time": 457.0879530906677
    },
    {
      "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,
      "entropy": 1.7035590310891469,
      "epoch": 0.5028735632183908,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1455024629831314,
      "learning_rate": 3e-06,
      "loss": -0.0005,
      "step": 1050,
      "step_time": 7.377466075122356
    },
    {
      "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,
      "entropy": 1.3119673295454546,
      "epoch": 0.5033524904214559,
      "fork_head/loss": 2.7248597145080566,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.058209918439388275,
      "grad_norm": 0.573043704032898,
      "learning_rate": 3e-06,
      "loss": 0.6825,
      "step": 1051,
      "step_time": 7.63057753443718
    },
    {
      "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,
      "entropy": 1.7047797242800395,
      "epoch": 0.5038314176245211,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1479671150445938,
      "learning_rate": 3e-06,
      "loss": -0.0006,
      "step": 1052,
      "step_time": 8.162165448069572
    },
    {
      "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,
      "entropy": 1.3359375,
      "epoch": 0.5043103448275862,
      "fork_head/loss": 2.708179473876953,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.061848293989896774,
      "grad_norm": 0.47428008913993835,
      "learning_rate": 3e-06,
      "loss": 0.6771,
      "step": 1053,
      "step_time": 7.079803958535194
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.694471577803294,
      "epoch": 0.5047892720306514,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09463685005903244,
      "learning_rate": 3e-06,
      "loss": -0.0012,
      "step": 1054,
      "step_time": 7.685120195150375
    },
    {
      "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,
      "entropy": 1.3322088068181819,
      "epoch": 0.5052681992337165,
      "fork_head/loss": 2.6899919509887695,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.07988680154085159,
      "grad_norm": 0.5005366802215576,
      "learning_rate": 3e-06,
      "loss": 0.6718,
      "step": 1055,
      "step_time": 7.688336551189423
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.027777778605620067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06944444527228673,
      "entropy": 1.6820746461550395,
      "epoch": 0.5057471264367817,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06218256428837776,
      "learning_rate": 3e-06,
      "loss": -0.0023,
      "step": 1056,
      "step_time": 7.3633184134960175
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 219.396484375,
      "completions/mean_terminated_length": 219.32484436035156,
      "completions/min_length": 72.0,
      "completions/min_terminated_length": 72.0,
      "entropy": 1.7172111706300215,
      "epoch": 0.5062260536398467,
      "fork/depth_mean": 0.2215193659067154,
      "fork/depth_std": 0.18242472410202026,
      "fork_head/loss": 2.7222964763641357,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.07970308512449265,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.42452317476272583,
      "learning_rate": 3e-06,
      "loss": 0.6796,
      "num_tokens": 27700072.0,
      "reward": 1.0074366331100464,
      "reward_std": 0.6146315932273865,
      "rewards/code_format_reward/mean": 0.740234375,
      "rewards/code_format_reward/std": 0.37130746245384216,
      "rewards/code_reward/mean": 0.267202228307724,
      "rewards/code_reward/std": 0.37563201785087585,
      "step": 1057,
      "step_time": 487.860076777637
    },
    {
      "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,
      "entropy": 1.5401278354904868,
      "epoch": 0.5067049808429118,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13299734890460968,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 1058,
      "step_time": 7.640148304402828
    },
    {
      "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,
      "entropy": 1.724579778584567,
      "epoch": 0.507183908045977,
      "fork_head/loss": 2.716782331466675,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.08367161452770233,
      "grad_norm": 0.4003957509994507,
      "learning_rate": 3e-06,
      "loss": 0.6777,
      "step": 1059,
      "step_time": 7.681398928165436
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.5369022271849893,
      "epoch": 0.5076628352490421,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13061557710170746,
      "learning_rate": 3e-06,
      "loss": -0.0045,
      "step": 1060,
      "step_time": 7.4745801612734795
    },
    {
      "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,
      "entropy": 1.7194602272727273,
      "epoch": 0.5081417624521073,
      "fork_head/loss": 2.7089011669158936,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.08381468057632446,
      "grad_norm": 0.3730641305446625,
      "learning_rate": 3e-06,
      "loss": 0.6746,
      "step": 1061,
      "step_time": 8.530044697225094
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.5465790629386902,
      "epoch": 0.5086206896551724,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1308056265115738,
      "learning_rate": 3e-06,
      "loss": -0.0058,
      "step": 1062,
      "step_time": 7.3197702169418335
    },
    {
      "clip_ratio/high_max": 0.18181818181818182,
      "clip_ratio/high_mean": 0.07575757666067644,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07575757666067644,
      "entropy": 1.7265477072108875,
      "epoch": 0.5090996168582376,
      "fork_head/loss": 2.6979362964630127,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.0818641185760498,
      "grad_norm": 0.34365713596343994,
      "learning_rate": 3e-06,
      "loss": 0.671,
      "step": 1063,
      "step_time": 7.517737329006195
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1060606065121564,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.1060606065121564,
      "entropy": 1.5251538916067644,
      "epoch": 0.5095785440613027,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11708634346723557,
      "learning_rate": 3e-06,
      "loss": -0.0068,
      "step": 1064,
      "step_time": 7.692078180611134
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 212.482421875,
      "completions/mean_terminated_length": 212.482421875,
      "completions/min_length": 66.0,
      "completions/min_terminated_length": 66.0,
      "entropy": 1.68427734375,
      "epoch": 0.5100574712643678,
      "fork/depth_mean": 0.21176216006278992,
      "fork/depth_std": 0.20097897946834564,
      "fork_head/loss": 2.7190775871276855,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.06798931211233139,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5393384695053101,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "num_tokens": 27903663.0,
      "reward": 0.9169465899467468,
      "reward_std": 0.5986756086349487,
      "rewards/code_format_reward/mean": 0.71875,
      "rewards/code_format_reward/std": 0.3894398510456085,
      "rewards/code_reward/mean": 0.19819660484790802,
      "rewards/code_reward/std": 0.3283262550830841,
      "step": 1065,
      "step_time": 515.7147946953773
    },
    {
      "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,
      "entropy": 1.7926106810569764,
      "epoch": 0.5105363984674329,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2201218456029892,
      "learning_rate": 3e-06,
      "loss": -0.0,
      "step": 1066,
      "step_time": 7.934501461684704
    },
    {
      "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,
      "entropy": 1.6951171875,
      "epoch": 0.5110153256704981,
      "fork_head/loss": 2.7070250511169434,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.07211549580097198,
      "grad_norm": 0.49236857891082764,
      "learning_rate": 3e-06,
      "loss": 0.6753,
      "step": 1067,
      "step_time": 7.950182668864727
    },
    {
      "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,
      "entropy": 1.78408203125,
      "epoch": 0.5114942528735632,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12144500017166138,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 1068,
      "step_time": 7.811611481010914
    },
    {
      "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,
      "entropy": 1.6973307251930236,
      "epoch": 0.5119731800766284,
      "fork_head/loss": 2.6891846656799316,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.11003945767879486,
      "grad_norm": 0.4700159728527069,
      "learning_rate": 3e-06,
      "loss": 0.6701,
      "step": 1069,
      "step_time": 7.2186233177781105
    },
    {
      "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,
      "entropy": 1.7728190064430236,
      "epoch": 0.5124521072796935,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1249096691608429,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 1070,
      "step_time": 7.606897689402103
    },
    {
      "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,
      "entropy": 1.7244791626930236,
      "epoch": 0.5129310344827587,
      "fork_head/loss": 2.6681466102600098,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.11288657784461975,
      "grad_norm": 0.4665851294994354,
      "learning_rate": 3e-06,
      "loss": 0.6644,
      "step": 1071,
      "step_time": 7.683606319129467
    },
    {
      "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,
      "entropy": 1.7567708373069764,
      "epoch": 0.5134099616858238,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13141722977161407,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 1072,
      "step_time": 7.407381817698479
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 214.7109375,
      "completions/mean_terminated_length": 214.63014221191406,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.4148491720358531,
      "epoch": 0.5138888888888888,
      "fork/depth_mean": 0.2355760931968689,
      "fork/depth_std": 0.21495850384235382,
      "fork_head/loss": 2.731394052505493,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08632883429527283,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.750490128993988,
      "learning_rate": 3e-06,
      "loss": 0.6778,
      "num_tokens": 28107035.0,
      "reward": 0.8163448572158813,
      "reward_std": 0.6191757321357727,
      "rewards/code_format_reward/mean": 0.6259765625,
      "rewards/code_format_reward/std": 0.41527703404426575,
      "rewards/code_reward/mean": 0.19036832451820374,
      "rewards/code_reward/std": 0.34852465987205505,
      "step": 1073,
      "step_time": 481.1307238265872
    },
    {
      "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,
      "entropy": 1.7010091145833333,
      "epoch": 0.514367816091954,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09246460348367691,
      "learning_rate": 3e-06,
      "loss": 0.0049,
      "step": 1074,
      "step_time": 7.573017597198486
    },
    {
      "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,
      "entropy": 1.4154052833716075,
      "epoch": 0.5148467432950191,
      "fork_head/loss": 2.711878538131714,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08353869616985321,
      "grad_norm": 0.6379687786102295,
      "learning_rate": 3e-06,
      "loss": 0.6718,
      "step": 1075,
      "step_time": 8.472323149442673
    },
    {
      "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,
      "entropy": 1.6803385416666667,
      "epoch": 0.5153256704980843,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09965392202138901,
      "learning_rate": 3e-06,
      "loss": 0.0047,
      "step": 1076,
      "step_time": 7.393253058195114
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.020833333333333332,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.020833333333333332,
      "entropy": 1.4164496461550395,
      "epoch": 0.5158045977011494,
      "fork_head/loss": 2.6862905025482178,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.10811351984739304,
      "grad_norm": 0.5288858413696289,
      "learning_rate": 3e-06,
      "loss": 0.665,
      "step": 1077,
      "step_time": 7.519454315304756
    },
    {
      "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,
      "entropy": 1.7079264322916667,
      "epoch": 0.5162835249042146,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09745121002197266,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 1078,
      "step_time": 7.7799790650606155
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.441053609053294,
      "epoch": 0.5167624521072797,
      "fork_head/loss": 2.6623306274414062,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.10307392477989197,
      "grad_norm": 0.5156726241111755,
      "learning_rate": 3e-06,
      "loss": 0.658,
      "step": 1079,
      "step_time": 7.3714594617486
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.7216796875,
      "epoch": 0.5172413793103449,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09026535600423813,
      "learning_rate": 3e-06,
      "loss": 0.0035,
      "step": 1080,
      "step_time": 7.948899999260902
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.015625,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.7734375,
      "completions/mean_terminated_length": 205.57647705078125,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.650482177734375,
      "epoch": 0.51772030651341,
      "fork/depth_mean": 0.2250799834728241,
      "fork/depth_std": 0.21446430683135986,
      "fork_head/loss": 2.7524001598358154,
      "fork_head/num_positive": 12.90625,
      "fork_head/pred_top1_mass": 0.06236470490694046,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7678264379501343,
      "learning_rate": 3e-06,
      "loss": 0.6945,
      "num_tokens": 28303415.0,
      "reward": 0.9323825836181641,
      "reward_std": 0.624844491481781,
      "rewards/code_format_reward/mean": 0.701171875,
      "rewards/code_format_reward/std": 0.3876248598098755,
      "rewards/code_reward/mean": 0.23121073842048645,
      "rewards/code_reward/std": 0.36294516921043396,
      "step": 1081,
      "step_time": 539.1063435897231
    },
    {
      "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,
      "entropy": 1.751207135617733,
      "epoch": 0.5181992337164751,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16862882673740387,
      "learning_rate": 3e-06,
      "loss": -0.0041,
      "step": 1082,
      "step_time": 8.940055407583714
    },
    {
      "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,
      "entropy": 1.6287943497300148,
      "epoch": 0.5186781609195402,
      "fork_head/loss": 2.7381763458251953,
      "fork_head/num_positive": 12.90625,
      "fork_head/pred_top1_mass": 0.06644558906555176,
      "grad_norm": 0.5379469990730286,
      "learning_rate": 3e-06,
      "loss": 0.6905,
      "step": 1083,
      "step_time": 7.766145043075085
    },
    {
      "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,
      "entropy": 1.7581583559513092,
      "epoch": 0.5191570881226054,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1870550960302353,
      "learning_rate": 3e-06,
      "loss": -0.0061,
      "step": 1084,
      "step_time": 7.870892323553562
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.015625,
      "entropy": 1.6451263427734375,
      "epoch": 0.5196360153256705,
      "fork_head/loss": 2.72719144821167,
      "fork_head/num_positive": 12.90625,
      "fork_head/pred_top1_mass": 0.07181701064109802,
      "grad_norm": 0.5683131217956543,
      "learning_rate": 3e-06,
      "loss": 0.6865,
      "step": 1085,
      "step_time": 8.179285481572151
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.02083333395421505,
      "clip_ratio/low_mean": 0.0416666679084301,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06250000186264515,
      "entropy": 1.762708880007267,
      "epoch": 0.5201149425287356,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1429254114627838,
      "learning_rate": 3e-06,
      "loss": -0.0076,
      "step": 1086,
      "step_time": 7.470239765942097
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.078125,
      "entropy": 1.6455637589097023,
      "epoch": 0.5205938697318008,
      "fork_head/loss": 2.714344024658203,
      "fork_head/num_positive": 12.90625,
      "fork_head/pred_top1_mass": 0.07889991253614426,
      "grad_norm": 0.5451633334159851,
      "learning_rate": 3e-06,
      "loss": 0.6827,
      "step": 1087,
      "step_time": 7.857953809201717
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0729166679084301,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0729166679084301,
      "entropy": 1.7595079243183136,
      "epoch": 0.5210727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12415889650583267,
      "learning_rate": 3e-06,
      "loss": -0.0082,
      "step": 1088,
      "step_time": 7.905882075428963
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 202.916015625,
      "completions/mean_terminated_length": 202.70785522460938,
      "completions/min_length": 66.0,
      "completions/min_terminated_length": 66.0,
      "entropy": 1.7826822837193808,
      "epoch": 0.521551724137931,
      "fork/depth_mean": 0.29645344614982605,
      "fork/depth_std": 0.2470807433128357,
      "fork_head/loss": 2.737107038497925,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.0664454773068428,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6711857914924622,
      "learning_rate": 3e-06,
      "loss": 0.6854,
      "num_tokens": 28506652.0,
      "reward": 0.9791938066482544,
      "reward_std": 0.580791711807251,
      "rewards/code_format_reward/mean": 0.775390625,
      "rewards/code_format_reward/std": 0.35020241141319275,
      "rewards/code_reward/mean": 0.2038031667470932,
      "rewards/code_reward/std": 0.3589105010032654,
      "step": 1089,
      "step_time": 507.9173001125455
    },
    {
      "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,
      "entropy": 1.9145100968224662,
      "epoch": 0.5220306513409961,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12093126773834229,
      "learning_rate": 3e-06,
      "loss": 0.0031,
      "step": 1090,
      "step_time": 7.5160786136984825
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.778710945447286,
      "epoch": 0.5225095785440613,
      "fork_head/loss": 2.7202072143554688,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.06470581889152527,
      "grad_norm": 0.5705659985542297,
      "learning_rate": 3e-06,
      "loss": 0.6814,
      "step": 1091,
      "step_time": 9.19686795026064
    },
    {
      "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,
      "entropy": 1.9147832989692688,
      "epoch": 0.5229885057471264,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1196548119187355,
      "learning_rate": 3e-06,
      "loss": 0.0028,
      "step": 1092,
      "step_time": 7.681105799973011
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.771679691473643,
      "epoch": 0.5234674329501916,
      "fork_head/loss": 2.6980831623077393,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.07938181608915329,
      "grad_norm": 0.48531031608581543,
      "learning_rate": 3e-06,
      "loss": 0.675,
      "step": 1093,
      "step_time": 8.121135734021664
    },
    {
      "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,
      "entropy": 1.9232352205685206,
      "epoch": 0.5239463601532567,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11308173090219498,
      "learning_rate": 3e-06,
      "loss": 0.0023,
      "step": 1094,
      "step_time": 8.18608283251524
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.044444445768992105,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.044444445768992105,
      "entropy": 1.7587565104166667,
      "epoch": 0.5244252873563219,
      "fork_head/loss": 2.6759145259857178,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.09106356650590897,
      "grad_norm": 0.4448702037334442,
      "learning_rate": 3e-06,
      "loss": 0.669,
      "step": 1095,
      "step_time": 8.17656983435154
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.954950247492109,
      "epoch": 0.524904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11571018397808075,
      "learning_rate": 3e-06,
      "loss": 0.0011,
      "step": 1096,
      "step_time": 7.738945908844471
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.06666666666666667,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 193.72265625,
      "completions/mean_terminated_length": 193.72265625,
      "completions/min_length": 37.0,
      "completions/min_terminated_length": 37.0,
      "entropy": 1.4699674447377522,
      "epoch": 0.5253831417624522,
      "fork/depth_mean": 0.28291478753089905,
      "fork/depth_std": 0.25693780183792114,
      "fork_head/loss": 2.7236809730529785,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.04807655140757561,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6434763669967651,
      "learning_rate": 3e-06,
      "loss": 0.6801,
      "num_tokens": 28697790.0,
      "reward": 0.8958501219749451,
      "reward_std": 0.6035638451576233,
      "rewards/code_format_reward/mean": 0.708984375,
      "rewards/code_format_reward/std": 0.38600435853004456,
      "rewards/code_reward/mean": 0.18686577677726746,
      "rewards/code_reward/std": 0.34092605113983154,
      "step": 1097,
      "step_time": 520.9348837211728
    },
    {
      "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,
      "entropy": 1.5998757156458767,
      "epoch": 0.5258620689655172,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1550596058368683,
      "learning_rate": 3e-06,
      "loss": 0.0005,
      "step": 1098,
      "step_time": 7.982303060591221
    },
    {
      "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,
      "entropy": 1.4654947916666667,
      "epoch": 0.5263409961685823,
      "fork_head/loss": 2.713207244873047,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.0625501200556755,
      "grad_norm": 0.6289548873901367,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "step": 1099,
      "step_time": 8.057472825050354
    },
    {
      "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,
      "entropy": 1.585996687412262,
      "epoch": 0.5268199233716475,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15890724956989288,
      "learning_rate": 3e-06,
      "loss": -0.0,
      "step": 1100,
      "step_time": 7.523484937846661
    },
    {
      "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,
      "entropy": 1.4783658822377523,
      "epoch": 0.5272988505747126,
      "fork_head/loss": 2.6937384605407715,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.0783175602555275,
      "grad_norm": 0.5881325602531433,
      "learning_rate": 3e-06,
      "loss": 0.6696,
      "step": 1101,
      "step_time": 7.6416409611701965
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.5910866531458767,
      "epoch": 0.5277777777777778,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13895690441131592,
      "learning_rate": 3e-06,
      "loss": -0.0011,
      "step": 1102,
      "step_time": 7.3843662813305855
    },
    {
      "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,
      "entropy": 1.4820117155710857,
      "epoch": 0.5282567049808429,
      "fork_head/loss": 2.6706013679504395,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.09364748001098633,
      "grad_norm": 0.5368005633354187,
      "learning_rate": 3e-06,
      "loss": 0.6618,
      "step": 1103,
      "step_time": 7.7272935807704926
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.01515151560306549,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.060606061057610947,
      "entropy": 1.5826231078668074,
      "epoch": 0.5287356321839081,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13343249261379242,
      "learning_rate": 3e-06,
      "loss": -0.0023,
      "step": 1104,
      "step_time": 7.134063370525837
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 195.814453125,
      "completions/mean_terminated_length": 195.814453125,
      "completions/min_length": 53.0,
      "completions/min_terminated_length": 53.0,
      "entropy": 1.5961631973584494,
      "epoch": 0.5292145593869731,
      "fork/depth_mean": 0.30094268918037415,
      "fork/depth_std": 0.2631970942020416,
      "fork_head/loss": 2.7243802547454834,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.07429881393909454,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5606684684753418,
      "learning_rate": 3e-06,
      "loss": 0.6836,
      "num_tokens": 28885151.0,
      "reward": 1.0181922912597656,
      "reward_std": 0.6590224504470825,
      "rewards/code_format_reward/mean": 0.732421875,
      "rewards/code_format_reward/std": 0.37884828448295593,
      "rewards/code_reward/mean": 0.285770446062088,
      "rewards/code_reward/std": 0.3971784710884094,
      "step": 1105,
      "step_time": 428.4990861490369
    },
    {
      "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,
      "entropy": 1.4626410563786825,
      "epoch": 0.5296934865900383,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14966872334480286,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 1106,
      "step_time": 7.37593998759985
    },
    {
      "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,
      "entropy": 1.592504334449768,
      "epoch": 0.5301724137931034,
      "fork_head/loss": 2.712674140930176,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.07647187262773514,
      "grad_norm": 0.48815861344337463,
      "learning_rate": 3e-06,
      "loss": 0.6799,
      "step": 1107,
      "step_time": 7.768675923347473
    },
    {
      "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,
      "entropy": 1.4440863688786825,
      "epoch": 0.5306513409961686,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14847590029239655,
      "learning_rate": 3e-06,
      "loss": -0.006,
      "step": 1108,
      "step_time": 7.437440790235996
    },
    {
      "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,
      "entropy": 1.5875260432561238,
      "epoch": 0.5311302681992337,
      "fork_head/loss": 2.697458028793335,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.07790730893611908,
      "grad_norm": 0.4014560282230377,
      "learning_rate": 3e-06,
      "loss": 0.6761,
      "step": 1109,
      "step_time": 7.125586777925491
    },
    {
      "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,
      "entropy": 1.4720811605453492,
      "epoch": 0.5316091954022989,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1547497659921646,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 1110,
      "step_time": 7.906190246343613
    },
    {
      "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,
      "entropy": 1.587669269243876,
      "epoch": 0.532088122605364,
      "fork_head/loss": 2.6823205947875977,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.08444351702928543,
      "grad_norm": 0.37369388341903687,
      "learning_rate": 3e-06,
      "loss": 0.6711,
      "step": 1111,
      "step_time": 6.974545426666737
    },
    {
      "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,
      "entropy": 1.442979598045349,
      "epoch": 0.5325670498084292,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13861441612243652,
      "learning_rate": 3e-06,
      "loss": -0.0077,
      "step": 1112,
      "step_time": 7.629594132304192
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 203.99609375,
      "completions/mean_terminated_length": 203.89431762695312,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.0945963472127915,
      "epoch": 0.5330459770114943,
      "fork/depth_mean": 0.3509463667869568,
      "fork/depth_std": 0.27247288823127747,
      "fork_head/loss": 2.7355637550354004,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.06317316740751266,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5583657026290894,
      "learning_rate": 3e-06,
      "loss": 0.681,
      "num_tokens": 29091725.0,
      "reward": 0.9803815484046936,
      "reward_std": 0.5107677578926086,
      "rewards/code_format_reward/mean": 0.787109375,
      "rewards/code_format_reward/std": 0.3363058269023895,
      "rewards/code_reward/mean": 0.1932721883058548,
      "rewards/code_reward/std": 0.3203173875808716,
      "step": 1113,
      "step_time": 452.6423316821456
    },
    {
      "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,
      "entropy": 1.5699055989583333,
      "epoch": 0.5335249042145593,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19165337085723877,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 1114,
      "step_time": 8.83633741736412
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.0897460967302321,
      "epoch": 0.5340038314176245,
      "fork_head/loss": 2.723864793777466,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.06304239481687546,
      "grad_norm": 0.5444360971450806,
      "learning_rate": 3e-06,
      "loss": 0.6778,
      "step": 1115,
      "step_time": 7.7522116005420685
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5608317057291667,
      "epoch": 0.5344827586206896,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10019698739051819,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 1116,
      "step_time": 7.225129470229149
    },
    {
      "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,
      "entropy": 1.0671875089406968,
      "epoch": 0.5349616858237548,
      "fork_head/loss": 2.7029194831848145,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.06199207901954651,
      "grad_norm": 0.537812352180481,
      "learning_rate": 3e-06,
      "loss": 0.6726,
      "step": 1117,
      "step_time": 7.739657238125801
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5496419270833333,
      "epoch": 0.5354406130268199,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11165056377649307,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 1118,
      "step_time": 7.394567087292671
    },
    {
      "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,
      "entropy": 1.0648111969232559,
      "epoch": 0.5359195402298851,
      "fork_head/loss": 2.6762166023254395,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.10494841635227203,
      "grad_norm": 0.5244402885437012,
      "learning_rate": 3e-06,
      "loss": 0.6653,
      "step": 1119,
      "step_time": 8.105601593852043
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5640869140625,
      "epoch": 0.5363984674329502,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0928414985537529,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 1120,
      "step_time": 7.431268736720085
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 218.57421875,
      "completions/mean_terminated_length": 218.57421875,
      "completions/min_length": 18.0,
      "completions/min_terminated_length": 18.0,
      "entropy": 1.5152633057700262,
      "epoch": 0.5368773946360154,
      "fork/depth_mean": 0.28535187244415283,
      "fork/depth_std": 0.2333620935678482,
      "fork_head/loss": 2.723414897918701,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.07136040925979614,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6097408533096313,
      "learning_rate": 3e-06,
      "loss": 0.6746,
      "num_tokens": 29302627.0,
      "reward": 0.9384679794311523,
      "reward_std": 0.611319899559021,
      "rewards/code_format_reward/mean": 0.728515625,
      "rewards/code_format_reward/std": 0.3825030028820038,
      "rewards/code_reward/mean": 0.20995241403579712,
      "rewards/code_reward/std": 0.3568902611732483,
      "step": 1121,
      "step_time": 616.1465182006359
    },
    {
      "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,
      "entropy": 1.0736762152777777,
      "epoch": 0.5373563218390804,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05038433521986008,
      "learning_rate": 3e-06,
      "loss": 0.0074,
      "step": 1122,
      "step_time": 7.693771153688431
    },
    {
      "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,
      "entropy": 1.5428964164521959,
      "epoch": 0.5378352490421456,
      "fork_head/loss": 2.7126872539520264,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.07355057448148727,
      "grad_norm": 0.5535516738891602,
      "learning_rate": 3e-06,
      "loss": 0.672,
      "step": 1123,
      "step_time": 8.234244868159294
    },
    {
      "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,
      "entropy": 1.0764973958333333,
      "epoch": 0.5383141762452107,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.048899609595537186,
      "learning_rate": 3e-06,
      "loss": 0.0072,
      "step": 1124,
      "step_time": 7.1108148619532585
    },
    {
      "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,
      "entropy": 1.5269820557700262,
      "epoch": 0.5387931034482759,
      "fork_head/loss": 2.6941208839416504,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.09064742177724838,
      "grad_norm": 0.4958550035953522,
      "learning_rate": 3e-06,
      "loss": 0.6672,
      "step": 1125,
      "step_time": 7.682939499616623
    },
    {
      "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,
      "entropy": 1.0595703125,
      "epoch": 0.539272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05012451484799385,
      "learning_rate": 3e-06,
      "loss": 0.0072,
      "step": 1126,
      "step_time": 7.5866566225886345
    },
    {
      "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,
      "entropy": 1.5155526664521959,
      "epoch": 0.5397509578544061,
      "fork_head/loss": 2.6743528842926025,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.08692239224910736,
      "grad_norm": 0.4372263550758362,
      "learning_rate": 3e-06,
      "loss": 0.6616,
      "step": 1127,
      "step_time": 7.281546652317047
    },
    {
      "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,
      "entropy": 1.0762080483966403,
      "epoch": 0.5402298850574713,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.051416970789432526,
      "learning_rate": 3e-06,
      "loss": 0.0071,
      "step": 1128,
      "step_time": 8.10206762701273
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.953125,
      "completions/mean_terminated_length": 205.8551788330078,
      "completions/min_length": 78.0,
      "completions/min_terminated_length": 78.0,
      "entropy": 1.2919921875,
      "epoch": 0.5407088122605364,
      "fork/depth_mean": 0.31159624457359314,
      "fork/depth_std": 0.25438758730888367,
      "fork_head/loss": 2.721834182739258,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.07588762044906616,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5232945680618286,
      "learning_rate": 3e-06,
      "loss": 0.6843,
      "num_tokens": 29511051.0,
      "reward": 1.0489161014556885,
      "reward_std": 0.5854722857475281,
      "rewards/code_format_reward/mean": 0.7939453125,
      "rewards/code_format_reward/std": 0.3383834958076477,
      "rewards/code_reward/mean": 0.2549707889556885,
      "rewards/code_reward/std": 0.3775743544101715,
      "step": 1129,
      "step_time": 445.67474737763405
    },
    {
      "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,
      "entropy": 1.3541341125965118,
      "epoch": 0.5411877394636015,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05787672474980354,
      "learning_rate": 3e-06,
      "loss": -0.0048,
      "step": 1130,
      "step_time": 9.02983894944191
    },
    {
      "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,
      "entropy": 1.2890625,
      "epoch": 0.5416666666666666,
      "fork_head/loss": 2.712735414505005,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.06848495453596115,
      "grad_norm": 0.4985482692718506,
      "learning_rate": 3e-06,
      "loss": 0.6818,
      "step": 1131,
      "step_time": 7.8284857124090195
    },
    {
      "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,
      "entropy": 1.344205731153488,
      "epoch": 0.5421455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.058308690786361694,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 1132,
      "step_time": 8.428980618715286
    },
    {
      "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,
      "entropy": 1.3001030782858531,
      "epoch": 0.5426245210727969,
      "fork_head/loss": 2.6974072456359863,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.0776357650756836,
      "grad_norm": 0.4696182906627655,
      "learning_rate": 3e-06,
      "loss": 0.6775,
      "step": 1133,
      "step_time": 7.414341852068901
    },
    {
      "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,
      "entropy": 1.3430664002895356,
      "epoch": 0.5431034482758621,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.057979073375463486,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 1134,
      "step_time": 7.9423698633909225
    },
    {
      "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,
      "entropy": 1.2893880208333333,
      "epoch": 0.5435823754789272,
      "fork_head/loss": 2.678722858428955,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.10182242095470428,
      "grad_norm": 0.43209537863731384,
      "learning_rate": 3e-06,
      "loss": 0.6726,
      "step": 1135,
      "step_time": 8.256889142096043
    },
    {
      "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,
      "entropy": 1.3385416686534881,
      "epoch": 0.5440613026819924,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05752446502447128,
      "learning_rate": 3e-06,
      "loss": -0.0053,
      "step": 1136,
      "step_time": 7.909924030303955
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 211.9453125,
      "completions/mean_terminated_length": 211.85910034179688,
      "completions/min_length": 83.0,
      "completions/min_terminated_length": 83.0,
      "entropy": 1.3977864583333333,
      "epoch": 0.5445402298850575,
      "fork/depth_mean": 0.30177590250968933,
      "fork/depth_std": 0.24180418252944946,
      "fork_head/loss": 2.743443489074707,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.07621581107378006,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.793299674987793,
      "learning_rate": 3e-06,
      "loss": 0.6922,
      "num_tokens": 29705855.0,
      "reward": 0.9994184970855713,
      "reward_std": 0.6684604287147522,
      "rewards/code_format_reward/mean": 0.7119140625,
      "rewards/code_format_reward/std": 0.40121468901634216,
      "rewards/code_reward/mean": 0.2875044643878937,
      "rewards/code_reward/std": 0.40185701847076416,
      "step": 1137,
      "step_time": 439.8290464952588
    },
    {
      "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,
      "entropy": 1.1063058035714286,
      "epoch": 0.5450191570881227,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04877922311425209,
      "learning_rate": 3e-06,
      "loss": -0.0042,
      "step": 1138,
      "step_time": 7.248103782534599
    },
    {
      "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,
      "entropy": 1.3856336805555556,
      "epoch": 0.5454980842911877,
      "fork_head/loss": 2.7255492210388184,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.07730969786643982,
      "grad_norm": 0.6511058807373047,
      "learning_rate": 3e-06,
      "loss": 0.6881,
      "step": 1139,
      "step_time": 8.389904379844666
    },
    {
      "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,
      "entropy": 1.1277901785714286,
      "epoch": 0.5459770114942529,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.047554224729537964,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 1140,
      "step_time": 7.614111497998238
    },
    {
      "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,
      "entropy": 1.46875,
      "epoch": 0.546455938697318,
      "fork_head/loss": 2.7043137550354004,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.0771561786532402,
      "grad_norm": 0.48419293761253357,
      "learning_rate": 3e-06,
      "loss": 0.6807,
      "step": 1141,
      "step_time": 7.193496912717819
    },
    {
      "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,
      "entropy": 1.1236049107142858,
      "epoch": 0.5469348659003831,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.049245089292526245,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 1142,
      "step_time": 7.62225054204464
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.4907769097222223,
      "epoch": 0.5474137931034483,
      "fork_head/loss": 2.6852545738220215,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.08825831115245819,
      "grad_norm": 0.45619040727615356,
      "learning_rate": 3e-06,
      "loss": 0.6755,
      "step": 1143,
      "step_time": 7.62576662003994
    },
    {
      "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,
      "entropy": 1.1277901785714286,
      "epoch": 0.5478927203065134,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.046499259769916534,
      "learning_rate": 3e-06,
      "loss": -0.0044,
      "step": 1144,
      "step_time": 7.846836484968662
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 214.75,
      "completions/mean_terminated_length": 214.75,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "entropy": 1.1487087706724803,
      "epoch": 0.5483716475095786,
      "fork/depth_mean": 0.26157769560813904,
      "fork/depth_std": 0.21412797272205353,
      "fork_head/loss": 2.7199807167053223,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.07153349369764328,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8931811451911926,
      "learning_rate": 3e-06,
      "loss": 0.6838,
      "num_tokens": 29908799.0,
      "reward": 1.0149638652801514,
      "reward_std": 0.5697383284568787,
      "rewards/code_format_reward/mean": 0.796875,
      "rewards/code_format_reward/std": 0.34373053908348083,
      "rewards/code_reward/mean": 0.21808885037899017,
      "rewards/code_reward/std": 0.3547295033931732,
      "step": 1145,
      "step_time": 477.6219480037689
    },
    {
      "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,
      "entropy": 1.1527157766478402,
      "epoch": 0.5488505747126436,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20867471396923065,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 1146,
      "step_time": 8.124475434422493
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.041666666666666664,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.1404622346162796,
      "epoch": 0.5493295019157088,
      "fork_head/loss": 2.706667900085449,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.07771118730306625,
      "grad_norm": 0.6776565909385681,
      "learning_rate": 3e-06,
      "loss": 0.6799,
      "step": 1147,
      "step_time": 8.144721999764442
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.1370210221835546,
      "epoch": 0.5498084291187739,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.056051310151815414,
      "learning_rate": 3e-06,
      "loss": -0.0068,
      "step": 1148,
      "step_time": 8.191717453300953
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.041666666666666664,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.1577419688304265,
      "epoch": 0.5502873563218391,
      "fork_head/loss": 2.689925193786621,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.07156842201948166,
      "grad_norm": 0.4487818479537964,
      "learning_rate": 3e-06,
      "loss": 0.676,
      "step": 1149,
      "step_time": 7.405213266611099
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.1354399153164454,
      "epoch": 0.5507662835249042,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04804816469550133,
      "learning_rate": 3e-06,
      "loss": -0.007,
      "step": 1150,
      "step_time": 7.753325246274471
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.041666666666666664,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.1660427500804265,
      "epoch": 0.5512452107279694,
      "fork_head/loss": 2.6780190467834473,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.09404663741588593,
      "grad_norm": 0.5612626671791077,
      "learning_rate": 3e-06,
      "loss": 0.6723,
      "step": 1151,
      "step_time": 8.127248793840408
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.124918622629983,
      "epoch": 0.5517241379310345,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06142627075314522,
      "learning_rate": 3e-06,
      "loss": -0.0072,
      "step": 1152,
      "step_time": 7.393701404333115
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 208.01171875,
      "completions/mean_terminated_length": 208.01171875,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.4472005208333334,
      "epoch": 0.5522030651340997,
      "fork/depth_mean": 0.24434737861156464,
      "fork/depth_std": 0.18885290622711182,
      "fork_head/loss": 2.7262489795684814,
      "fork_head/num_positive": 11.22580623626709,
      "fork_head/pred_top1_mass": 0.0738171935081482,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6631020903587341,
      "learning_rate": 3e-06,
      "loss": 0.6833,
      "num_tokens": 30108037.0,
      "reward": 1.042567491531372,
      "reward_std": 0.5379015803337097,
      "rewards/code_format_reward/mean": 0.8173828125,
      "rewards/code_format_reward/std": 0.33449816703796387,
      "rewards/code_reward/mean": 0.22518478333950043,
      "rewards/code_reward/std": 0.3350582420825958,
      "step": 1153,
      "step_time": 575.0161463990808
    },
    {
      "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,
      "entropy": 1.2885488986968994,
      "epoch": 0.5526819923371648,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1501377373933792,
      "learning_rate": 3e-06,
      "loss": 0.0051,
      "step": 1154,
      "step_time": 8.111935310065746
    },
    {
      "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,
      "entropy": 1.4619574705759684,
      "epoch": 0.5531609195402298,
      "fork_head/loss": 2.714817523956299,
      "fork_head/num_positive": 11.22580623626709,
      "fork_head/pred_top1_mass": 0.07809489220380783,
      "grad_norm": 0.6285907030105591,
      "learning_rate": 3e-06,
      "loss": 0.6786,
      "step": 1155,
      "step_time": 8.198688223958015
    },
    {
      "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,
      "entropy": 1.2922887841860453,
      "epoch": 0.553639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1429741233587265,
      "learning_rate": 3e-06,
      "loss": 0.0036,
      "step": 1156,
      "step_time": 7.488676503300667
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.055555556217829385,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07777777910232545,
      "entropy": 1.460460074742635,
      "epoch": 0.5541187739463601,
      "fork_head/loss": 2.6975574493408203,
      "fork_head/num_positive": 11.22580623626709,
      "fork_head/pred_top1_mass": 0.07732202857732773,
      "grad_norm": 0.48931461572647095,
      "learning_rate": 3e-06,
      "loss": 0.673,
      "step": 1157,
      "step_time": 7.509223625063896
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.2786603053410848,
      "epoch": 0.5545977011494253,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11621524393558502,
      "learning_rate": 3e-06,
      "loss": 0.0017,
      "step": 1158,
      "step_time": 8.374464854598045
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.055555556217829385,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07777777910232545,
      "entropy": 1.4916015625,
      "epoch": 0.5550766283524904,
      "fork_head/loss": 2.6786293983459473,
      "fork_head/num_positive": 11.22580623626709,
      "fork_head/pred_top1_mass": 0.1026933491230011,
      "grad_norm": 0.41769111156463623,
      "learning_rate": 3e-06,
      "loss": 0.6665,
      "step": 1159,
      "step_time": 7.239317238330841
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.2959418376286824,
      "epoch": 0.5555555555555556,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11914685368537903,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 1160,
      "step_time": 8.183499127626419
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.0625,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 206.6640625,
      "completions/mean_terminated_length": 206.6640625,
      "completions/min_length": 66.0,
      "completions/min_terminated_length": 66.0,
      "entropy": 1.303271483629942,
      "epoch": 0.5560344827586207,
      "fork/depth_mean": 0.23753352463245392,
      "fork/depth_std": 0.214760884642601,
      "fork_head/loss": 2.7408573627471924,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.06025591492652893,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6029261350631714,
      "learning_rate": 3e-06,
      "loss": 0.6947,
      "num_tokens": 30307481.0,
      "reward": 1.088289499282837,
      "reward_std": 0.6008219718933105,
      "rewards/code_format_reward/mean": 0.81640625,
      "rewards/code_format_reward/std": 0.3447742164134979,
      "rewards/code_reward/mean": 0.2718833386898041,
      "rewards/code_reward/std": 0.38039395213127136,
      "step": 1161,
      "step_time": 492.4608489796519
    },
    {
      "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,
      "entropy": 1.195917044367109,
      "epoch": 0.5565134099616859,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13284681737422943,
      "learning_rate": 3e-06,
      "loss": -0.0091,
      "step": 1162,
      "step_time": 8.019711785018444
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.315810140222311,
      "epoch": 0.5569923371647509,
      "fork_head/loss": 2.7292990684509277,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.06679696589708328,
      "grad_norm": 0.4991438090801239,
      "learning_rate": 3e-06,
      "loss": 0.6912,
      "step": 1163,
      "step_time": 7.279047340154648
    },
    {
      "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,
      "entropy": 1.2041248083114624,
      "epoch": 0.5574712643678161,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13946571946144104,
      "learning_rate": 3e-06,
      "loss": -0.0094,
      "step": 1164,
      "step_time": 8.142554491758347
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.3150878921151161,
      "epoch": 0.5579501915708812,
      "fork_head/loss": 2.7164809703826904,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.0696786567568779,
      "grad_norm": 0.5351805686950684,
      "learning_rate": 3e-06,
      "loss": 0.6874,
      "step": 1165,
      "step_time": 7.780182860791683
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.193603515625,
      "epoch": 0.5584291187739464,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09242653846740723,
      "learning_rate": 3e-06,
      "loss": -0.0106,
      "step": 1166,
      "step_time": 7.348979279398918
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.3188883438706398,
      "epoch": 0.5589080459770115,
      "fork_head/loss": 2.7014482021331787,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.07644268125295639,
      "grad_norm": 0.5663687586784363,
      "learning_rate": 3e-06,
      "loss": 0.6823,
      "step": 1167,
      "step_time": 8.208418250083923
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.2031831315585546,
      "epoch": 0.5593869731800766,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0889187604188919,
      "learning_rate": 3e-06,
      "loss": -0.0111,
      "step": 1168,
      "step_time": 7.508331783115864
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 212.41796875,
      "completions/mean_terminated_length": 212.33267211914062,
      "completions/min_length": 68.0,
      "completions/min_terminated_length": 68.0,
      "entropy": 1.4506096189672297,
      "epoch": 0.5598659003831418,
      "fork/depth_mean": 0.23007777333259583,
      "fork/depth_std": 0.20967106521129608,
      "fork_head/loss": 2.7144389152526855,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08785288035869598,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5329833030700684,
      "learning_rate": 3e-06,
      "loss": 0.6841,
      "num_tokens": 30511871.0,
      "reward": 1.1441493034362793,
      "reward_std": 0.6653321385383606,
      "rewards/code_format_reward/mean": 0.80078125,
      "rewards/code_format_reward/std": 0.3502296805381775,
      "rewards/code_reward/mean": 0.3433680236339569,
      "rewards/code_reward/std": 0.4328913688659668,
      "step": 1169,
      "step_time": 511.3078159019351
    },
    {
      "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,
      "entropy": 1.3150173584620157,
      "epoch": 0.5603448275862069,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12407517433166504,
      "learning_rate": 3e-06,
      "loss": -0.0013,
      "step": 1170,
      "step_time": 7.606714777648449
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.4855883121490479,
      "epoch": 0.560823754789272,
      "fork_head/loss": 2.704644203186035,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08164317905902863,
      "grad_norm": 0.47929710149765015,
      "learning_rate": 3e-06,
      "loss": 0.6807,
      "step": 1171,
      "step_time": 7.936922542750835
    },
    {
      "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,
      "entropy": 1.3096137126286824,
      "epoch": 0.5613026819923371,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1369147002696991,
      "learning_rate": 3e-06,
      "loss": -0.0015,
      "step": 1172,
      "step_time": 7.507490061223507
    },
    {
      "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,
      "entropy": 1.4764737215909092,
      "epoch": 0.5617816091954023,
      "fork_head/loss": 2.6920132637023926,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07395877689123154,
      "grad_norm": 0.4706690311431885,
      "learning_rate": 3e-06,
      "loss": 0.6768,
      "step": 1173,
      "step_time": 7.2264560386538506
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.06666666666666667,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.3149305582046509,
      "epoch": 0.5622605363984674,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09802954643964767,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 1174,
      "step_time": 8.402007140219212
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.472508278760043,
      "epoch": 0.5627394636015326,
      "fork_head/loss": 2.6783289909362793,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08285442739725113,
      "grad_norm": 0.43052852153778076,
      "learning_rate": 3e-06,
      "loss": 0.6725,
      "step": 1175,
      "step_time": 7.9302521869540215
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.08888888955116273,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08888888955116273,
      "entropy": 1.3016710042953492,
      "epoch": 0.5632183908045977,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08181533962488174,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 1176,
      "step_time": 7.6793602630496025
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.724609375,
      "completions/mean_terminated_length": 205.626220703125,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.2795003255208333,
      "epoch": 0.5636973180076629,
      "fork/depth_mean": 0.23563574254512787,
      "fork/depth_std": 0.20408764481544495,
      "fork_head/loss": 2.7241945266723633,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.0704195573925972,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5368164777755737,
      "learning_rate": 3e-06,
      "loss": 0.6767,
      "num_tokens": 30709202.0,
      "reward": 1.1115392446517944,
      "reward_std": 0.6065499186515808,
      "rewards/code_format_reward/mean": 0.8037109375,
      "rewards/code_format_reward/std": 0.32212281227111816,
      "rewards/code_reward/mean": 0.30782827734947205,
      "rewards/code_reward/std": 0.40675878524780273,
      "step": 1177,
      "step_time": 532.6802942082286
    },
    {
      "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,
      "entropy": 1.1590983072916667,
      "epoch": 0.564176245210728,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08891890943050385,
      "learning_rate": 3e-06,
      "loss": 0.0041,
      "step": 1178,
      "step_time": 8.315753199160099
    },
    {
      "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,
      "entropy": 1.2773980001608531,
      "epoch": 0.5646551724137931,
      "fork_head/loss": 2.714662551879883,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.06616104394197464,
      "grad_norm": 0.5067700147628784,
      "learning_rate": 3e-06,
      "loss": 0.6732,
      "step": 1179,
      "step_time": 7.284523598849773
    },
    {
      "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,
      "entropy": 1.1656901041666667,
      "epoch": 0.5651340996168582,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08772671222686768,
      "learning_rate": 3e-06,
      "loss": 0.0036,
      "step": 1180,
      "step_time": 7.597567342221737
    },
    {
      "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,
      "entropy": 1.2819010416666667,
      "epoch": 0.5656130268199234,
      "fork_head/loss": 2.6990444660186768,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.07260269671678543,
      "grad_norm": 0.4799690842628479,
      "learning_rate": 3e-06,
      "loss": 0.6688,
      "step": 1181,
      "step_time": 7.581791922450066
    },
    {
      "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,
      "entropy": 1.1641438802083333,
      "epoch": 0.5660919540229885,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09047119319438934,
      "learning_rate": 3e-06,
      "loss": 0.0031,
      "step": 1182,
      "step_time": 7.1711417362093925
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.027777778605620067,
      "clip_ratio/low_mean": 0.06944444527228673,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0972222238779068,
      "entropy": 1.2891845703125,
      "epoch": 0.5665708812260536,
      "fork_head/loss": 2.6805505752563477,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.10024632513523102,
      "grad_norm": 0.419123113155365,
      "learning_rate": 3e-06,
      "loss": 0.6634,
      "step": 1183,
      "step_time": 7.805322349071503
    },
    {
      "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,
      "entropy": 1.1714680989583333,
      "epoch": 0.5670498084291188,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08734393119812012,
      "learning_rate": 3e-06,
      "loss": 0.0023,
      "step": 1184,
      "step_time": 7.403753086924553
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 211.921875,
      "completions/mean_terminated_length": 211.921875,
      "completions/min_length": 60.0,
      "completions/min_terminated_length": 60.0,
      "entropy": 1.1397298127412796,
      "epoch": 0.5675287356321839,
      "fork/depth_mean": 0.18646426498889923,
      "fork/depth_std": 0.17187722027301788,
      "fork_head/loss": 2.7369658946990967,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.06458143889904022,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.39918798208236694,
      "learning_rate": 3e-06,
      "loss": 0.6956,
      "num_tokens": 30914026.0,
      "reward": 0.9907362461090088,
      "reward_std": 0.5877758860588074,
      "rewards/code_format_reward/mean": 0.7646484375,
      "rewards/code_format_reward/std": 0.34234753251075745,
      "rewards/code_reward/mean": 0.22608788311481476,
      "rewards/code_reward/std": 0.3515779376029968,
      "step": 1185,
      "step_time": 530.9824638515711
    },
    {
      "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,
      "entropy": 1.0900390625,
      "epoch": 0.5680076628352491,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10033371299505234,
      "learning_rate": 3e-06,
      "loss": -0.0092,
      "step": 1186,
      "step_time": 7.264442212879658
    },
    {
      "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,
      "entropy": 1.1422119140625,
      "epoch": 0.5684865900383141,
      "fork_head/loss": 2.7306454181671143,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.06189839541912079,
      "grad_norm": 0.3783790171146393,
      "learning_rate": 3e-06,
      "loss": 0.6935,
      "step": 1187,
      "step_time": 8.390573672950268
    },
    {
      "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,
      "entropy": 1.0892578125,
      "epoch": 0.5689655172413793,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16485761106014252,
      "learning_rate": 3e-06,
      "loss": -0.0094,
      "step": 1188,
      "step_time": 7.316713936626911
    },
    {
      "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,
      "entropy": 1.1453450471162796,
      "epoch": 0.5694444444444444,
      "fork_head/loss": 2.7210962772369385,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.06573666632175446,
      "grad_norm": 0.3695511817932129,
      "learning_rate": 3e-06,
      "loss": 0.6907,
      "step": 1189,
      "step_time": 7.402496963739395
    },
    {
      "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,
      "entropy": 1.0716796875,
      "epoch": 0.5699233716475096,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10008382797241211,
      "learning_rate": 3e-06,
      "loss": -0.0101,
      "step": 1190,
      "step_time": 7.871283948421478
    },
    {
      "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,
      "entropy": 1.1680908203125,
      "epoch": 0.5704022988505747,
      "fork_head/loss": 2.709078073501587,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.06521498411893845,
      "grad_norm": 0.35742270946502686,
      "learning_rate": 3e-06,
      "loss": 0.6875,
      "step": 1191,
      "step_time": 7.3598495200276375
    },
    {
      "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,
      "entropy": 1.07919921875,
      "epoch": 0.5708812260536399,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10514120012521744,
      "learning_rate": 3e-06,
      "loss": -0.0108,
      "step": 1192,
      "step_time": 7.441029280424118
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 204.55859375,
      "completions/mean_terminated_length": 204.4579315185547,
      "completions/min_length": 10.0,
      "completions/min_terminated_length": 10.0,
      "entropy": 1.4377811280163852,
      "epoch": 0.571360153256705,
      "fork/depth_mean": 0.19633103907108307,
      "fork/depth_std": 0.17213495075702667,
      "fork_head/loss": 2.714294672012329,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.09081510454416275,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.44634997844696045,
      "learning_rate": 3e-06,
      "loss": 0.6774,
      "num_tokens": 31116616.0,
      "reward": 0.9241606593132019,
      "reward_std": 0.5630743503570557,
      "rewards/code_format_reward/mean": 0.75,
      "rewards/code_format_reward/std": 0.3687923848628998,
      "rewards/code_reward/mean": 0.1741606444120407,
      "rewards/code_reward/std": 0.31139808893203735,
      "step": 1193,
      "step_time": 569.6938766390085
    },
    {
      "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,
      "entropy": 1.4508611505681819,
      "epoch": 0.5718390804597702,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.28132954239845276,
      "learning_rate": 3e-06,
      "loss": 0.0011,
      "step": 1194,
      "step_time": 8.377124570310116
    },
    {
      "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,
      "entropy": 1.4276012073863635,
      "epoch": 0.5723180076628352,
      "fork_head/loss": 2.7073755264282227,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.0893775224685669,
      "grad_norm": 0.4199484586715698,
      "learning_rate": 3e-06,
      "loss": 0.6752,
      "step": 1195,
      "step_time": 8.130125522613525
    },
    {
      "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,
      "entropy": 1.4474875710227273,
      "epoch": 0.5727969348659003,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.25173208117485046,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 1196,
      "step_time": 7.414397306740284
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.4790482954545454,
      "epoch": 0.5732758620689655,
      "fork_head/loss": 2.6959445476531982,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.10117421299219131,
      "grad_norm": 0.3760857880115509,
      "learning_rate": 3e-06,
      "loss": 0.6718,
      "step": 1197,
      "step_time": 8.32177960127592
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.022727272727272728,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022727272727272728,
      "entropy": 1.4534801136363635,
      "epoch": 0.5737547892720306,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16870561242103577,
      "learning_rate": 3e-06,
      "loss": -0.0022,
      "step": 1198,
      "step_time": 7.187861613929272
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.4937559149482034,
      "epoch": 0.5742337164750958,
      "fork_head/loss": 2.682337760925293,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.10010357946157455,
      "grad_norm": 0.3532992899417877,
      "learning_rate": 3e-06,
      "loss": 0.6672,
      "step": 1199,
      "step_time": 8.125095032155514
    },
    {
      "clip_ratio/high_max": 0.36363636363636365,
      "clip_ratio/high_mean": 0.1590909090909091,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.4374778053977273,
      "epoch": 0.5747126436781609,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08400231599807739,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 1200,
      "step_time": 7.841033406555653
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 202.689453125,
      "completions/mean_terminated_length": 202.689453125,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.463230935010043,
      "epoch": 0.5751915708812261,
      "fork/depth_mean": 0.19575466215610504,
      "fork/depth_std": 0.1894097626209259,
      "fork_head/loss": 2.7352097034454346,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.07902181148529053,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5796050429344177,
      "learning_rate": 3e-06,
      "loss": 0.6982,
      "num_tokens": 31316761.0,
      "reward": 1.0372190475463867,
      "reward_std": 0.5873136520385742,
      "rewards/code_format_reward/mean": 0.783203125,
      "rewards/code_format_reward/std": 0.3551013469696045,
      "rewards/code_reward/mean": 0.25401586294174194,
      "rewards/code_reward/std": 0.37339454889297485,
      "step": 1201,
      "step_time": 511.9747292101383
    },
    {
      "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,
      "entropy": 1.6534468978643417,
      "epoch": 0.5756704980842912,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19731275737285614,
      "learning_rate": 3e-06,
      "loss": -0.0119,
      "step": 1202,
      "step_time": 7.877321191132069
    },
    {
      "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,
      "entropy": 1.466648903760043,
      "epoch": 0.5761494252873564,
      "fork_head/loss": 2.7241902351379395,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.07932259142398834,
      "grad_norm": 0.5561451315879822,
      "learning_rate": 3e-06,
      "loss": 0.6941,
      "step": 1203,
      "step_time": 7.993929848074913
    },
    {
      "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,
      "entropy": 1.6183087279399235,
      "epoch": 0.5766283524904214,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21597129106521606,
      "learning_rate": 3e-06,
      "loss": -0.0138,
      "step": 1204,
      "step_time": 7.78944056481123
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.12121212211522189,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.12121212211522189,
      "entropy": 1.5098543925718828,
      "epoch": 0.5771072796934866,
      "fork_head/loss": 2.7070813179016113,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.07743734866380692,
      "grad_norm": 0.47033989429473877,
      "learning_rate": 3e-06,
      "loss": 0.6883,
      "step": 1205,
      "step_time": 7.488701768219471
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5816695590813954,
      "epoch": 0.5775862068965517,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11388923972845078,
      "learning_rate": 3e-06,
      "loss": -0.0162,
      "step": 1206,
      "step_time": 7.950137108564377
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07575757666067644,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07575757666067644,
      "entropy": 1.544374411756342,
      "epoch": 0.5780651340996169,
      "fork_head/loss": 2.6868419647216797,
      "fork_head/num_positive": 12.15625,
      "fork_head/pred_top1_mass": 0.0792449489235878,
      "grad_norm": 0.43858015537261963,
      "learning_rate": 3e-06,
      "loss": 0.6829,
      "step": 1207,
      "step_time": 7.26382190734148
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.16666666666666666,
      "clip_ratio/low_mean": 0.009259259328246117,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.17592592599491277,
      "entropy": 1.5481680383284886,
      "epoch": 0.578544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09408188611268997,
      "learning_rate": 3e-06,
      "loss": -0.0168,
      "step": 1208,
      "step_time": 7.774660333991051
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 207.134765625,
      "completions/mean_terminated_length": 207.0391387939453,
      "completions/min_length": 51.0,
      "completions/min_terminated_length": 51.0,
      "entropy": 1.75703125,
      "epoch": 0.5790229885057471,
      "fork/depth_mean": 0.23373126983642578,
      "fork/depth_std": 0.21945959329605103,
      "fork_head/loss": 2.7175917625427246,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07207492738962173,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4456835985183716,
      "learning_rate": 3e-06,
      "loss": 0.6774,
      "num_tokens": 31521470.0,
      "reward": 0.9504466652870178,
      "reward_std": 0.5829663872718811,
      "rewards/code_format_reward/mean": 0.7568359375,
      "rewards/code_format_reward/std": 0.35244762897491455,
      "rewards/code_reward/mean": 0.19361069798469543,
      "rewards/code_reward/std": 0.3521755635738373,
      "step": 1209,
      "step_time": 501.2677293717861
    },
    {
      "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,
      "entropy": 1.634765625,
      "epoch": 0.5795019157088123,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.050546448677778244,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 1210,
      "step_time": 8.311654105782509
    },
    {
      "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,
      "entropy": 1.7515625,
      "epoch": 0.5799808429118773,
      "fork_head/loss": 2.7109897136688232,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07057107985019684,
      "grad_norm": 0.43463680148124695,
      "learning_rate": 3e-06,
      "loss": 0.6759,
      "step": 1211,
      "step_time": 7.467784583568573
    },
    {
      "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,
      "entropy": 1.6484375,
      "epoch": 0.5804597701149425,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05060596391558647,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 1212,
      "step_time": 7.58904230594635
    },
    {
      "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,
      "entropy": 1.7751953125,
      "epoch": 0.5809386973180076,
      "fork_head/loss": 2.698601245880127,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07611119747161865,
      "grad_norm": 0.4103969633579254,
      "learning_rate": 3e-06,
      "loss": 0.6724,
      "step": 1213,
      "step_time": 8.174460798501968
    },
    {
      "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,
      "entropy": 1.6546875,
      "epoch": 0.5814176245210728,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05035419017076492,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 1214,
      "step_time": 7.271344691514969
    },
    {
      "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,
      "entropy": 1.7650390625,
      "epoch": 0.5818965517241379,
      "fork_head/loss": 2.6829586029052734,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.08631376922130585,
      "grad_norm": 0.3740735650062561,
      "learning_rate": 3e-06,
      "loss": 0.6682,
      "step": 1215,
      "step_time": 7.844158440828323
    },
    {
      "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,
      "entropy": 1.66796875,
      "epoch": 0.5823754789272031,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05021122843027115,
      "learning_rate": 3e-06,
      "loss": 0.0018,
      "step": 1216,
      "step_time": 7.47797878831625
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 216.72265625,
      "completions/mean_terminated_length": 216.72265625,
      "completions/min_length": 50.0,
      "completions/min_terminated_length": 50.0,
      "entropy": 0.9825148752757481,
      "epoch": 0.5828544061302682,
      "fork/depth_mean": 0.28651049733161926,
      "fork/depth_std": 0.24545645713806152,
      "fork_head/loss": 2.737083673477173,
      "fork_head/num_positive": 11.322580337524414,
      "fork_head/pred_top1_mass": 0.0707007348537445,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5463359951972961,
      "learning_rate": 3e-06,
      "loss": 0.6696,
      "num_tokens": 31720784.0,
      "reward": 0.9295758008956909,
      "reward_std": 0.5952340960502625,
      "rewards/code_format_reward/mean": 0.740234375,
      "rewards/code_format_reward/std": 0.3739333748817444,
      "rewards/code_reward/mean": 0.18934139609336853,
      "rewards/code_reward/std": 0.34046462178230286,
      "step": 1217,
      "step_time": 430.3127967566252
    },
    {
      "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,
      "entropy": 1.5005425347222223,
      "epoch": 0.5833333333333334,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10507620126008987,
      "learning_rate": 3e-06,
      "loss": 0.0148,
      "step": 1218,
      "step_time": 7.304134123027325
    },
    {
      "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,
      "entropy": 0.9780040979385376,
      "epoch": 0.5838122605363985,
      "fork_head/loss": 2.728609085083008,
      "fork_head/num_positive": 11.322580337524414,
      "fork_head/pred_top1_mass": 0.07774487137794495,
      "grad_norm": 0.5212793946266174,
      "learning_rate": 3e-06,
      "loss": 0.6676,
      "step": 1219,
      "step_time": 8.262968011200428
    },
    {
      "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,
      "entropy": 1.5206163194444444,
      "epoch": 0.5842911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10173754394054413,
      "learning_rate": 3e-06,
      "loss": 0.0142,
      "step": 1220,
      "step_time": 7.352088890969753
    },
    {
      "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,
      "entropy": 0.9901878663471767,
      "epoch": 0.5847701149425287,
      "fork_head/loss": 2.713090181350708,
      "fork_head/num_positive": 11.322580337524414,
      "fork_head/pred_top1_mass": 0.07925166189670563,
      "grad_norm": 0.4768601953983307,
      "learning_rate": 3e-06,
      "loss": 0.6634,
      "step": 1221,
      "step_time": 7.119022712111473
    },
    {
      "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,
      "entropy": 1.5159505208333333,
      "epoch": 0.5852490421455939,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10220044106245041,
      "learning_rate": 3e-06,
      "loss": 0.0137,
      "step": 1222,
      "step_time": 8.058827638626099
    },
    {
      "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,
      "entropy": 0.9813058035714286,
      "epoch": 0.585727969348659,
      "fork_head/loss": 2.6942203044891357,
      "fork_head/num_positive": 11.322580337524414,
      "fork_head/pred_top1_mass": 0.08366601169109344,
      "grad_norm": 0.4334302842617035,
      "learning_rate": 3e-06,
      "loss": 0.6587,
      "step": 1223,
      "step_time": 7.460855647921562
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.05555555555555555,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.5472005208333333,
      "epoch": 0.5862068965517241,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0733422115445137,
      "learning_rate": 3e-06,
      "loss": 0.0133,
      "step": 1224,
      "step_time": 7.302926287055016
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.986328125,
      "completions/mean_terminated_length": 205.986328125,
      "completions/min_length": 24.0,
      "completions/min_terminated_length": 24.0,
      "entropy": 1.6000162661075592,
      "epoch": 0.5866858237547893,
      "fork/depth_mean": 0.2495345175266266,
      "fork/depth_std": 0.22396200895309448,
      "fork_head/loss": 2.732815742492676,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.05913162976503372,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5410925149917603,
      "learning_rate": 3e-06,
      "loss": 0.6822,
      "num_tokens": 31923641.0,
      "reward": 0.9860805869102478,
      "reward_std": 0.575463056564331,
      "rewards/code_format_reward/mean": 0.767578125,
      "rewards/code_format_reward/std": 0.34788089990615845,
      "rewards/code_reward/mean": 0.21850243210792542,
      "rewards/code_reward/std": 0.35081747174263,
      "step": 1225,
      "step_time": 460.10744915902615
    },
    {
      "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,
      "entropy": 0.9222935267857143,
      "epoch": 0.5871647509578544,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.055605385452508926,
      "learning_rate": 3e-06,
      "loss": 0.0013,
      "step": 1226,
      "step_time": 7.590301752090454
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.5904134213924408,
      "epoch": 0.5876436781609196,
      "fork_head/loss": 2.722034215927124,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.055441781878471375,
      "grad_norm": 0.4928164780139923,
      "learning_rate": 3e-06,
      "loss": 0.679,
      "step": 1227,
      "step_time": 7.979963034391403
    },
    {
      "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,
      "entropy": 0.9444754464285714,
      "epoch": 0.5881226053639846,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05535980686545372,
      "learning_rate": 3e-06,
      "loss": 0.0011,
      "step": 1228,
      "step_time": 7.373115114867687
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.6153157651424408,
      "epoch": 0.5886015325670498,
      "fork_head/loss": 2.7047009468078613,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.06094115972518921,
      "grad_norm": 0.4501667618751526,
      "learning_rate": 3e-06,
      "loss": 0.6748,
      "step": 1229,
      "step_time": 7.757383048534393
    },
    {
      "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,
      "entropy": 0.9310825892857143,
      "epoch": 0.5890804597701149,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05708025023341179,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 1230,
      "step_time": 7.710626184940338
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.6163736879825592,
      "epoch": 0.5895593869731801,
      "fork_head/loss": 2.6854915618896484,
      "fork_head/num_positive": 12.3125,
      "fork_head/pred_top1_mass": 0.06756097078323364,
      "grad_norm": 0.40553826093673706,
      "learning_rate": 3e-06,
      "loss": 0.6693,
      "step": 1231,
      "step_time": 7.4539161548018456
    },
    {
      "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,
      "entropy": 0.93359375,
      "epoch": 0.5900383141762452,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05511977896094322,
      "learning_rate": 3e-06,
      "loss": 0.0005,
      "step": 1232,
      "step_time": 7.572506166994572
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 204.458984375,
      "completions/mean_terminated_length": 204.458984375,
      "completions/min_length": 39.0,
      "completions/min_terminated_length": 39.0,
      "entropy": 1.5786132911841075,
      "epoch": 0.5905172413793104,
      "fork/depth_mean": 0.2783034145832062,
      "fork/depth_std": 0.23671114444732666,
      "fork_head/loss": 2.7302005290985107,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.06978774815797806,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7465428113937378,
      "learning_rate": 3e-06,
      "loss": 0.6908,
      "num_tokens": 32116644.0,
      "reward": 0.9516594409942627,
      "reward_std": 0.5735039114952087,
      "rewards/code_format_reward/mean": 0.7578125,
      "rewards/code_format_reward/std": 0.35242727398872375,
      "rewards/code_reward/mean": 0.1938469409942627,
      "rewards/code_reward/std": 0.3332540690898895,
      "step": 1233,
      "step_time": 485.08003237843513
    },
    {
      "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,
      "entropy": 1.5908647017045454,
      "epoch": 0.5909961685823755,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08991478383541107,
      "learning_rate": 3e-06,
      "loss": -0.0059,
      "step": 1234,
      "step_time": 7.752601005136967
    },
    {
      "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,
      "entropy": 1.5780436197916667,
      "epoch": 0.5914750957854407,
      "fork_head/loss": 2.7145371437072754,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.07784140110015869,
      "grad_norm": 0.6843044757843018,
      "learning_rate": 3e-06,
      "loss": 0.6861,
      "step": 1235,
      "step_time": 7.760538846254349
    },
    {
      "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,
      "entropy": 1.5924627130681819,
      "epoch": 0.5919540229885057,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09807665646076202,
      "learning_rate": 3e-06,
      "loss": -0.0061,
      "step": 1236,
      "step_time": 7.8073302656412125
    },
    {
      "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,
      "entropy": 1.5731879373391469,
      "epoch": 0.5924329501915708,
      "fork_head/loss": 2.6900479793548584,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.08100918680429459,
      "grad_norm": 0.5948209166526794,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 1237,
      "step_time": 7.785027660429478
    },
    {
      "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,
      "entropy": 1.5886008522727273,
      "epoch": 0.592911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09047724306583405,
      "learning_rate": 3e-06,
      "loss": -0.0068,
      "step": 1238,
      "step_time": 7.676516771316528
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777778605620067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777778605620067,
      "entropy": 1.579671214024226,
      "epoch": 0.5933908045977011,
      "fork_head/loss": 2.662510395050049,
      "fork_head/num_positive": 10.8125,
      "fork_head/pred_top1_mass": 0.11774115264415741,
      "grad_norm": 0.5064089298248291,
      "learning_rate": 3e-06,
      "loss": 0.6711,
      "step": 1239,
      "step_time": 7.38732223957777
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.5928622159090908,
      "epoch": 0.5938697318007663,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08194725215435028,
      "learning_rate": 3e-06,
      "loss": -0.0071,
      "step": 1240,
      "step_time": 7.48304259032011
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 201.99609375,
      "completions/mean_terminated_length": 201.89041137695312,
      "completions/min_length": 52.0,
      "completions/min_terminated_length": 52.0,
      "entropy": 1.190039058526357,
      "epoch": 0.5943486590038314,
      "fork/depth_mean": 0.2629584074020386,
      "fork/depth_std": 0.22794362902641296,
      "fork_head/loss": 2.723748207092285,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07718798518180847,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4912604093551636,
      "learning_rate": 3e-06,
      "loss": 0.6921,
      "num_tokens": 32320386.0,
      "reward": 1.1224695444107056,
      "reward_std": 0.6034237146377563,
      "rewards/code_format_reward/mean": 0.818359375,
      "rewards/code_format_reward/std": 0.30841881036758423,
      "rewards/code_reward/mean": 0.30411016941070557,
      "rewards/code_reward/std": 0.4022022485733032,
      "step": 1241,
      "step_time": 592.5921499356627
    },
    {
      "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,
      "entropy": 1.6124674479166667,
      "epoch": 0.5948275862068966,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.22768168151378632,
      "learning_rate": 3e-06,
      "loss": -0.0147,
      "step": 1242,
      "step_time": 7.938585497438908
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.2083984335263571,
      "epoch": 0.5953065134099617,
      "fork_head/loss": 2.716857433319092,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07861856371164322,
      "grad_norm": 0.4817172586917877,
      "learning_rate": 3e-06,
      "loss": 0.6892,
      "step": 1243,
      "step_time": 8.262786738574505
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.5755750834941864,
      "epoch": 0.5957854406130269,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16609647870063782,
      "learning_rate": 3e-06,
      "loss": -0.0164,
      "step": 1244,
      "step_time": 7.778308480978012
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.2181315024693806,
      "epoch": 0.5962643678160919,
      "fork_head/loss": 2.704087495803833,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.0776442363858223,
      "grad_norm": 0.45363911986351013,
      "learning_rate": 3e-06,
      "loss": 0.6856,
      "step": 1245,
      "step_time": 8.145023755729198
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.6130099892616272,
      "epoch": 0.5967432950191571,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.26538702845573425,
      "learning_rate": 3e-06,
      "loss": -0.0178,
      "step": 1246,
      "step_time": 7.366597436368465
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.016666666666666666,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.2370008786519369,
      "epoch": 0.5972222222222222,
      "fork_head/loss": 2.6880369186401367,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07868736982345581,
      "grad_norm": 0.42826032638549805,
      "learning_rate": 3e-06,
      "loss": 0.6805,
      "step": 1247,
      "step_time": 8.129316419363022
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.1111111119389534,
      "clip_ratio/low_mean": 0.09722222263614337,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.20833333457509676,
      "entropy": 1.538194437821706,
      "epoch": 0.5977011494252874,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11436054110527039,
      "learning_rate": 3e-06,
      "loss": -0.0185,
      "step": 1248,
      "step_time": 7.844606176018715
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 209.033203125,
      "completions/mean_terminated_length": 209.033203125,
      "completions/min_length": 47.0,
      "completions/min_terminated_length": 47.0,
      "entropy": 1.4983072996139526,
      "epoch": 0.5981800766283525,
      "fork/depth_mean": 0.20877237617969513,
      "fork/depth_std": 0.19773024320602417,
      "fork_head/loss": 2.7221181392669678,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07116174697875977,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4946906268596649,
      "learning_rate": 3e-06,
      "loss": 0.6852,
      "num_tokens": 32532179.0,
      "reward": 0.9082882404327393,
      "reward_std": 0.5310388207435608,
      "rewards/code_format_reward/mean": 0.75390625,
      "rewards/code_format_reward/std": 0.35387757420539856,
      "rewards/code_reward/mean": 0.15438202023506165,
      "rewards/code_reward/std": 0.2967389225959778,
      "step": 1249,
      "step_time": 608.6532326117158
    },
    {
      "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,
      "entropy": 1.8461588501930237,
      "epoch": 0.5986590038314177,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11045681685209274,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 1250,
      "step_time": 8.07956600934267
    },
    {
      "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,
      "entropy": 1.516796875,
      "epoch": 0.5991379310344828,
      "fork_head/loss": 2.713318109512329,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07175112515687943,
      "grad_norm": 0.4808366596698761,
      "learning_rate": 3e-06,
      "loss": 0.6818,
      "step": 1251,
      "step_time": 7.909421317279339
    },
    {
      "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,
      "entropy": 1.8412760496139526,
      "epoch": 0.5996168582375478,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10212449729442596,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 1252,
      "step_time": 7.954533897340298
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.5218098878860473,
      "epoch": 0.600095785440613,
      "fork_head/loss": 2.699192523956299,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07588115334510803,
      "grad_norm": 0.4581367075443268,
      "learning_rate": 3e-06,
      "loss": 0.6774,
      "step": 1253,
      "step_time": 7.157216787338257
    },
    {
      "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,
      "entropy": 1.8578776121139526,
      "epoch": 0.6005747126436781,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09305202215909958,
      "learning_rate": 3e-06,
      "loss": -0.0042,
      "step": 1254,
      "step_time": 7.836392283439636
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.15,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.15,
      "entropy": 1.539013671875,
      "epoch": 0.6010536398467433,
      "fork_head/loss": 2.6828274726867676,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07270865887403488,
      "grad_norm": 0.405984491109848,
      "learning_rate": 3e-06,
      "loss": 0.6725,
      "step": 1255,
      "step_time": 7.878757558763027
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.8531901001930238,
      "epoch": 0.6015325670498084,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07773242145776749,
      "learning_rate": 3e-06,
      "loss": -0.0048,
      "step": 1256,
      "step_time": 7.1511385217309
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 199.806640625,
      "completions/mean_terminated_length": 199.806640625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.4240944602272727,
      "epoch": 0.6020114942528736,
      "fork/depth_mean": 0.25877660512924194,
      "fork/depth_std": 0.2432098686695099,
      "fork_head/loss": 2.737292528152466,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.06590335816144943,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5890001058578491,
      "learning_rate": 3e-06,
      "loss": 0.6955,
      "num_tokens": 32728272.0,
      "reward": 0.9675644636154175,
      "reward_std": 0.61105877161026,
      "rewards/code_format_reward/mean": 0.7314453125,
      "rewards/code_format_reward/std": 0.3683244287967682,
      "rewards/code_reward/mean": 0.23611918091773987,
      "rewards/code_reward/std": 0.38217657804489136,
      "step": 1257,
      "step_time": 509.58246909826994
    },
    {
      "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,
      "entropy": 1.760860562324524,
      "epoch": 0.6024904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.120705746114254,
      "learning_rate": 3e-06,
      "loss": -0.013,
      "step": 1258,
      "step_time": 7.712301559746265
    },
    {
      "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,
      "entropy": 1.4272904829545454,
      "epoch": 0.6029693486590039,
      "fork_head/loss": 2.728769302368164,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.06543350964784622,
      "grad_norm": 0.4651166796684265,
      "learning_rate": 3e-06,
      "loss": 0.692,
      "step": 1259,
      "step_time": 7.95987705886364
    },
    {
      "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,
      "entropy": 1.7362393465909092,
      "epoch": 0.603448275862069,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1319141834974289,
      "learning_rate": 3e-06,
      "loss": -0.0148,
      "step": 1260,
      "step_time": 7.26655075699091
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.13636363636363635,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.4356356534090908,
      "epoch": 0.6039272030651341,
      "fork_head/loss": 2.7173542976379395,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.06879982352256775,
      "grad_norm": 0.42004886269569397,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "step": 1261,
      "step_time": 7.648788847029209
    },
    {
      "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,
      "entropy": 1.710878307169134,
      "epoch": 0.6044061302681992,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1374126821756363,
      "learning_rate": 3e-06,
      "loss": -0.0155,
      "step": 1262,
      "step_time": 8.35694470256567
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.22727272727272727,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.3181818181818182,
      "entropy": 1.4476207386363635,
      "epoch": 0.6048850574712644,
      "fork_head/loss": 2.704364776611328,
      "fork_head/num_positive": 12.5,
      "fork_head/pred_top1_mass": 0.07576461136341095,
      "grad_norm": 0.37124577164649963,
      "learning_rate": 3e-06,
      "loss": 0.6818,
      "step": 1263,
      "step_time": 7.345022700726986
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.03030303120613098,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.69140625,
      "epoch": 0.6053639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07436004281044006,
      "learning_rate": 3e-06,
      "loss": -0.0161,
      "step": 1264,
      "step_time": 7.853051483631134
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 191.751953125,
      "completions/mean_terminated_length": 191.626220703125,
      "completions/min_length": 45.0,
      "completions/min_terminated_length": 45.0,
      "entropy": 1.3824425827373157,
      "epoch": 0.6058429118773946,
      "fork/depth_mean": 0.26095542311668396,
      "fork/depth_std": 0.23214635252952576,
      "fork_head/loss": 2.7252845764160156,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.0712287425994873,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5810605883598328,
      "learning_rate": 3e-06,
      "loss": 0.6898,
      "num_tokens": 32909377.0,
      "reward": 1.0104905366897583,
      "reward_std": 0.5322887301445007,
      "rewards/code_format_reward/mean": 0.8037109375,
      "rewards/code_format_reward/std": 0.30815067887306213,
      "rewards/code_reward/mean": 0.2067796289920807,
      "rewards/code_reward/std": 0.3389192223548889,
      "step": 1265,
      "step_time": 508.80919655412436
    },
    {
      "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,
      "entropy": 1.68931748966376,
      "epoch": 0.6063218390804598,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20311018824577332,
      "learning_rate": 3e-06,
      "loss": -0.0034,
      "step": 1266,
      "step_time": 8.431553229689598
    },
    {
      "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,
      "entropy": 1.3957346894524314,
      "epoch": 0.6068007662835249,
      "fork_head/loss": 2.7125091552734375,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.07835260778665543,
      "grad_norm": 0.5340934991836548,
      "learning_rate": 3e-06,
      "loss": 0.6853,
      "step": 1267,
      "step_time": 7.6036840453743935
    },
    {
      "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,
      "entropy": 1.7072211255629857,
      "epoch": 0.60727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14145171642303467,
      "learning_rate": 3e-06,
      "loss": -0.005,
      "step": 1268,
      "step_time": 7.3613101541996
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.4084891947832974,
      "epoch": 0.6077586206896551,
      "fork_head/loss": 2.692551851272583,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.07085265219211578,
      "grad_norm": 0.4727090001106262,
      "learning_rate": 3e-06,
      "loss": 0.6802,
      "step": 1269,
      "step_time": 7.542431563138962
    },
    {
      "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,
      "entropy": 1.6804199268420537,
      "epoch": 0.6082375478927203,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1356913298368454,
      "learning_rate": 3e-06,
      "loss": -0.0057,
      "step": 1270,
      "step_time": 7.832564763724804
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.07575757666067644,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07575757666067644,
      "entropy": 1.4212584820660679,
      "epoch": 0.6087164750957854,
      "fork_head/loss": 2.670293092727661,
      "fork_head/num_positive": 11.53125,
      "fork_head/pred_top1_mass": 0.0852895975112915,
      "grad_norm": 0.42028990387916565,
      "learning_rate": 3e-06,
      "loss": 0.6743,
      "step": 1271,
      "step_time": 7.718599081039429
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777778605620067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777778605620067,
      "entropy": 1.6851399888594945,
      "epoch": 0.6091954022988506,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11023169010877609,
      "learning_rate": 3e-06,
      "loss": -0.0065,
      "step": 1272,
      "step_time": 7.48003064095974
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 200.6640625,
      "completions/mean_terminated_length": 200.55577087402344,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.3355902791023255,
      "epoch": 0.6096743295019157,
      "fork/depth_mean": 0.3021516501903534,
      "fork/depth_std": 0.24048469960689545,
      "fork_head/loss": 2.7296500205993652,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.060847677290439606,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6643913984298706,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "num_tokens": 33108277.0,
      "reward": 1.0627479553222656,
      "reward_std": 0.6091307401657104,
      "rewards/code_format_reward/mean": 0.79296875,
      "rewards/code_format_reward/std": 0.33706074953079224,
      "rewards/code_reward/mean": 0.2697790861129761,
      "rewards/code_reward/std": 0.39489662647247314,
      "step": 1273,
      "step_time": 425.5319680944085
    },
    {
      "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,
      "entropy": 1.6472005367279052,
      "epoch": 0.6101532567049809,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21941567957401276,
      "learning_rate": 3e-06,
      "loss": 0.0036,
      "step": 1274,
      "step_time": 8.138485558331013
    },
    {
      "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,
      "entropy": 1.3454210082689921,
      "epoch": 0.610632183908046,
      "fork_head/loss": 2.717421770095825,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.06437502801418304,
      "grad_norm": 0.5996183753013611,
      "learning_rate": 3e-06,
      "loss": 0.6717,
      "step": 1275,
      "step_time": 8.190852604806423
    },
    {
      "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,
      "entropy": 1.6291449626286825,
      "epoch": 0.6111111111111112,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.30141469836235046,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1276,
      "step_time": 7.391383722424507
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.3483561197916667,
      "epoch": 0.6115900383141762,
      "fork_head/loss": 2.6977639198303223,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.07313107699155807,
      "grad_norm": 0.5123305916786194,
      "learning_rate": 3e-06,
      "loss": 0.6653,
      "step": 1277,
      "step_time": 8.622748255729675
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.06666666666666667,
      "clip_ratio/low_mean": 0.044444445768992105,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.11111111243565877,
      "entropy": 1.5926866292953492,
      "epoch": 0.6120689655172413,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1973695456981659,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 1278,
      "step_time": 7.750348217785358
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.016666666666666666,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.016666666666666666,
      "entropy": 1.327734375,
      "epoch": 0.6125478927203065,
      "fork_head/loss": 2.676365613937378,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.08809874206781387,
      "grad_norm": 0.4688320457935333,
      "learning_rate": 3e-06,
      "loss": 0.6586,
      "step": 1279,
      "step_time": 7.874524615705013
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.08888888955116273,
      "clip_ratio/low_mean": 0.08888889054457347,
      "clip_ratio/low_min": 0.03333333333333333,
      "clip_ratio/region_mean": 0.1777777800957362,
      "entropy": 1.5768663167953492,
      "epoch": 0.6130268199233716,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15050119161605835,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 1280,
      "step_time": 7.638955935835838
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 196.669921875,
      "completions/mean_terminated_length": 196.43727111816406,
      "completions/min_length": 71.0,
      "completions/min_terminated_length": 71.0,
      "entropy": 1.4100341846545537,
      "epoch": 0.6135057471264368,
      "fork/depth_mean": 0.25556275248527527,
      "fork/depth_std": 0.2364576756954193,
      "fork_head/loss": 2.727785587310791,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.0739019438624382,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7222157120704651,
      "learning_rate": 3e-06,
      "loss": 0.6733,
      "num_tokens": 33306508.0,
      "reward": 0.9159694910049438,
      "reward_std": 0.602653980255127,
      "rewards/code_format_reward/mean": 0.7177734375,
      "rewards/code_format_reward/std": 0.3855721354484558,
      "rewards/code_reward/mean": 0.19819605350494385,
      "rewards/code_reward/std": 0.35171300172805786,
      "step": 1281,
      "step_time": 614.4978936761618
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.022222222884496053,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.6012228727340698,
      "epoch": 0.6139846743295019,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20048366487026215,
      "learning_rate": 3e-06,
      "loss": 0.0028,
      "step": 1282,
      "step_time": 8.237967193126678
    },
    {
      "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,
      "entropy": 1.3992648671070735,
      "epoch": 0.6144636015325671,
      "fork_head/loss": 2.711284875869751,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.09539781510829926,
      "grad_norm": 0.6685845851898193,
      "learning_rate": 3e-06,
      "loss": 0.6683,
      "step": 1283,
      "step_time": 7.594122089445591
    },
    {
      "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,
      "entropy": 1.5734939257303873,
      "epoch": 0.6149425287356322,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19208170473575592,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 1284,
      "step_time": 7.766638949513435
    },
    {
      "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,
      "entropy": 1.4056125183900197,
      "epoch": 0.6154214559386973,
      "fork_head/loss": 2.686264753341675,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.09230707585811615,
      "grad_norm": 0.5716957449913025,
      "learning_rate": 3e-06,
      "loss": 0.6617,
      "step": 1285,
      "step_time": 7.697835944592953
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.013333333532015483,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.013333333532015483,
      "entropy": 1.583642578125,
      "epoch": 0.6159003831417624,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19839312136173248,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 1286,
      "step_time": 7.750511854887009
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.020833333333333332,
      "clip_ratio/low_mean": 0.027777778605620067,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0486111119389534,
      "entropy": 1.4039984792470932,
      "epoch": 0.6163793103448276,
      "fork_head/loss": 2.6605756282806396,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.0869695395231247,
      "grad_norm": 0.46972954273223877,
      "learning_rate": 3e-06,
      "loss": 0.6536,
      "step": 1287,
      "step_time": 7.645483992993832
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.03333333333333333,
      "clip_ratio/low_mean": 0.022222222884496053,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.055555556217829385,
      "entropy": 1.5957219362258912,
      "epoch": 0.6168582375478927,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14829620718955994,
      "learning_rate": 3e-06,
      "loss": -0.0005,
      "step": 1288,
      "step_time": 7.546638913452625
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 197.759765625,
      "completions/mean_terminated_length": 197.6457977294922,
      "completions/min_length": 66.0,
      "completions/min_terminated_length": 66.0,
      "entropy": 1.682996968428294,
      "epoch": 0.6173371647509579,
      "fork/depth_mean": 0.23611529171466827,
      "fork/depth_std": 0.2007407248020172,
      "fork_head/loss": 2.7270777225494385,
      "fork_head/num_positive": 12.322580337524414,
      "fork_head/pred_top1_mass": 0.06804967671632767,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5282750725746155,
      "learning_rate": 3e-06,
      "loss": 0.6781,
      "num_tokens": 33497617.0,
      "reward": 1.0611259937286377,
      "reward_std": 0.6262097954750061,
      "rewards/code_format_reward/mean": 0.77734375,
      "rewards/code_format_reward/std": 0.3458368182182312,
      "rewards/code_reward/mean": 0.2837821841239929,
      "rewards/code_reward/std": 0.39427992701530457,
      "step": 1289,
      "step_time": 460.18395460397005
    },
    {
      "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,
      "entropy": 1.3912760416666667,
      "epoch": 0.617816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2338375747203827,
      "learning_rate": 3e-06,
      "loss": 0.006,
      "step": 1290,
      "step_time": 8.455193139612675
    },
    {
      "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,
      "entropy": 1.670030375321706,
      "epoch": 0.6182950191570882,
      "fork_head/loss": 2.7197420597076416,
      "fork_head/num_positive": 12.322580337524414,
      "fork_head/pred_top1_mass": 0.06522884964942932,
      "grad_norm": 0.5162252187728882,
      "learning_rate": 3e-06,
      "loss": 0.6752,
      "step": 1291,
      "step_time": 7.867668755352497
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.06666666666666667,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.4013780355453491,
      "epoch": 0.6187739463601533,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19219473004341125,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 1292,
      "step_time": 7.691857345402241
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.6275499165058136,
      "epoch": 0.6192528735632183,
      "fork_head/loss": 2.705883264541626,
      "fork_head/num_positive": 12.322580337524414,
      "fork_head/pred_top1_mass": 0.06972744315862656,
      "grad_norm": 0.4715343415737152,
      "learning_rate": 3e-06,
      "loss": 0.6702,
      "step": 1293,
      "step_time": 7.920236706733704
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.06666666666666667,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13333333333333333,
      "entropy": 1.4227647542953492,
      "epoch": 0.6197318007662835,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14290593564510345,
      "learning_rate": 3e-06,
      "loss": 0.0029,
      "step": 1294,
      "step_time": 7.357550732791424
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.6082899272441864,
      "epoch": 0.6202107279693486,
      "fork_head/loss": 2.6893744468688965,
      "fork_head/num_positive": 12.322580337524414,
      "fork_head/pred_top1_mass": 0.08003610372543335,
      "grad_norm": 0.4151245653629303,
      "learning_rate": 3e-06,
      "loss": 0.6646,
      "step": 1295,
      "step_time": 7.712658226490021
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.11666666666666667,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.2,
      "entropy": 1.421202254295349,
      "epoch": 0.6206896551724138,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06969907134771347,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 1296,
      "step_time": 7.628276094794273
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 212.751953125,
      "completions/mean_terminated_length": 212.6673126220703,
      "completions/min_length": 64.0,
      "completions/min_terminated_length": 64.0,
      "entropy": 1.9760298295454546,
      "epoch": 0.6211685823754789,
      "fork/depth_mean": 0.2996405065059662,
      "fork/depth_std": 0.26254910230636597,
      "fork_head/loss": 2.7238240242004395,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06334158778190613,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.470645546913147,
      "learning_rate": 3e-06,
      "loss": 0.6739,
      "num_tokens": 33702242.0,
      "reward": 0.9960950613021851,
      "reward_std": 0.6330500841140747,
      "rewards/code_format_reward/mean": 0.7314453125,
      "rewards/code_format_reward/std": 0.39025238156318665,
      "rewards/code_reward/mean": 0.26464974880218506,
      "rewards/code_reward/std": 0.39547500014305115,
      "step": 1297,
      "step_time": 372.27783570438623
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.745008691151937,
      "epoch": 0.6216475095785441,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09446755796670914,
      "learning_rate": 3e-06,
      "loss": 0.0068,
      "step": 1298,
      "step_time": 7.626780115067959
    },
    {
      "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,
      "entropy": 1.9665897326035933,
      "epoch": 0.6221264367816092,
      "fork_head/loss": 2.7135119438171387,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07492092996835709,
      "grad_norm": 0.4510795474052429,
      "learning_rate": 3e-06,
      "loss": 0.6708,
      "step": 1299,
      "step_time": 7.57541523873806
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.7490885337193807,
      "epoch": 0.6226053639846744,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09731665253639221,
      "learning_rate": 3e-06,
      "loss": 0.0069,
      "step": 1300,
      "step_time": 8.208527691662312
    },
    {
      "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,
      "entropy": 1.9367305798964067,
      "epoch": 0.6230842911877394,
      "fork_head/loss": 2.6975185871124268,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07923697680234909,
      "grad_norm": 0.42991945147514343,
      "learning_rate": 3e-06,
      "loss": 0.6663,
      "step": 1301,
      "step_time": 7.569353483617306
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.7662218014399211,
      "epoch": 0.6235632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09277286380529404,
      "learning_rate": 3e-06,
      "loss": 0.0059,
      "step": 1302,
      "step_time": 7.782640606164932
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.9245087666945024,
      "epoch": 0.6240421455938697,
      "fork_head/loss": 2.678539276123047,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.08186698704957962,
      "grad_norm": 0.40507280826568604,
      "learning_rate": 3e-06,
      "loss": 0.661,
      "step": 1303,
      "step_time": 7.8225531578063965
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.7602322181065877,
      "epoch": 0.6245210727969349,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09465325623750687,
      "learning_rate": 3e-06,
      "loss": 0.0054,
      "step": 1304,
      "step_time": 7.448671638965607
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 198.580078125,
      "completions/mean_terminated_length": 198.580078125,
      "completions/min_length": 26.0,
      "completions/min_terminated_length": 26.0,
      "entropy": 1.508130786816279,
      "epoch": 0.625,
      "fork/depth_mean": 0.34862709045410156,
      "fork/depth_std": 0.2831815779209137,
      "fork_head/loss": 2.7278056144714355,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.06747990846633911,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.47953352332115173,
      "learning_rate": 3e-06,
      "loss": 0.6742,
      "num_tokens": 33894955.0,
      "reward": 0.9802225828170776,
      "reward_std": 0.5934088826179504,
      "rewards/code_format_reward/mean": 0.7587890625,
      "rewards/code_format_reward/std": 0.3338691294193268,
      "rewards/code_reward/mean": 0.22143349051475525,
      "rewards/code_reward/std": 0.3761115074157715,
      "step": 1305,
      "step_time": 528.0989399328828
    },
    {
      "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,
      "entropy": 1.8264973958333333,
      "epoch": 0.6254789272030651,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14369140565395355,
      "learning_rate": 3e-06,
      "loss": 0.013,
      "step": 1306,
      "step_time": 7.503056980669498
    },
    {
      "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,
      "entropy": 1.4923828125,
      "epoch": 0.6259578544061303,
      "fork_head/loss": 2.7209112644195557,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.07955241203308105,
      "grad_norm": 0.4793788492679596,
      "learning_rate": 3e-06,
      "loss": 0.6714,
      "step": 1307,
      "step_time": 8.51169328391552
    },
    {
      "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,
      "entropy": 1.8270670572916667,
      "epoch": 0.6264367816091954,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.137701615691185,
      "learning_rate": 3e-06,
      "loss": 0.0129,
      "step": 1308,
      "step_time": 7.366413816809654
    },
    {
      "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,
      "entropy": 1.4812138338883718,
      "epoch": 0.6269157088122606,
      "fork_head/loss": 2.7085001468658447,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.08192708343267441,
      "grad_norm": 0.45767804980278015,
      "learning_rate": 3e-06,
      "loss": 0.6662,
      "step": 1309,
      "step_time": 8.364381320774555
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.859375,
      "epoch": 0.6273946360153256,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12888748943805695,
      "learning_rate": 3e-06,
      "loss": 0.0104,
      "step": 1310,
      "step_time": 7.081967145204544
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.06666666666666667,
      "clip_ratio/low_mean": 0.022222222884496053,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08888888955116273,
      "entropy": 1.4787109375,
      "epoch": 0.6278735632183908,
      "fork_head/loss": 2.6913654804229736,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.0938798114657402,
      "grad_norm": 0.40800946950912476,
      "learning_rate": 3e-06,
      "loss": 0.6595,
      "step": 1311,
      "step_time": 8.260319404304028
    },
    {
      "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,
      "entropy": 1.8624674479166667,
      "epoch": 0.6283524904214559,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1480175405740738,
      "learning_rate": 3e-06,
      "loss": 0.0099,
      "step": 1312,
      "step_time": 7.442023858428001
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 191.25390625,
      "completions/mean_terminated_length": 191.127197265625,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "entropy": 1.7154651988636365,
      "epoch": 0.6288314176245211,
      "fork/depth_mean": 0.34289440512657166,
      "fork/depth_std": 0.2700771689414978,
      "fork_head/loss": 2.720181703567505,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.07064446806907654,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5531618595123291,
      "learning_rate": 3e-06,
      "loss": 0.69,
      "num_tokens": 34083405.0,
      "reward": 1.0373616218566895,
      "reward_std": 0.6234892010688782,
      "rewards/code_format_reward/mean": 0.78125,
      "rewards/code_format_reward/std": 0.34550508856773376,
      "rewards/code_reward/mean": 0.25611165165901184,
      "rewards/code_reward/std": 0.4009593427181244,
      "step": 1313,
      "step_time": 456.22875718027353
    },
    {
      "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,
      "entropy": 1.49111328125,
      "epoch": 0.6293103448275862,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.408583402633667,
      "learning_rate": 3e-06,
      "loss": -0.0058,
      "step": 1314,
      "step_time": 7.654555328190327
    },
    {
      "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,
      "entropy": 1.7181285511363635,
      "epoch": 0.6297892720306514,
      "fork_head/loss": 2.707509994506836,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.08282244205474854,
      "grad_norm": 0.5102949738502502,
      "learning_rate": 3e-06,
      "loss": 0.6869,
      "step": 1315,
      "step_time": 7.850399762392044
    },
    {
      "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,
      "entropy": 1.4916015625,
      "epoch": 0.6302681992337165,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19268769025802612,
      "learning_rate": 3e-06,
      "loss": -0.0083,
      "step": 1316,
      "step_time": 7.819357484579086
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.022727272727272728,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022727272727272728,
      "entropy": 1.7147105823863635,
      "epoch": 0.6307471264367817,
      "fork_head/loss": 2.6904921531677246,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.0824677050113678,
      "grad_norm": 0.4633443057537079,
      "learning_rate": 3e-06,
      "loss": 0.6827,
      "step": 1317,
      "step_time": 7.0584172904491425
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.49052734375,
      "epoch": 0.6312260536398467,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15438207983970642,
      "learning_rate": 3e-06,
      "loss": -0.0096,
      "step": 1318,
      "step_time": 7.723993271589279
    },
    {
      "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,
      "entropy": 1.70654296875,
      "epoch": 0.6317049808429118,
      "fork_head/loss": 2.670279026031494,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.0963497906923294,
      "grad_norm": 0.42504778504371643,
      "learning_rate": 3e-06,
      "loss": 0.677,
      "step": 1319,
      "step_time": 7.382658712565899
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.15,
      "entropy": 1.46787109375,
      "epoch": 0.632183908045977,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11202124506235123,
      "learning_rate": 3e-06,
      "loss": -0.0109,
      "step": 1320,
      "step_time": 7.322004958987236
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 196.9453125,
      "completions/mean_terminated_length": 196.71372985839844,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 2.0063920454545454,
      "epoch": 0.6326628352490421,
      "fork/depth_mean": 0.3078047037124634,
      "fork/depth_std": 0.2584734857082367,
      "fork_head/loss": 2.720284938812256,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.07190078496932983,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.44805392622947693,
      "learning_rate": 3e-06,
      "loss": 0.67,
      "num_tokens": 34279281.0,
      "reward": 1.048086404800415,
      "reward_std": 0.655335545539856,
      "rewards/code_format_reward/mean": 0.759765625,
      "rewards/code_format_reward/std": 0.36196672916412354,
      "rewards/code_reward/mean": 0.28832072019577026,
      "rewards/code_reward/std": 0.4019969403743744,
      "step": 1321,
      "step_time": 387.38564989715815
    },
    {
      "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,
      "entropy": 1.5415482954545454,
      "epoch": 0.6331417624521073,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07247398793697357,
      "learning_rate": 3e-06,
      "loss": 0.0108,
      "step": 1322,
      "step_time": 7.504920341074467
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.045454545454545456,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 2.0172230113636362,
      "epoch": 0.6336206896551724,
      "fork_head/loss": 2.7119369506835938,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.07583064585924149,
      "grad_norm": 0.4202962815761566,
      "learning_rate": 3e-06,
      "loss": 0.6678,
      "step": 1323,
      "step_time": 8.647040642797947
    },
    {
      "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,
      "entropy": 1.5432350852272727,
      "epoch": 0.6340996168582376,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07120251655578613,
      "learning_rate": 3e-06,
      "loss": 0.0105,
      "step": 1324,
      "step_time": 7.3127778097987175
    },
    {
      "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,
      "entropy": 2.020744562149048,
      "epoch": 0.6345785440613027,
      "fork_head/loss": 2.6995997428894043,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.0822952538728714,
      "grad_norm": 0.4070238769054413,
      "learning_rate": 3e-06,
      "loss": 0.6643,
      "step": 1325,
      "step_time": 7.585196897387505
    },
    {
      "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,
      "entropy": 1.5500118298964067,
      "epoch": 0.6350574712643678,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06976013630628586,
      "learning_rate": 3e-06,
      "loss": 0.01,
      "step": 1326,
      "step_time": 8.099083371460438
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.045454545454545456,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 2.0238813920454546,
      "epoch": 0.6355363984674329,
      "fork_head/loss": 2.683549642562866,
      "fork_head/num_positive": 11.84375,
      "fork_head/pred_top1_mass": 0.08429475873708725,
      "grad_norm": 0.3962523341178894,
      "learning_rate": 3e-06,
      "loss": 0.6604,
      "step": 1327,
      "step_time": 7.319538049399853
    },
    {
      "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,
      "entropy": 1.5487985394217751,
      "epoch": 0.6360153256704981,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06856725364923477,
      "learning_rate": 3e-06,
      "loss": 0.0096,
      "step": 1328,
      "step_time": 7.607748366892338
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 209.50390625,
      "completions/mean_terminated_length": 209.4129180908203,
      "completions/min_length": 41.0,
      "completions/min_terminated_length": 41.0,
      "entropy": 1.58994140625,
      "epoch": 0.6364942528735632,
      "fork/depth_mean": 0.25671660900115967,
      "fork/depth_std": 0.22559943795204163,
      "fork_head/loss": 2.7317347526550293,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07228148728609085,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6298647522926331,
      "learning_rate": 3e-06,
      "loss": 0.679,
      "num_tokens": 34478963.0,
      "reward": 1.051497459411621,
      "reward_std": 0.56549471616745,
      "rewards/code_format_reward/mean": 0.8125,
      "rewards/code_format_reward/std": 0.3143657445907593,
      "rewards/code_reward/mean": 0.23899739980697632,
      "rewards/code_reward/std": 0.3773312270641327,
      "step": 1329,
      "step_time": 443.28919779509306
    },
    {
      "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,
      "entropy": 1.9476318359375,
      "epoch": 0.6369731800766284,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09876218438148499,
      "learning_rate": 3e-06,
      "loss": 0.0043,
      "step": 1330,
      "step_time": 8.689488887786865
    },
    {
      "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,
      "entropy": 1.577734375,
      "epoch": 0.6374521072796935,
      "fork_head/loss": 2.7193925380706787,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07787003368139267,
      "grad_norm": 0.5501748919487,
      "learning_rate": 3e-06,
      "loss": 0.6757,
      "step": 1331,
      "step_time": 7.42064618319273
    },
    {
      "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,
      "entropy": 1.98876953125,
      "epoch": 0.6379310344827587,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09590617567300797,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 1332,
      "step_time": 7.373455718159676
    },
    {
      "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,
      "entropy": 1.57021484375,
      "epoch": 0.6384099616858238,
      "fork_head/loss": 2.7020468711853027,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.0775848850607872,
      "grad_norm": 0.4505503475666046,
      "learning_rate": 3e-06,
      "loss": 0.6703,
      "step": 1333,
      "step_time": 8.002647273242474
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.9700927734375,
      "epoch": 0.6388888888888888,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05777670815587044,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1334,
      "step_time": 7.037509873509407
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.56875,
      "epoch": 0.639367816091954,
      "fork_head/loss": 2.6839230060577393,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07926002144813538,
      "grad_norm": 0.41374653577804565,
      "learning_rate": 3e-06,
      "loss": 0.665,
      "step": 1335,
      "step_time": 7.9503347128629684
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.9610595703125,
      "epoch": 0.6398467432950191,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.057757195085287094,
      "learning_rate": 3e-06,
      "loss": 0.0029,
      "step": 1336,
      "step_time": 7.370636627078056
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 204.275390625,
      "completions/mean_terminated_length": 204.275390625,
      "completions/min_length": 60.0,
      "completions/min_terminated_length": 60.0,
      "entropy": 1.2659801136363635,
      "epoch": 0.6403256704980843,
      "fork/depth_mean": 0.24053026735782623,
      "fork/depth_std": 0.19587865471839905,
      "fork_head/loss": 2.7221503257751465,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.056847795844078064,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5248513221740723,
      "learning_rate": 3e-06,
      "loss": 0.679,
      "num_tokens": 34672880.0,
      "reward": 1.0259957313537598,
      "reward_std": 0.5805543065071106,
      "rewards/code_format_reward/mean": 0.798828125,
      "rewards/code_format_reward/std": 0.34629517793655396,
      "rewards/code_reward/mean": 0.22716768085956573,
      "rewards/code_reward/std": 0.3573475182056427,
      "step": 1337,
      "step_time": 511.79695146530867
    },
    {
      "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,
      "entropy": 1.29345703125,
      "epoch": 0.6408045977011494,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15720747411251068,
      "learning_rate": 3e-06,
      "loss": 0.0042,
      "step": 1338,
      "step_time": 7.418395444750786
    },
    {
      "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,
      "entropy": 1.2419211647727273,
      "epoch": 0.6412835249042146,
      "fork_head/loss": 2.7131752967834473,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.06537280976772308,
      "grad_norm": 0.5176728367805481,
      "learning_rate": 3e-06,
      "loss": 0.6768,
      "step": 1339,
      "step_time": 7.794783897697926
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.3009114503860473,
      "epoch": 0.6417624521072797,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15908409655094147,
      "learning_rate": 3e-06,
      "loss": 0.0029,
      "step": 1340,
      "step_time": 7.445944212377071
    },
    {
      "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,
      "entropy": 1.2565696022727273,
      "epoch": 0.6422413793103449,
      "fork_head/loss": 2.6971495151519775,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08490607142448425,
      "grad_norm": 0.48782533407211304,
      "learning_rate": 3e-06,
      "loss": 0.6719,
      "step": 1341,
      "step_time": 7.471238784492016
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.2878580689430237,
      "epoch": 0.64272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.157088041305542,
      "learning_rate": 3e-06,
      "loss": 0.0016,
      "step": 1342,
      "step_time": 7.603275537490845
    },
    {
      "clip_ratio/high_max": 0.045454545454545456,
      "clip_ratio/high_mean": 0.022727272727272728,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.11363636363636363,
      "entropy": 1.2866654829545454,
      "epoch": 0.6431992337164751,
      "fork_head/loss": 2.675053834915161,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.09175540506839752,
      "grad_norm": 0.44934937357902527,
      "learning_rate": 3e-06,
      "loss": 0.6657,
      "step": 1343,
      "step_time": 8.067017644643784
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.15,
      "entropy": 1.2801432251930236,
      "epoch": 0.6436781609195402,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10160642117261887,
      "learning_rate": 3e-06,
      "loss": 0.0007,
      "step": 1344,
      "step_time": 7.609432891011238
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 200.4140625,
      "completions/mean_terminated_length": 200.30528259277344,
      "completions/min_length": 47.0,
      "completions/min_terminated_length": 47.0,
      "entropy": 1.3998480902777777,
      "epoch": 0.6441570881226054,
      "fork/depth_mean": 0.2402515709400177,
      "fork/depth_std": 0.21460725367069244,
      "fork_head/loss": 2.720726728439331,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.07807105779647827,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4034910500049591,
      "learning_rate": 3e-06,
      "loss": 0.6778,
      "num_tokens": 34869620.0,
      "reward": 1.0007567405700684,
      "reward_std": 0.6004472374916077,
      "rewards/code_format_reward/mean": 0.7607421875,
      "rewards/code_format_reward/std": 0.3308676779270172,
      "rewards/code_reward/mean": 0.24001452326774597,
      "rewards/code_reward/std": 0.394407719373703,
      "step": 1345,
      "step_time": 437.89963145554066
    },
    {
      "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,
      "entropy": 1.84951171875,
      "epoch": 0.6446360153256705,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10568832606077194,
      "learning_rate": 3e-06,
      "loss": 0.004,
      "step": 1346,
      "step_time": 7.994163788855076
    },
    {
      "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,
      "entropy": 1.3454861111111112,
      "epoch": 0.6451149425287356,
      "fork_head/loss": 2.7148094177246094,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.07864190638065338,
      "grad_norm": 0.3851127028465271,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "step": 1347,
      "step_time": 8.161041356623173
    },
    {
      "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,
      "entropy": 1.83447265625,
      "epoch": 0.6455938697318008,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08703666925430298,
      "learning_rate": 3e-06,
      "loss": 0.0035,
      "step": 1348,
      "step_time": 7.207193121314049
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05555555555555555,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.3659939236111112,
      "epoch": 0.6460727969348659,
      "fork_head/loss": 2.704807996749878,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.07627778500318527,
      "grad_norm": 0.3757297992706299,
      "learning_rate": 3e-06,
      "loss": 0.6727,
      "step": 1349,
      "step_time": 7.8086492866277695
    },
    {
      "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,
      "entropy": 1.86083984375,
      "epoch": 0.646551724137931,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0945829451084137,
      "learning_rate": 3e-06,
      "loss": 0.0032,
      "step": 1350,
      "step_time": 7.372463844716549
    },
    {
      "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,
      "entropy": 1.3366970486111112,
      "epoch": 0.6470306513409961,
      "fork_head/loss": 2.6920316219329834,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.08022256195545197,
      "grad_norm": 0.35716670751571655,
      "learning_rate": 3e-06,
      "loss": 0.6686,
      "step": 1351,
      "step_time": 7.593595437705517
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.85478515625,
      "epoch": 0.6475095785440613,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08135408908128738,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 1352,
      "step_time": 7.79392122477293
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 208.083984375,
      "completions/mean_terminated_length": 208.083984375,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.30810546875,
      "epoch": 0.6479885057471264,
      "fork/depth_mean": 0.23418034613132477,
      "fork/depth_std": 0.20491275191307068,
      "fork_head/loss": 2.720308780670166,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.07311040908098221,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.423724889755249,
      "learning_rate": 3e-06,
      "loss": 0.6815,
      "num_tokens": 35068383.0,
      "reward": 0.9141353368759155,
      "reward_std": 0.5489258766174316,
      "rewards/code_format_reward/mean": 0.751953125,
      "rewards/code_format_reward/std": 0.3538937568664551,
      "rewards/code_reward/mean": 0.1621822565793991,
      "rewards/code_reward/std": 0.3162064254283905,
      "step": 1353,
      "step_time": 488.64906959980726
    },
    {
      "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,
      "entropy": 1.797119140625,
      "epoch": 0.6484674329501916,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06275929510593414,
      "learning_rate": 3e-06,
      "loss": -0.0,
      "step": 1354,
      "step_time": 7.278231367468834
    },
    {
      "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,
      "entropy": 1.333984375,
      "epoch": 0.6489463601532567,
      "fork_head/loss": 2.7125062942504883,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.07666046917438507,
      "grad_norm": 0.4002985954284668,
      "learning_rate": 3e-06,
      "loss": 0.6792,
      "step": 1355,
      "step_time": 7.604305103421211
    },
    {
      "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,
      "entropy": 1.757080078125,
      "epoch": 0.6494252873563219,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11932183057069778,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 1356,
      "step_time": 7.801400855183601
    },
    {
      "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,
      "entropy": 1.287109375,
      "epoch": 0.649904214559387,
      "fork_head/loss": 2.7010202407836914,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.08277522772550583,
      "grad_norm": 0.3753034770488739,
      "learning_rate": 3e-06,
      "loss": 0.6755,
      "step": 1357,
      "step_time": 7.0915310978889465
    },
    {
      "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,
      "entropy": 1.7275390625,
      "epoch": 0.6503831417624522,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11305497586727142,
      "learning_rate": 3e-06,
      "loss": -0.0015,
      "step": 1358,
      "step_time": 7.8838231563568115
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.125,
      "clip_ratio/low_min": 0.125,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.28515625,
      "epoch": 0.6508620689655172,
      "fork_head/loss": 2.6867854595184326,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.08323990553617477,
      "grad_norm": 0.327163964509964,
      "learning_rate": 3e-06,
      "loss": 0.6713,
      "step": 1359,
      "step_time": 7.496409811079502
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.73681640625,
      "epoch": 0.6513409961685823,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04057007282972336,
      "learning_rate": 3e-06,
      "loss": -0.002,
      "step": 1360,
      "step_time": 7.115173667669296
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 202.142578125,
      "completions/mean_terminated_length": 202.142578125,
      "completions/min_length": 27.0,
      "completions/min_terminated_length": 27.0,
      "entropy": 1.5644235394217751,
      "epoch": 0.6518199233716475,
      "fork/depth_mean": 0.2587094306945801,
      "fork/depth_std": 0.23499368131160736,
      "fork_head/loss": 2.7255311012268066,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.06523667275905609,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6302865147590637,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "num_tokens": 35265176.0,
      "reward": 1.0215351581573486,
      "reward_std": 0.5922224521636963,
      "rewards/code_format_reward/mean": 0.7783203125,
      "rewards/code_format_reward/std": 0.32833969593048096,
      "rewards/code_reward/mean": 0.24321481585502625,
      "rewards/code_reward/std": 0.367257297039032,
      "step": 1361,
      "step_time": 492.22538690268993
    },
    {
      "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,
      "entropy": 1.59462890625,
      "epoch": 0.6522988505747126,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20379450917243958,
      "learning_rate": 3e-06,
      "loss": 0.0041,
      "step": 1362,
      "step_time": 8.023186780512333
    },
    {
      "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,
      "entropy": 1.5639796473763206,
      "epoch": 0.6527777777777778,
      "fork_head/loss": 2.7103257179260254,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.06799086928367615,
      "grad_norm": 0.5835075378417969,
      "learning_rate": 3e-06,
      "loss": 0.6746,
      "step": 1363,
      "step_time": 7.623768337070942
    },
    {
      "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,
      "entropy": 1.5833984375,
      "epoch": 0.6532567049808429,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18448969721794128,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1364,
      "step_time": 7.786944665014744
    },
    {
      "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,
      "entropy": 1.5572916594418613,
      "epoch": 0.6537356321839081,
      "fork_head/loss": 2.686514377593994,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.09294313192367554,
      "grad_norm": 0.5093080401420593,
      "learning_rate": 3e-06,
      "loss": 0.6679,
      "step": 1365,
      "step_time": 7.326099716126919
    },
    {
      "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,
      "entropy": 1.6095703125,
      "epoch": 0.6542145593869731,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17919521033763885,
      "learning_rate": 3e-06,
      "loss": 0.0029,
      "step": 1366,
      "step_time": 7.200638599693775
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.5982481132854114,
      "epoch": 0.6546934865900383,
      "fork_head/loss": 2.6614086627960205,
      "fork_head/num_positive": 10.96875,
      "fork_head/pred_top1_mass": 0.13619579374790192,
      "grad_norm": 0.4436998665332794,
      "learning_rate": 3e-06,
      "loss": 0.6611,
      "step": 1367,
      "step_time": 7.658790618181229
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.2,
      "clip_ratio/low_min": 0.2,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.65380859375,
      "epoch": 0.6551724137931034,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05776969715952873,
      "learning_rate": 3e-06,
      "loss": 0.0017,
      "step": 1368,
      "step_time": 7.79339087754488
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 193.119140625,
      "completions/mean_terminated_length": 193.119140625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.4318214654922485,
      "epoch": 0.6556513409961686,
      "fork/depth_mean": 0.2934594452381134,
      "fork/depth_std": 0.2565016746520996,
      "fork_head/loss": 2.7346749305725098,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.06974847614765167,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6653174757957458,
      "learning_rate": 3e-06,
      "loss": 0.6946,
      "num_tokens": 35456117.0,
      "reward": 0.9083816409111023,
      "reward_std": 0.5884016752243042,
      "rewards/code_format_reward/mean": 0.7216796875,
      "rewards/code_format_reward/std": 0.3650304973125458,
      "rewards/code_reward/mean": 0.1867019683122635,
      "rewards/code_reward/std": 0.3299090266227722,
      "step": 1369,
      "step_time": 438.617430485785
    },
    {
      "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,
      "entropy": 1.9922526001930236,
      "epoch": 0.6561302681992337,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 1.1069127321243286,
      "learning_rate": 3e-06,
      "loss": -0.0091,
      "step": 1370,
      "step_time": 7.911059461534023
    },
    {
      "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,
      "entropy": 1.4452401664521959,
      "epoch": 0.6566091954022989,
      "fork_head/loss": 2.7285189628601074,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.07372351735830307,
      "grad_norm": 0.6320387125015259,
      "learning_rate": 3e-06,
      "loss": 0.6921,
      "step": 1371,
      "step_time": 7.965831279754639
    },
    {
      "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,
      "entropy": 2.0173502564430237,
      "epoch": 0.657088122605364,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.26139193773269653,
      "learning_rate": 3e-06,
      "loss": -0.0137,
      "step": 1372,
      "step_time": 7.377800442278385
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.4771412081188626,
      "epoch": 0.6575670498084292,
      "fork_head/loss": 2.7172343730926514,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.07141725718975067,
      "grad_norm": 0.4234224557876587,
      "learning_rate": 3e-06,
      "loss": 0.688,
      "step": 1373,
      "step_time": 7.553964242339134
    },
    {
      "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,
      "entropy": 2.0050130248069764,
      "epoch": 0.6580459770114943,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.25751468539237976,
      "learning_rate": 3e-06,
      "loss": -0.0135,
      "step": 1374,
      "step_time": 7.8962889686226845
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.5254629585478041,
      "epoch": 0.6585249042145593,
      "fork_head/loss": 2.7034904956817627,
      "fork_head/num_positive": 12.09375,
      "fork_head/pred_top1_mass": 0.08107049018144608,
      "grad_norm": 0.3898079991340637,
      "learning_rate": 3e-06,
      "loss": 0.6838,
      "step": 1375,
      "step_time": 7.677472658455372
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.03333333432674408,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333432674409,
      "entropy": 1.9916015625,
      "epoch": 0.6590038314176245,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20686423778533936,
      "learning_rate": 3e-06,
      "loss": -0.0154,
      "step": 1376,
      "step_time": 7.311256058514118
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 198.787109375,
      "completions/mean_terminated_length": 198.4499053955078,
      "completions/min_length": 76.0,
      "completions/min_terminated_length": 76.0,
      "entropy": 1.4498697916666667,
      "epoch": 0.6594827586206896,
      "fork/depth_mean": 0.29643839597702026,
      "fork/depth_std": 0.2622925341129303,
      "fork_head/loss": 2.7321863174438477,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.057723890990018845,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5111808776855469,
      "learning_rate": 3e-06,
      "loss": 0.6781,
      "num_tokens": 35661864.0,
      "reward": 0.9385378360748291,
      "reward_std": 0.5670013427734375,
      "rewards/code_format_reward/mean": 0.7587890625,
      "rewards/code_format_reward/std": 0.3579142987728119,
      "rewards/code_reward/mean": 0.1797487586736679,
      "rewards/code_reward/std": 0.33114179968833923,
      "step": 1377,
      "step_time": 426.8060734793544
    },
    {
      "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,
      "entropy": 1.3952094217141469,
      "epoch": 0.6599616858237548,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.3648955821990967,
      "learning_rate": 3e-06,
      "loss": 0.001,
      "step": 1378,
      "step_time": 7.557579047977924
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.4291811386744182,
      "epoch": 0.6604406130268199,
      "fork_head/loss": 2.7221832275390625,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.057710129767656326,
      "grad_norm": 0.4977321922779083,
      "learning_rate": 3e-06,
      "loss": 0.6754,
      "step": 1379,
      "step_time": 8.220669500529766
    },
    {
      "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,
      "entropy": 1.3927680154641469,
      "epoch": 0.6609195402298851,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.4409365952014923,
      "learning_rate": 3e-06,
      "loss": -0.0001,
      "step": 1380,
      "step_time": 7.44830197095871
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.4575376245710585,
      "epoch": 0.6613984674329502,
      "fork_head/loss": 2.704706907272339,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.07498567551374435,
      "grad_norm": 0.4558364152908325,
      "learning_rate": 3e-06,
      "loss": 0.6705,
      "step": 1381,
      "step_time": 7.51410111784935
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.4064127629001935,
      "epoch": 0.6618773946360154,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1637144237756729,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 1382,
      "step_time": 7.739915989339352
    },
    {
      "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,
      "entropy": 1.4670138888888888,
      "epoch": 0.6623563218390804,
      "fork_head/loss": 2.6844091415405273,
      "fork_head/num_positive": 11.258064270019531,
      "fork_head/pred_top1_mass": 0.09640350192785263,
      "grad_norm": 0.4800805449485779,
      "learning_rate": 3e-06,
      "loss": 0.6654,
      "step": 1383,
      "step_time": 7.044495448470116
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.4165581564108531,
      "epoch": 0.6628352490421456,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13618537783622742,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 1384,
      "step_time": 8.019125908613205
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 205.228515625,
      "completions/mean_terminated_length": 205.129150390625,
      "completions/min_length": 17.0,
      "completions/min_terminated_length": 17.0,
      "entropy": 1.5069986879825592,
      "epoch": 0.6633141762452107,
      "fork/depth_mean": 0.2583203613758087,
      "fork/depth_std": 0.24045157432556152,
      "fork_head/loss": 2.7334585189819336,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07472139596939087,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7135023474693298,
      "learning_rate": 3e-06,
      "loss": 0.6939,
      "num_tokens": 35864013.0,
      "reward": 0.9736538529396057,
      "reward_std": 0.6004277467727661,
      "rewards/code_format_reward/mean": 0.751953125,
      "rewards/code_format_reward/std": 0.3552735149860382,
      "rewards/code_reward/mean": 0.22170069813728333,
      "rewards/code_reward/std": 0.356412798166275,
      "step": 1385,
      "step_time": 518.7055416330695
    },
    {
      "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,
      "entropy": 1.5531668535300664,
      "epoch": 0.6637931034482759,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.29638347029685974,
      "learning_rate": 3e-06,
      "loss": -0.0174,
      "step": 1386,
      "step_time": 8.00100302696228
    },
    {
      "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,
      "entropy": 1.5266994933287303,
      "epoch": 0.664272030651341,
      "fork_head/loss": 2.719074249267578,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07854604721069336,
      "grad_norm": 0.6369419693946838,
      "learning_rate": 3e-06,
      "loss": 0.6898,
      "step": 1387,
      "step_time": 8.690398186445236
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.014285714498588018,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.014285714498588018,
      "entropy": 1.5627697152750832,
      "epoch": 0.6647509578544061,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.25704696774482727,
      "learning_rate": 3e-06,
      "loss": -0.0189,
      "step": 1388,
      "step_time": 7.642225451767445
    },
    {
      "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,
      "entropy": 1.5254041850566864,
      "epoch": 0.6652298850574713,
      "fork_head/loss": 2.6982781887054443,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07638681679964066,
      "grad_norm": 0.5231743454933167,
      "learning_rate": 3e-06,
      "loss": 0.6837,
      "step": 1389,
      "step_time": 8.567752599716187
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.03809523901769093,
      "clip_ratio/low_mean": 0.14285714285714285,
      "clip_ratio/low_min": 0.14285714285714285,
      "clip_ratio/region_mean": 0.18095238187483378,
      "entropy": 1.5372256338596344,
      "epoch": 0.6657088122605364,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16101393103599548,
      "learning_rate": 3e-06,
      "loss": -0.0213,
      "step": 1390,
      "step_time": 7.345813170075417
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0555555559694767,
      "clip_ratio/low_min": 0.041666666666666664,
      "clip_ratio/region_mean": 0.0555555559694767,
      "entropy": 1.5630764464537303,
      "epoch": 0.6661877394636015,
      "fork_head/loss": 2.6773650646209717,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.1029171571135521,
      "grad_norm": 0.4152030646800995,
      "learning_rate": 3e-06,
      "loss": 0.6776,
      "step": 1391,
      "step_time": 8.336527548730373
    },
    {
      "clip_ratio/high_max": 0.2857142857142857,
      "clip_ratio/high_mean": 0.13333333496536529,
      "clip_ratio/low_mean": 0.14285714285714285,
      "clip_ratio/low_min": 0.14285714285714285,
      "clip_ratio/region_mean": 0.2761904778225081,
      "entropy": 1.5441917734486716,
      "epoch": 0.6666666666666666,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10576190054416656,
      "learning_rate": 3e-06,
      "loss": -0.0222,
      "step": 1392,
      "step_time": 7.88690260797739
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 199.119140625,
      "completions/mean_terminated_length": 199.119140625,
      "completions/min_length": 20.0,
      "completions/min_terminated_length": 20.0,
      "entropy": 1.9425307729027488,
      "epoch": 0.6671455938697318,
      "fork/depth_mean": 0.296577513217926,
      "fork/depth_std": 0.2506187856197357,
      "fork_head/loss": 2.697432518005371,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.08801718801259995,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5065756440162659,
      "learning_rate": 3e-06,
      "loss": 0.6621,
      "num_tokens": 36064490.0,
      "reward": 0.9311908483505249,
      "reward_std": 0.531613826751709,
      "rewards/code_format_reward/mean": 0.759765625,
      "rewards/code_format_reward/std": 0.3173125088214874,
      "rewards/code_reward/mean": 0.1714252531528473,
      "rewards/code_reward/std": 0.31924355030059814,
      "step": 1393,
      "step_time": 441.4194323942065
    },
    {
      "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,
      "entropy": 1.3673502604166667,
      "epoch": 0.6676245210727969,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1033051609992981,
      "learning_rate": 3e-06,
      "loss": 0.012,
      "step": 1394,
      "step_time": 7.838689759373665
    },
    {
      "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,
      "entropy": 1.9225852272727273,
      "epoch": 0.6681034482758621,
      "fork_head/loss": 2.6868796348571777,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.10144855827093124,
      "grad_norm": 0.5050684213638306,
      "learning_rate": 3e-06,
      "loss": 0.6594,
      "step": 1395,
      "step_time": 8.090821400284767
    },
    {
      "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,
      "entropy": 1.3829752604166667,
      "epoch": 0.6685823754789272,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0992671549320221,
      "learning_rate": 3e-06,
      "loss": 0.0116,
      "step": 1396,
      "step_time": 7.412431575357914
    },
    {
      "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,
      "entropy": 1.9019886363636365,
      "epoch": 0.6690613026819924,
      "fork_head/loss": 2.6694858074188232,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.10750554502010345,
      "grad_norm": 0.5068361759185791,
      "learning_rate": 3e-06,
      "loss": 0.6548,
      "step": 1397,
      "step_time": 7.633307680487633
    },
    {
      "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,
      "entropy": 1.40234375,
      "epoch": 0.6695402298850575,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0923180878162384,
      "learning_rate": 3e-06,
      "loss": 0.0111,
      "step": 1398,
      "step_time": 7.602883614599705
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.9430042613636365,
      "epoch": 0.6700191570881227,
      "fork_head/loss": 2.647641658782959,
      "fork_head/num_positive": 10.483870506286621,
      "fork_head/pred_top1_mass": 0.10606937110424042,
      "grad_norm": 0.4915046989917755,
      "learning_rate": 3e-06,
      "loss": 0.6489,
      "step": 1399,
      "step_time": 7.573417991399765
    },
    {
      "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,
      "entropy": 1.4034288194444444,
      "epoch": 0.6704980842911877,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08941534906625748,
      "learning_rate": 3e-06,
      "loss": 0.011,
      "step": 1400,
      "step_time": 7.639555923640728
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 186.779296875,
      "completions/mean_terminated_length": 186.64382934570312,
      "completions/min_length": 58.0,
      "completions/min_terminated_length": 58.0,
      "entropy": 1.46240234375,
      "epoch": 0.6709770114942529,
      "fork/depth_mean": 0.2925666272640228,
      "fork/depth_std": 0.22777502238750458,
      "fork_head/loss": 2.739542245864868,
      "fork_head/num_positive": 10.433333396911621,
      "fork_head/pred_top1_mass": 0.061298809945583344,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.6649423241615295,
      "learning_rate": 3e-06,
      "loss": 0.6773,
      "num_tokens": 36256761.0,
      "reward": 1.083742618560791,
      "reward_std": 0.6127785444259644,
      "rewards/code_format_reward/mean": 0.7919921875,
      "rewards/code_format_reward/std": 0.3177623748779297,
      "rewards/code_reward/mean": 0.29175037145614624,
      "rewards/code_reward/std": 0.41999733448028564,
      "step": 1401,
      "step_time": 458.2949330583215
    },
    {
      "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,
      "entropy": 1.40185546875,
      "epoch": 0.671455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17754249274730682,
      "learning_rate": 3e-06,
      "loss": 0.0056,
      "step": 1402,
      "step_time": 8.481785155832767
    },
    {
      "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,
      "entropy": 1.4696723123391469,
      "epoch": 0.6719348659003831,
      "fork_head/loss": 2.718863010406494,
      "fork_head/num_positive": 10.433333396911621,
      "fork_head/pred_top1_mass": 0.0671248733997345,
      "grad_norm": 0.6348344683647156,
      "learning_rate": 3e-06,
      "loss": 0.6722,
      "step": 1403,
      "step_time": 7.9291331470012665
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.412109375,
      "epoch": 0.6724137931034483,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12226544320583344,
      "learning_rate": 3e-06,
      "loss": 0.0048,
      "step": 1404,
      "step_time": 7.415308825671673
    },
    {
      "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,
      "entropy": 1.4553222606579463,
      "epoch": 0.6728927203065134,
      "fork_head/loss": 2.6898484230041504,
      "fork_head/num_positive": 10.433333396911621,
      "fork_head/pred_top1_mass": 0.0909736230969429,
      "grad_norm": 0.586983859539032,
      "learning_rate": 3e-06,
      "loss": 0.6638,
      "step": 1405,
      "step_time": 7.702978111803532
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.40517578125,
      "epoch": 0.6733716475095786,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19131262600421906,
      "learning_rate": 3e-06,
      "loss": 0.0046,
      "step": 1406,
      "step_time": 7.418585039675236
    },
    {
      "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,
      "entropy": 1.4566243539253871,
      "epoch": 0.6738505747126436,
      "fork_head/loss": 2.6557693481445312,
      "fork_head/num_positive": 10.433333396911621,
      "fork_head/pred_top1_mass": 0.09702623635530472,
      "grad_norm": 0.5332108736038208,
      "learning_rate": 3e-06,
      "loss": 0.6551,
      "step": 1407,
      "step_time": 8.236035332083702
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.15,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.2,
      "entropy": 1.438671875,
      "epoch": 0.6743295019157088,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0704173743724823,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 1408,
      "step_time": 7.497564762830734
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 202.150390625,
      "completions/mean_terminated_length": 202.150390625,
      "completions/min_length": 9.0,
      "completions/min_terminated_length": 9.0,
      "entropy": 1.4279296875,
      "epoch": 0.6748084291187739,
      "fork/depth_mean": 0.30274683237075806,
      "fork/depth_std": 0.24691052734851837,
      "fork_head/loss": 2.7395670413970947,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.06706482172012329,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 1.2298401594161987,
      "learning_rate": 3e-06,
      "loss": 0.6868,
      "num_tokens": 36452662.0,
      "reward": 0.9561575651168823,
      "reward_std": 0.5769464373588562,
      "rewards/code_format_reward/mean": 0.751953125,
      "rewards/code_format_reward/std": 0.36073973774909973,
      "rewards/code_reward/mean": 0.20420442521572113,
      "rewards/code_reward/std": 0.318168580532074,
      "step": 1409,
      "step_time": 634.6694694235921
    },
    {
      "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,
      "entropy": 1.1962193080357142,
      "epoch": 0.6752873563218391,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14714445173740387,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 1410,
      "step_time": 7.646161913871765
    },
    {
      "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,
      "entropy": 1.4345703125,
      "epoch": 0.6757662835249042,
      "fork_head/loss": 2.717240333557129,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.0770740956068039,
      "grad_norm": 1.084122657775879,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "step": 1411,
      "step_time": 8.378074519336224
    },
    {
      "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,
      "entropy": 1.2035667768546514,
      "epoch": 0.6762452107279694,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13627170026302338,
      "learning_rate": 3e-06,
      "loss": -0.002,
      "step": 1412,
      "step_time": 7.190284229815006
    },
    {
      "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,
      "entropy": 1.4228515625,
      "epoch": 0.6767241379310345,
      "fork_head/loss": 2.6792385578155518,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.1394628882408142,
      "grad_norm": 0.8437270522117615,
      "learning_rate": 3e-06,
      "loss": 0.6699,
      "step": 1413,
      "step_time": 7.057223416864872
    },
    {
      "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,
      "entropy": 1.2150181361607142,
      "epoch": 0.6772030651340997,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16315120458602905,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 1414,
      "step_time": 8.079334795475006
    },
    {
      "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,
      "entropy": 1.4322265625,
      "epoch": 0.6776819923371648,
      "fork_head/loss": 2.6413683891296387,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.14987535774707794,
      "grad_norm": 0.6715210676193237,
      "learning_rate": 3e-06,
      "loss": 0.6588,
      "step": 1415,
      "step_time": 7.553252376616001
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.200485953262874,
      "epoch": 0.6781609195402298,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07871872186660767,
      "learning_rate": 3e-06,
      "loss": -0.0036,
      "step": 1416,
      "step_time": 7.334107339382172
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 185.078125,
      "completions/mean_terminated_length": 184.9393310546875,
      "completions/min_length": 35.0,
      "completions/min_terminated_length": 35.0,
      "entropy": 1.2005859375,
      "epoch": 0.678639846743295,
      "fork/depth_mean": 0.3005317747592926,
      "fork/depth_std": 0.2670457661151886,
      "fork_head/loss": 2.741230010986328,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.07416325062513351,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.558203399181366,
      "learning_rate": 3e-06,
      "loss": 0.6889,
      "num_tokens": 36636238.0,
      "reward": 0.9668709635734558,
      "reward_std": 0.6293718218803406,
      "rewards/code_format_reward/mean": 0.7236328125,
      "rewards/code_format_reward/std": 0.36517706513404846,
      "rewards/code_reward/mean": 0.24323809146881104,
      "rewards/code_reward/std": 0.37072739005088806,
      "step": 1417,
      "step_time": 482.44050401449203
    },
    {
      "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,
      "entropy": 2.125759548611111,
      "epoch": 0.6791187739463601,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09866639971733093,
      "learning_rate": 3e-06,
      "loss": -0.0023,
      "step": 1418,
      "step_time": 7.2981982454657555
    },
    {
      "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,
      "entropy": 1.226171875,
      "epoch": 0.6795977011494253,
      "fork_head/loss": 2.7353203296661377,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.07358857989311218,
      "grad_norm": 0.555182695388794,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "step": 1419,
      "step_time": 8.824774466454983
    },
    {
      "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,
      "entropy": 2.126736111111111,
      "epoch": 0.6800766283524904,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09486979246139526,
      "learning_rate": 3e-06,
      "loss": -0.0021,
      "step": 1420,
      "step_time": 6.966759406030178
    },
    {
      "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,
      "entropy": 1.20986328125,
      "epoch": 0.6805555555555556,
      "fork_head/loss": 2.722454071044922,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.07355596125125885,
      "grad_norm": 0.541753888130188,
      "learning_rate": 3e-06,
      "loss": 0.684,
      "step": 1421,
      "step_time": 8.102120332419872
    },
    {
      "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,
      "entropy": 2.1255425347222223,
      "epoch": 0.6810344827586207,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10386858135461807,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 1422,
      "step_time": 7.665256127715111
    },
    {
      "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,
      "entropy": 1.21982421875,
      "epoch": 0.6815134099616859,
      "fork_head/loss": 2.705817937850952,
      "fork_head/num_positive": 12.1875,
      "fork_head/pred_top1_mass": 0.07274998724460602,
      "grad_norm": 0.4972435534000397,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "step": 1423,
      "step_time": 7.366883836686611
    },
    {
      "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,
      "entropy": 2.135525173611111,
      "epoch": 0.6819923371647509,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09522092342376709,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 1424,
      "step_time": 7.825736880302429
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 194.638671875,
      "completions/mean_terminated_length": 194.638671875,
      "completions/min_length": 44.0,
      "completions/min_terminated_length": 44.0,
      "entropy": 1.2896484375,
      "epoch": 0.6824712643678161,
      "fork/depth_mean": 0.3142189681529999,
      "fork/depth_std": 0.2635904550552368,
      "fork_head/loss": 2.735713481903076,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.06339726597070694,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6976447105407715,
      "learning_rate": 3e-06,
      "loss": 0.6866,
      "num_tokens": 36825701.0,
      "reward": 0.9716391563415527,
      "reward_std": 0.6192626357078552,
      "rewards/code_format_reward/mean": 0.734375,
      "rewards/code_format_reward/std": 0.3617607355117798,
      "rewards/code_reward/mean": 0.23726418614387512,
      "rewards/code_reward/std": 0.37662389874458313,
      "step": 1425,
      "step_time": 465.8406205698848
    },
    {
      "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,
      "entropy": 1.7466796875,
      "epoch": 0.6829501915708812,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0668109804391861,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 1426,
      "step_time": 8.056603394448757
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.025,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.025,
      "entropy": 1.301171875,
      "epoch": 0.6834291187739464,
      "fork_head/loss": 2.7224349975585938,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.07737080752849579,
      "grad_norm": 0.6739128232002258,
      "learning_rate": 3e-06,
      "loss": 0.6822,
      "step": 1427,
      "step_time": 7.378807440400124
    },
    {
      "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,
      "entropy": 1.7647786378860473,
      "epoch": 0.6839080459770115,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05805273354053497,
      "learning_rate": 3e-06,
      "loss": -0.0038,
      "step": 1428,
      "step_time": 7.919040605425835
    },
    {
      "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,
      "entropy": 1.3056640625,
      "epoch": 0.6843869731800766,
      "fork_head/loss": 2.6978538036346436,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.10094684362411499,
      "grad_norm": 0.6068020462989807,
      "learning_rate": 3e-06,
      "loss": 0.6758,
      "step": 1429,
      "step_time": 7.596444271504879
    },
    {
      "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,
      "entropy": 1.76953125,
      "epoch": 0.6848659003831418,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06090360879898071,
      "learning_rate": 3e-06,
      "loss": -0.004,
      "step": 1430,
      "step_time": 7.443889111280441
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.025,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.075,
      "entropy": 1.312646484375,
      "epoch": 0.6853448275862069,
      "fork_head/loss": 2.6695287227630615,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.09300817549228668,
      "grad_norm": 0.5204740166664124,
      "learning_rate": 3e-06,
      "loss": 0.6681,
      "step": 1431,
      "step_time": 7.382092021405697
    },
    {
      "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,
      "entropy": 1.7633463621139527,
      "epoch": 0.685823754789272,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05983824282884598,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 1432,
      "step_time": 7.717966414988041
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.005859375,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 199.607421875,
      "completions/mean_terminated_length": 199.27505493164062,
      "completions/min_length": 65.0,
      "completions/min_terminated_length": 65.0,
      "entropy": 1.68900553509593,
      "epoch": 0.6863026819923371,
      "fork/depth_mean": 0.29542630910873413,
      "fork/depth_std": 0.25622066855430603,
      "fork_head/loss": 2.7322070598602295,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.07098288089036942,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5219731330871582,
      "learning_rate": 3e-06,
      "loss": 0.6741,
      "num_tokens": 37025996.0,
      "reward": 0.9510269165039062,
      "reward_std": 0.5758591294288635,
      "rewards/code_format_reward/mean": 0.7490234375,
      "rewards/code_format_reward/std": 0.35112208127975464,
      "rewards/code_reward/mean": 0.20200343430042267,
      "rewards/code_reward/std": 0.3269713521003723,
      "step": 1433,
      "step_time": 471.97587487101555
    },
    {
      "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,
      "entropy": 1.531220403584567,
      "epoch": 0.6867816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08378790318965912,
      "learning_rate": 3e-06,
      "loss": 0.0068,
      "step": 1434,
      "step_time": 7.998877689242363
    },
    {
      "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,
      "entropy": 1.6873779259622097,
      "epoch": 0.6872605363984674,
      "fork_head/loss": 2.7270073890686035,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.06906108558177948,
      "grad_norm": 0.5230662226676941,
      "learning_rate": 3e-06,
      "loss": 0.6722,
      "step": 1435,
      "step_time": 7.540785871446133
    },
    {
      "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,
      "entropy": 1.5542436079545454,
      "epoch": 0.6877394636015326,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08075089752674103,
      "learning_rate": 3e-06,
      "loss": 0.0065,
      "step": 1436,
      "step_time": 7.325361117720604
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.6643676795065403,
      "epoch": 0.6882183908045977,
      "fork_head/loss": 2.713780164718628,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.07071059942245483,
      "grad_norm": 0.4618673622608185,
      "learning_rate": 3e-06,
      "loss": 0.6683,
      "step": 1437,
      "step_time": 7.776523053646088
    },
    {
      "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,
      "entropy": 1.5215731534090908,
      "epoch": 0.6886973180076629,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08503401279449463,
      "learning_rate": 3e-06,
      "loss": 0.0065,
      "step": 1438,
      "step_time": 7.5768387615680695
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.03125,
      "clip_ratio/low_min": 0.03125,
      "clip_ratio/region_mean": 0.09375,
      "entropy": 1.65250651165843,
      "epoch": 0.689176245210728,
      "fork_head/loss": 2.6977763175964355,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.07933369278907776,
      "grad_norm": 0.40294474363327026,
      "learning_rate": 3e-06,
      "loss": 0.6641,
      "step": 1439,
      "step_time": 7.997237220406532
    },
    {
      "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,
      "entropy": 1.5444779829545454,
      "epoch": 0.6896551724137931,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08520954102277756,
      "learning_rate": 3e-06,
      "loss": 0.0059,
      "step": 1440,
      "step_time": 7.522975668311119
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 189.119140625,
      "completions/mean_terminated_length": 189.119140625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.5632398236881604,
      "epoch": 0.6901340996168582,
      "fork/depth_mean": 0.4063381254673004,
      "fork/depth_std": 0.2965775430202484,
      "fork_head/loss": 2.7320399284362793,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.07418684661388397,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7822938561439514,
      "learning_rate": 3e-06,
      "loss": 0.6732,
      "num_tokens": 37216089.0,
      "reward": 0.8707991242408752,
      "reward_std": 0.5628657937049866,
      "rewards/code_format_reward/mean": 0.7197265625,
      "rewards/code_format_reward/std": 0.36353009939193726,
      "rewards/code_reward/mean": 0.15107253193855286,
      "rewards/code_reward/std": 0.31711241602897644,
      "step": 1441,
      "step_time": 389.72493001818657
    },
    {
      "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,
      "entropy": 1.461901536354652,
      "epoch": 0.6906130268199234,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2461722195148468,
      "learning_rate": 3e-06,
      "loss": 0.0064,
      "step": 1442,
      "step_time": 7.192356362938881
    },
    {
      "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,
      "entropy": 1.5455729094418613,
      "epoch": 0.6910919540229885,
      "fork_head/loss": 2.7137513160705566,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.08630926162004471,
      "grad_norm": 0.6731169819831848,
      "learning_rate": 3e-06,
      "loss": 0.6687,
      "step": 1443,
      "step_time": 8.092235267162323
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.10256410332826468,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.10256410332826468,
      "entropy": 1.4795422645715566,
      "epoch": 0.6915708812260536,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11826719343662262,
      "learning_rate": 3e-06,
      "loss": 0.0046,
      "step": 1444,
      "step_time": 7.62542812526226
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.03030303120613098,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.533143943006342,
      "epoch": 0.6920498084291188,
      "fork_head/loss": 2.689884662628174,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.08650080859661102,
      "grad_norm": 0.5277546644210815,
      "learning_rate": 3e-06,
      "loss": 0.6607,
      "step": 1445,
      "step_time": 7.469564579427242
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.12820512973345244,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.12820512973345244,
      "entropy": 1.4887820482254028,
      "epoch": 0.6925287356321839,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07350471615791321,
      "learning_rate": 3e-06,
      "loss": 0.0042,
      "step": 1446,
      "step_time": 7.95102533698082
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.03030303120613098,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.5073390115391125,
      "epoch": 0.6930076628352491,
      "fork_head/loss": 2.6670196056365967,
      "fork_head/num_positive": 10.5625,
      "fork_head/pred_top1_mass": 0.09406482428312302,
      "grad_norm": 0.48783716559410095,
      "learning_rate": 3e-06,
      "loss": 0.6549,
      "step": 1447,
      "step_time": 7.506313681602478
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.12820512973345244,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.12820512973345244,
      "entropy": 1.520758216197674,
      "epoch": 0.6934865900383141,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08367009460926056,
      "learning_rate": 3e-06,
      "loss": 0.004,
      "step": 1448,
      "step_time": 7.33135936409235
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 196.74609375,
      "completions/mean_terminated_length": 196.63014221191406,
      "completions/min_length": 36.0,
      "completions/min_terminated_length": 36.0,
      "entropy": 1.7676669034090908,
      "epoch": 0.6939655172413793,
      "fork/depth_mean": 0.40368011593818665,
      "fork/depth_std": 0.29304829239845276,
      "fork_head/loss": 2.747210741043091,
      "fork_head/num_positive": 11.399999618530273,
      "fork_head/pred_top1_mass": 0.07342169433832169,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.7471319437026978,
      "learning_rate": 3e-06,
      "loss": 0.6944,
      "num_tokens": 37404247.0,
      "reward": 1.0698868036270142,
      "reward_std": 0.5854098200798035,
      "rewards/code_format_reward/mean": 0.7958984375,
      "rewards/code_format_reward/std": 0.303022563457489,
      "rewards/code_reward/mean": 0.27398836612701416,
      "rewards/code_reward/std": 0.3982892632484436,
      "step": 1449,
      "step_time": 405.4045415595174
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.3434244827790693,
      "epoch": 0.6944444444444444,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2757726311683655,
      "learning_rate": 3e-06,
      "loss": -0.0064,
      "step": 1450,
      "step_time": 7.985459618270397
    },
    {
      "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,
      "entropy": 1.7661280848763206,
      "epoch": 0.6949233716475096,
      "fork_head/loss": 2.732020616531372,
      "fork_head/num_positive": 11.399999618530273,
      "fork_head/pred_top1_mass": 0.07415793091058731,
      "grad_norm": 0.6195501685142517,
      "learning_rate": 3e-06,
      "loss": 0.6899,
      "step": 1451,
      "step_time": 7.829889044165611
    },
    {
      "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,
      "entropy": 1.3254912333054976,
      "epoch": 0.6954022988505747,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.29838359355926514,
      "learning_rate": 3e-06,
      "loss": -0.0078,
      "step": 1452,
      "step_time": 7.446549706161022
    },
    {
      "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,
      "entropy": 1.7998342730782249,
      "epoch": 0.6958812260536399,
      "fork_head/loss": 2.7125637531280518,
      "fork_head/num_positive": 11.399999618530273,
      "fork_head/pred_top1_mass": 0.0884866937994957,
      "grad_norm": 0.5178653597831726,
      "learning_rate": 3e-06,
      "loss": 0.6825,
      "step": 1453,
      "step_time": 7.804129436612129
    },
    {
      "clip_ratio/high_max": 0.18181818181818182,
      "clip_ratio/high_mean": 0.13636363636363635,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.22727272727272727,
      "entropy": 1.3546993298964067,
      "epoch": 0.696360153256705,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2225257009267807,
      "learning_rate": 3e-06,
      "loss": -0.0109,
      "step": 1454,
      "step_time": 7.314799651503563
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.18181818181818182,
      "clip_ratio/low_min": 0.18181818181818182,
      "clip_ratio/region_mean": 0.18181818181818182,
      "entropy": 1.850733908739957,
      "epoch": 0.6968390804597702,
      "fork_head/loss": 2.695573091506958,
      "fork_head/num_positive": 11.399999618530273,
      "fork_head/pred_top1_mass": 0.09428577870130539,
      "grad_norm": 0.516975998878479,
      "learning_rate": 3e-06,
      "loss": 0.6774,
      "step": 1455,
      "step_time": 7.665839344263077
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.12121212211522189,
      "entropy": 1.3503491986881604,
      "epoch": 0.6973180076628352,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2528851628303528,
      "learning_rate": 3e-06,
      "loss": -0.0126,
      "step": 1456,
      "step_time": 7.795031689107418
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 193.763671875,
      "completions/mean_terminated_length": 193.763671875,
      "completions/min_length": 23.0,
      "completions/min_terminated_length": 23.0,
      "entropy": 1.9846413352272727,
      "epoch": 0.6977969348659003,
      "fork/depth_mean": 0.3380690813064575,
      "fork/depth_std": 0.2645202875137329,
      "fork_head/loss": 2.7510509490966797,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.04655063524842262,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7103862762451172,
      "learning_rate": 3e-06,
      "loss": 0.6837,
      "num_tokens": 37598366.0,
      "reward": 1.0343388319015503,
      "reward_std": 0.5787485241889954,
      "rewards/code_format_reward/mean": 0.787109375,
      "rewards/code_format_reward/std": 0.3229478895664215,
      "rewards/code_reward/mean": 0.24722951650619507,
      "rewards/code_reward/std": 0.38644111156463623,
      "step": 1457,
      "step_time": 510.77899381518364
    },
    {
      "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,
      "entropy": 1.7527373378927058,
      "epoch": 0.6982758620689655,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0859786868095398,
      "learning_rate": 3e-06,
      "loss": 0.0037,
      "step": 1458,
      "step_time": 7.446927085518837
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.9859730113636365,
      "epoch": 0.6987547892720306,
      "fork_head/loss": 2.7407333850860596,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.0453898124396801,
      "grad_norm": 0.6822317242622375,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "step": 1459,
      "step_time": 7.156584784388542
    },
    {
      "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,
      "entropy": 1.7488754770972512,
      "epoch": 0.6992337164750958,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0849049910902977,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1460,
      "step_time": 8.005798295140266
    },
    {
      "clip_ratio/high_max": 0.18181818181818182,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.18181818181818182,
      "entropy": 2.000266335227273,
      "epoch": 0.6997126436781609,
      "fork_head/loss": 2.7181389331817627,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.04793929308652878,
      "grad_norm": 0.614717423915863,
      "learning_rate": 3e-06,
      "loss": 0.6742,
      "step": 1461,
      "step_time": 7.129568263888359
    },
    {
      "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,
      "entropy": 1.7478101361881604,
      "epoch": 0.7001915708812261,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07656670361757278,
      "learning_rate": 3e-06,
      "loss": 0.0028,
      "step": 1462,
      "step_time": 7.744229651987553
    },
    {
      "clip_ratio/high_max": 0.2727272727272727,
      "clip_ratio/high_mean": 0.18181818181818182,
      "clip_ratio/low_mean": 0.2727272727272727,
      "clip_ratio/low_min": 0.2727272727272727,
      "clip_ratio/region_mean": 0.45454545454545453,
      "entropy": 2.0485174005681817,
      "epoch": 0.7006704980842912,
      "fork_head/loss": 2.690368413925171,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.05081874504685402,
      "grad_norm": 0.5404395461082458,
      "learning_rate": 3e-06,
      "loss": 0.6661,
      "step": 1463,
      "step_time": 7.806676663458347
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.7689393975517966,
      "epoch": 0.7011494252873564,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08926545828580856,
      "learning_rate": 3e-06,
      "loss": 0.0018,
      "step": 1464,
      "step_time": 7.2653549015522
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 182.076171875,
      "completions/mean_terminated_length": 182.076171875,
      "completions/min_length": 21.0,
      "completions/min_terminated_length": 21.0,
      "entropy": 1.783203125,
      "epoch": 0.7016283524904214,
      "fork/depth_mean": 0.32930219173431396,
      "fork/depth_std": 0.2795391380786896,
      "fork_head/loss": 2.734384298324585,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.06567609310150146,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5111980438232422,
      "learning_rate": 3e-06,
      "loss": 0.69,
      "num_tokens": 37783189.0,
      "reward": 0.9497822523117065,
      "reward_std": 0.6009857654571533,
      "rewards/code_format_reward/mean": 0.7294921875,
      "rewards/code_format_reward/std": 0.3463103473186493,
      "rewards/code_reward/mean": 0.22029003500938416,
      "rewards/code_reward/std": 0.35530632734298706,
      "step": 1465,
      "step_time": 473.6247968971729
    },
    {
      "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,
      "entropy": 1.22998046875,
      "epoch": 0.7021072796934866,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09811525791883469,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 1466,
      "step_time": 7.564504057168961
    },
    {
      "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,
      "entropy": 1.776160023429177,
      "epoch": 0.7025862068965517,
      "fork_head/loss": 2.7287707328796387,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.067698173224926,
      "grad_norm": 0.5270439982414246,
      "learning_rate": 3e-06,
      "loss": 0.6877,
      "step": 1467,
      "step_time": 8.168501287698746
    },
    {
      "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,
      "entropy": 1.2332589285714286,
      "epoch": 0.7030651340996169,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1029156818985939,
      "learning_rate": 3e-06,
      "loss": -0.005,
      "step": 1468,
      "step_time": 7.65222692489624
    },
    {
      "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,
      "entropy": 1.7899502949281172,
      "epoch": 0.703544061302682,
      "fork_head/loss": 2.716705799102783,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.07849329710006714,
      "grad_norm": 0.4897667467594147,
      "learning_rate": 3e-06,
      "loss": 0.6844,
      "step": 1469,
      "step_time": 7.412097878754139
    },
    {
      "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,
      "entropy": 1.2141694554260798,
      "epoch": 0.7040229885057471,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10426180064678192,
      "learning_rate": 3e-06,
      "loss": -0.0059,
      "step": 1470,
      "step_time": 7.898303434252739
    },
    {
      "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,
      "entropy": 1.7676965106617322,
      "epoch": 0.7045019157088123,
      "fork_head/loss": 2.7001101970672607,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.08226881921291351,
      "grad_norm": 0.4484749436378479,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "step": 1471,
      "step_time": 7.221885517239571
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.07142857142857142,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.2193312857832228,
      "epoch": 0.7049808429118773,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0727488175034523,
      "learning_rate": 3e-06,
      "loss": -0.0066,
      "step": 1472,
      "step_time": 8.115140989422798
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 189.830078125,
      "completions/mean_terminated_length": 189.5706024169922,
      "completions/min_length": 58.0,
      "completions/min_terminated_length": 58.0,
      "entropy": 1.65458984375,
      "epoch": 0.7054597701149425,
      "fork/depth_mean": 0.3349766135215759,
      "fork/depth_std": 0.2737233340740204,
      "fork_head/loss": 2.743448257446289,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.05980851501226425,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5404109954833984,
      "learning_rate": 3e-06,
      "loss": 0.693,
      "num_tokens": 37976446.0,
      "reward": 0.9964802861213684,
      "reward_std": 0.6031015515327454,
      "rewards/code_format_reward/mean": 0.7490234375,
      "rewards/code_format_reward/std": 0.3235671818256378,
      "rewards/code_reward/mean": 0.2474568784236908,
      "rewards/code_reward/std": 0.3674238622188568,
      "step": 1473,
      "step_time": 491.014720633626
    },
    {
      "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,
      "entropy": 1.7676668925718828,
      "epoch": 0.7059386973180076,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19916638731956482,
      "learning_rate": 3e-06,
      "loss": -0.0083,
      "step": 1474,
      "step_time": 7.929130047559738
    },
    {
      "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,
      "entropy": 1.6672526121139526,
      "epoch": 0.7064176245210728,
      "fork_head/loss": 2.7332546710968018,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.0584498792886734,
      "grad_norm": 0.5174863934516907,
      "learning_rate": 3e-06,
      "loss": 0.6904,
      "step": 1475,
      "step_time": 7.32404313236475
    },
    {
      "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,
      "entropy": 1.7721650058572942,
      "epoch": 0.7068965517241379,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.20303477346897125,
      "learning_rate": 3e-06,
      "loss": -0.009,
      "step": 1476,
      "step_time": 7.880098305642605
    },
    {
      "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,
      "entropy": 1.6617838621139527,
      "epoch": 0.7073754789272031,
      "fork_head/loss": 2.7165186405181885,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.0646061897277832,
      "grad_norm": 0.4735065698623657,
      "learning_rate": 3e-06,
      "loss": 0.6863,
      "step": 1477,
      "step_time": 7.510357819497585
    },
    {
      "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,
      "entropy": 1.7695312391627918,
      "epoch": 0.7078544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1664612740278244,
      "learning_rate": 3e-06,
      "loss": -0.01,
      "step": 1478,
      "step_time": 8.321472600102425
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.6755859375,
      "epoch": 0.7083333333333334,
      "fork_head/loss": 2.6964712142944336,
      "fork_head/num_positive": 11.15625,
      "fork_head/pred_top1_mass": 0.06696254760026932,
      "grad_norm": 0.4195403754711151,
      "learning_rate": 3e-06,
      "loss": 0.6802,
      "step": 1479,
      "step_time": 7.775007739663124
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07575757666067644,
      "entropy": 1.786073630506342,
      "epoch": 0.7088122605363985,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10697503387928009,
      "learning_rate": 3e-06,
      "loss": -0.0109,
      "step": 1480,
      "step_time": 7.180888697504997
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 188.310546875,
      "completions/mean_terminated_length": 188.310546875,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.9265407986111112,
      "epoch": 0.7092911877394636,
      "fork/depth_mean": 0.319924920797348,
      "fork/depth_std": 0.2728624939918518,
      "fork_head/loss": 2.7243711948394775,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.07514957338571548,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5340588092803955,
      "learning_rate": 3e-06,
      "loss": 0.6812,
      "num_tokens": 38160941.0,
      "reward": 1.0301024913787842,
      "reward_std": 0.5737175345420837,
      "rewards/code_format_reward/mean": 0.775390625,
      "rewards/code_format_reward/std": 0.3373951017856598,
      "rewards/code_reward/mean": 0.2547118365764618,
      "rewards/code_reward/std": 0.3486918807029724,
      "step": 1481,
      "step_time": 432.5974672883749
    },
    {
      "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,
      "entropy": 1.2082868303571428,
      "epoch": 0.7097701149425287,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10139384120702744,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 1482,
      "step_time": 6.956057287752628
    },
    {
      "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,
      "entropy": 1.9128689236111112,
      "epoch": 0.7102490421455939,
      "fork_head/loss": 2.7130305767059326,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.08862610161304474,
      "grad_norm": 0.491874098777771,
      "learning_rate": 3e-06,
      "loss": 0.6786,
      "step": 1483,
      "step_time": 8.81648714095354
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.07142857142857142,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.2061941964285714,
      "epoch": 0.710727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10230117291212082,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 1484,
      "step_time": 7.190805725753307
    },
    {
      "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,
      "entropy": 1.9054904513888888,
      "epoch": 0.7112068965517241,
      "fork_head/loss": 2.6975932121276855,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.09389691799879074,
      "grad_norm": 0.4567605257034302,
      "learning_rate": 3e-06,
      "loss": 0.6743,
      "step": 1485,
      "step_time": 7.590588428080082
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.07142857142857142,
      "clip_ratio/low_mean": 0.14285714285714285,
      "clip_ratio/low_min": 0.14285714285714285,
      "clip_ratio/region_mean": 0.21428571428571427,
      "entropy": 1.2144252232142858,
      "epoch": 0.7116858237547893,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14653438329696655,
      "learning_rate": 3e-06,
      "loss": -0.001,
      "step": 1486,
      "step_time": 7.680543780326843
    },
    {
      "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,
      "entropy": 1.9034288194444444,
      "epoch": 0.7121647509578544,
      "fork_head/loss": 2.6790831089019775,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.10602312535047531,
      "grad_norm": 0.431194543838501,
      "learning_rate": 3e-06,
      "loss": 0.6693,
      "step": 1487,
      "step_time": 7.287034049630165
    },
    {
      "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,
      "entropy": 1.2039620535714286,
      "epoch": 0.7126436781609196,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11989593505859375,
      "learning_rate": 3e-06,
      "loss": -0.0012,
      "step": 1488,
      "step_time": 7.952828735113144
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 194.197265625,
      "completions/mean_terminated_length": 194.197265625,
      "completions/min_length": 70.0,
      "completions/min_terminated_length": 70.0,
      "entropy": 1.58828125,
      "epoch": 0.7131226053639846,
      "fork/depth_mean": 0.2963114082813263,
      "fork/depth_std": 0.23414696753025055,
      "fork_head/loss": 2.721353054046631,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.08232082426548004,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5887291431427002,
      "learning_rate": 3e-06,
      "loss": 0.6828,
      "num_tokens": 38360114.0,
      "reward": 0.9562423229217529,
      "reward_std": 0.5483870506286621,
      "rewards/code_format_reward/mean": 0.7646484375,
      "rewards/code_format_reward/std": 0.3108920753002167,
      "rewards/code_reward/mean": 0.1915939450263977,
      "rewards/code_reward/std": 0.3401045501232147,
      "step": 1489,
      "step_time": 455.5402759462595
    },
    {
      "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,
      "entropy": 1.4164496527777777,
      "epoch": 0.7136015325670498,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07314696907997131,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 1490,
      "step_time": 8.21780577301979
    },
    {
      "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,
      "entropy": 1.580078125,
      "epoch": 0.7140804597701149,
      "fork_head/loss": 2.71073055267334,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.0861649289727211,
      "grad_norm": 0.5345377326011658,
      "learning_rate": 3e-06,
      "loss": 0.6802,
      "step": 1491,
      "step_time": 7.9681561216712
    },
    {
      "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,
      "entropy": 1.4181857638888888,
      "epoch": 0.7145593869731801,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06522709131240845,
      "learning_rate": 3e-06,
      "loss": -0.003,
      "step": 1492,
      "step_time": 7.553822435438633
    },
    {
      "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,
      "entropy": 1.5515625,
      "epoch": 0.7150383141762452,
      "fork_head/loss": 2.6940836906433105,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.09033020585775375,
      "grad_norm": 0.4627848267555237,
      "learning_rate": 3e-06,
      "loss": 0.676,
      "step": 1493,
      "step_time": 7.276139251887798
    },
    {
      "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,
      "entropy": 1.44921875,
      "epoch": 0.7155172413793104,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0716366395354271,
      "learning_rate": 3e-06,
      "loss": -0.0031,
      "step": 1494,
      "step_time": 7.385489858686924
    },
    {
      "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,
      "entropy": 1.56171875,
      "epoch": 0.7159961685823755,
      "fork_head/loss": 2.6754846572875977,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.1239749863743782,
      "grad_norm": 0.41974666714668274,
      "learning_rate": 3e-06,
      "loss": 0.6713,
      "step": 1495,
      "step_time": 7.79998105019331
    },
    {
      "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,
      "entropy": 1.4377170138888888,
      "epoch": 0.7164750957854407,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07236219942569733,
      "learning_rate": 3e-06,
      "loss": -0.0031,
      "step": 1496,
      "step_time": 7.663121409714222
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 195.740234375,
      "completions/mean_terminated_length": 195.622314453125,
      "completions/min_length": 56.0,
      "completions/min_terminated_length": 56.0,
      "entropy": 1.6149088541666667,
      "epoch": 0.7169540229885057,
      "fork/depth_mean": 0.333900511264801,
      "fork/depth_std": 0.2766574025154114,
      "fork_head/loss": 2.7162251472473145,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.07516415417194366,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6716023087501526,
      "learning_rate": 3e-06,
      "loss": 0.6793,
      "num_tokens": 38568365.0,
      "reward": 1.0808155536651611,
      "reward_std": 0.5706558227539062,
      "rewards/code_format_reward/mean": 0.8203125,
      "rewards/code_format_reward/std": 0.2999306917190552,
      "rewards/code_reward/mean": 0.2605031430721283,
      "rewards/code_reward/std": 0.3766803741455078,
      "step": 1497,
      "step_time": 536.8236894309521
    },
    {
      "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,
      "entropy": 1.4912832710478041,
      "epoch": 0.7174329501915708,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16425877809524536,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 1498,
      "step_time": 8.29213684797287
    },
    {
      "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,
      "entropy": 1.6171875,
      "epoch": 0.717911877394636,
      "fork_head/loss": 2.7030832767486572,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.07889816164970398,
      "grad_norm": 0.6415725946426392,
      "learning_rate": 3e-06,
      "loss": 0.6761,
      "step": 1499,
      "step_time": 8.050395287573338
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.4705222845077515,
      "epoch": 0.7183908045977011,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09235037118196487,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 1500,
      "step_time": 7.705100163817406
    },
    {
      "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,
      "entropy": 1.5777994791666667,
      "epoch": 0.7188697318007663,
      "fork_head/loss": 2.6796369552612305,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.14644774794578552,
      "grad_norm": 0.5832608342170715,
      "learning_rate": 3e-06,
      "loss": 0.6698,
      "step": 1501,
      "step_time": 7.912817560136318
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.4646990696589153,
      "epoch": 0.7193486590038314,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09737823903560638,
      "learning_rate": 3e-06,
      "loss": -0.0029,
      "step": 1502,
      "step_time": 7.7259553745388985
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.16666666666666666,
      "entropy": 1.6853841145833333,
      "epoch": 0.7198275862068966,
      "fork_head/loss": 2.65336537361145,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.14644774794578552,
      "grad_norm": 0.5108456611633301,
      "learning_rate": 3e-06,
      "loss": 0.6629,
      "step": 1503,
      "step_time": 7.809218838810921
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.2222222222222222,
      "entropy": 1.5062934160232544,
      "epoch": 0.7203065134099617,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0796150341629982,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 1504,
      "step_time": 7.688022583723068
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 201.71875,
      "completions/mean_terminated_length": 201.71875,
      "completions/min_length": 66.0,
      "completions/min_terminated_length": 66.0,
      "entropy": 1.562109375,
      "epoch": 0.7207854406130269,
      "fork/depth_mean": 0.292351096868515,
      "fork/depth_std": 0.22331808507442474,
      "fork_head/loss": 2.7406976222991943,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.06480318307876587,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5853738784790039,
      "learning_rate": 3e-06,
      "loss": 0.6809,
      "num_tokens": 38769005.0,
      "reward": 1.1105862855911255,
      "reward_std": 0.56151282787323,
      "rewards/code_format_reward/mean": 0.8349609375,
      "rewards/code_format_reward/std": 0.2755680978298187,
      "rewards/code_reward/mean": 0.2756253480911255,
      "rewards/code_reward/std": 0.3914823830127716,
      "step": 1505,
      "step_time": 476.35089968144894
    },
    {
      "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,
      "entropy": 1.7593994140625,
      "epoch": 0.7212643678160919,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06757554411888123,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 1506,
      "step_time": 8.30424927920103
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.551171875,
      "epoch": 0.7217432950191571,
      "fork_head/loss": 2.7303130626678467,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.07263685017824173,
      "grad_norm": 0.5380768775939941,
      "learning_rate": 3e-06,
      "loss": 0.6773,
      "step": 1507,
      "step_time": 7.634451545774937
    },
    {
      "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,
      "entropy": 1.736328125,
      "epoch": 0.7222222222222222,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07266451418399811,
      "learning_rate": 3e-06,
      "loss": 0.0014,
      "step": 1508,
      "step_time": 7.695490472018719
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.15,
      "entropy": 1.52138671875,
      "epoch": 0.7227011494252874,
      "fork_head/loss": 2.712846040725708,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.07437114417552948,
      "grad_norm": 0.4877777695655823,
      "learning_rate": 3e-06,
      "loss": 0.6728,
      "step": 1509,
      "step_time": 7.772065348923206
    },
    {
      "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,
      "entropy": 1.749755859375,
      "epoch": 0.7231800766283525,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06310731917619705,
      "learning_rate": 3e-06,
      "loss": 0.0008,
      "step": 1510,
      "step_time": 7.272068381309509
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.15,
      "entropy": 1.513671875,
      "epoch": 0.7236590038314177,
      "fork_head/loss": 2.6926991939544678,
      "fork_head/num_positive": 10.516129493713379,
      "fork_head/pred_top1_mass": 0.07506543397903442,
      "grad_norm": 0.4463857412338257,
      "learning_rate": 3e-06,
      "loss": 0.6674,
      "step": 1511,
      "step_time": 7.629418835043907
    },
    {
      "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,
      "entropy": 1.77294921875,
      "epoch": 0.7241379310344828,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06374835222959518,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 1512,
      "step_time": 7.724256299436092
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 184.46875,
      "completions/mean_terminated_length": 184.46875,
      "completions/min_length": 13.0,
      "completions/min_terminated_length": 13.0,
      "entropy": 1.7274889854284434,
      "epoch": 0.7246168582375478,
      "fork/depth_mean": 0.2541565001010895,
      "fork/depth_std": 0.1976403295993805,
      "fork_head/loss": 2.7238590717315674,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.0771736279129982,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.49231454730033875,
      "learning_rate": 3e-06,
      "loss": 0.6838,
      "num_tokens": 38948141.0,
      "reward": 1.1019482612609863,
      "reward_std": 0.6163280606269836,
      "rewards/code_format_reward/mean": 0.796875,
      "rewards/code_format_reward/std": 0.3321489095687866,
      "rewards/code_reward/mean": 0.3050733208656311,
      "rewards/code_reward/std": 0.407625287771225,
      "step": 1513,
      "step_time": 437.294837474823
    },
    {
      "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,
      "entropy": 1.9550419516033597,
      "epoch": 0.725095785440613,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15828631818294525,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 1514,
      "step_time": 7.337834171950817
    },
    {
      "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,
      "entropy": 1.721692213645348,
      "epoch": 0.7255747126436781,
      "fork_head/loss": 2.7175607681274414,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.0765923261642456,
      "grad_norm": 0.47515353560447693,
      "learning_rate": 3e-06,
      "loss": 0.6811,
      "step": 1515,
      "step_time": 8.449060529470444
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.037037038140826754,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.037037038140826754,
      "entropy": 1.9929832352532282,
      "epoch": 0.7260536398467433,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12728291749954224,
      "learning_rate": 3e-06,
      "loss": -0.0006,
      "step": 1516,
      "step_time": 8.003713384270668
    },
    {
      "clip_ratio/high_max": 0.15384615384615385,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.6884765670849726,
      "epoch": 0.7265325670498084,
      "fork_head/loss": 2.7065718173980713,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.07298362255096436,
      "grad_norm": 0.4307793974876404,
      "learning_rate": 3e-06,
      "loss": 0.6766,
      "step": 1517,
      "step_time": 7.733816482126713
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 2.0160590410232544,
      "epoch": 0.7270114942528736,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07761317491531372,
      "learning_rate": 3e-06,
      "loss": -0.0013,
      "step": 1518,
      "step_time": 7.545086309313774
    },
    {
      "clip_ratio/high_max": 0.15384615384615385,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.699375258042262,
      "epoch": 0.7274904214559387,
      "fork_head/loss": 2.693310499191284,
      "fork_head/num_positive": 11.71875,
      "fork_head/pred_top1_mass": 0.07106167078018188,
      "grad_norm": 0.4182277321815491,
      "learning_rate": 3e-06,
      "loss": 0.6721,
      "step": 1519,
      "step_time": 7.3830355778336525
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.14814814925193787,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.14814814925193787,
      "entropy": 2.0304181045956082,
      "epoch": 0.7279693486590039,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07311449199914932,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 1520,
      "step_time": 7.318667002022266
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 178.697265625,
      "completions/mean_terminated_length": 178.697265625,
      "completions/min_length": 60.0,
      "completions/min_terminated_length": 60.0,
      "entropy": 1.5791015625,
      "epoch": 0.728448275862069,
      "fork/depth_mean": 0.22041495144367218,
      "fork/depth_std": 0.18644854426383972,
      "fork_head/loss": 2.7470436096191406,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.0775047317147255,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.510956346988678,
      "learning_rate": 3e-06,
      "loss": 0.6976,
      "num_tokens": 39130962.0,
      "reward": 0.9987048506736755,
      "reward_std": 0.5972860455513,
      "rewards/code_format_reward/mean": 0.7685546875,
      "rewards/code_format_reward/std": 0.3407248258590698,
      "rewards/code_reward/mean": 0.23015020787715912,
      "rewards/code_reward/std": 0.37125229835510254,
      "step": 1521,
      "step_time": 457.6775895357132
    },
    {
      "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,
      "entropy": 1.5545572936534882,
      "epoch": 0.7289272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16265106201171875,
      "learning_rate": 3e-06,
      "loss": -0.0078,
      "step": 1522,
      "step_time": 7.930737316608429
    },
    {
      "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,
      "entropy": 1.56767578125,
      "epoch": 0.7294061302681992,
      "fork_head/loss": 2.7385289669036865,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.07870008796453476,
      "grad_norm": 0.4635392129421234,
      "learning_rate": 3e-06,
      "loss": 0.695,
      "step": 1523,
      "step_time": 7.561041735112667
    },
    {
      "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,
      "entropy": 1.5602864563465118,
      "epoch": 0.7298850574712644,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.162272647023201,
      "learning_rate": 3e-06,
      "loss": -0.0083,
      "step": 1524,
      "step_time": 7.348002806305885
    },
    {
      "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,
      "entropy": 1.59814453125,
      "epoch": 0.7303639846743295,
      "fork_head/loss": 2.7261481285095215,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.07913022488355637,
      "grad_norm": 0.43207111954689026,
      "learning_rate": 3e-06,
      "loss": 0.6916,
      "step": 1525,
      "step_time": 7.823541060090065
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.15,
      "entropy": 1.5803059875965118,
      "epoch": 0.7308429118773946,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1223580613732338,
      "learning_rate": 3e-06,
      "loss": -0.01,
      "step": 1526,
      "step_time": 7.417480327188969
    },
    {
      "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,
      "entropy": 1.62978515625,
      "epoch": 0.7313218390804598,
      "fork_head/loss": 2.7109038829803467,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.08402419835329056,
      "grad_norm": 0.40449804067611694,
      "learning_rate": 3e-06,
      "loss": 0.6867,
      "step": 1527,
      "step_time": 7.765740290284157
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.03333333432674408,
      "clip_ratio/low_mean": 0.15,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1833333343267441,
      "entropy": 1.5858723938465118,
      "epoch": 0.7318007662835249,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0941624715924263,
      "learning_rate": 3e-06,
      "loss": -0.0107,
      "step": 1528,
      "step_time": 7.704613797366619
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 186.984375,
      "completions/mean_terminated_length": 186.984375,
      "completions/min_length": 70.0,
      "completions/min_terminated_length": 70.0,
      "entropy": 1.4423828125,
      "epoch": 0.73227969348659,
      "fork/depth_mean": 0.23908153176307678,
      "fork/depth_std": 0.21137286722660065,
      "fork_head/loss": 2.7359890937805176,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.06754648685455322,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5045578479766846,
      "learning_rate": 3e-06,
      "loss": 0.691,
      "num_tokens": 39321594.0,
      "reward": 1.0841717720031738,
      "reward_std": 0.5556795597076416,
      "rewards/code_format_reward/mean": 0.82421875,
      "rewards/code_format_reward/std": 0.30705878138542175,
      "rewards/code_reward/mean": 0.25995293259620667,
      "rewards/code_reward/std": 0.3615990877151489,
      "step": 1529,
      "step_time": 532.6936872825027
    },
    {
      "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,
      "entropy": 1.3605143249034881,
      "epoch": 0.7327586206896551,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1804383397102356,
      "learning_rate": 3e-06,
      "loss": -0.0082,
      "step": 1530,
      "step_time": 7.72706476598978
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.4639811217784882,
      "epoch": 0.7332375478927203,
      "fork_head/loss": 2.7275514602661133,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.07093271613121033,
      "grad_norm": 0.49840301275253296,
      "learning_rate": 3e-06,
      "loss": 0.6879,
      "step": 1531,
      "step_time": 8.100192405283451
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.03333333432674408,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333432674408,
      "entropy": 1.3622070252895355,
      "epoch": 0.7337164750957854,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13839560747146606,
      "learning_rate": 3e-06,
      "loss": -0.0089,
      "step": 1532,
      "step_time": 7.736477427184582
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.020833333333333332,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.455411781867345,
      "epoch": 0.7341954022988506,
      "fork_head/loss": 2.715019702911377,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.07297392189502716,
      "grad_norm": 0.4556269347667694,
      "learning_rate": 3e-06,
      "loss": 0.6835,
      "step": 1533,
      "step_time": 7.393910974264145
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.03333333432674408,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333432674408,
      "entropy": 1.3454427123069763,
      "epoch": 0.7346743295019157,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1384958177804947,
      "learning_rate": 3e-06,
      "loss": -0.0099,
      "step": 1534,
      "step_time": 7.817169182002544
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.05833333358168602,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.10000000024835269,
      "entropy": 1.4595865905284882,
      "epoch": 0.7351532567049809,
      "fork_head/loss": 2.6993093490600586,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.08516833186149597,
      "grad_norm": 0.40738075971603394,
      "learning_rate": 3e-06,
      "loss": 0.6786,
      "step": 1535,
      "step_time": 7.495090700685978
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.03333333432674408,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333432674408,
      "entropy": 1.3576822936534882,
      "epoch": 0.735632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14352485537528992,
      "learning_rate": 3e-06,
      "loss": -0.0114,
      "step": 1536,
      "step_time": 7.172332227230072
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 173.609375,
      "completions/mean_terminated_length": 173.44813537597656,
      "completions/min_length": 66.0,
      "completions/min_terminated_length": 66.0,
      "entropy": 1.47958984375,
      "epoch": 0.7361111111111112,
      "fork/depth_mean": 0.2179853916168213,
      "fork/depth_std": 0.1876557320356369,
      "fork_head/loss": 2.726149320602417,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.07600525766611099,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.48058968782424927,
      "learning_rate": 3e-06,
      "loss": 0.6853,
      "num_tokens": 39512690.0,
      "reward": 0.9953204989433289,
      "reward_std": 0.6001764535903931,
      "rewards/code_format_reward/mean": 0.7578125,
      "rewards/code_format_reward/std": 0.33970388770103455,
      "rewards/code_reward/mean": 0.23750793933868408,
      "rewards/code_reward/std": 0.37742751836776733,
      "step": 1537,
      "step_time": 540.8782968074083
    },
    {
      "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,
      "entropy": 2.1831597222222223,
      "epoch": 0.7365900383141762,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12678876519203186,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 1538,
      "step_time": 7.669106155633926
    },
    {
      "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,
      "entropy": 1.471875,
      "epoch": 0.7370689655172413,
      "fork_head/loss": 2.7189247608184814,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.08261339366436005,
      "grad_norm": 0.42631977796554565,
      "learning_rate": 3e-06,
      "loss": 0.6834,
      "step": 1539,
      "step_time": 7.38191407173872
    },
    {
      "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,
      "entropy": 2.164822048611111,
      "epoch": 0.7375478927203065,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13477595150470734,
      "learning_rate": 3e-06,
      "loss": -0.0041,
      "step": 1540,
      "step_time": 7.508913055062294
    },
    {
      "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,
      "entropy": 1.4744140625,
      "epoch": 0.7380268199233716,
      "fork_head/loss": 2.707718849182129,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.07842180877923965,
      "grad_norm": 0.3962675631046295,
      "learning_rate": 3e-06,
      "loss": 0.68,
      "step": 1541,
      "step_time": 7.422584943473339
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 2.137767646047804,
      "epoch": 0.7385057471264368,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07106178998947144,
      "learning_rate": 3e-06,
      "loss": -0.005,
      "step": 1542,
      "step_time": 7.4754040986299515
    },
    {
      "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,
      "entropy": 1.4685546875,
      "epoch": 0.7389846743295019,
      "fork_head/loss": 2.695218086242676,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.08724720031023026,
      "grad_norm": 0.37353235483169556,
      "learning_rate": 3e-06,
      "loss": 0.6771,
      "step": 1543,
      "step_time": 7.940148748457432
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.2222222222222222,
      "entropy": 2.1631944444444446,
      "epoch": 0.7394636015325671,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05056702345609665,
      "learning_rate": 3e-06,
      "loss": -0.0057,
      "step": 1544,
      "step_time": 7.185071907937527
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 182.443359375,
      "completions/mean_terminated_length": 182.29940795898438,
      "completions/min_length": 54.0,
      "completions/min_terminated_length": 54.0,
      "entropy": 2.345703125,
      "epoch": 0.7399425287356322,
      "fork/depth_mean": 0.2908271551132202,
      "fork/depth_std": 0.24797216057777405,
      "fork_head/loss": 2.741117477416992,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.05216376110911369,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7383875846862793,
      "learning_rate": 3e-06,
      "loss": 0.6906,
      "num_tokens": 39706229.0,
      "reward": 0.9452080726623535,
      "reward_std": 0.5519225597381592,
      "rewards/code_format_reward/mean": 0.763671875,
      "rewards/code_format_reward/std": 0.3395181894302368,
      "rewards/code_reward/mean": 0.1815362125635147,
      "rewards/code_reward/std": 0.33093270659446716,
      "step": 1545,
      "step_time": 465.0169722288847
    },
    {
      "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,
      "entropy": 1.3404947916666667,
      "epoch": 0.7404214559386973,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08623725175857544,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 1546,
      "step_time": 7.540657974779606
    },
    {
      "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,
      "entropy": 2.369140625,
      "epoch": 0.7409003831417624,
      "fork_head/loss": 2.7246105670928955,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.05936941131949425,
      "grad_norm": 0.6647422909736633,
      "learning_rate": 3e-06,
      "loss": 0.6861,
      "step": 1547,
      "step_time": 7.790370836853981
    },
    {
      "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,
      "entropy": 1.3385416666666667,
      "epoch": 0.7413793103448276,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10554826259613037,
      "learning_rate": 3e-06,
      "loss": -0.0044,
      "step": 1548,
      "step_time": 7.926804229617119
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.25,
      "entropy": 2.4296875,
      "epoch": 0.7418582375478927,
      "fork_head/loss": 2.697840690612793,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.11690133064985275,
      "grad_norm": 0.561761736869812,
      "learning_rate": 3e-06,
      "loss": 0.6786,
      "step": 1549,
      "step_time": 7.378112368285656
    },
    {
      "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,
      "entropy": 1.3305121527777777,
      "epoch": 0.7423371647509579,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09118472784757614,
      "learning_rate": 3e-06,
      "loss": -0.0048,
      "step": 1550,
      "step_time": 7.33522342890501
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.25,
      "entropy": 2.5625,
      "epoch": 0.742816091954023,
      "fork_head/loss": 2.6707749366760254,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.1328607052564621,
      "grad_norm": 0.531539261341095,
      "learning_rate": 3e-06,
      "loss": 0.6715,
      "step": 1551,
      "step_time": 7.431642383337021
    },
    {
      "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,
      "entropy": 1.3314887152777777,
      "epoch": 0.7432950191570882,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11794424802064896,
      "learning_rate": 3e-06,
      "loss": -0.0051,
      "step": 1552,
      "step_time": 7.644943252205849
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 188.904296875,
      "completions/mean_terminated_length": 188.7729949951172,
      "completions/min_length": 47.0,
      "completions/min_terminated_length": 47.0,
      "entropy": 1.959765625,
      "epoch": 0.7437739463601533,
      "fork/depth_mean": 0.2944221496582031,
      "fork/depth_std": 0.2420850247144699,
      "fork_head/loss": 2.731923818588257,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.06658589094877243,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5838572382926941,
      "learning_rate": 3e-06,
      "loss": 0.6809,
      "num_tokens": 39904804.0,
      "reward": 0.9273373484611511,
      "reward_std": 0.5688993334770203,
      "rewards/code_format_reward/mean": 0.736328125,
      "rewards/code_format_reward/std": 0.3452339768409729,
      "rewards/code_reward/mean": 0.1910092830657959,
      "rewards/code_reward/std": 0.34217655658721924,
      "step": 1553,
      "step_time": 533.2300861775875
    },
    {
      "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,
      "entropy": 1.760892432469588,
      "epoch": 0.7442528735632183,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2209196537733078,
      "learning_rate": 3e-06,
      "loss": 0.0024,
      "step": 1554,
      "step_time": 8.856486156582832
    },
    {
      "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,
      "entropy": 1.99560546875,
      "epoch": 0.7447318007662835,
      "fork_head/loss": 2.723381519317627,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.07080215215682983,
      "grad_norm": 0.5147295594215393,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "step": 1555,
      "step_time": 7.350439645349979
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.11538461538461539,
      "entropy": 1.7603665911234343,
      "epoch": 0.7452107279693486,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2302352339029312,
      "learning_rate": 3e-06,
      "loss": 0.0019,
      "step": 1556,
      "step_time": 7.769287057220936
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.05,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.99375,
      "epoch": 0.7456896551724138,
      "fork_head/loss": 2.7092955112457275,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.06859320402145386,
      "grad_norm": 0.4890981614589691,
      "learning_rate": 3e-06,
      "loss": 0.6749,
      "step": 1557,
      "step_time": 7.724744789302349
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.06410256486672622,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.10256410332826468,
      "entropy": 1.7649990045107329,
      "epoch": 0.7461685823754789,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07261817902326584,
      "learning_rate": 3e-06,
      "loss": 0.0002,
      "step": 1558,
      "step_time": 7.100544258952141
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.05,
      "clip_ratio/region_mean": 0.1,
      "entropy": 2.01220703125,
      "epoch": 0.7466475095785441,
      "fork_head/loss": 2.690814256668091,
      "fork_head/num_positive": 12.25,
      "fork_head/pred_top1_mass": 0.0757717490196228,
      "grad_norm": 0.4651314318180084,
      "learning_rate": 3e-06,
      "loss": 0.6702,
      "step": 1559,
      "step_time": 8.264982737600803
    },
    {
      "clip_ratio/high_max": 0.15384615384615385,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.051282052810375504,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.12820512973345244,
      "entropy": 1.7518279277361357,
      "epoch": 0.7471264367816092,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0677623599767685,
      "learning_rate": 3e-06,
      "loss": -0.0002,
      "step": 1560,
      "step_time": 7.671792931854725
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 177.43359375,
      "completions/mean_terminated_length": 177.43359375,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.3845703184604645,
      "epoch": 0.7476053639846744,
      "fork/depth_mean": 0.2944263219833374,
      "fork/depth_std": 0.23423147201538086,
      "fork_head/loss": 2.721487045288086,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.06934444606304169,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4467160403728485,
      "learning_rate": 3e-06,
      "loss": 0.6816,
      "num_tokens": 40087506.0,
      "reward": 0.974919319152832,
      "reward_std": 0.6335763335227966,
      "rewards/code_format_reward/mean": 0.7236328125,
      "rewards/code_format_reward/std": 0.35704824328422546,
      "rewards/code_reward/mean": 0.2512865960597992,
      "rewards/code_reward/std": 0.3864017426967621,
      "step": 1561,
      "step_time": 429.65822388231754
    },
    {
      "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,
      "entropy": 1.9866071428571428,
      "epoch": 0.7480842911877394,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06711108237504959,
      "learning_rate": 3e-06,
      "loss": 0.0007,
      "step": 1562,
      "step_time": 7.220076039433479
    },
    {
      "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,
      "entropy": 1.3937174499034881,
      "epoch": 0.7485632183908046,
      "fork_head/loss": 2.714625597000122,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07403703033924103,
      "grad_norm": 0.3912059962749481,
      "learning_rate": 3e-06,
      "loss": 0.6795,
      "step": 1563,
      "step_time": 8.238652341067791
    },
    {
      "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,
      "entropy": 1.95703125,
      "epoch": 0.7490421455938697,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07454177737236023,
      "learning_rate": 3e-06,
      "loss": 0.0006,
      "step": 1564,
      "step_time": 7.864592231810093
    },
    {
      "clip_ratio/high_max": 0.03333333432674408,
      "clip_ratio/high_mean": 0.01666666716337204,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.01666666716337204,
      "entropy": 1.3826822936534882,
      "epoch": 0.7495210727969349,
      "fork_head/loss": 2.705512523651123,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07773233950138092,
      "grad_norm": 0.34894436597824097,
      "learning_rate": 3e-06,
      "loss": 0.6766,
      "step": 1565,
      "step_time": 7.410274565219879
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.14285714285714285,
      "clip_ratio/low_min": 0.14285714285714285,
      "clip_ratio/region_mean": 0.14285714285714285,
      "entropy": 1.99609375,
      "epoch": 0.75,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06991235911846161,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 1566,
      "step_time": 7.725941866636276
    },
    {
      "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,
      "entropy": 1.4072916626930236,
      "epoch": 0.7504789272030651,
      "fork_head/loss": 2.694561004638672,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.08683434873819351,
      "grad_norm": 0.33692270517349243,
      "learning_rate": 3e-06,
      "loss": 0.6734,
      "step": 1567,
      "step_time": 8.018931858241558
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.14285714285714285,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.14285714285714285,
      "entropy": 1.955078125,
      "epoch": 0.7509578544061303,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04096390679478645,
      "learning_rate": 3e-06,
      "loss": -0.0003,
      "step": 1568,
      "step_time": 7.420079402625561
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 183.427734375,
      "completions/mean_terminated_length": 183.427734375,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.42468631809408,
      "epoch": 0.7514367816091954,
      "fork/depth_mean": 0.3988223969936371,
      "fork/depth_std": 0.2738179862499237,
      "fork_head/loss": 2.719836711883545,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.06734199076890945,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4766174256801605,
      "learning_rate": 3e-06,
      "loss": 0.6795,
      "num_tokens": 40277741.0,
      "reward": 1.1701842546463013,
      "reward_std": 0.5885690450668335,
      "rewards/code_format_reward/mean": 0.833984375,
      "rewards/code_format_reward/std": 0.2650136649608612,
      "rewards/code_reward/mean": 0.33619993925094604,
      "rewards/code_reward/std": 0.4190303385257721,
      "step": 1569,
      "step_time": 407.73472432792187
    },
    {
      "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,
      "entropy": 1.5233191360126843,
      "epoch": 0.7519157088122606,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13004596531391144,
      "learning_rate": 3e-06,
      "loss": 0.0003,
      "step": 1570,
      "step_time": 8.24436179548502
    },
    {
      "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,
      "entropy": 1.423739341172305,
      "epoch": 0.7523946360153256,
      "fork_head/loss": 2.713428497314453,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.0709504559636116,
      "grad_norm": 0.4547756314277649,
      "learning_rate": 3e-06,
      "loss": 0.6772,
      "step": 1571,
      "step_time": 8.121442757546902
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.5349195003509521,
      "epoch": 0.7528735632183908,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12710770964622498,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 1572,
      "step_time": 7.28192101418972
    },
    {
      "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,
      "entropy": 1.4387724778868936,
      "epoch": 0.7533524904214559,
      "fork_head/loss": 2.7014689445495605,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.07753404974937439,
      "grad_norm": 0.42741307616233826,
      "learning_rate": 3e-06,
      "loss": 0.6732,
      "step": 1573,
      "step_time": 8.042969673871994
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.5150331475517966,
      "epoch": 0.7538314176245211,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13711875677108765,
      "learning_rate": 3e-06,
      "loss": -0.0005,
      "step": 1574,
      "step_time": 7.77912050485611
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.441287875175476,
      "epoch": 0.7543103448275862,
      "fork_head/loss": 2.687001943588257,
      "fork_head/num_positive": 11.90625,
      "fork_head/pred_top1_mass": 0.0753159150481224,
      "grad_norm": 0.39357078075408936,
      "learning_rate": 3e-06,
      "loss": 0.6686,
      "step": 1575,
      "step_time": 7.624381899833679
    },
    {
      "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,
      "entropy": 1.5250651077790693,
      "epoch": 0.7547892720306514,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13314273953437805,
      "learning_rate": 3e-06,
      "loss": -0.0018,
      "step": 1576,
      "step_time": 7.528708450496197
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 172.333984375,
      "completions/mean_terminated_length": 172.333984375,
      "completions/min_length": 56.0,
      "completions/min_terminated_length": 56.0,
      "entropy": 1.3157848011363635,
      "epoch": 0.7552681992337165,
      "fork/depth_mean": 0.3136504292488098,
      "fork/depth_std": 0.26150432229042053,
      "fork_head/loss": 2.732785224914551,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.06755885481834412,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5499385595321655,
      "learning_rate": 3e-06,
      "loss": 0.687,
      "num_tokens": 40466408.0,
      "reward": 1.022377848625183,
      "reward_std": 0.5438774228096008,
      "rewards/code_format_reward/mean": 0.7958984375,
      "rewards/code_format_reward/std": 0.30782803893089294,
      "rewards/code_reward/mean": 0.2264794111251831,
      "rewards/code_reward/std": 0.37379828095436096,
      "step": 1577,
      "step_time": 407.37196407467127
    },
    {
      "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,
      "entropy": 1.7777876420454546,
      "epoch": 0.7557471264367817,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.051120575517416,
      "learning_rate": 3e-06,
      "loss": -0.0035,
      "step": 1578,
      "step_time": 7.648734018206596
    },
    {
      "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,
      "entropy": 1.3212002840909092,
      "epoch": 0.7562260536398467,
      "fork_head/loss": 2.721381664276123,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08211956918239594,
      "grad_norm": 0.528831422328949,
      "learning_rate": 3e-06,
      "loss": 0.6842,
      "step": 1579,
      "step_time": 8.200476631522179
    },
    {
      "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,
      "entropy": 1.7702710736881604,
      "epoch": 0.7567049808429118,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06140183284878731,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 1580,
      "step_time": 7.5093115493655205
    },
    {
      "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,
      "entropy": 1.3103693181818181,
      "epoch": 0.757183908045977,
      "fork_head/loss": 2.7020926475524902,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08259810507297516,
      "grad_norm": 0.5004136562347412,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 1581,
      "step_time": 7.234812535345554
    },
    {
      "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,
      "entropy": 1.7895655740391125,
      "epoch": 0.7576628352490421,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05189323425292969,
      "learning_rate": 3e-06,
      "loss": -0.0035,
      "step": 1582,
      "step_time": 8.019521340727806
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.3298117897727273,
      "epoch": 0.7581417624521073,
      "fork_head/loss": 2.679436206817627,
      "fork_head/num_positive": 10.78125,
      "fork_head/pred_top1_mass": 0.08859021961688995,
      "grad_norm": 0.45338207483291626,
      "learning_rate": 3e-06,
      "loss": 0.6727,
      "step": 1583,
      "step_time": 7.184541553258896
    },
    {
      "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,
      "entropy": 1.7768406759608875,
      "epoch": 0.7586206896551724,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11317520588636398,
      "learning_rate": 3e-06,
      "loss": -0.0037,
      "step": 1584,
      "step_time": 7.020590655505657
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 181.384765625,
      "completions/mean_terminated_length": 181.384765625,
      "completions/min_length": 48.0,
      "completions/min_terminated_length": 48.0,
      "entropy": 1.1538628472222223,
      "epoch": 0.7590996168582376,
      "fork/depth_mean": 0.28927820920944214,
      "fork/depth_std": 0.2527294456958771,
      "fork_head/loss": 2.727832317352295,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.0675850510597229,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6783609390258789,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "num_tokens": 40657117.0,
      "reward": 1.0538794994354248,
      "reward_std": 0.611422598361969,
      "rewards/code_format_reward/mean": 0.7880859375,
      "rewards/code_format_reward/std": 0.339094340801239,
      "rewards/code_reward/mean": 0.26579350233078003,
      "rewards/code_reward/std": 0.3846275508403778,
      "step": 1585,
      "step_time": 424.14311503618956
    },
    {
      "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,
      "entropy": 1.80078125,
      "epoch": 0.7595785440613027,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11054841428995132,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1586,
      "step_time": 8.259677529335022
    },
    {
      "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,
      "entropy": 1.1417100694444444,
      "epoch": 0.7600574712643678,
      "fork_head/loss": 2.7133219242095947,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.07306847721338272,
      "grad_norm": 0.5589765310287476,
      "learning_rate": 3e-06,
      "loss": 0.675,
      "step": 1587,
      "step_time": 8.202637784183025
    },
    {
      "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,
      "entropy": 1.8091517857142858,
      "epoch": 0.7605363984674329,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10814685374498367,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1588,
      "step_time": 7.326293401420116
    },
    {
      "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,
      "entropy": 1.1489800347222223,
      "epoch": 0.7610153256704981,
      "fork_head/loss": 2.695648431777954,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.0984872430562973,
      "grad_norm": 0.47788888216018677,
      "learning_rate": 3e-06,
      "loss": 0.6705,
      "step": 1589,
      "step_time": 7.665240876376629
    },
    {
      "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,
      "entropy": 1.8074776785714286,
      "epoch": 0.7614942528735632,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11563403904438019,
      "learning_rate": 3e-06,
      "loss": 0.002,
      "step": 1590,
      "step_time": 7.939075857400894
    },
    {
      "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,
      "entropy": 1.1424696180555556,
      "epoch": 0.7619731800766284,
      "fork_head/loss": 2.675914764404297,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.1387021839618683,
      "grad_norm": 0.4719749391078949,
      "learning_rate": 3e-06,
      "loss": 0.6657,
      "step": 1591,
      "step_time": 7.503116883337498
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.14285714285714285,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.21428571428571427,
      "entropy": 1.7979910714285714,
      "epoch": 0.7624521072796935,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03353405371308327,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 1592,
      "step_time": 8.33813988417387
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 176.947265625,
      "completions/mean_terminated_length": 176.947265625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.0649233195516798,
      "epoch": 0.7629310344827587,
      "fork/depth_mean": 0.25528427958488464,
      "fork/depth_std": 0.23279763758182526,
      "fork_head/loss": 2.712087392807007,
      "fork_head/num_positive": 11.533333778381348,
      "fork_head/pred_top1_mass": 0.0796191468834877,
      "frac_reward_zero_std": 0.0625,
      "grad_norm": 0.5354710817337036,
      "learning_rate": 3e-06,
      "loss": 0.676,
      "num_tokens": 40846210.0,
      "reward": 1.0278927087783813,
      "reward_std": 0.5848143696784973,
      "rewards/code_format_reward/mean": 0.7919921875,
      "rewards/code_format_reward/std": 0.3371840715408325,
      "rewards/code_reward/mean": 0.23590049147605896,
      "rewards/code_reward/std": 0.36053356528282166,
      "step": 1593,
      "step_time": 529.3112192824483
    },
    {
      "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,
      "entropy": 1.59619140625,
      "epoch": 0.7634099616858238,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10514330118894577,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 1594,
      "step_time": 8.462634563446045
    },
    {
      "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,
      "entropy": 1.0652488436963823,
      "epoch": 0.7638888888888888,
      "fork_head/loss": 2.7051730155944824,
      "fork_head/num_positive": 11.533333778381348,
      "fork_head/pred_top1_mass": 0.07840313762426376,
      "grad_norm": 0.5402235388755798,
      "learning_rate": 3e-06,
      "loss": 0.6744,
      "step": 1595,
      "step_time": 7.146320939064026
    },
    {
      "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,
      "entropy": 1.5966796875,
      "epoch": 0.764367816091954,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10047408938407898,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1596,
      "step_time": 7.643926940858364
    },
    {
      "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,
      "entropy": 1.0687210692299738,
      "epoch": 0.7648467432950191,
      "fork_head/loss": 2.689962148666382,
      "fork_head/num_positive": 11.533333778381348,
      "fork_head/pred_top1_mass": 0.08121655136346817,
      "grad_norm": 0.4963892996311188,
      "learning_rate": 3e-06,
      "loss": 0.6693,
      "step": 1597,
      "step_time": 8.01333124935627
    },
    {
      "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,
      "entropy": 1.6009657118055556,
      "epoch": 0.7653256704980843,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10026224702596664,
      "learning_rate": 3e-06,
      "loss": 0.0027,
      "step": 1598,
      "step_time": 7.3458321839571
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.037037038140826754,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.037037038140826754,
      "entropy": 1.0700954894224803,
      "epoch": 0.7658045977011494,
      "fork_head/loss": 2.6714065074920654,
      "fork_head/num_positive": 11.533333778381348,
      "fork_head/pred_top1_mass": 0.10449806600809097,
      "grad_norm": 0.42423170804977417,
      "learning_rate": 3e-06,
      "loss": 0.6641,
      "step": 1599,
      "step_time": 7.617575563490391
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.6256510416666667,
      "epoch": 0.7662835249042146,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06979271024465561,
      "learning_rate": 3e-06,
      "loss": 0.0019,
      "step": 1600,
      "step_time": 7.531307727098465
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 184.8984375,
      "completions/mean_terminated_length": 184.8984375,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.8614760854027488,
      "epoch": 0.7667624521072797,
      "fork/depth_mean": 0.28053104877471924,
      "fork/depth_std": 0.24156111478805542,
      "fork_head/loss": 2.7359020709991455,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.07400497049093246,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5573444366455078,
      "learning_rate": 3e-06,
      "loss": 0.6935,
      "num_tokens": 41034782.0,
      "reward": 1.119775414466858,
      "reward_std": 0.6340643167495728,
      "rewards/code_format_reward/mean": 0.7998046875,
      "rewards/code_format_reward/std": 0.3404555022716522,
      "rewards/code_reward/mean": 0.31997066736221313,
      "rewards/code_reward/std": 0.41275832056999207,
      "step": 1601,
      "step_time": 457.77370584756136
    },
    {
      "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,
      "entropy": 1.4722222222222223,
      "epoch": 0.7672413793103449,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1258172243833542,
      "learning_rate": 3e-06,
      "loss": -0.0076,
      "step": 1602,
      "step_time": 7.257066175341606
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.8826941360126843,
      "epoch": 0.76772030651341,
      "fork_head/loss": 2.724625587463379,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.07402918487787247,
      "grad_norm": 0.5311563014984131,
      "learning_rate": 3e-06,
      "loss": 0.6901,
      "step": 1603,
      "step_time": 8.325623951852322
    },
    {
      "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,
      "entropy": 1.4542100694444444,
      "epoch": 0.7681992337164751,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13136976957321167,
      "learning_rate": 3e-06,
      "loss": -0.0072,
      "step": 1604,
      "step_time": 7.27550945430994
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.12121212211522189,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.16666666756976733,
      "entropy": 1.8729285110126843,
      "epoch": 0.7686781609195402,
      "fork_head/loss": 2.707385540008545,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.07306776195764542,
      "grad_norm": 0.47770464420318604,
      "learning_rate": 3e-06,
      "loss": 0.6845,
      "step": 1605,
      "step_time": 7.484113156795502
    },
    {
      "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,
      "entropy": 1.4862196180555556,
      "epoch": 0.7691570881226054,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11529039591550827,
      "learning_rate": 3e-06,
      "loss": -0.0084,
      "step": 1606,
      "step_time": 7.998063765466213
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.12121212211522189,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.16666666756976733,
      "entropy": 1.8794981132854114,
      "epoch": 0.7696360153256705,
      "fork_head/loss": 2.6864733695983887,
      "fork_head/num_positive": 10.875,
      "fork_head/pred_top1_mass": 0.13312768936157227,
      "grad_norm": 0.4522073566913605,
      "learning_rate": 3e-06,
      "loss": 0.6783,
      "step": 1607,
      "step_time": 6.878000549972057
    },
    {
      "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,
      "entropy": 1.4713541666666667,
      "epoch": 0.7701149425287356,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09842885285615921,
      "learning_rate": 3e-06,
      "loss": -0.0092,
      "step": 1608,
      "step_time": 7.835959166288376
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 163.197265625,
      "completions/mean_terminated_length": 163.01565551757812,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "entropy": 1.819365987410912,
      "epoch": 0.7705938697318008,
      "fork/depth_mean": 0.3186684846878052,
      "fork/depth_std": 0.26566094160079956,
      "fork_head/loss": 2.7304415702819824,
      "fork_head/num_positive": 11.903225898742676,
      "fork_head/pred_top1_mass": 0.06434106081724167,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.45929229259490967,
      "learning_rate": 3e-06,
      "loss": 0.6845,
      "num_tokens": 41219187.0,
      "reward": 0.9811039566993713,
      "reward_std": 0.6120336055755615,
      "rewards/code_format_reward/mean": 0.75,
      "rewards/code_format_reward/std": 0.3497273027896881,
      "rewards/code_reward/mean": 0.23110392689704895,
      "rewards/code_reward/std": 0.37180283665657043,
      "step": 1609,
      "step_time": 356.15161672234535
    },
    {
      "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,
      "entropy": 2.2943793402777777,
      "epoch": 0.7710727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12757502496242523,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 1610,
      "step_time": 8.40633425116539
    },
    {
      "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,
      "entropy": 1.8238982397776384,
      "epoch": 0.771551724137931,
      "fork_head/loss": 2.7234041690826416,
      "fork_head/num_positive": 11.903225898742676,
      "fork_head/pred_top1_mass": 0.06746984273195267,
      "grad_norm": 0.43504995107650757,
      "learning_rate": 3e-06,
      "loss": 0.6824,
      "step": 1611,
      "step_time": 7.503612004220486
    },
    {
      "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,
      "entropy": 2.2965494791666665,
      "epoch": 0.7720306513409961,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13095828890800476,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 1612,
      "step_time": 7.1745225712656975
    },
    {
      "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,
      "entropy": 1.8246494348232563,
      "epoch": 0.7725095785440613,
      "fork_head/loss": 2.711479902267456,
      "fork_head/num_positive": 11.903225898742676,
      "fork_head/pred_top1_mass": 0.06891966611146927,
      "grad_norm": 0.39772483706474304,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 1613,
      "step_time": 8.11818528175354
    },
    {
      "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,
      "entropy": 2.27734375,
      "epoch": 0.7729885057471264,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12609705328941345,
      "learning_rate": 3e-06,
      "loss": -0.0031,
      "step": 1614,
      "step_time": 6.940619312226772
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.15384615384615385,
      "entropy": 1.8388822078704834,
      "epoch": 0.7734674329501916,
      "fork_head/loss": 2.6974520683288574,
      "fork_head/num_positive": 11.903225898742676,
      "fork_head/pred_top1_mass": 0.08358265459537506,
      "grad_norm": 0.3710062503814697,
      "learning_rate": 3e-06,
      "loss": 0.6738,
      "step": 1615,
      "step_time": 7.560499355196953
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.1111111111111111,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.2777777777777778,
      "entropy": 2.2447916666666665,
      "epoch": 0.7739463601532567,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07654251158237457,
      "learning_rate": 3e-06,
      "loss": -0.0039,
      "step": 1616,
      "step_time": 7.940674193203449
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 178.873046875,
      "completions/mean_terminated_length": 178.873046875,
      "completions/min_length": 15.0,
      "completions/min_terminated_length": 15.0,
      "entropy": 1.91982421875,
      "epoch": 0.7744252873563219,
      "fork/depth_mean": 0.30634480714797974,
      "fork/depth_std": 0.27521196007728577,
      "fork_head/loss": 2.728524684906006,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.06809713691473007,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.47033923864364624,
      "learning_rate": 3e-06,
      "loss": 0.6762,
      "num_tokens": 41401490.0,
      "reward": 0.9715394377708435,
      "reward_std": 0.5866793394088745,
      "rewards/code_format_reward/mean": 0.744140625,
      "rewards/code_format_reward/std": 0.3368509113788605,
      "rewards/code_reward/mean": 0.2273988276720047,
      "rewards/code_reward/std": 0.3473547399044037,
      "step": 1617,
      "step_time": 516.3739641085267
    },
    {
      "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,
      "entropy": 1.8555131392045454,
      "epoch": 0.774904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19955341517925262,
      "learning_rate": 3e-06,
      "loss": 0.0012,
      "step": 1618,
      "step_time": 8.12548315525055
    },
    {
      "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,
      "entropy": 1.9017578125,
      "epoch": 0.7753831417624522,
      "fork_head/loss": 2.7201690673828125,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07037343084812164,
      "grad_norm": 0.46683573722839355,
      "learning_rate": 3e-06,
      "loss": 0.6739,
      "step": 1619,
      "step_time": 7.5804593116045
    },
    {
      "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,
      "entropy": 1.8318093039772727,
      "epoch": 0.7758620689655172,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1952708214521408,
      "learning_rate": 3e-06,
      "loss": -0.0003,
      "step": 1620,
      "step_time": 7.635256610810757
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.2,
      "clip_ratio/low_min": 0.2,
      "clip_ratio/region_mean": 0.3,
      "entropy": 1.8984375,
      "epoch": 0.7763409961685823,
      "fork_head/loss": 2.7057673931121826,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.071192666888237,
      "grad_norm": 0.4361971318721771,
      "learning_rate": 3e-06,
      "loss": 0.6696,
      "step": 1621,
      "step_time": 7.462408743798733
    },
    {
      "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,
      "entropy": 1.8082830255681819,
      "epoch": 0.7768199233716475,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.325668603181839,
      "learning_rate": 3e-06,
      "loss": -0.0012,
      "step": 1622,
      "step_time": 7.659833632409573
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.2,
      "entropy": 1.88671875,
      "epoch": 0.7772988505747126,
      "fork_head/loss": 2.6873857975006104,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.08234211802482605,
      "grad_norm": 0.41317376494407654,
      "learning_rate": 3e-06,
      "loss": 0.6649,
      "step": 1623,
      "step_time": 7.696011573076248
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.8338068181818181,
      "epoch": 0.7777777777777778,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13904428482055664,
      "learning_rate": 3e-06,
      "loss": -0.0022,
      "step": 1624,
      "step_time": 7.390308246016502
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 179.10546875,
      "completions/mean_terminated_length": 179.10546875,
      "completions/min_length": 23.0,
      "completions/min_terminated_length": 23.0,
      "entropy": 1.7071155394826616,
      "epoch": 0.7782567049808429,
      "fork/depth_mean": 0.31029394268989563,
      "fork/depth_std": 0.2644100487232208,
      "fork_head/loss": 2.732328414916992,
      "fork_head/num_positive": 12.21875,
      "fork_head/pred_top1_mass": 0.0703577771782875,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4721582233905792,
      "learning_rate": 3e-06,
      "loss": 0.6731,
      "num_tokens": 41586440.0,
      "reward": 0.9836258888244629,
      "reward_std": 0.5722676515579224,
      "rewards/code_format_reward/mean": 0.7802734375,
      "rewards/code_format_reward/std": 0.32965248823165894,
      "rewards/code_reward/mean": 0.20335251092910767,
      "rewards/code_reward/std": 0.35955628752708435,
      "step": 1625,
      "step_time": 424.9916955307126
    },
    {
      "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,
      "entropy": 1.7947591145833333,
      "epoch": 0.7787356321839081,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0649082288146019,
      "learning_rate": 3e-06,
      "loss": 0.0076,
      "step": 1626,
      "step_time": 7.744886063039303
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.6840645926339286,
      "epoch": 0.7792145593869731,
      "fork_head/loss": 2.7259535789489746,
      "fork_head/num_positive": 12.21875,
      "fork_head/pred_top1_mass": 0.07366352528333664,
      "grad_norm": 0.38023504614830017,
      "learning_rate": 3e-06,
      "loss": 0.671,
      "step": 1627,
      "step_time": 8.696444600820541
    },
    {
      "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,
      "entropy": 1.7921549479166667,
      "epoch": 0.7796934865900383,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.057383954524993896,
      "learning_rate": 3e-06,
      "loss": 0.0074,
      "step": 1628,
      "step_time": 7.223318912088871
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.656996956893376,
      "epoch": 0.7801724137931034,
      "fork_head/loss": 2.714792013168335,
      "fork_head/num_positive": 12.21875,
      "fork_head/pred_top1_mass": 0.07602488994598389,
      "grad_norm": 0.362493097782135,
      "learning_rate": 3e-06,
      "loss": 0.6674,
      "step": 1629,
      "step_time": 7.804230116307735
    },
    {
      "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,
      "entropy": 1.7984212239583333,
      "epoch": 0.7806513409961686,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06141570582985878,
      "learning_rate": 3e-06,
      "loss": 0.0071,
      "step": 1630,
      "step_time": 8.281326480209827
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.6534394706998552,
      "epoch": 0.7811302681992337,
      "fork_head/loss": 2.70135235786438,
      "fork_head/num_positive": 12.21875,
      "fork_head/pred_top1_mass": 0.09066841006278992,
      "grad_norm": 0.3349497616291046,
      "learning_rate": 3e-06,
      "loss": 0.664,
      "step": 1631,
      "step_time": 7.738295331597328
    },
    {
      "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,
      "entropy": 1.8262532552083333,
      "epoch": 0.7816091954022989,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1121542900800705,
      "learning_rate": 3e-06,
      "loss": 0.0066,
      "step": 1632,
      "step_time": 7.440846614539623
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 174.4453125,
      "completions/mean_terminated_length": 174.4453125,
      "completions/min_length": 24.0,
      "completions/min_terminated_length": 24.0,
      "entropy": 1.8101167253085546,
      "epoch": 0.782088122605364,
      "fork/depth_mean": 0.29044535756111145,
      "fork/depth_std": 0.2561399042606354,
      "fork_head/loss": 2.7232892513275146,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.06890150904655457,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.60362309217453,
      "learning_rate": 3e-06,
      "loss": 0.6862,
      "num_tokens": 41767404.0,
      "reward": 0.9158552885055542,
      "reward_std": 0.5600640773773193,
      "rewards/code_format_reward/mean": 0.73828125,
      "rewards/code_format_reward/std": 0.367276132106781,
      "rewards/code_reward/mean": 0.1775740683078766,
      "rewards/code_reward/std": 0.3259078562259674,
      "step": 1633,
      "step_time": 465.97080973535776
    },
    {
      "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,
      "entropy": 1.9056114783653846,
      "epoch": 0.7825670498084292,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.32191362977027893,
      "learning_rate": 3e-06,
      "loss": -0.0085,
      "step": 1634,
      "step_time": 7.900352627038956
    },
    {
      "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,
      "entropy": 1.8050595351627894,
      "epoch": 0.7830459770114943,
      "fork_head/loss": 2.713627815246582,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.06699281930923462,
      "grad_norm": 0.5804851651191711,
      "learning_rate": 3e-06,
      "loss": 0.6815,
      "step": 1635,
      "step_time": 7.461205936968327
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.8123121995192308,
      "epoch": 0.7835249042145593,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08857124298810959,
      "learning_rate": 3e-06,
      "loss": -0.0091,
      "step": 1636,
      "step_time": 7.742298401892185
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.011904762259551458,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.04761904797383717,
      "entropy": 1.792317705495017,
      "epoch": 0.7840038314176245,
      "fork_head/loss": 2.699040174484253,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.08563944697380066,
      "grad_norm": 0.4631795585155487,
      "learning_rate": 3e-06,
      "loss": 0.6754,
      "step": 1637,
      "step_time": 7.726798474788666
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.7661508413461537,
      "epoch": 0.7844827586206896,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10194876790046692,
      "learning_rate": 3e-06,
      "loss": -0.0099,
      "step": 1638,
      "step_time": 7.438102848827839
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.029761905116694316,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06547619083098002,
      "entropy": 1.786667593887874,
      "epoch": 0.7849616858237548,
      "fork_head/loss": 2.6834661960601807,
      "fork_head/num_positive": 11.6875,
      "fork_head/pred_top1_mass": 0.09062463045120239,
      "grad_norm": 0.4291606843471527,
      "learning_rate": 3e-06,
      "loss": 0.6698,
      "step": 1639,
      "step_time": 7.911218672990799
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.73095703125,
      "epoch": 0.7854406130268199,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09268602728843689,
      "learning_rate": 3e-06,
      "loss": -0.0103,
      "step": 1640,
      "step_time": 6.933789849281311
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 182.669921875,
      "completions/mean_terminated_length": 182.669921875,
      "completions/min_length": 34.0,
      "completions/min_terminated_length": 34.0,
      "entropy": 1.65380859375,
      "epoch": 0.7859195402298851,
      "fork/depth_mean": 0.2566729187965393,
      "fork/depth_std": 0.22188422083854675,
      "fork_head/loss": 2.743476390838623,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.060103170573711395,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6810645461082458,
      "learning_rate": 3e-06,
      "loss": 0.6881,
      "num_tokens": 41953651.0,
      "reward": 1.035825490951538,
      "reward_std": 0.619810163974762,
      "rewards/code_format_reward/mean": 0.7744140625,
      "rewards/code_format_reward/std": 0.3301391005516052,
      "rewards/code_reward/mean": 0.2614113688468933,
      "rewards/code_reward/std": 0.40175071358680725,
      "step": 1641,
      "step_time": 477.09335224330425
    },
    {
      "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,
      "entropy": 1.66243490907881,
      "epoch": 0.7863984674329502,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14308346807956696,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 1642,
      "step_time": 7.798128940165043
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.6474880675474803,
      "epoch": 0.7868773946360154,
      "fork_head/loss": 2.730006217956543,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.07163127511739731,
      "grad_norm": 0.6506496667861938,
      "learning_rate": 3e-06,
      "loss": 0.6847,
      "step": 1643,
      "step_time": 7.693417340517044
    },
    {
      "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,
      "entropy": 1.6651837428410847,
      "epoch": 0.7873563218390804,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14347976446151733,
      "learning_rate": 3e-06,
      "loss": -0.0035,
      "step": 1644,
      "step_time": 7.8422635570168495
    },
    {
      "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,
      "entropy": 1.6459689637025197,
      "epoch": 0.7878352490421456,
      "fork_head/loss": 2.7101197242736816,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.08347414433956146,
      "grad_norm": 0.6215928196907043,
      "learning_rate": 3e-06,
      "loss": 0.6794,
      "step": 1645,
      "step_time": 7.672035649418831
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.037037038140826754,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.037037038140826754,
      "entropy": 1.6724175347222223,
      "epoch": 0.7883141762452107,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09694896638393402,
      "learning_rate": 3e-06,
      "loss": -0.0046,
      "step": 1646,
      "step_time": 7.881716229021549
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06944444527228673,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06944444527228673,
      "entropy": 1.6648491819699605,
      "epoch": 0.7887931034482759,
      "fork_head/loss": 2.685800552368164,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.11022897064685822,
      "grad_norm": 0.5190877318382263,
      "learning_rate": 3e-06,
      "loss": 0.6726,
      "step": 1647,
      "step_time": 7.719199858605862
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.14814814925193787,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.14814814925193787,
      "entropy": 1.684859659936693,
      "epoch": 0.789272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0781940445303917,
      "learning_rate": 3e-06,
      "loss": -0.0052,
      "step": 1648,
      "step_time": 7.29811754822731
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 168.31640625,
      "completions/mean_terminated_length": 168.31640625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.4275535345077515,
      "epoch": 0.7897509578544061,
      "fork/depth_mean": 0.27575457096099854,
      "fork/depth_std": 0.23484788835048676,
      "fork_head/loss": 2.7229535579681396,
      "fork_head/num_positive": 10.125,
      "fork_head/pred_top1_mass": 0.05392061173915863,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6228688955307007,
      "learning_rate": 3e-06,
      "loss": 0.6795,
      "num_tokens": 42128405.0,
      "reward": 1.0585319995880127,
      "reward_std": 0.6522288918495178,
      "rewards/code_format_reward/mean": 0.75390625,
      "rewards/code_format_reward/std": 0.33687928318977356,
      "rewards/code_reward/mean": 0.3046257197856903,
      "rewards/code_reward/std": 0.41124799847602844,
      "step": 1649,
      "step_time": 432.309296682477
    },
    {
      "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,
      "entropy": 1.8818655230782249,
      "epoch": 0.7902298850574713,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.23973043262958527,
      "learning_rate": 3e-06,
      "loss": 0.0023,
      "step": 1650,
      "step_time": 7.9381518959999084
    },
    {
      "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,
      "entropy": 1.4183666043811374,
      "epoch": 0.7907088122605364,
      "fork_head/loss": 2.711296558380127,
      "fork_head/num_positive": 10.125,
      "fork_head/pred_top1_mass": 0.05632155016064644,
      "grad_norm": 0.5877904891967773,
      "learning_rate": 3e-06,
      "loss": 0.6765,
      "step": 1651,
      "step_time": 7.305283881723881
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.8506747159090908,
      "epoch": 0.7911877394636015,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1971426159143448,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 1652,
      "step_time": 8.007600240409374
    },
    {
      "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,
      "entropy": 1.4242259793811374,
      "epoch": 0.7916666666666666,
      "fork_head/loss": 2.690875768661499,
      "fork_head/num_positive": 10.125,
      "fork_head/pred_top1_mass": 0.08323263376951218,
      "grad_norm": 0.5402908325195312,
      "learning_rate": 3e-06,
      "loss": 0.6709,
      "step": 1653,
      "step_time": 7.869535021483898
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.09090909090909091,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.8747928576035933,
      "epoch": 0.7921455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1955091506242752,
      "learning_rate": 3e-06,
      "loss": -0.0001,
      "step": 1654,
      "step_time": 7.368319615721703
    },
    {
      "clip_ratio/high_max": 0.1111111111111111,
      "clip_ratio/high_mean": 0.05555555555555555,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.4148943821589153,
      "epoch": 0.7926245210727969,
      "fork_head/loss": 2.6673812866210938,
      "fork_head/num_positive": 10.125,
      "fork_head/pred_top1_mass": 0.10265588760375977,
      "grad_norm": 0.49417561292648315,
      "learning_rate": 3e-06,
      "loss": 0.6641,
      "step": 1655,
      "step_time": 7.55899977684021
    },
    {
      "clip_ratio/high_max": 0.18181818181818182,
      "clip_ratio/high_mean": 0.13636363636363635,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.8536931818181819,
      "epoch": 0.7931034482758621,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1075291633605957,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 1656,
      "step_time": 7.612168237566948
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 178.859375,
      "completions/mean_terminated_length": 178.859375,
      "completions/min_length": 17.0,
      "completions/min_terminated_length": 17.0,
      "entropy": 1.4391818576388888,
      "epoch": 0.7935823754789272,
      "fork/depth_mean": 0.278293251991272,
      "fork/depth_std": 0.24490945041179657,
      "fork_head/loss": 2.7350080013275146,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.05894673243165016,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.8190966248512268,
      "learning_rate": 3e-06,
      "loss": 0.6853,
      "num_tokens": 42312413.0,
      "reward": 1.0040861368179321,
      "reward_std": 0.5885089039802551,
      "rewards/code_format_reward/mean": 0.75390625,
      "rewards/code_format_reward/std": 0.37273040413856506,
      "rewards/code_reward/mean": 0.2501799166202545,
      "rewards/code_reward/std": 0.3918035924434662,
      "step": 1657,
      "step_time": 386.5576263666153
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.025641026405187752,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06410256486672622,
      "entropy": 1.6982922645715566,
      "epoch": 0.7940613026819924,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12824732065200806,
      "learning_rate": 3e-06,
      "loss": -0.0011,
      "step": 1658,
      "step_time": 7.338604807853699
    },
    {
      "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,
      "entropy": 1.4414785835478041,
      "epoch": 0.7945402298850575,
      "fork_head/loss": 2.7125847339630127,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07206963002681732,
      "grad_norm": 0.7570093870162964,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 1659,
      "step_time": 7.955242455005646
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.7015975530330951,
      "epoch": 0.7950191570881227,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15082955360412598,
      "learning_rate": 3e-06,
      "loss": -0.0017,
      "step": 1660,
      "step_time": 7.936320520937443
    },
    {
      "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,
      "entropy": 1.4397967259089153,
      "epoch": 0.7954980842911877,
      "fork_head/loss": 2.680145025253296,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.11024956405162811,
      "grad_norm": 0.6608654260635376,
      "learning_rate": 3e-06,
      "loss": 0.6692,
      "step": 1661,
      "step_time": 7.087606683373451
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.038461538461538464,
      "clip_ratio/low_mean": 0.07692307692307693,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.11538461538461539,
      "entropy": 1.7027243559177105,
      "epoch": 0.7959770114942529,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1291087120771408,
      "learning_rate": 3e-06,
      "loss": -0.0026,
      "step": 1662,
      "step_time": 7.782496199011803
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777777777777776,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777777777777776,
      "entropy": 1.4406467013888888,
      "epoch": 0.796455938697318,
      "fork_head/loss": 2.646028757095337,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.11786328256130219,
      "grad_norm": 0.544061541557312,
      "learning_rate": 3e-06,
      "loss": 0.6603,
      "step": 1663,
      "step_time": 7.425881914794445
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.14102564178980315,
      "clip_ratio/low_min": 0.07692307692307693,
      "clip_ratio/region_mean": 0.14102564178980315,
      "entropy": 1.7241586538461537,
      "epoch": 0.7969348659003831,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15652510523796082,
      "learning_rate": 3e-06,
      "loss": -0.0036,
      "step": 1664,
      "step_time": 7.466271348297596
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 191.630859375,
      "completions/mean_terminated_length": 191.50489807128906,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.0981987847222223,
      "epoch": 0.7974137931034483,
      "fork/depth_mean": 0.30759233236312866,
      "fork/depth_std": 0.2662152051925659,
      "fork_head/loss": 2.7323505878448486,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.07271755486726761,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5668928027153015,
      "learning_rate": 3e-06,
      "loss": 0.6836,
      "num_tokens": 42508752.0,
      "reward": 0.9426342844963074,
      "reward_std": 0.5483670830726624,
      "rewards/code_format_reward/mean": 0.75390625,
      "rewards/code_format_reward/std": 0.3339621126651764,
      "rewards/code_reward/mean": 0.18872803449630737,
      "rewards/code_reward/std": 0.31552228331565857,
      "step": 1665,
      "step_time": 534.3964474648237
    },
    {
      "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,
      "entropy": 1.8526785714285714,
      "epoch": 0.7978927203065134,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05867677927017212,
      "learning_rate": 3e-06,
      "loss": -0.0012,
      "step": 1666,
      "step_time": 8.303844630718231
    },
    {
      "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,
      "entropy": 1.0986328125,
      "epoch": 0.7983716475095786,
      "fork_head/loss": 2.722517967224121,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.07366964221000671,
      "grad_norm": 0.4895974397659302,
      "learning_rate": 3e-06,
      "loss": 0.6812,
      "step": 1667,
      "step_time": 7.996964298188686
    },
    {
      "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,
      "entropy": 1.859002981867109,
      "epoch": 0.7988505747126436,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0606057345867157,
      "learning_rate": 3e-06,
      "loss": -0.0015,
      "step": 1668,
      "step_time": 7.759949065744877
    },
    {
      "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,
      "entropy": 1.107421875,
      "epoch": 0.7993295019157088,
      "fork_head/loss": 2.7086071968078613,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.07225207984447479,
      "grad_norm": 0.4139837622642517,
      "learning_rate": 3e-06,
      "loss": 0.6776,
      "step": 1669,
      "step_time": 7.560967355966568
    },
    {
      "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,
      "entropy": 1.868768606867109,
      "epoch": 0.7998084291187739,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0638652816414833,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 1670,
      "step_time": 7.594862677156925
    },
    {
      "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,
      "entropy": 1.1089409722222223,
      "epoch": 0.8002873563218391,
      "fork_head/loss": 2.694568157196045,
      "fork_head/num_positive": 12.28125,
      "fork_head/pred_top1_mass": 0.08430754393339157,
      "grad_norm": 0.3779997229576111,
      "learning_rate": 3e-06,
      "loss": 0.6739,
      "step": 1671,
      "step_time": 7.4797787219285965
    },
    {
      "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,
      "entropy": 1.8709077324186052,
      "epoch": 0.8007662835249042,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0597514733672142,
      "learning_rate": 3e-06,
      "loss": -0.002,
      "step": 1672,
      "step_time": 7.858847543597221
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0416666679084301,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0416666679084301,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 182.853515625,
      "completions/mean_terminated_length": 182.853515625,
      "completions/min_length": 53.0,
      "completions/min_terminated_length": 53.0,
      "entropy": 1.39892578125,
      "epoch": 0.8012452107279694,
      "fork/depth_mean": 0.32360872626304626,
      "fork/depth_std": 0.26501229405403137,
      "fork_head/loss": 2.7389559745788574,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.05273577570915222,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5505276918411255,
      "learning_rate": 3e-06,
      "loss": 0.682,
      "num_tokens": 42698101.0,
      "reward": 1.0608234405517578,
      "reward_std": 0.545071005821228,
      "rewards/code_format_reward/mean": 0.8173828125,
      "rewards/code_format_reward/std": 0.28728872537612915,
      "rewards/code_reward/mean": 0.2434406578540802,
      "rewards/code_reward/std": 0.36511433124542236,
      "step": 1673,
      "step_time": 470.7398479282856
    },
    {
      "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,
      "entropy": 1.4582331730769231,
      "epoch": 0.8017241379310345,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17829293012619019,
      "learning_rate": 3e-06,
      "loss": 0.0016,
      "step": 1674,
      "step_time": 8.279980912804604
    },
    {
      "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,
      "entropy": 1.4052734375,
      "epoch": 0.8022030651340997,
      "fork_head/loss": 2.7293214797973633,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.05226830393075943,
      "grad_norm": 0.5132107138633728,
      "learning_rate": 3e-06,
      "loss": 0.6794,
      "step": 1675,
      "step_time": 7.124364048242569
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.4571814903846154,
      "epoch": 0.8026819923371648,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1637042760848999,
      "learning_rate": 3e-06,
      "loss": 0.0005,
      "step": 1676,
      "step_time": 7.628060266375542
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0416666679084301,
      "clip_ratio/low_mean": 0.125,
      "clip_ratio/low_min": 0.125,
      "clip_ratio/region_mean": 0.1666666679084301,
      "entropy": 1.3894856795668602,
      "epoch": 0.8031609195402298,
      "fork_head/loss": 2.7142205238342285,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.05710005760192871,
      "grad_norm": 0.4375266134738922,
      "learning_rate": 3e-06,
      "loss": 0.6751,
      "step": 1677,
      "step_time": 7.631436295807362
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.4689002403846154,
      "epoch": 0.803639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16822029650211334,
      "learning_rate": 3e-06,
      "loss": -0.0009,
      "step": 1678,
      "step_time": 7.512751705944538
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0416666679084301,
      "clip_ratio/low_mean": 0.125,
      "clip_ratio/low_min": 0.125,
      "clip_ratio/region_mean": 0.1666666679084301,
      "entropy": 1.3564453125,
      "epoch": 0.8041187739463601,
      "fork_head/loss": 2.697742223739624,
      "fork_head/num_positive": 12.03125,
      "fork_head/pred_top1_mass": 0.0669177770614624,
      "grad_norm": 0.39069443941116333,
      "learning_rate": 3e-06,
      "loss": 0.6706,
      "step": 1679,
      "step_time": 7.734044693410397
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.038461538461538464,
      "entropy": 1.4815204326923077,
      "epoch": 0.8045977011494253,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1739567369222641,
      "learning_rate": 3e-06,
      "loss": -0.0021,
      "step": 1680,
      "step_time": 7.195019096136093
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.125,
      "clip_ratio/low_min": 0.125,
      "clip_ratio/region_mean": 0.125,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 194.931640625,
      "completions/mean_terminated_length": 194.931640625,
      "completions/min_length": 59.0,
      "completions/min_terminated_length": 59.0,
      "entropy": 1.3995361328125,
      "epoch": 0.8050766283524904,
      "fork/depth_mean": 0.2931597828865051,
      "fork/depth_std": 0.23074522614479065,
      "fork_head/loss": 2.7470405101776123,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.05722810700535774,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6291385889053345,
      "learning_rate": 3e-06,
      "loss": 0.6858,
      "num_tokens": 42894370.0,
      "reward": 0.9548913240432739,
      "reward_std": 0.496175616979599,
      "rewards/code_format_reward/mean": 0.8125,
      "rewards/code_format_reward/std": 0.30966177582740784,
      "rewards/code_reward/mean": 0.14239129424095154,
      "rewards/code_reward/std": 0.311124712228775,
      "step": 1681,
      "step_time": 507.66042163968086
    },
    {
      "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,
      "entropy": 1.568359375,
      "epoch": 0.8055555555555556,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13207697868347168,
      "learning_rate": 3e-06,
      "loss": 0.0035,
      "step": 1682,
      "step_time": 8.051269605755806
    },
    {
      "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,
      "entropy": 1.3760986328125,
      "epoch": 0.8060344827586207,
      "fork_head/loss": 2.7336432933807373,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.0621204748749733,
      "grad_norm": 0.54183429479599,
      "learning_rate": 3e-06,
      "loss": 0.6816,
      "step": 1683,
      "step_time": 7.951900668442249
    },
    {
      "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,
      "entropy": 1.58515625,
      "epoch": 0.8065134099616859,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13948644697666168,
      "learning_rate": 3e-06,
      "loss": 0.0037,
      "step": 1684,
      "step_time": 7.121768973767757
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.125,
      "clip_ratio/low_min": 0.125,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.389404296875,
      "epoch": 0.8069923371647509,
      "fork_head/loss": 2.7157344818115234,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.0698925107717514,
      "grad_norm": 0.4433601200580597,
      "learning_rate": 3e-06,
      "loss": 0.6769,
      "step": 1685,
      "step_time": 8.275402672588825
    },
    {
      "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,
      "entropy": 1.628515625,
      "epoch": 0.8074712643678161,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10825586318969727,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 1686,
      "step_time": 7.58610487729311
    },
    {
      "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,
      "entropy": 1.3416748046875,
      "epoch": 0.8079501915708812,
      "fork_head/loss": 2.698239326477051,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.0816265344619751,
      "grad_norm": 0.4030522406101227,
      "learning_rate": 3e-06,
      "loss": 0.672,
      "step": 1687,
      "step_time": 7.121706768870354
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.2,
      "clip_ratio/low_min": 0.2,
      "clip_ratio/region_mean": 0.2,
      "entropy": 1.680859375,
      "epoch": 0.8084291187739464,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03828394040465355,
      "learning_rate": 3e-06,
      "loss": 0.0023,
      "step": 1688,
      "step_time": 8.327973507344723
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 192.052734375,
      "completions/mean_terminated_length": 192.052734375,
      "completions/min_length": 59.0,
      "completions/min_terminated_length": 59.0,
      "entropy": 0.94599609375,
      "epoch": 0.8089080459770115,
      "fork/depth_mean": 0.3348939120769501,
      "fork/depth_std": 0.22374509274959564,
      "fork_head/loss": 2.7027597427368164,
      "fork_head/num_positive": 10.59375,
      "fork_head/pred_top1_mass": 0.08859985321760178,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5726222395896912,
      "learning_rate": 3e-06,
      "loss": 0.677,
      "num_tokens": 43083389.0,
      "reward": 1.1277841329574585,
      "reward_std": 0.5802088379859924,
      "rewards/code_format_reward/mean": 0.794921875,
      "rewards/code_format_reward/std": 0.29828864336013794,
      "rewards/code_reward/mean": 0.33286231756210327,
      "rewards/code_reward/std": 0.4046121835708618,
      "step": 1689,
      "step_time": 471.8238093033433
    },
    {
      "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,
      "entropy": 1.2159830729166667,
      "epoch": 0.8093869731800766,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0955977812409401,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 1690,
      "step_time": 8.30879508703947
    },
    {
      "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,
      "entropy": 0.96162109375,
      "epoch": 0.8098659003831418,
      "fork_head/loss": 2.6920082569122314,
      "fork_head/num_positive": 10.59375,
      "fork_head/pred_top1_mass": 0.09229012578725815,
      "grad_norm": 0.5715081095695496,
      "learning_rate": 3e-06,
      "loss": 0.6739,
      "step": 1691,
      "step_time": 7.19587492197752
    },
    {
      "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,
      "entropy": 1.2080078125,
      "epoch": 0.8103448275862069,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09429975599050522,
      "learning_rate": 3e-06,
      "loss": -0.0027,
      "step": 1692,
      "step_time": 7.891079485416412
    },
    {
      "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,
      "entropy": 0.9462890625,
      "epoch": 0.810823754789272,
      "fork_head/loss": 2.6725096702575684,
      "fork_head/num_positive": 10.59375,
      "fork_head/pred_top1_mass": 0.09762676060199738,
      "grad_norm": 0.5607989430427551,
      "learning_rate": 3e-06,
      "loss": 0.6693,
      "step": 1693,
      "step_time": 7.877622723579407
    },
    {
      "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,
      "entropy": 1.1934407552083333,
      "epoch": 0.8113026819923371,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09995875507593155,
      "learning_rate": 3e-06,
      "loss": -0.0032,
      "step": 1694,
      "step_time": 6.9349254965782166
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 0.9501953125,
      "epoch": 0.8117816091954023,
      "fork_head/loss": 2.646657943725586,
      "fork_head/num_positive": 10.59375,
      "fork_head/pred_top1_mass": 0.10215619206428528,
      "grad_norm": 0.523724377155304,
      "learning_rate": 3e-06,
      "loss": 0.6617,
      "step": 1695,
      "step_time": 7.680436410009861
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.1979166666666667,
      "epoch": 0.8122605363984674,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07100048661231995,
      "learning_rate": 3e-06,
      "loss": -0.0038,
      "step": 1696,
      "step_time": 7.60769160836935
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 189.900390625,
      "completions/mean_terminated_length": 189.900390625,
      "completions/min_length": 70.0,
      "completions/min_terminated_length": 70.0,
      "entropy": 1.3173828125,
      "epoch": 0.8127394636015326,
      "fork/depth_mean": 0.3286548852920532,
      "fork/depth_std": 0.2254074364900589,
      "fork_head/loss": 2.7283365726470947,
      "fork_head/num_positive": 11.645161628723145,
      "fork_head/pred_top1_mass": 0.07483305037021637,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5569212436676025,
      "learning_rate": 3e-06,
      "loss": 0.6879,
      "num_tokens": 43278874.0,
      "reward": 1.1698100566864014,
      "reward_std": 0.5237036943435669,
      "rewards/code_format_reward/mean": 0.8662109375,
      "rewards/code_format_reward/std": 0.24861828982830048,
      "rewards/code_reward/mean": 0.30359917879104614,
      "rewards/code_reward/std": 0.36850443482398987,
      "step": 1697,
      "step_time": 443.60256500542164
    },
    {
      "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,
      "entropy": 1.1485877403846154,
      "epoch": 0.8132183908045977,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08715919405221939,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 1698,
      "step_time": 7.053438648581505
    },
    {
      "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,
      "entropy": 1.315478515625,
      "epoch": 0.8136973180076629,
      "fork_head/loss": 2.717130661010742,
      "fork_head/num_positive": 11.645161628723145,
      "fork_head/pred_top1_mass": 0.08159855008125305,
      "grad_norm": 0.5253411531448364,
      "learning_rate": 3e-06,
      "loss": 0.685,
      "step": 1699,
      "step_time": 8.107347451150417
    },
    {
      "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,
      "entropy": 1.1628605769230769,
      "epoch": 0.814176245210728,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12427154928445816,
      "learning_rate": 3e-06,
      "loss": -0.0038,
      "step": 1700,
      "step_time": 7.841774083673954
    },
    {
      "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,
      "entropy": 1.3196940064430236,
      "epoch": 0.8146551724137931,
      "fork_head/loss": 2.6996538639068604,
      "fork_head/num_positive": 11.645161628723145,
      "fork_head/pred_top1_mass": 0.09683857858181,
      "grad_norm": 0.48968398571014404,
      "learning_rate": 3e-06,
      "loss": 0.6799,
      "step": 1701,
      "step_time": 7.266111880540848
    },
    {
      "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,
      "entropy": 1.1853590745192308,
      "epoch": 0.8151340996168582,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1738494485616684,
      "learning_rate": 3e-06,
      "loss": -0.0044,
      "step": 1702,
      "step_time": 8.200465448200703
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.34111328125,
      "epoch": 0.8156130268199234,
      "fork_head/loss": 2.6794724464416504,
      "fork_head/num_positive": 11.645161628723145,
      "fork_head/pred_top1_mass": 0.10635598748922348,
      "grad_norm": 0.45648908615112305,
      "learning_rate": 3e-06,
      "loss": 0.6732,
      "step": 1703,
      "step_time": 7.756812885403633
    },
    {
      "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,
      "entropy": 1.1882887620192308,
      "epoch": 0.8160919540229885,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09297064691781998,
      "learning_rate": 3e-06,
      "loss": -0.0046,
      "step": 1704,
      "step_time": 7.472486011683941
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 189.28515625,
      "completions/mean_terminated_length": 189.28515625,
      "completions/min_length": 41.0,
      "completions/min_terminated_length": 41.0,
      "entropy": 1.4768880208333333,
      "epoch": 0.8165708812260536,
      "fork/depth_mean": 0.3433726131916046,
      "fork/depth_std": 0.24788202345371246,
      "fork_head/loss": 2.730684280395508,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.06651991605758667,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.49349263310432434,
      "learning_rate": 3e-06,
      "loss": 0.6825,
      "num_tokens": 43467116.0,
      "reward": 1.1635591983795166,
      "reward_std": 0.5629668235778809,
      "rewards/code_format_reward/mean": 0.85546875,
      "rewards/code_format_reward/std": 0.24947965145111084,
      "rewards/code_reward/mean": 0.30809053778648376,
      "rewards/code_reward/std": 0.40820834040641785,
      "step": 1705,
      "step_time": 525.7649584710598
    },
    {
      "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,
      "entropy": 1.8474121178899492,
      "epoch": 0.8170498084291188,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12360148131847382,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 1706,
      "step_time": 7.962917946279049
    },
    {
      "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,
      "entropy": 1.4833984375,
      "epoch": 0.8175287356321839,
      "fork_head/loss": 2.722294569015503,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.06408809870481491,
      "grad_norm": 0.4688494801521301,
      "learning_rate": 3e-06,
      "loss": 0.6796,
      "step": 1707,
      "step_time": 8.02228931337595
    },
    {
      "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,
      "entropy": 1.852736702987126,
      "epoch": 0.8180076628352491,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12917467951774597,
      "learning_rate": 3e-06,
      "loss": 0.0018,
      "step": 1708,
      "step_time": 7.407250039279461
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.013333333532015483,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.046666666865348816,
      "entropy": 1.4813671906789143,
      "epoch": 0.8184865900383141,
      "fork_head/loss": 2.7082419395446777,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.06686510145664215,
      "grad_norm": 0.41968944668769836,
      "learning_rate": 3e-06,
      "loss": 0.6749,
      "step": 1709,
      "step_time": 7.993480958044529
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.07142857142857142,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.8333216990743364,
      "epoch": 0.8189655172413793,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11044571548700333,
      "learning_rate": 3e-06,
      "loss": 0.0005,
      "step": 1710,
      "step_time": 7.107736743986607
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.046666666865348816,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.046666666865348816,
      "entropy": 1.4804513891537985,
      "epoch": 0.8194444444444444,
      "fork_head/loss": 2.690962553024292,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.09528446942567825,
      "grad_norm": 0.3970760703086853,
      "learning_rate": 3e-06,
      "loss": 0.67,
      "step": 1711,
      "step_time": 8.137358285486698
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.07142857142857142,
      "clip_ratio/low_mean": 0.07142857142857142,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.14285714285714285,
      "entropy": 1.8474353551864624,
      "epoch": 0.8199233716475096,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10380705446004868,
      "learning_rate": 3e-06,
      "loss": -0.0001,
      "step": 1712,
      "step_time": 7.345376834273338
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 175.01953125,
      "completions/mean_terminated_length": 175.01953125,
      "completions/min_length": 62.0,
      "completions/min_terminated_length": 62.0,
      "entropy": 1.15625,
      "epoch": 0.8204022988505747,
      "fork/depth_mean": 0.3370177447795868,
      "fork/depth_std": 0.27087196707725525,
      "fork_head/loss": 2.730753183364868,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.06688083708286285,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5376905798912048,
      "learning_rate": 3e-06,
      "loss": 0.6895,
      "num_tokens": 43646630.0,
      "reward": 1.0501283407211304,
      "reward_std": 0.6051070690155029,
      "rewards/code_format_reward/mean": 0.779296875,
      "rewards/code_format_reward/std": 0.3034715950489044,
      "rewards/code_reward/mean": 0.270831435918808,
      "rewards/code_reward/std": 0.39372140169143677,
      "step": 1713,
      "step_time": 513.1775993034244
    },
    {
      "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,
      "entropy": 1.6639293323863635,
      "epoch": 0.8208812260536399,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1358344405889511,
      "learning_rate": 3e-06,
      "loss": -0.0061,
      "step": 1714,
      "step_time": 8.028253182768822
    },
    {
      "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,
      "entropy": 1.1405029296875,
      "epoch": 0.821360153256705,
      "fork_head/loss": 2.719226837158203,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.055453110486269,
      "grad_norm": 0.49818354845046997,
      "learning_rate": 3e-06,
      "loss": 0.6855,
      "step": 1715,
      "step_time": 7.18964809179306
    },
    {
      "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,
      "entropy": 1.6539417613636365,
      "epoch": 0.8218390804597702,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14353959262371063,
      "learning_rate": 3e-06,
      "loss": -0.007,
      "step": 1716,
      "step_time": 7.9436585903167725
    },
    {
      "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,
      "entropy": 1.153564453125,
      "epoch": 0.8223180076628352,
      "fork_head/loss": 2.7019097805023193,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.0669836699962616,
      "grad_norm": 0.4626806378364563,
      "learning_rate": 3e-06,
      "loss": 0.6804,
      "step": 1717,
      "step_time": 7.127514831721783
    },
    {
      "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,
      "entropy": 1.6564275568181819,
      "epoch": 0.8227969348659003,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14357106387615204,
      "learning_rate": 3e-06,
      "loss": -0.0081,
      "step": 1718,
      "step_time": 7.853924199938774
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.125,
      "clip_ratio/region_mean": 0.3125,
      "entropy": 1.229736328125,
      "epoch": 0.8232758620689655,
      "fork_head/loss": 2.6823248863220215,
      "fork_head/num_positive": 11.0,
      "fork_head/pred_top1_mass": 0.0930696651339531,
      "grad_norm": 0.412383496761322,
      "learning_rate": 3e-06,
      "loss": 0.6741,
      "step": 1719,
      "step_time": 7.933532796800137
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.6677468039772727,
      "epoch": 0.8237547892720306,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11638367176055908,
      "learning_rate": 3e-06,
      "loss": -0.0084,
      "step": 1720,
      "step_time": 7.198164641857147
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 178.7109375,
      "completions/mean_terminated_length": 178.5596923828125,
      "completions/min_length": 59.0,
      "completions/min_terminated_length": 59.0,
      "entropy": 1.4511718658300548,
      "epoch": 0.8242337164750958,
      "fork/depth_mean": 0.40630343556404114,
      "fork/depth_std": 0.2939899265766144,
      "fork_head/loss": 2.7348220348358154,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.0590275377035141,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5743808150291443,
      "learning_rate": 3e-06,
      "loss": 0.6779,
      "num_tokens": 43832642.0,
      "reward": 0.9879850149154663,
      "reward_std": 0.5559214353561401,
      "rewards/code_format_reward/mean": 0.763671875,
      "rewards/code_format_reward/std": 0.3262917995452881,
      "rewards/code_reward/mean": 0.22431311011314392,
      "rewards/code_reward/std": 0.350944459438324,
      "step": 1721,
      "step_time": 367.1875958517194
    },
    {
      "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,
      "entropy": 1.3748520395972512,
      "epoch": 0.8247126436781609,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17254547774791718,
      "learning_rate": 3e-06,
      "loss": 0.0077,
      "step": 1722,
      "step_time": 7.522021263837814
    },
    {
      "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,
      "entropy": 1.4452123458568866,
      "epoch": 0.8251915708812261,
      "fork_head/loss": 2.7239716053009033,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.06358569115400314,
      "grad_norm": 0.5485330820083618,
      "learning_rate": 3e-06,
      "loss": 0.6748,
      "step": 1723,
      "step_time": 8.079354390501976
    },
    {
      "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,
      "entropy": 1.3863044543699785,
      "epoch": 0.8256704980842912,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16081349551677704,
      "learning_rate": 3e-06,
      "loss": 0.0065,
      "step": 1724,
      "step_time": 7.165771082043648
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.025641026405187752,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.025641026405187752,
      "entropy": 1.44936898121467,
      "epoch": 0.8261494252873564,
      "fork_head/loss": 2.706298589706421,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.06357390433549881,
      "grad_norm": 0.48894378542900085,
      "learning_rate": 3e-06,
      "loss": 0.6692,
      "step": 1725,
      "step_time": 7.551666423678398
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.3941169543699785,
      "epoch": 0.8266283524904214,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10924766212701797,
      "learning_rate": 3e-06,
      "loss": 0.0057,
      "step": 1726,
      "step_time": 7.779080502688885
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.025641026405187752,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.025641026405187752,
      "entropy": 1.4483172985223622,
      "epoch": 0.8271072796934866,
      "fork_head/loss": 2.684333324432373,
      "fork_head/num_positive": 11.125,
      "fork_head/pred_top1_mass": 0.0836537554860115,
      "grad_norm": 0.4704294204711914,
      "learning_rate": 3e-06,
      "loss": 0.6624,
      "step": 1727,
      "step_time": 7.2699834406375885
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.4057469151236794,
      "epoch": 0.8275862068965517,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11020220816135406,
      "learning_rate": 3e-06,
      "loss": 0.0051,
      "step": 1728,
      "step_time": 7.555858485400677
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 187.654296875,
      "completions/mean_terminated_length": 187.5205535888672,
      "completions/min_length": 61.0,
      "completions/min_terminated_length": 61.0,
      "entropy": 1.137939453125,
      "epoch": 0.8280651340996169,
      "fork/depth_mean": 0.3666301667690277,
      "fork/depth_std": 0.2923791706562042,
      "fork_head/loss": 2.7332944869995117,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.06117723882198334,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4896426498889923,
      "learning_rate": 3e-06,
      "loss": 0.6825,
      "num_tokens": 44019969.0,
      "reward": 0.9478482007980347,
      "reward_std": 0.5528872013092041,
      "rewards/code_format_reward/mean": 0.7578125,
      "rewards/code_format_reward/std": 0.32195815443992615,
      "rewards/code_reward/mean": 0.1900356560945511,
      "rewards/code_reward/std": 0.3323415219783783,
      "step": 1729,
      "step_time": 361.89285247027874
    },
    {
      "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,
      "entropy": 1.5416666666666667,
      "epoch": 0.828544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03438634052872658,
      "learning_rate": 3e-06,
      "loss": -0.0011,
      "step": 1730,
      "step_time": 8.178371414542198
    },
    {
      "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,
      "entropy": 1.1243489533662796,
      "epoch": 0.8290229885057471,
      "fork_head/loss": 2.7256014347076416,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.06269823014736176,
      "grad_norm": 0.4659962058067322,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "step": 1731,
      "step_time": 7.550453826785088
    },
    {
      "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,
      "entropy": 1.4947916666666667,
      "epoch": 0.8295019157088123,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03611206263303757,
      "learning_rate": 3e-06,
      "loss": -0.001,
      "step": 1732,
      "step_time": 7.605315461754799
    },
    {
      "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,
      "entropy": 1.1347249299287796,
      "epoch": 0.8299808429118773,
      "fork_head/loss": 2.7118887901306152,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.07689490169286728,
      "grad_norm": 0.42926546931266785,
      "learning_rate": 3e-06,
      "loss": 0.6773,
      "step": 1733,
      "step_time": 7.596202529966831
    },
    {
      "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,
      "entropy": 1.546875,
      "epoch": 0.8304597701149425,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03384893760085106,
      "learning_rate": 3e-06,
      "loss": -0.0012,
      "step": 1734,
      "step_time": 7.457046538591385
    },
    {
      "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,
      "entropy": 1.1338704377412796,
      "epoch": 0.8309386973180076,
      "fork_head/loss": 2.694791555404663,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.08371073752641678,
      "grad_norm": 0.39531785249710083,
      "learning_rate": 3e-06,
      "loss": 0.6727,
      "step": 1735,
      "step_time": 7.8681280836462975
    },
    {
      "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,
      "entropy": 1.4322916666666667,
      "epoch": 0.8314176245210728,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.029110310599207878,
      "learning_rate": 3e-06,
      "loss": -0.0007,
      "step": 1736,
      "step_time": 7.455036461353302
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 178.20703125,
      "completions/mean_terminated_length": 177.9019775390625,
      "completions/min_length": 6.0,
      "completions/min_terminated_length": 6.0,
      "entropy": 1.26396484375,
      "epoch": 0.8318965517241379,
      "fork/depth_mean": 0.39410123229026794,
      "fork/depth_std": 0.32622087001800537,
      "fork_head/loss": 2.722435474395752,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.07789134234189987,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5114188194274902,
      "learning_rate": 3e-06,
      "loss": 0.668,
      "num_tokens": 44204795.0,
      "reward": 1.0037897825241089,
      "reward_std": 0.530603289604187,
      "rewards/code_format_reward/mean": 0.794921875,
      "rewards/code_format_reward/std": 0.2933269441127777,
      "rewards/code_reward/mean": 0.20886793732643127,
      "rewards/code_reward/std": 0.33715689182281494,
      "step": 1737,
      "step_time": 389.1636358425021
    },
    {
      "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,
      "entropy": 1.419921875,
      "epoch": 0.8323754789272031,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07610714435577393,
      "learning_rate": 3e-06,
      "loss": 0.0115,
      "step": 1738,
      "step_time": 7.544346630573273
    },
    {
      "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,
      "entropy": 1.25048828125,
      "epoch": 0.8328544061302682,
      "fork_head/loss": 2.71351957321167,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.07808981835842133,
      "grad_norm": 0.4803083837032318,
      "learning_rate": 3e-06,
      "loss": 0.665,
      "step": 1739,
      "step_time": 7.777800336480141
    },
    {
      "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,
      "entropy": 1.4427083333333333,
      "epoch": 0.8333333333333334,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0721912682056427,
      "learning_rate": 3e-06,
      "loss": 0.011,
      "step": 1740,
      "step_time": 7.970320418477058
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.2486328125,
      "epoch": 0.8338122605363985,
      "fork_head/loss": 2.699148178100586,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.08296602964401245,
      "grad_norm": 0.42295801639556885,
      "learning_rate": 3e-06,
      "loss": 0.6599,
      "step": 1741,
      "step_time": 7.153035886585712
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.16666666666666666,
      "entropy": 1.4833984375,
      "epoch": 0.8342911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0314658023416996,
      "learning_rate": 3e-06,
      "loss": 0.0105,
      "step": 1742,
      "step_time": 8.007494956254959
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.23193359375,
      "epoch": 0.8347701149425287,
      "fork_head/loss": 2.6838135719299316,
      "fork_head/num_positive": 11.46875,
      "fork_head/pred_top1_mass": 0.08567246794700623,
      "grad_norm": 0.36267635226249695,
      "learning_rate": 3e-06,
      "loss": 0.6551,
      "step": 1743,
      "step_time": 7.428311087191105
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.3333333333333333,
      "clip_ratio/low_min": 0.3333333333333333,
      "clip_ratio/region_mean": 0.3333333333333333,
      "entropy": 1.5299479166666667,
      "epoch": 0.8352490421455939,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.028663136065006256,
      "learning_rate": 3e-06,
      "loss": 0.0104,
      "step": 1744,
      "step_time": 7.878395900130272
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 190.501953125,
      "completions/mean_terminated_length": 190.501953125,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.351086744895348,
      "epoch": 0.835727969348659,
      "fork/depth_mean": 0.35423383116722107,
      "fork/depth_std": 0.27723801136016846,
      "fork_head/loss": 2.7145373821258545,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.07227487862110138,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.43134334683418274,
      "learning_rate": 3e-06,
      "loss": 0.674,
      "num_tokens": 44386828.0,
      "reward": 1.0651499032974243,
      "reward_std": 0.5790050625801086,
      "rewards/code_format_reward/mean": 0.80078125,
      "rewards/code_format_reward/std": 0.2923676371574402,
      "rewards/code_reward/mean": 0.2643686532974243,
      "rewards/code_reward/std": 0.3828522264957428,
      "step": 1745,
      "step_time": 364.42691527307034
    },
    {
      "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,
      "entropy": 1.22490234375,
      "epoch": 0.8362068965517241,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07654975354671478,
      "learning_rate": 3e-06,
      "loss": 0.0072,
      "step": 1746,
      "step_time": 7.6312079429626465
    },
    {
      "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,
      "entropy": 1.355644024335421,
      "epoch": 0.8366858237547893,
      "fork_head/loss": 2.7084689140319824,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.0728173702955246,
      "grad_norm": 0.40093255043029785,
      "learning_rate": 3e-06,
      "loss": 0.6717,
      "step": 1747,
      "step_time": 8.77837124466896
    },
    {
      "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,
      "entropy": 1.22255859375,
      "epoch": 0.8371647509578544,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10877234488725662,
      "learning_rate": 3e-06,
      "loss": 0.0068,
      "step": 1748,
      "step_time": 7.141087286174297
    },
    {
      "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,
      "entropy": 1.3503856108738825,
      "epoch": 0.8376436781609196,
      "fork_head/loss": 2.697819948196411,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.07801204174757004,
      "grad_norm": 0.3771112859249115,
      "learning_rate": 3e-06,
      "loss": 0.6684,
      "step": 1749,
      "step_time": 7.565739281475544
    },
    {
      "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,
      "entropy": 1.23193359375,
      "epoch": 0.8381226053639846,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07760516554117203,
      "learning_rate": 3e-06,
      "loss": 0.0068,
      "step": 1750,
      "step_time": 7.70088405162096
    },
    {
      "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,
      "entropy": 1.3536658653846154,
      "epoch": 0.8386015325670498,
      "fork_head/loss": 2.6848177909851074,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.08245369791984558,
      "grad_norm": 0.36130696535110474,
      "learning_rate": 3e-06,
      "loss": 0.6651,
      "step": 1751,
      "step_time": 7.467967741191387
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.05,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.2455078125,
      "epoch": 0.8390804597701149,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03427048400044441,
      "learning_rate": 3e-06,
      "loss": 0.0064,
      "step": 1752,
      "step_time": 8.116762526333332
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 175.806640625,
      "completions/mean_terminated_length": 175.806640625,
      "completions/min_length": 53.0,
      "completions/min_terminated_length": 53.0,
      "entropy": 1.659765625,
      "epoch": 0.8395593869731801,
      "fork/depth_mean": 0.41563791036605835,
      "fork/depth_std": 0.2977336049079895,
      "fork_head/loss": 2.721287727355957,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.07509365677833557,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6652722358703613,
      "learning_rate": 3e-06,
      "loss": 0.6801,
      "num_tokens": 44566921.0,
      "reward": 1.0375633239746094,
      "reward_std": 0.5541945099830627,
      "rewards/code_format_reward/mean": 0.8037109375,
      "rewards/code_format_reward/std": 0.2901614010334015,
      "rewards/code_reward/mean": 0.23385238647460938,
      "rewards/code_reward/std": 0.3692070543766022,
      "step": 1753,
      "step_time": 420.0595871284604
    },
    {
      "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,
      "entropy": 1.358203125,
      "epoch": 0.8400383141762452,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05048543959856033,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 1754,
      "step_time": 8.204821050167084
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.62236328125,
      "epoch": 0.8405172413793104,
      "fork_head/loss": 2.706986665725708,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.0788654312491417,
      "grad_norm": 0.5467221736907959,
      "learning_rate": 3e-06,
      "loss": 0.6761,
      "step": 1755,
      "step_time": 7.4287847727537155
    },
    {
      "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,
      "entropy": 1.36884765625,
      "epoch": 0.8409961685823755,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.049420066177845,
      "learning_rate": 3e-06,
      "loss": 0.0016,
      "step": 1756,
      "step_time": 7.869111493229866
    },
    {
      "clip_ratio/high_max": 0.2,
      "clip_ratio/high_mean": 0.2,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.2,
      "entropy": 1.57998046875,
      "epoch": 0.8414750957854407,
      "fork_head/loss": 2.686915397644043,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.10143189877271652,
      "grad_norm": 0.47680625319480896,
      "learning_rate": 3e-06,
      "loss": 0.6706,
      "step": 1757,
      "step_time": 7.4680114313960075
    },
    {
      "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,
      "entropy": 1.3654296875,
      "epoch": 0.8419540229885057,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04774697124958038,
      "learning_rate": 3e-06,
      "loss": 0.0012,
      "step": 1758,
      "step_time": 7.433919459581375
    },
    {
      "clip_ratio/high_max": 0.2,
      "clip_ratio/high_mean": 0.2,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.2,
      "entropy": 1.61044921875,
      "epoch": 0.8424329501915708,
      "fork_head/loss": 2.665290117263794,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.13921129703521729,
      "grad_norm": 0.4091050326824188,
      "learning_rate": 3e-06,
      "loss": 0.6649,
      "step": 1759,
      "step_time": 7.387947201728821
    },
    {
      "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,
      "entropy": 1.3529296875,
      "epoch": 0.842911877394636,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04850362241268158,
      "learning_rate": 3e-06,
      "loss": 0.0012,
      "step": 1760,
      "step_time": 7.2350097224116325
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 171.044921875,
      "completions/mean_terminated_length": 170.71177673339844,
      "completions/min_length": 58.0,
      "completions/min_terminated_length": 58.0,
      "entropy": 1.3850911458333333,
      "epoch": 0.8433908045977011,
      "fork/depth_mean": 0.5147525668144226,
      "fork/depth_std": 0.3145754039287567,
      "fork_head/loss": 2.718479871749878,
      "fork_head/num_positive": 10.46875,
      "fork_head/pred_top1_mass": 0.07089004665613174,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5468909740447998,
      "learning_rate": 3e-06,
      "loss": 0.6793,
      "num_tokens": 44748000.0,
      "reward": 0.9588813185691833,
      "reward_std": 0.5542901158332825,
      "rewards/code_format_reward/mean": 0.7568359375,
      "rewards/code_format_reward/std": 0.3127307891845703,
      "rewards/code_reward/mean": 0.20204538106918335,
      "rewards/code_reward/std": 0.33509549498558044,
      "step": 1761,
      "step_time": 371.09616968780756
    },
    {
      "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,
      "entropy": 1.5888671875,
      "epoch": 0.8438697318007663,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07511966675519943,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 1762,
      "step_time": 6.950195550918579
    },
    {
      "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,
      "entropy": 1.3470052083333333,
      "epoch": 0.8443486590038314,
      "fork_head/loss": 2.7100322246551514,
      "fork_head/num_positive": 10.46875,
      "fork_head/pred_top1_mass": 0.08128438889980316,
      "grad_norm": 0.5385843515396118,
      "learning_rate": 3e-06,
      "loss": 0.6769,
      "step": 1763,
      "step_time": 8.332321047782898
    },
    {
      "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,
      "entropy": 1.5712890625,
      "epoch": 0.8448275862068966,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07544727623462677,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 1764,
      "step_time": 7.550758481025696
    },
    {
      "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,
      "entropy": 1.3424479166666667,
      "epoch": 0.8453065134099617,
      "fork_head/loss": 2.693164110183716,
      "fork_head/num_positive": 10.46875,
      "fork_head/pred_top1_mass": 0.09719545394182205,
      "grad_norm": 0.505386471748352,
      "learning_rate": 3e-06,
      "loss": 0.6724,
      "step": 1765,
      "step_time": 7.1922077015042305
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.5966796875,
      "epoch": 0.8457854406130269,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.040074728429317474,
      "learning_rate": 3e-06,
      "loss": 0.0001,
      "step": 1766,
      "step_time": 7.826809145510197
    },
    {
      "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,
      "entropy": 1.3606770833333333,
      "epoch": 0.8462643678160919,
      "fork_head/loss": 2.6707682609558105,
      "fork_head/num_positive": 10.46875,
      "fork_head/pred_top1_mass": 0.09909690171480179,
      "grad_norm": 0.4558461010456085,
      "learning_rate": 3e-06,
      "loss": 0.6668,
      "step": 1767,
      "step_time": 7.51395720988512
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.6201171875,
      "epoch": 0.8467432950191571,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04736287146806717,
      "learning_rate": 3e-06,
      "loss": -0.0003,
      "step": 1768,
      "step_time": 7.313026808202267
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 174.865234375,
      "completions/mean_terminated_length": 174.70645141601562,
      "completions/min_length": 37.0,
      "completions/min_terminated_length": 37.0,
      "entropy": 1.7494140625,
      "epoch": 0.8472222222222222,
      "fork/depth_mean": 0.4610847532749176,
      "fork/depth_std": 0.2988463342189789,
      "fork_head/loss": 2.735118865966797,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.05819857865571976,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.515457808971405,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "num_tokens": 44930139.0,
      "reward": 0.9311182498931885,
      "reward_std": 0.5102359056472778,
      "rewards/code_format_reward/mean": 0.7734375,
      "rewards/code_format_reward/std": 0.30558645725250244,
      "rewards/code_reward/mean": 0.15768073499202728,
      "rewards/code_reward/std": 0.31294918060302734,
      "step": 1769,
      "step_time": 373.03447703272104
    },
    {
      "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,
      "entropy": 1.45546875,
      "epoch": 0.8477011494252874,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07082739472389221,
      "learning_rate": 3e-06,
      "loss": 0.0064,
      "step": 1770,
      "step_time": 7.51860161870718
    },
    {
      "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,
      "entropy": 1.75859375,
      "epoch": 0.8481800766283525,
      "fork_head/loss": 2.7260372638702393,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.05694909766316414,
      "grad_norm": 0.4867427349090576,
      "learning_rate": 3e-06,
      "loss": 0.6762,
      "step": 1771,
      "step_time": 8.466601729393005
    },
    {
      "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,
      "entropy": 1.46640625,
      "epoch": 0.8486590038314177,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12783364951610565,
      "learning_rate": 3e-06,
      "loss": 0.0062,
      "step": 1772,
      "step_time": 6.811240404844284
    },
    {
      "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,
      "entropy": 1.737890625,
      "epoch": 0.8491379310344828,
      "fork_head/loss": 2.7106213569641113,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.06644035875797272,
      "grad_norm": 0.4363928735256195,
      "learning_rate": 3e-06,
      "loss": 0.672,
      "step": 1773,
      "step_time": 7.931365251541138
    },
    {
      "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,
      "entropy": 1.49375,
      "epoch": 0.8496168582375478,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06649966537952423,
      "learning_rate": 3e-06,
      "loss": 0.0058,
      "step": 1774,
      "step_time": 7.757053196430206
    },
    {
      "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,
      "entropy": 1.675,
      "epoch": 0.850095785440613,
      "fork_head/loss": 2.692768096923828,
      "fork_head/num_positive": 11.59375,
      "fork_head/pred_top1_mass": 0.0728229209780693,
      "grad_norm": 0.4008762538433075,
      "learning_rate": 3e-06,
      "loss": 0.667,
      "step": 1775,
      "step_time": 7.266289547085762
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.2,
      "clip_ratio/low_min": 0.2,
      "clip_ratio/region_mean": 0.2,
      "entropy": 1.546875,
      "epoch": 0.8505747126436781,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.055284012109041214,
      "learning_rate": 3e-06,
      "loss": 0.0055,
      "step": 1776,
      "step_time": 7.777600109577179
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 179.931640625,
      "completions/mean_terminated_length": 179.931640625,
      "completions/min_length": 34.0,
      "completions/min_terminated_length": 34.0,
      "entropy": 1.83984375,
      "epoch": 0.8510536398467433,
      "fork/depth_mean": 0.3511047661304474,
      "fork/depth_std": 0.28072744607925415,
      "fork_head/loss": 2.7212209701538086,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.09032492339611053,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4650612771511078,
      "learning_rate": 3e-06,
      "loss": 0.6918,
      "num_tokens": 45122120.0,
      "reward": 0.9493803381919861,
      "reward_std": 0.6338837146759033,
      "rewards/code_format_reward/mean": 0.70703125,
      "rewards/code_format_reward/std": 0.35405033826828003,
      "rewards/code_reward/mean": 0.2423490583896637,
      "rewards/code_reward/std": 0.37151816487312317,
      "step": 1777,
      "step_time": 374.66581558436155
    },
    {
      "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,
      "entropy": 1.2963324652777777,
      "epoch": 0.8515325670498084,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0857936218380928,
      "learning_rate": 3e-06,
      "loss": -0.0098,
      "step": 1778,
      "step_time": 8.221673287451267
    },
    {
      "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,
      "entropy": 1.8423122862974803,
      "epoch": 0.8520114942528736,
      "fork_head/loss": 2.7141542434692383,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.09232120215892792,
      "grad_norm": 0.4617348909378052,
      "learning_rate": 3e-06,
      "loss": 0.6885,
      "step": 1779,
      "step_time": 7.630509734153748
    },
    {
      "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,
      "entropy": 1.2833116319444444,
      "epoch": 0.8524904214559387,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08602269738912582,
      "learning_rate": 3e-06,
      "loss": -0.0098,
      "step": 1780,
      "step_time": 7.6538093239068985
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.8409288227558136,
      "epoch": 0.8529693486590039,
      "fork_head/loss": 2.701080799102783,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.0962679460644722,
      "grad_norm": 0.4122597277164459,
      "learning_rate": 3e-06,
      "loss": 0.6838,
      "step": 1781,
      "step_time": 7.5987429320812225
    },
    {
      "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,
      "entropy": 1.2961154513888888,
      "epoch": 0.853448275862069,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08957609534263611,
      "learning_rate": 3e-06,
      "loss": -0.0099,
      "step": 1782,
      "step_time": 7.384443439543247
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333333333333,
      "entropy": 1.8370768229166667,
      "epoch": 0.8539272030651341,
      "fork_head/loss": 2.6858103275299072,
      "fork_head/num_positive": 11.96875,
      "fork_head/pred_top1_mass": 0.10199446231126785,
      "grad_norm": 0.3645326495170593,
      "learning_rate": 3e-06,
      "loss": 0.6785,
      "step": 1783,
      "step_time": 8.10536578297615
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05555555555555555,
      "clip_ratio/low_min": 0.05555555555555555,
      "clip_ratio/region_mean": 0.05555555555555555,
      "entropy": 1.2898220486111112,
      "epoch": 0.8544061302681992,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07331640273332596,
      "learning_rate": 3e-06,
      "loss": -0.0106,
      "step": 1784,
      "step_time": 7.209994427859783
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 161.677734375,
      "completions/mean_terminated_length": 161.677734375,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.58203125,
      "epoch": 0.8548850574712644,
      "fork/depth_mean": 0.3954583406448364,
      "fork/depth_std": 0.29577234387397766,
      "fork_head/loss": 2.7215723991394043,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.06731783598661423,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6186858415603638,
      "learning_rate": 3e-06,
      "loss": 0.6783,
      "num_tokens": 45297347.0,
      "reward": 0.9566037654876709,
      "reward_std": 0.6194646954536438,
      "rewards/code_format_reward/mean": 0.7216796875,
      "rewards/code_format_reward/std": 0.35826653242111206,
      "rewards/code_reward/mean": 0.23492401838302612,
      "rewards/code_reward/std": 0.3831101655960083,
      "step": 1785,
      "step_time": 337.66560039669275
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.16666666666666666,
      "entropy": 1.6180013020833333,
      "epoch": 0.8553639846743295,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.24476730823516846,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 1786,
      "step_time": 6.845875956118107
    },
    {
      "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,
      "entropy": 1.592529296875,
      "epoch": 0.8558429118773946,
      "fork_head/loss": 2.7071681022644043,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.06611152738332748,
      "grad_norm": 0.5645313262939453,
      "learning_rate": 3e-06,
      "loss": 0.6745,
      "step": 1787,
      "step_time": 8.155555635690689
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.16666666666666666,
      "entropy": 1.6277669270833333,
      "epoch": 0.8563218390804598,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.27580180764198303,
      "learning_rate": 3e-06,
      "loss": 0.0015,
      "step": 1788,
      "step_time": 7.731114387512207
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.5234375,
      "epoch": 0.8568007662835249,
      "fork_head/loss": 2.685920238494873,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.09978516399860382,
      "grad_norm": 0.5086996555328369,
      "learning_rate": 3e-06,
      "loss": 0.6691,
      "step": 1789,
      "step_time": 6.929079480469227
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.08333333333333333,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.5847981770833333,
      "epoch": 0.85727969348659,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.04737140238285065,
      "learning_rate": 3e-06,
      "loss": 0.0012,
      "step": 1790,
      "step_time": 7.6722414046525955
    },
    {
      "clip_ratio/high_max": 0.375,
      "clip_ratio/high_mean": 0.375,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.375,
      "entropy": 1.5205078125,
      "epoch": 0.8577586206896551,
      "fork_head/loss": 2.6605911254882812,
      "fork_head/num_positive": 11.40625,
      "fork_head/pred_top1_mass": 0.10080157220363617,
      "grad_norm": 0.47319766879081726,
      "learning_rate": 3e-06,
      "loss": 0.6628,
      "step": 1791,
      "step_time": 7.743199810385704
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.16666666666666666,
      "clip_ratio/low_mean": 0.16666666666666666,
      "clip_ratio/low_min": 0.16666666666666666,
      "clip_ratio/region_mean": 0.3333333333333333,
      "entropy": 1.61572265625,
      "epoch": 0.8582375478927203,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.03178822994232178,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 1792,
      "step_time": 7.39870660007
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 177.73046875,
      "completions/mean_terminated_length": 177.73046875,
      "completions/min_length": 56.0,
      "completions/min_terminated_length": 56.0,
      "entropy": 1.2828125,
      "epoch": 0.8587164750957854,
      "fork/depth_mean": 0.38777175545692444,
      "fork/depth_std": 0.2996022701263428,
      "fork_head/loss": 2.7200279235839844,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.08343689888715744,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5368065237998962,
      "learning_rate": 3e-06,
      "loss": 0.6808,
      "num_tokens": 45485897.0,
      "reward": 0.8965902328491211,
      "reward_std": 0.5929865837097168,
      "rewards/code_format_reward/mean": 0.697265625,
      "rewards/code_format_reward/std": 0.3554886281490326,
      "rewards/code_reward/mean": 0.1993246078491211,
      "rewards/code_reward/std": 0.35169869661331177,
      "step": 1793,
      "step_time": 401.5219871997833
    },
    {
      "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,
      "entropy": 1.507421875,
      "epoch": 0.8591954022988506,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09637828171253204,
      "learning_rate": 3e-06,
      "loss": -0.001,
      "step": 1794,
      "step_time": 7.913275457918644
    },
    {
      "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,
      "entropy": 1.333984375,
      "epoch": 0.8596743295019157,
      "fork_head/loss": 2.7098727226257324,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.08903679996728897,
      "grad_norm": 0.50602787733078,
      "learning_rate": 3e-06,
      "loss": 0.6779,
      "step": 1795,
      "step_time": 7.859169542789459
    },
    {
      "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,
      "entropy": 1.511328125,
      "epoch": 0.8601532567049809,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07520931214094162,
      "learning_rate": 3e-06,
      "loss": -0.0013,
      "step": 1796,
      "step_time": 6.929112412035465
    },
    {
      "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,
      "entropy": 1.31796875,
      "epoch": 0.860632183908046,
      "fork_head/loss": 2.6923999786376953,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.08708610385656357,
      "grad_norm": 0.44842639565467834,
      "learning_rate": 3e-06,
      "loss": 0.6736,
      "step": 1797,
      "step_time": 8.057665050029755
    },
    {
      "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,
      "entropy": 1.509375,
      "epoch": 0.8611111111111112,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07752242684364319,
      "learning_rate": 3e-06,
      "loss": -0.0016,
      "step": 1798,
      "step_time": 7.348737969994545
    },
    {
      "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,
      "entropy": 1.369140625,
      "epoch": 0.8615900383141762,
      "fork_head/loss": 2.6728765964508057,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.10259447991847992,
      "grad_norm": 0.3936537802219391,
      "learning_rate": 3e-06,
      "loss": 0.6683,
      "step": 1799,
      "step_time": 7.067150957882404
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.4,
      "clip_ratio/low_min": 0.4,
      "clip_ratio/region_mean": 0.4,
      "entropy": 1.55546875,
      "epoch": 0.8620689655172413,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.033074360340833664,
      "learning_rate": 3e-06,
      "loss": -0.002,
      "step": 1800,
      "step_time": 8.246410928666592
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 174.515625,
      "completions/mean_terminated_length": 174.515625,
      "completions/min_length": 12.0,
      "completions/min_terminated_length": 12.0,
      "entropy": 1.4052734375,
      "epoch": 0.8625478927203065,
      "fork/depth_mean": 0.3950399160385132,
      "fork/depth_std": 0.2851546108722687,
      "fork_head/loss": 2.722015619277954,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.058658014982938766,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.4727039039134979,
      "learning_rate": 3e-06,
      "loss": 0.6785,
      "num_tokens": 45663489.0,
      "reward": 0.9558480978012085,
      "reward_std": 0.5891363024711609,
      "rewards/code_format_reward/mean": 0.7373046875,
      "rewards/code_format_reward/std": 0.348093718290329,
      "rewards/code_reward/mean": 0.2185433954000473,
      "rewards/code_reward/std": 0.3557001054286957,
      "step": 1801,
      "step_time": 425.62458776682615
    },
    {
      "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,
      "entropy": 1.4612165178571428,
      "epoch": 0.8630268199233716,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06505949795246124,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 1802,
      "step_time": 7.888839848339558
    },
    {
      "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,
      "entropy": 1.39697265625,
      "epoch": 0.8635057471264368,
      "fork_head/loss": 2.712008476257324,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.05946527048945427,
      "grad_norm": 0.406217485666275,
      "learning_rate": 3e-06,
      "loss": 0.6762,
      "step": 1803,
      "step_time": 7.470268540084362
    },
    {
      "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,
      "entropy": 1.4313616071428572,
      "epoch": 0.8639846743295019,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06882815062999725,
      "learning_rate": 3e-06,
      "loss": 0.0021,
      "step": 1804,
      "step_time": 7.527650259435177
    },
    {
      "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,
      "entropy": 1.38720703125,
      "epoch": 0.8644636015325671,
      "fork_head/loss": 2.699416160583496,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.06203332543373108,
      "grad_norm": 0.37276145815849304,
      "learning_rate": 3e-06,
      "loss": 0.6728,
      "step": 1805,
      "step_time": 7.852101676166058
    },
    {
      "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,
      "entropy": 1.4478236607142858,
      "epoch": 0.8649425287356322,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07014673203229904,
      "learning_rate": 3e-06,
      "loss": 0.002,
      "step": 1806,
      "step_time": 7.388694658875465
    },
    {
      "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,
      "entropy": 1.392578125,
      "epoch": 0.8654214559386973,
      "fork_head/loss": 2.685873031616211,
      "fork_head/num_positive": 11.5,
      "fork_head/pred_top1_mass": 0.07736188173294067,
      "grad_norm": 0.3703729212284088,
      "learning_rate": 3e-06,
      "loss": 0.6692,
      "step": 1807,
      "step_time": 7.770945712924004
    },
    {
      "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,
      "entropy": 1.4313616071428572,
      "epoch": 0.8659003831417624,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07244925945997238,
      "learning_rate": 3e-06,
      "loss": 0.0016,
      "step": 1808,
      "step_time": 7.645722642540932
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 164.001953125,
      "completions/mean_terminated_length": 164.001953125,
      "completions/min_length": 35.0,
      "completions/min_terminated_length": 35.0,
      "entropy": 1.53369140625,
      "epoch": 0.8663793103448276,
      "fork/depth_mean": 0.38527214527130127,
      "fork/depth_std": 0.27433741092681885,
      "fork_head/loss": 2.732236623764038,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.06770980358123779,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6002530455589294,
      "learning_rate": 3e-06,
      "loss": 0.6863,
      "num_tokens": 45842370.0,
      "reward": 1.0286142826080322,
      "reward_std": 0.5837862491607666,
      "rewards/code_format_reward/mean": 0.78125,
      "rewards/code_format_reward/std": 0.29510024189949036,
      "rewards/code_reward/mean": 0.24736428260803223,
      "rewards/code_reward/std": 0.3811674118041992,
      "step": 1809,
      "step_time": 331.21103997528553
    },
    {
      "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,
      "entropy": 1.1515764508928572,
      "epoch": 0.8668582375478927,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07008857280015945,
      "learning_rate": 3e-06,
      "loss": -0.0033,
      "step": 1810,
      "step_time": 6.9592699855566025
    },
    {
      "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,
      "entropy": 1.532470703125,
      "epoch": 0.8673371647509579,
      "fork_head/loss": 2.721816301345825,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.07923897355794907,
      "grad_norm": 0.5036424994468689,
      "learning_rate": 3e-06,
      "loss": 0.6828,
      "step": 1811,
      "step_time": 7.9908547922968864
    },
    {
      "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,
      "entropy": 1.1249302455357142,
      "epoch": 0.867816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06834066659212112,
      "learning_rate": 3e-06,
      "loss": -0.0035,
      "step": 1812,
      "step_time": 7.786233797669411
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.5255126953125,
      "epoch": 0.8682950191570882,
      "fork_head/loss": 2.7073681354522705,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.08952999114990234,
      "grad_norm": 0.40941688418388367,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "step": 1813,
      "step_time": 7.284088499844074
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.1396484375,
      "epoch": 0.8687739463601533,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0760452151298523,
      "learning_rate": 3e-06,
      "loss": -0.0038,
      "step": 1814,
      "step_time": 8.0750987008214
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.5206298828125,
      "epoch": 0.8692528735632183,
      "fork_head/loss": 2.6920480728149414,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.09742949903011322,
      "grad_norm": 0.3861252963542938,
      "learning_rate": 3e-06,
      "loss": 0.6739,
      "step": 1815,
      "step_time": 7.510050386190414
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.1288364955357142,
      "epoch": 0.8697318007662835,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06926261633634567,
      "learning_rate": 3e-06,
      "loss": -0.0042,
      "step": 1816,
      "step_time": 7.611213773488998
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 191.943359375,
      "completions/mean_terminated_length": 191.943359375,
      "completions/min_length": 35.0,
      "completions/min_terminated_length": 35.0,
      "entropy": 1.6382125298182169,
      "epoch": 0.8702107279693486,
      "fork/depth_mean": 0.3535064160823822,
      "fork/depth_std": 0.2755790948867798,
      "fork_head/loss": 2.748997688293457,
      "fork_head/num_positive": 10.71875,
      "fork_head/pred_top1_mass": 0.0613025464117527,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7249987721443176,
      "learning_rate": 3e-06,
      "loss": 0.6859,
      "num_tokens": 46032405.0,
      "reward": 1.1188186407089233,
      "reward_std": 0.5639982223510742,
      "rewards/code_format_reward/mean": 0.83984375,
      "rewards/code_format_reward/std": 0.24972465634346008,
      "rewards/code_reward/mean": 0.27897489070892334,
      "rewards/code_reward/std": 0.4100664556026459,
      "step": 1817,
      "step_time": 428.274243183434
    },
    {
      "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,
      "entropy": 1.4017144123713174,
      "epoch": 0.8706896551724138,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14068542420864105,
      "learning_rate": 3e-06,
      "loss": 0.0004,
      "step": 1818,
      "step_time": 8.311047233641148
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.6588216145833334,
      "epoch": 0.8711685823754789,
      "fork_head/loss": 2.7321057319641113,
      "fork_head/num_positive": 10.71875,
      "fork_head/pred_top1_mass": 0.06470954418182373,
      "grad_norm": 0.6934947967529297,
      "learning_rate": 3e-06,
      "loss": 0.6789,
      "step": 1819,
      "step_time": 8.470478817820549
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.4257486899693808,
      "epoch": 0.8716475095785441,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.18249328434467316,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 1820,
      "step_time": 7.455411218106747
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.016666666666666666,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.016666666666666666,
      "entropy": 1.630595342318217,
      "epoch": 0.8721264367816092,
      "fork_head/loss": 2.7047064304351807,
      "fork_head/num_positive": 10.71875,
      "fork_head/pred_top1_mass": 0.08271777629852295,
      "grad_norm": 0.6542117595672607,
      "learning_rate": 3e-06,
      "loss": 0.6704,
      "step": 1821,
      "step_time": 8.45177212357521
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.06666666666666667,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06666666666666667,
      "entropy": 1.4196831623713175,
      "epoch": 0.8726053639846744,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1171095222234726,
      "learning_rate": 3e-06,
      "loss": -0.0019,
      "step": 1822,
      "step_time": 7.5368767231702805
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1277777781089147,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.1277777781089147,
      "entropy": 1.6502459605534872,
      "epoch": 0.8730842911877394,
      "fork_head/loss": 2.6720407009124756,
      "fork_head/num_positive": 10.71875,
      "fork_head/pred_top1_mass": 0.11948247998952866,
      "grad_norm": 0.5851904153823853,
      "learning_rate": 3e-06,
      "loss": 0.6597,
      "step": 1823,
      "step_time": 7.819558002054691
    },
    {
      "clip_ratio/high_max": 0.2,
      "clip_ratio/high_mean": 0.07222222288449605,
      "clip_ratio/low_mean": 0.07777777910232545,
      "clip_ratio/low_min": 0.022222222884496053,
      "clip_ratio/region_mean": 0.15000000198682148,
      "entropy": 1.4276909748713176,
      "epoch": 0.8735632183908046,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16423694789409637,
      "learning_rate": 3e-06,
      "loss": -0.003,
      "step": 1824,
      "step_time": 8.183344319462776
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 173.962890625,
      "completions/mean_terminated_length": 173.962890625,
      "completions/min_length": 16.0,
      "completions/min_terminated_length": 16.0,
      "entropy": 1.7491122159090908,
      "epoch": 0.8740421455938697,
      "fork/depth_mean": 0.30616435408592224,
      "fork/depth_std": 0.2605513632297516,
      "fork_head/loss": 2.743685483932495,
      "fork_head/num_positive": 11.45161247253418,
      "fork_head/pred_top1_mass": 0.06605062633752823,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5554107427597046,
      "learning_rate": 3e-06,
      "loss": 0.6901,
      "num_tokens": 46210482.0,
      "reward": 1.0240557193756104,
      "reward_std": 0.5582324862480164,
      "rewards/code_format_reward/mean": 0.802734375,
      "rewards/code_format_reward/std": 0.3146027624607086,
      "rewards/code_reward/mean": 0.22132126986980438,
      "rewards/code_reward/std": 0.3551701009273529,
      "step": 1825,
      "step_time": 399.0111090838909
    },
    {
      "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,
      "entropy": 1.67646484375,
      "epoch": 0.8745210727969349,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19063767790794373,
      "learning_rate": 3e-06,
      "loss": -0.0038,
      "step": 1826,
      "step_time": 8.314379759132862
    },
    {
      "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,
      "entropy": 1.7732599431818181,
      "epoch": 0.875,
      "fork_head/loss": 2.7334229946136475,
      "fork_head/num_positive": 11.45161247253418,
      "fork_head/pred_top1_mass": 0.06623461097478867,
      "grad_norm": 0.5321997404098511,
      "learning_rate": 3e-06,
      "loss": 0.6867,
      "step": 1827,
      "step_time": 7.514283001422882
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.68837890625,
      "epoch": 0.8754789272030651,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12345686554908752,
      "learning_rate": 3e-06,
      "loss": -0.0052,
      "step": 1828,
      "step_time": 7.532141551375389
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.09090909090909091,
      "entropy": 1.7771661931818181,
      "epoch": 0.8759578544061303,
      "fork_head/loss": 2.7161786556243896,
      "fork_head/num_positive": 11.45161247253418,
      "fork_head/pred_top1_mass": 0.06831970810890198,
      "grad_norm": 0.5133740305900574,
      "learning_rate": 3e-06,
      "loss": 0.6819,
      "step": 1829,
      "step_time": 7.419032528996468
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.689453125,
      "epoch": 0.8764367816091954,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12146399170160294,
      "learning_rate": 3e-06,
      "loss": -0.0057,
      "step": 1830,
      "step_time": 7.576378367841244
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.8004261363636365,
      "epoch": 0.8769157088122606,
      "fork_head/loss": 2.6940009593963623,
      "fork_head/num_positive": 11.45161247253418,
      "fork_head/pred_top1_mass": 0.08094344288110733,
      "grad_norm": 0.4935895800590515,
      "learning_rate": 3e-06,
      "loss": 0.6755,
      "step": 1831,
      "step_time": 7.517265304923058
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.1,
      "clip_ratio/low_min": 0.1,
      "clip_ratio/region_mean": 0.15,
      "entropy": 1.682421875,
      "epoch": 0.8773946360153256,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1066981703042984,
      "learning_rate": 3e-06,
      "loss": -0.0064,
      "step": 1832,
      "step_time": 7.403790272772312
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 183.69921875,
      "completions/mean_terminated_length": 183.69921875,
      "completions/min_length": 32.0,
      "completions/min_terminated_length": 32.0,
      "entropy": 1.234375007947286,
      "epoch": 0.8778735632183908,
      "fork/depth_mean": 0.30047017335891724,
      "fork/depth_std": 0.2635124623775482,
      "fork_head/loss": 2.7317001819610596,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.0765690729022026,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6126608848571777,
      "learning_rate": 3e-06,
      "loss": 0.6824,
      "num_tokens": 46394776.0,
      "reward": 1.0422207117080688,
      "reward_std": 0.5838432908058167,
      "rewards/code_format_reward/mean": 0.791015625,
      "rewards/code_format_reward/std": 0.3416280746459961,
      "rewards/code_reward/mean": 0.25120508670806885,
      "rewards/code_reward/std": 0.3658998906612396,
      "step": 1833,
      "step_time": 474.44532623142004
    },
    {
      "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,
      "entropy": 1.4390345896993364,
      "epoch": 0.8783524904214559,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19237647950649261,
      "learning_rate": 3e-06,
      "loss": 0.0005,
      "step": 1834,
      "step_time": 7.663116283714771
    },
    {
      "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,
      "entropy": 1.2373263955116272,
      "epoch": 0.8788314176245211,
      "fork_head/loss": 2.722970485687256,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.08091438561677933,
      "grad_norm": 0.5827015042304993,
      "learning_rate": 3e-06,
      "loss": 0.679,
      "step": 1835,
      "step_time": 8.069175399839878
    },
    {
      "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,
      "entropy": 1.4062732458114624,
      "epoch": 0.8793103448275862,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19795387983322144,
      "learning_rate": 3e-06,
      "loss": -0.0004,
      "step": 1836,
      "step_time": 7.579083979129791
    },
    {
      "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,
      "entropy": 1.2496093789736429,
      "epoch": 0.8797892720306514,
      "fork_head/loss": 2.7082786560058594,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.0843655988574028,
      "grad_norm": 0.5217430591583252,
      "learning_rate": 3e-06,
      "loss": 0.6719,
      "step": 1837,
      "step_time": 7.672431014478207
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.4272809624671936,
      "epoch": 0.8802681992337165,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15271103382110596,
      "learning_rate": 3e-06,
      "loss": -0.0018,
      "step": 1838,
      "step_time": 7.961123801767826
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.07777777910232545,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.10000000198682149,
      "entropy": 1.2484157959620157,
      "epoch": 0.8807471264367817,
      "fork_head/loss": 2.688565492630005,
      "fork_head/num_positive": 11.34375,
      "fork_head/pred_top1_mass": 0.106740802526474,
      "grad_norm": 0.4550882577896118,
      "learning_rate": 3e-06,
      "loss": 0.6647,
      "step": 1839,
      "step_time": 7.583956748247147
    },
    {
      "clip_ratio/high_max": 0.17857142857142858,
      "clip_ratio/high_mean": 0.09523809594767434,
      "clip_ratio/low_mean": 0.011904762259551458,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.10714285714285714,
      "entropy": 1.439255484512874,
      "epoch": 0.8812260536398467,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0966431051492691,
      "learning_rate": 3e-06,
      "loss": -0.0032,
      "step": 1840,
      "step_time": 7.670131206512451
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 158.236328125,
      "completions/mean_terminated_length": 158.236328125,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.641380671943937,
      "epoch": 0.8817049808429118,
      "fork/depth_mean": 0.2644617259502411,
      "fork/depth_std": 0.22686246037483215,
      "fork_head/loss": 2.7260706424713135,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.06646822392940521,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5733960866928101,
      "learning_rate": 3e-06,
      "loss": 0.6754,
      "num_tokens": 46572113.0,
      "reward": 1.0415091514587402,
      "reward_std": 0.6405970454216003,
      "rewards/code_format_reward/mean": 0.7607421875,
      "rewards/code_format_reward/std": 0.35786089301109314,
      "rewards/code_reward/mean": 0.280767023563385,
      "rewards/code_reward/std": 0.3893580436706543,
      "step": 1841,
      "step_time": 420.4660912454128
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.5306206583976745,
      "epoch": 0.882183908045977,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.23023511469364166,
      "learning_rate": 3e-06,
      "loss": 0.0032,
      "step": 1842,
      "step_time": 8.605161182582378
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.05357142857142857,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08928571428571429,
      "entropy": 1.627906437431063,
      "epoch": 0.8826628352490421,
      "fork_head/loss": 2.7160582542419434,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.06819144636392593,
      "grad_norm": 0.622704267501831,
      "learning_rate": 3e-06,
      "loss": 0.6728,
      "step": 1843,
      "step_time": 7.001322381198406
    },
    {
      "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,
      "entropy": 1.5484591960906982,
      "epoch": 0.8831417624521073,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.22026731073856354,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 1844,
      "step_time": 7.969638705253601
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.6354224809578486,
      "epoch": 0.8836206896551724,
      "fork_head/loss": 2.698211193084717,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.07127612084150314,
      "grad_norm": 0.5320491194725037,
      "learning_rate": 3e-06,
      "loss": 0.667,
      "step": 1845,
      "step_time": 7.864672914147377
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.11666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.11666666666666667,
      "entropy": 1.5679036418596903,
      "epoch": 0.8840996168582376,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15782766044139862,
      "learning_rate": 3e-06,
      "loss": 0.0007,
      "step": 1846,
      "step_time": 7.698912911117077
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07142857142857142,
      "entropy": 1.6263543963432312,
      "epoch": 0.8845785440613027,
      "fork_head/loss": 2.6775424480438232,
      "fork_head/num_positive": 11.875,
      "fork_head/pred_top1_mass": 0.09649039804935455,
      "grad_norm": 0.4450169503688812,
      "learning_rate": 3e-06,
      "loss": 0.6606,
      "step": 1847,
      "step_time": 7.835430175065994
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.11666666666666667,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.11666666666666667,
      "entropy": 1.574327258268992,
      "epoch": 0.8850574712643678,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14436982572078705,
      "learning_rate": 3e-06,
      "loss": -0.0008,
      "step": 1848,
      "step_time": 7.540453620254993
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 172.705078125,
      "completions/mean_terminated_length": 172.705078125,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.4439057878085546,
      "epoch": 0.8855363984674329,
      "fork/depth_mean": 0.3307799696922302,
      "fork/depth_std": 0.26446670293807983,
      "fork_head/loss": 2.7309682369232178,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.07738344371318817,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.7073980569839478,
      "learning_rate": 3e-06,
      "loss": 0.6924,
      "num_tokens": 46761802.0,
      "reward": 0.9832098484039307,
      "reward_std": 0.5754454135894775,
      "rewards/code_format_reward/mean": 0.7578125,
      "rewards/code_format_reward/std": 0.33970388770103455,
      "rewards/code_reward/mean": 0.22539734840393066,
      "rewards/code_reward/std": 0.3661334812641144,
      "step": 1849,
      "step_time": 408.59951639920473
    },
    {
      "clip_ratio/high_max": 0.1,
      "clip_ratio/high_mean": 0.05,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.2028971374034882,
      "epoch": 0.8860153256704981,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.25839048624038696,
      "learning_rate": 3e-06,
      "loss": -0.0105,
      "step": 1850,
      "step_time": 7.297603629529476
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.4658900669642858,
      "epoch": 0.8864942528735632,
      "fork_head/loss": 2.714817762374878,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.0803895890712738,
      "grad_norm": 0.672944962978363,
      "learning_rate": 3e-06,
      "loss": 0.6877,
      "step": 1851,
      "step_time": 7.790516525506973
    },
    {
      "clip_ratio/high_max": 0.2,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.1752278625965118,
      "epoch": 0.8869731800766284,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1776382327079773,
      "learning_rate": 3e-06,
      "loss": -0.0125,
      "step": 1852,
      "step_time": 8.246307522058487
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.4732724172728402,
      "epoch": 0.8874521072796935,
      "fork_head/loss": 2.6880130767822266,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.11193352192640305,
      "grad_norm": 0.6344205141067505,
      "learning_rate": 3e-06,
      "loss": 0.6805,
      "step": 1853,
      "step_time": 7.240733407437801
    },
    {
      "clip_ratio/high_max": 0.2,
      "clip_ratio/high_mean": 0.1,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1,
      "entropy": 1.183072918653488,
      "epoch": 0.8879310344827587,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1738264262676239,
      "learning_rate": 3e-06,
      "loss": -0.0126,
      "step": 1854,
      "step_time": 7.860413946211338
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.125,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.490769156387874,
      "epoch": 0.8884099616858238,
      "fork_head/loss": 2.658403158187866,
      "fork_head/num_positive": 10.45161247253418,
      "fork_head/pred_top1_mass": 0.08967424184083939,
      "grad_norm": 0.5272842645645142,
      "learning_rate": 3e-06,
      "loss": 0.6724,
      "step": 1855,
      "step_time": 7.4691591411828995
    },
    {
      "clip_ratio/high_max": 0.3,
      "clip_ratio/high_mean": 0.15,
      "clip_ratio/low_mean": 0.03333333432674408,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.1833333343267441,
      "entropy": 1.1806315064430237,
      "epoch": 0.8888888888888888,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0790441557765007,
      "learning_rate": 3e-06,
      "loss": -0.0135,
      "step": 1856,
      "step_time": 7.601013168692589
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 166.705078125,
      "completions/mean_terminated_length": 166.705078125,
      "completions/min_length": 16.0,
      "completions/min_terminated_length": 16.0,
      "entropy": 1.58984375,
      "epoch": 0.889367816091954,
      "fork/depth_mean": 0.2570313811302185,
      "fork/depth_std": 0.2281252145767212,
      "fork_head/loss": 2.731959581375122,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.0695209875702858,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5241669416427612,
      "learning_rate": 3e-06,
      "loss": 0.6784,
      "num_tokens": 46940467.0,
      "reward": 1.1013622283935547,
      "reward_std": 0.6391882300376892,
      "rewards/code_format_reward/mean": 0.7587890625,
      "rewards/code_format_reward/std": 0.3592785894870758,
      "rewards/code_reward/mean": 0.34257322549819946,
      "rewards/code_reward/std": 0.4093513786792755,
      "step": 1857,
      "step_time": 352.0367464646697
    },
    {
      "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,
      "entropy": 1.4994377385486255,
      "epoch": 0.8898467432950191,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10685953497886658,
      "learning_rate": 3e-06,
      "loss": 0.0035,
      "step": 1858,
      "step_time": 7.9229458048939705
    },
    {
      "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,
      "entropy": 1.583984375,
      "epoch": 0.8903256704980843,
      "fork_head/loss": 2.7257394790649414,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.06848738342523575,
      "grad_norm": 0.5263566970825195,
      "learning_rate": 3e-06,
      "loss": 0.6757,
      "step": 1859,
      "step_time": 8.068633459508419
    },
    {
      "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,
      "entropy": 1.52160274440592,
      "epoch": 0.8908045977011494,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09394814819097519,
      "learning_rate": 3e-06,
      "loss": 0.003,
      "step": 1860,
      "step_time": 7.326377309858799
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.55859375,
      "epoch": 0.8912835249042146,
      "fork_head/loss": 2.7137527465820312,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.07505198568105698,
      "grad_norm": 0.46978452801704407,
      "learning_rate": 3e-06,
      "loss": 0.6719,
      "step": 1861,
      "step_time": 7.780171498656273
    },
    {
      "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,
      "entropy": 1.5178740512241016,
      "epoch": 0.8917624521072797,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09752379357814789,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 1862,
      "step_time": 7.630619652569294
    },
    {
      "clip_ratio/high_max": 0.125,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.5234375,
      "epoch": 0.8922413793103449,
      "fork_head/loss": 2.6983375549316406,
      "fork_head/num_positive": 12.125,
      "fork_head/pred_top1_mass": 0.08026771992444992,
      "grad_norm": 0.4137095510959625,
      "learning_rate": 3e-06,
      "loss": 0.6681,
      "step": 1863,
      "step_time": 7.6350757628679276
    },
    {
      "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,
      "entropy": 1.5379083806818181,
      "epoch": 0.89272030651341,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09722406417131424,
      "learning_rate": 3e-06,
      "loss": 0.0017,
      "step": 1864,
      "step_time": 7.788514137268066
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 175.921875,
      "completions/mean_terminated_length": 175.921875,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.56164180690592,
      "epoch": 0.8931992337164751,
      "fork/depth_mean": 0.2862626314163208,
      "fork/depth_std": 0.26480618119239807,
      "fork_head/loss": 2.7373275756835938,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.06420324742794037,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6689591407775879,
      "learning_rate": 3e-06,
      "loss": 0.6809,
      "num_tokens": 47124251.0,
      "reward": 1.0133464336395264,
      "reward_std": 0.6342117786407471,
      "rewards/code_format_reward/mean": 0.7568359375,
      "rewards/code_format_reward/std": 0.3552129864692688,
      "rewards/code_reward/mean": 0.25651049613952637,
      "rewards/code_reward/std": 0.39144647121429443,
      "step": 1865,
      "step_time": 438.7898567467928
    },
    {
      "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,
      "entropy": 1.2223011309450322,
      "epoch": 0.8936781609195402,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12002206593751907,
      "learning_rate": 3e-06,
      "loss": 0.0041,
      "step": 1866,
      "step_time": 8.2990263402462
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.13636363636363635,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.5915897271849893,
      "epoch": 0.8941570881226054,
      "fork_head/loss": 2.7232141494750977,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.06481137871742249,
      "grad_norm": 0.6040025353431702,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "step": 1867,
      "step_time": 7.161419704556465
    },
    {
      "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,
      "entropy": 1.2188387729904868,
      "epoch": 0.8946360153256705,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12236090004444122,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 1868,
      "step_time": 8.024585284292698
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.13636363636363635,
      "entropy": 1.5981297330422835,
      "epoch": 0.8951149425287356,
      "fork_head/loss": 2.7011187076568604,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.06859560310840607,
      "grad_norm": 0.5192272663116455,
      "learning_rate": 3e-06,
      "loss": 0.6685,
      "step": 1869,
      "step_time": 7.163428597152233
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.01515151560306549,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.01515151560306549,
      "entropy": 1.218661216172305,
      "epoch": 0.8955938697318008,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.109833724796772,
      "learning_rate": 3e-06,
      "loss": 0.0027,
      "step": 1870,
      "step_time": 7.337175287306309
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.13636363636363635,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.18181818181818182,
      "entropy": 1.6206498633731494,
      "epoch": 0.8960727969348659,
      "fork_head/loss": 2.6767072677612305,
      "fork_head/num_positive": 11.25,
      "fork_head/pred_top1_mass": 0.07695873081684113,
      "grad_norm": 0.4654580354690552,
      "learning_rate": 3e-06,
      "loss": 0.6618,
      "step": 1871,
      "step_time": 8.337641566991806
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.045454545454545456,
      "clip_ratio/low_mean": 0.01515151560306549,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.060606061057610947,
      "entropy": 1.219667375087738,
      "epoch": 0.896551724137931,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10100460797548294,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 1872,
      "step_time": 6.888163506984711
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 172.564453125,
      "completions/mean_terminated_length": 172.564453125,
      "completions/min_length": 24.0,
      "completions/min_terminated_length": 24.0,
      "entropy": 1.417414367198944,
      "epoch": 0.8970306513409961,
      "fork/depth_mean": 0.24722501635551453,
      "fork/depth_std": 0.22301340103149414,
      "fork_head/loss": 2.7255730628967285,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.06921997666358948,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6561981439590454,
      "learning_rate": 3e-06,
      "loss": 0.6755,
      "num_tokens": 47302892.0,
      "reward": 0.9992384910583496,
      "reward_std": 0.634560227394104,
      "rewards/code_format_reward/mean": 0.73828125,
      "rewards/code_format_reward/std": 0.36190861463546753,
      "rewards/code_reward/mean": 0.2609572410583496,
      "rewards/code_reward/std": 0.3721035420894623,
      "step": 1873,
      "step_time": 425.70354261249304
    },
    {
      "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,
      "entropy": 1.5468866314206804,
      "epoch": 0.8975095785440613,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.3109261691570282,
      "learning_rate": 3e-06,
      "loss": 0.0027,
      "step": 1874,
      "step_time": 6.964692838490009
    },
    {
      "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,
      "entropy": 1.4062243483283303,
      "epoch": 0.8979885057471264,
      "fork_head/loss": 2.709256172180176,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.0622582733631134,
      "grad_norm": 0.5933750867843628,
      "learning_rate": 3e-06,
      "loss": 0.6706,
      "step": 1875,
      "step_time": 8.528663016855717
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.023809524519102915,
      "clip_ratio/low_mean": 0.09226190511669431,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.11607142963579722,
      "entropy": 1.5605817522321428,
      "epoch": 0.8984674329501916,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.23709021508693695,
      "learning_rate": 3e-06,
      "loss": 0.0008,
      "step": 1876,
      "step_time": 7.789073094725609
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.018181818452748386,
      "clip_ratio/low_mean": 0.022727272727272728,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.04090909118002111,
      "entropy": 1.4023477001623674,
      "epoch": 0.8989463601532567,
      "fork_head/loss": 2.686616897583008,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.0709264725446701,
      "grad_norm": 0.5453788042068481,
      "learning_rate": 3e-06,
      "loss": 0.6636,
      "step": 1877,
      "step_time": 7.5412262082099915
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.03571428571428571,
      "clip_ratio/low_mean": 0.09226190511669431,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.12797619083098002,
      "entropy": 1.5592157244682312,
      "epoch": 0.8994252873563219,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15578444302082062,
      "learning_rate": 3e-06,
      "loss": -0.0001,
      "step": 1878,
      "step_time": 7.863943062722683
    },
    {
      "clip_ratio/high_max": 0.18181818181818182,
      "clip_ratio/high_mean": 0.048484849658879364,
      "clip_ratio/low_mean": 0.022727272727272728,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0712121223861521,
      "entropy": 1.3983033462004228,
      "epoch": 0.899904214559387,
      "fork_head/loss": 2.6620748043060303,
      "fork_head/num_positive": 11.0625,
      "fork_head/pred_top1_mass": 0.09749637544155121,
      "grad_norm": 0.4812721908092499,
      "learning_rate": 3e-06,
      "loss": 0.6567,
      "step": 1879,
      "step_time": 7.616656884551048
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.05952381023338863,
      "clip_ratio/low_mean": 0.21726190511669433,
      "clip_ratio/low_min": 0.14285714285714285,
      "clip_ratio/region_mean": 0.2767857142857143,
      "entropy": 1.5839146205357142,
      "epoch": 0.9003831417624522,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12527680397033691,
      "learning_rate": 3e-06,
      "loss": -0.0024,
      "step": 1880,
      "step_time": 7.396390601992607
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 166.390625,
      "completions/mean_terminated_length": 166.390625,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.4514857700892858,
      "epoch": 0.9008620689655172,
      "fork/depth_mean": 0.27700671553611755,
      "fork/depth_std": 0.24561356008052826,
      "fork_head/loss": 2.730321168899536,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.0686953142285347,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5229273438453674,
      "learning_rate": 3e-06,
      "loss": 0.6967,
      "num_tokens": 47485108.0,
      "reward": 1.0271854400634766,
      "reward_std": 0.6120472550392151,
      "rewards/code_format_reward/mean": 0.7666015625,
      "rewards/code_format_reward/std": 0.3379426896572113,
      "rewards/code_reward/mean": 0.2605839967727661,
      "rewards/code_reward/std": 0.37985748052597046,
      "step": 1881,
      "step_time": 424.00046898424625
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.017857142857142856,
      "entropy": 1.4862444230488367,
      "epoch": 0.9013409961685823,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.3331870138645172,
      "learning_rate": 3e-06,
      "loss": -0.0186,
      "step": 1882,
      "step_time": 8.875872559845448
    },
    {
      "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,
      "entropy": 1.4562523365020752,
      "epoch": 0.9018199233716475,
      "fork_head/loss": 2.7252211570739746,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.0689963772892952,
      "grad_norm": 0.47554731369018555,
      "learning_rate": 3e-06,
      "loss": 0.693,
      "step": 1883,
      "step_time": 8.062989391386509
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.4877569334847587,
      "epoch": 0.9022988505747126,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2713257372379303,
      "learning_rate": 3e-06,
      "loss": -0.02,
      "step": 1884,
      "step_time": 7.985105484724045
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03571428571428571,
      "entropy": 1.4928792374474662,
      "epoch": 0.9027777777777778,
      "fork_head/loss": 2.714816093444824,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.0741049200296402,
      "grad_norm": 0.4241489768028259,
      "learning_rate": 3e-06,
      "loss": 0.6893,
      "step": 1885,
      "step_time": 8.080932222306728
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.10714285714285714,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.5147774815559387,
      "epoch": 0.9032567049808429,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17739556729793549,
      "learning_rate": 3e-06,
      "loss": -0.0223,
      "step": 1886,
      "step_time": 8.229330703616142
    },
    {
      "clip_ratio/high_max": 0.03571428571428571,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.05357142857142857,
      "entropy": 1.5134800502232142,
      "epoch": 0.9037356321839081,
      "fork_head/loss": 2.700204849243164,
      "fork_head/num_positive": 11.78125,
      "fork_head/pred_top1_mass": 0.07524453103542328,
      "grad_norm": 0.39720892906188965,
      "learning_rate": 3e-06,
      "loss": 0.6847,
      "step": 1887,
      "step_time": 7.748231880366802
    },
    {
      "clip_ratio/high_max": 0.21428571428571427,
      "clip_ratio/high_mean": 0.05595238187483379,
      "clip_ratio/low_mean": 0.10714285714285714,
      "clip_ratio/low_min": 0.07142857142857142,
      "clip_ratio/region_mean": 0.16309523901769094,
      "entropy": 1.5357433472360884,
      "epoch": 0.9042145593869731,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15871886909008026,
      "learning_rate": 3e-06,
      "loss": -0.0245,
      "step": 1888,
      "step_time": 8.276939675211906
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 146.875,
      "completions/mean_terminated_length": 146.6614532470703,
      "completions/min_length": 16.0,
      "completions/min_terminated_length": 16.0,
      "entropy": 1.2635044625827245,
      "epoch": 0.9046934865900383,
      "fork/depth_mean": 0.29100674390792847,
      "fork/depth_std": 0.2681611478328705,
      "fork_head/loss": 2.7362608909606934,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.063289575278759,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.6256260871887207,
      "learning_rate": 3e-06,
      "loss": 0.677,
      "num_tokens": 47648836.0,
      "reward": 0.984861433506012,
      "reward_std": 0.5884665250778198,
      "rewards/code_format_reward/mean": 0.7490234375,
      "rewards/code_format_reward/std": 0.33251550793647766,
      "rewards/code_reward/mean": 0.23583802580833435,
      "rewards/code_reward/std": 0.37079086899757385,
      "step": 1889,
      "step_time": 472.3136313855648
    },
    {
      "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,
      "entropy": 1.2198311941964286,
      "epoch": 0.9051724137931034,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2882499396800995,
      "learning_rate": 3e-06,
      "loss": 0.0097,
      "step": 1890,
      "step_time": 7.869675435125828
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.2499511752809798,
      "epoch": 0.9056513409961686,
      "fork_head/loss": 2.7255446910858154,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.060657501220703125,
      "grad_norm": 0.5029435157775879,
      "learning_rate": 3e-06,
      "loss": 0.6713,
      "step": 1891,
      "step_time": 8.299250803887844
    },
    {
      "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,
      "entropy": 1.2259579641478402,
      "epoch": 0.9061302681992337,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.27449461817741394,
      "learning_rate": 3e-06,
      "loss": 0.0081,
      "step": 1892,
      "step_time": 7.551197893917561
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.023809524519102915,
      "clip_ratio/low_mean": 0.023809524519102915,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.04761904903820583,
      "entropy": 1.2440871511186873,
      "epoch": 0.9066091954022989,
      "fork_head/loss": 2.712803363800049,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.0675322562456131,
      "grad_norm": 0.4459119737148285,
      "learning_rate": 3e-06,
      "loss": 0.6656,
      "step": 1893,
      "step_time": 7.520968914031982
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.1011904776096344,
      "clip_ratio/low_min": 0.03571428571428571,
      "clip_ratio/region_mean": 0.11904762046677726,
      "entropy": 1.2381262268338884,
      "epoch": 0.907088122605364,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.22413372993469238,
      "learning_rate": 3e-06,
      "loss": 0.0057,
      "step": 1894,
      "step_time": 8.174439013004303
    },
    {
      "clip_ratio/high_max": 0.14285714285714285,
      "clip_ratio/high_mean": 0.05952381023338863,
      "clip_ratio/low_mean": 0.04761904903820583,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.10714285927159446,
      "entropy": 1.240392484835216,
      "epoch": 0.9075670498084292,
      "fork_head/loss": 2.699986219406128,
      "fork_head/num_positive": 12.677419662475586,
      "fork_head/pred_top1_mass": 0.06780719012022018,
      "grad_norm": 0.42116716504096985,
      "learning_rate": 3e-06,
      "loss": 0.6594,
      "step": 1895,
      "step_time": 7.513641119003296
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.017857142857142856,
      "clip_ratio/low_mean": 0.11904762046677726,
      "clip_ratio/low_min": 0.03571428571428571,
      "clip_ratio/region_mean": 0.13690476332392013,
      "entropy": 1.2630227548735482,
      "epoch": 0.9080459770114943,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16271236538887024,
      "learning_rate": 3e-06,
      "loss": 0.0038,
      "step": 1896,
      "step_time": 7.560025438666344
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "completions/clipped_ratio": 0.001953125,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 162.375,
      "completions/mean_terminated_length": 162.1917724609375,
      "completions/min_length": 47.0,
      "completions/min_terminated_length": 47.0,
      "entropy": 1.5160518129666647,
      "epoch": 0.9085249042145593,
      "fork/depth_mean": 0.365882933139801,
      "fork/depth_std": 0.2685880959033966,
      "fork_head/loss": 2.7255663871765137,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.07302823662757874,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.578039288520813,
      "learning_rate": 3e-06,
      "loss": 0.6713,
      "num_tokens": 47818276.0,
      "reward": 1.063380241394043,
      "reward_std": 0.5197767615318298,
      "rewards/code_format_reward/mean": 0.8212890625,
      "rewards/code_format_reward/std": 0.28462886810302734,
      "rewards/code_reward/mean": 0.24209117889404297,
      "rewards/code_reward/std": 0.34946340322494507,
      "step": 1897,
      "step_time": 415.69323156774044
    },
    {
      "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,
      "entropy": 1.412323490778605,
      "epoch": 0.9090038314176245,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17793314158916473,
      "learning_rate": 3e-06,
      "loss": 0.0042,
      "step": 1898,
      "step_time": 8.434662893414497
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03333333333333333,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.5008246541023254,
      "epoch": 0.9094827586206896,
      "fork_head/loss": 2.715812921524048,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.0711435079574585,
      "grad_norm": 0.5352245569229126,
      "learning_rate": 3e-06,
      "loss": 0.6688,
      "step": 1899,
      "step_time": 8.930839367210865
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.03333333333333333,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03333333333333333,
      "entropy": 1.4181025664011637,
      "epoch": 0.9099616858237548,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.17056655883789062,
      "learning_rate": 3e-06,
      "loss": 0.0034,
      "step": 1900,
      "step_time": 8.337860494852066
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.061111111442248026,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.061111111442248026,
      "entropy": 1.538671871026357,
      "epoch": 0.9104406130268199,
      "fork_head/loss": 2.6981863975524902,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.08534223586320877,
      "grad_norm": 0.46599701046943665,
      "learning_rate": 3e-06,
      "loss": 0.6624,
      "step": 1901,
      "step_time": 8.071616396307945
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.055555556217829385,
      "clip_ratio/low_mean": 0.03000000019868215,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08555555641651154,
      "entropy": 1.4201117674509685,
      "epoch": 0.9109195402298851,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15827664732933044,
      "learning_rate": 3e-06,
      "loss": 0.0018,
      "step": 1902,
      "step_time": 8.270521871745586
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.016666666666666666,
      "clip_ratio/low_mean": 0.11111111144224803,
      "clip_ratio/low_min": 0.06666666666666667,
      "clip_ratio/region_mean": 0.1277777781089147,
      "entropy": 1.538642950852712,
      "epoch": 0.9113984674329502,
      "fork_head/loss": 2.6784849166870117,
      "fork_head/num_positive": 11.375,
      "fork_head/pred_top1_mass": 0.10200417041778564,
      "grad_norm": 0.44080448150634766,
      "learning_rate": 3e-06,
      "loss": 0.6567,
      "step": 1903,
      "step_time": 7.753599010407925
    },
    {
      "clip_ratio/high_max": 0.13333333333333333,
      "clip_ratio/high_mean": 0.055555556217829385,
      "clip_ratio/low_mean": 0.03888888955116272,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.09444444576899211,
      "entropy": 1.4331485191980997,
      "epoch": 0.9118773946360154,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.12027783691883087,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 1904,
      "step_time": 8.426918372511864
    },
    {
      "clip_ratio/high_max": 0.02083333395421505,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.010416666977107525,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 149.078125,
      "completions/mean_terminated_length": 149.078125,
      "completions/min_length": 34.0,
      "completions/min_terminated_length": 34.0,
      "entropy": 1.47517903521657,
      "epoch": 0.9123563218390804,
      "fork/depth_mean": 0.3123027980327606,
      "fork/depth_std": 0.2632388472557068,
      "fork_head/loss": 2.7308125495910645,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.06907062232494354,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6409235000610352,
      "learning_rate": 3e-06,
      "loss": 0.6768,
      "num_tokens": 47997420.0,
      "reward": 1.0667598247528076,
      "reward_std": 0.5551367402076721,
      "rewards/code_format_reward/mean": 0.814453125,
      "rewards/code_format_reward/std": 0.3044774830341339,
      "rewards/code_reward/mean": 0.25230661034584045,
      "rewards/code_reward/std": 0.3631918132305145,
      "step": 1905,
      "step_time": 420.51631554961205
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.055555556217829385,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.055555556217829385,
      "entropy": 1.5920334339141846,
      "epoch": 0.9128352490421456,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13152053952217102,
      "learning_rate": 3e-06,
      "loss": 0.009,
      "step": 1906,
      "step_time": 9.617921844124794
    },
    {
      "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,
      "entropy": 1.4679921492934227,
      "epoch": 0.9133141762452107,
      "fork_head/loss": 2.7209854125976562,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.06801984459161758,
      "grad_norm": 0.5562285780906677,
      "learning_rate": 3e-06,
      "loss": 0.6734,
      "step": 1907,
      "step_time": 8.106380090117455
    },
    {
      "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,
      "entropy": 1.5817534724871318,
      "epoch": 0.9137931034482759,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14180712401866913,
      "learning_rate": 3e-06,
      "loss": 0.0088,
      "step": 1908,
      "step_time": 8.428455151617527
    },
    {
      "clip_ratio/high_max": 0.02083333395421505,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.015625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.026041666977107525,
      "entropy": 1.4685618057847023,
      "epoch": 0.914272030651341,
      "fork_head/loss": 2.7060301303863525,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.06981489062309265,
      "grad_norm": 0.5192699432373047,
      "learning_rate": 3e-06,
      "loss": 0.6657,
      "step": 1909,
      "step_time": 8.316374093294144
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.022222222884496053,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.596080736319224,
      "epoch": 0.9147509578544061,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13349373638629913,
      "learning_rate": 3e-06,
      "loss": 0.0079,
      "step": 1910,
      "step_time": 8.62867970764637
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0416666679084301,
      "clip_ratio/low_mean": 0.046875,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0885416679084301,
      "entropy": 1.471633903682232,
      "epoch": 0.9152298850574713,
      "fork_head/loss": 2.6886146068573,
      "fork_head/num_positive": 11.625,
      "fork_head/pred_top1_mass": 0.06857769191265106,
      "grad_norm": 0.48247429728507996,
      "learning_rate": 3e-06,
      "loss": 0.6598,
      "step": 1911,
      "step_time": 8.405080616474152
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.08888888955116273,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08888888955116273,
      "entropy": 1.614891497294108,
      "epoch": 0.9157088122605364,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10927785933017731,
      "learning_rate": 3e-06,
      "loss": 0.0064,
      "step": 1912,
      "step_time": 8.872526966035366
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.02083333395421505,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.02083333395421505,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 164.23046875,
      "completions/mean_terminated_length": 164.23046875,
      "completions/min_length": 39.0,
      "completions/min_terminated_length": 39.0,
      "entropy": 1.4471191428601742,
      "epoch": 0.9161877394636015,
      "fork/depth_mean": 0.29667922854423523,
      "fork/depth_std": 0.24301199615001678,
      "fork_head/loss": 2.726388931274414,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.06194590777158737,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.504120409488678,
      "learning_rate": 3e-06,
      "loss": 0.6806,
      "num_tokens": 48181282.0,
      "reward": 0.9630024433135986,
      "reward_std": 0.528886079788208,
      "rewards/code_format_reward/mean": 0.783203125,
      "rewards/code_format_reward/std": 0.3233736753463745,
      "rewards/code_reward/mean": 0.17979930341243744,
      "rewards/code_reward/std": 0.3132753372192383,
      "step": 1913,
      "step_time": 362.1274154186249
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.014285714498588018,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.014285714498588018,
      "entropy": 1.674655863216945,
      "epoch": 0.9166666666666666,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.28098323941230774,
      "learning_rate": 3e-06,
      "loss": 0.0067,
      "step": 1914,
      "step_time": 8.45255671441555
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0416666679084301,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0416666679084301,
      "entropy": 1.4469380788505077,
      "epoch": 0.9171455938697318,
      "fork_head/loss": 2.717156171798706,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.06426342576742172,
      "grad_norm": 0.48546263575553894,
      "learning_rate": 3e-06,
      "loss": 0.6775,
      "step": 1915,
      "step_time": 7.546060420572758
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.023809524519102915,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.023809524519102915,
      "entropy": 1.6454566631998335,
      "epoch": 0.9176245210727969,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.24206747114658356,
      "learning_rate": 3e-06,
      "loss": 0.0054,
      "step": 1916,
      "step_time": 8.213897243142128
    },
    {
      "clip_ratio/high_max": 0.03125,
      "clip_ratio/high_mean": 0.010416666977107525,
      "clip_ratio/low_mean": 0.05208333395421505,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.4385904930531979,
      "epoch": 0.9181034482758621,
      "fork_head/loss": 2.7025837898254395,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.07379596680402756,
      "grad_norm": 0.4544762670993805,
      "learning_rate": 3e-06,
      "loss": 0.671,
      "step": 1917,
      "step_time": 7.491182334721088
    },
    {
      "clip_ratio/high_max": 0.07142857142857142,
      "clip_ratio/high_mean": 0.023809524519102915,
      "clip_ratio/low_mean": 0.017857142857142856,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.04166666737624577,
      "entropy": 1.6647088868277413,
      "epoch": 0.9185823754789272,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2118074595928192,
      "learning_rate": 3e-06,
      "loss": 0.0033,
      "step": 1918,
      "step_time": 7.953509360551834
    },
    {
      "clip_ratio/high_max": 0.15625,
      "clip_ratio/high_mean": 0.057291666977107525,
      "clip_ratio/low_mean": 0.1510416679084301,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.20833333395421505,
      "entropy": 1.4545878134667873,
      "epoch": 0.9190613026819924,
      "fork_head/loss": 2.6845743656158447,
      "fork_head/num_positive": 11.1875,
      "fork_head/pred_top1_mass": 0.08080393075942993,
      "grad_norm": 0.41982850432395935,
      "learning_rate": 3e-06,
      "loss": 0.6652,
      "step": 1919,
      "step_time": 7.986360877752304
    },
    {
      "clip_ratio/high_max": 0.21428571428571427,
      "clip_ratio/high_mean": 0.08333333475249154,
      "clip_ratio/low_mean": 0.03571428571428571,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.11904762046677726,
      "entropy": 1.6614397423607963,
      "epoch": 0.9195402298850575,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11306506395339966,
      "learning_rate": 3e-06,
      "loss": 0.0018,
      "step": 1920,
      "step_time": 7.742419436573982
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 155.546875,
      "completions/mean_terminated_length": 155.546875,
      "completions/min_length": 54.0,
      "completions/min_terminated_length": 54.0,
      "entropy": 1.4322048664093017,
      "epoch": 0.9200191570881227,
      "fork/depth_mean": 0.33119428157806396,
      "fork/depth_std": 0.2821493148803711,
      "fork_head/loss": 2.716120481491089,
      "fork_head/num_positive": 11.838709831237793,
      "fork_head/pred_top1_mass": 0.06566507369279861,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.519585907459259,
      "learning_rate": 3e-06,
      "loss": 0.6735,
      "num_tokens": 48352282.0,
      "reward": 1.0132849216461182,
      "reward_std": 0.5491307377815247,
      "rewards/code_format_reward/mean": 0.791015625,
      "rewards/code_format_reward/std": 0.2922303378582001,
      "rewards/code_reward/mean": 0.22226938605308533,
      "rewards/code_reward/std": 0.3547648787498474,
      "step": 1921,
      "step_time": 437.6258350238204
    },
    {
      "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,
      "entropy": 1.2923522537404841,
      "epoch": 0.9204980842911877,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.22042503952980042,
      "learning_rate": 3e-06,
      "loss": 0.0037,
      "step": 1922,
      "step_time": 7.950380057096481
    },
    {
      "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,
      "entropy": 1.4446614662806192,
      "epoch": 0.9209770114942529,
      "fork_head/loss": 2.7065982818603516,
      "fork_head/num_positive": 11.838709831237793,
      "fork_head/pred_top1_mass": 0.07631608843803406,
      "grad_norm": 0.47520020604133606,
      "learning_rate": 3e-06,
      "loss": 0.6708,
      "step": 1923,
      "step_time": 8.645217895507812
    },
    {
      "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,
      "entropy": 1.286127881570296,
      "epoch": 0.921455938697318,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.19191282987594604,
      "learning_rate": 3e-06,
      "loss": 0.0029,
      "step": 1924,
      "step_time": 8.13094425946474
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.055555556217829385,
      "clip_ratio/low_min": 0.022222222884496053,
      "clip_ratio/region_mean": 0.07777777910232545,
      "entropy": 1.4536675294240315,
      "epoch": 0.9219348659003831,
      "fork_head/loss": 2.693230390548706,
      "fork_head/num_positive": 11.838709831237793,
      "fork_head/pred_top1_mass": 0.0797845795750618,
      "grad_norm": 0.39439189434051514,
      "learning_rate": 3e-06,
      "loss": 0.6656,
      "step": 1925,
      "step_time": 8.355597212910652
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.045454545454545456,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.045454545454545456,
      "entropy": 1.2968454252589832,
      "epoch": 0.9224137931034483,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1810930371284485,
      "learning_rate": 3e-06,
      "loss": 0.002,
      "step": 1926,
      "step_time": 7.862555742263794
    },
    {
      "clip_ratio/high_max": 0.06666666666666667,
      "clip_ratio/high_mean": 0.022222222884496053,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.022222222884496053,
      "entropy": 1.4341579874356587,
      "epoch": 0.9228927203065134,
      "fork_head/loss": 2.679076671600342,
      "fork_head/num_positive": 11.838709831237793,
      "fork_head/pred_top1_mass": 0.07839152961969376,
      "grad_norm": 0.3574138581752777,
      "learning_rate": 3e-06,
      "loss": 0.6616,
      "step": 1927,
      "step_time": 8.752508990466595
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.12121212211522189,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.12121212211522189,
      "entropy": 1.3122681921178645,
      "epoch": 0.9233716475095786,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.21265186369419098,
      "learning_rate": 3e-06,
      "loss": 0.0009,
      "step": 1928,
      "step_time": 7.9115357249975204
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 148.123046875,
      "completions/mean_terminated_length": 148.123046875,
      "completions/min_length": 11.0,
      "completions/min_terminated_length": 11.0,
      "entropy": 1.4894501566886902,
      "epoch": 0.9238505747126436,
      "fork/depth_mean": 0.2965339720249176,
      "fork/depth_std": 0.29323264956474304,
      "fork_head/loss": 2.720953941345215,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.06777673959732056,
      "frac_reward_zero_std": 0.03125,
      "grad_norm": 0.5990156531333923,
      "learning_rate": 3e-06,
      "loss": 0.6825,
      "num_tokens": 48515273.0,
      "reward": 1.0061254501342773,
      "reward_std": 0.5709160566329956,
      "rewards/code_format_reward/mean": 0.78515625,
      "rewards/code_format_reward/std": 0.32164934277534485,
      "rewards/code_reward/mean": 0.22096920013427734,
      "rewards/code_reward/std": 0.3815155327320099,
      "step": 1929,
      "step_time": 439.54104113578796
    },
    {
      "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,
      "entropy": 1.7512019230769231,
      "epoch": 0.9243295019157088,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 1.0752289295196533,
      "learning_rate": 3e-06,
      "loss": -0.0061,
      "step": 1930,
      "step_time": 7.777521178126335
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.4787967584349893,
      "epoch": 0.9248084291187739,
      "fork_head/loss": 2.713820457458496,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.06582251191139221,
      "grad_norm": 0.5020179748535156,
      "learning_rate": 3e-06,
      "loss": 0.6811,
      "step": 1931,
      "step_time": 8.312488555908203
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.019230769230769232,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.019230769230769232,
      "entropy": 1.7821890024038463,
      "epoch": 0.9252873563218391,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1978655904531479,
      "learning_rate": 3e-06,
      "loss": -0.0067,
      "step": 1932,
      "step_time": 7.604778707027435
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.03030303120613098,
      "entropy": 1.4903971281918613,
      "epoch": 0.9257662835249042,
      "fork_head/loss": 2.7022557258605957,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.07463289052248001,
      "grad_norm": 0.44347813725471497,
      "learning_rate": 3e-06,
      "loss": 0.6767,
      "step": 1933,
      "step_time": 8.42561537027359
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.019230769230769232,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.019230769230769232,
      "entropy": 1.7567983774038463,
      "epoch": 0.9262452107279694,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16926737129688263,
      "learning_rate": 3e-06,
      "loss": -0.008,
      "step": 1934,
      "step_time": 7.249819114804268
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.03030303120613098,
      "clip_ratio/low_mean": 0.03030303120613098,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.06060606241226196,
      "entropy": 1.4923650513995776,
      "epoch": 0.9267241379310345,
      "fork_head/loss": 2.6883442401885986,
      "fork_head/num_positive": 12.0,
      "fork_head/pred_top1_mass": 0.0771927535533905,
      "grad_norm": 0.43228888511657715,
      "learning_rate": 3e-06,
      "loss": 0.6692,
      "step": 1935,
      "step_time": 7.861064650118351
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.7208533653846154,
      "epoch": 0.9272030651340997,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2134193331003189,
      "learning_rate": 3e-06,
      "loss": -0.0095,
      "step": 1936,
      "step_time": 7.7159438803792
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 155.90234375,
      "completions/mean_terminated_length": 155.5098114013672,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.395751953125,
      "epoch": 0.9276819923371648,
      "fork/depth_mean": 0.32154789566993713,
      "fork/depth_std": 0.29048675298690796,
      "fork_head/loss": 2.73842191696167,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.06749001145362854,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5546726584434509,
      "learning_rate": 3e-06,
      "loss": 0.6871,
      "num_tokens": 48685319.0,
      "reward": 0.9816254377365112,
      "reward_std": 0.5965516567230225,
      "rewards/code_format_reward/mean": 0.7509765625,
      "rewards/code_format_reward/std": 0.32807186245918274,
      "rewards/code_reward/mean": 0.23064887523651123,
      "rewards/code_reward/std": 0.3721805810928345,
      "step": 1937,
      "step_time": 437.0333309844136
    },
    {
      "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,
      "entropy": 1.9167887369791667,
      "epoch": 0.9281609195402298,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.2966855466365814,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 1938,
      "step_time": 7.5905503779649734
    },
    {
      "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,
      "entropy": 1.4149576822916667,
      "epoch": 0.928639846743295,
      "fork_head/loss": 2.7291836738586426,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07626700401306152,
      "grad_norm": 0.5000538229942322,
      "learning_rate": 3e-06,
      "loss": 0.6846,
      "step": 1939,
      "step_time": 7.844723284244537
    },
    {
      "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,
      "entropy": 1.9285074869791667,
      "epoch": 0.9291187739463601,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.29223453998565674,
      "learning_rate": 3e-06,
      "loss": -0.0069,
      "step": 1940,
      "step_time": 7.795609630644321
    },
    {
      "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,
      "entropy": 1.421630859375,
      "epoch": 0.9295977011494253,
      "fork_head/loss": 2.715059518814087,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.07749582827091217,
      "grad_norm": 0.4243602752685547,
      "learning_rate": 3e-06,
      "loss": 0.6812,
      "step": 1941,
      "step_time": 7.71520060300827
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.013888889302810034,
      "clip_ratio/low_mean": 0.08333333333333333,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.09722222263614337,
      "entropy": 1.9576280415058136,
      "epoch": 0.9300766283524904,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.23609083890914917,
      "learning_rate": 3e-06,
      "loss": -0.0087,
      "step": 1942,
      "step_time": 7.706417560577393
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.10416666666666667,
      "clip_ratio/low_min": 0.08333333333333333,
      "clip_ratio/region_mean": 0.10416666666666667,
      "entropy": 1.4200439453125,
      "epoch": 0.9305555555555556,
      "fork_head/loss": 2.6986207962036133,
      "fork_head/num_positive": 11.75,
      "fork_head/pred_top1_mass": 0.09286387264728546,
      "grad_norm": 0.38261473178863525,
      "learning_rate": 3e-06,
      "loss": 0.6767,
      "step": 1943,
      "step_time": 7.682610638439655
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.0416666679084301,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.2916666679084301,
      "entropy": 1.9893391877412796,
      "epoch": 0.9310344827586207,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.15271751582622528,
      "learning_rate": 3e-06,
      "loss": -0.0105,
      "step": 1944,
      "step_time": 7.2067103534936905
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 143.15234375,
      "completions/mean_terminated_length": 143.15234375,
      "completions/min_length": 32.0,
      "completions/min_terminated_length": 32.0,
      "entropy": 1.12900390625,
      "epoch": 0.9315134099616859,
      "fork/depth_mean": 0.3670128881931305,
      "fork/depth_std": 0.29545849561691284,
      "fork_head/loss": 2.728550434112549,
      "fork_head/num_positive": 12.4375,
      "fork_head/pred_top1_mass": 0.054341260343790054,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.3949643075466156,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "num_tokens": 48853541.0,
      "reward": 1.070020318031311,
      "reward_std": 0.6317546367645264,
      "rewards/code_format_reward/mean": 0.771484375,
      "rewards/code_format_reward/std": 0.3258698284626007,
      "rewards/code_reward/mean": 0.29853594303131104,
      "rewards/code_reward/std": 0.4012470543384552,
      "step": 1945,
      "step_time": 358.99415238946676
    },
    {
      "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,
      "entropy": 1.31298828125,
      "epoch": 0.9319923371647509,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11073314398527145,
      "learning_rate": 3e-06,
      "loss": 0.0067,
      "step": 1946,
      "step_time": 7.629308834671974
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.05,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.1296875,
      "epoch": 0.9324712643678161,
      "fork_head/loss": 2.7221643924713135,
      "fork_head/num_positive": 12.4375,
      "fork_head/pred_top1_mass": 0.05981622263789177,
      "grad_norm": 0.3762713670730591,
      "learning_rate": 3e-06,
      "loss": 0.6772,
      "step": 1947,
      "step_time": 7.775346830487251
    },
    {
      "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,
      "entropy": 1.336458338631524,
      "epoch": 0.9329501915708812,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10231012105941772,
      "learning_rate": 3e-06,
      "loss": 0.006,
      "step": 1948,
      "step_time": 7.714216783642769
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.05,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.13271484375,
      "epoch": 0.9334291187739464,
      "fork_head/loss": 2.711524486541748,
      "fork_head/num_positive": 12.4375,
      "fork_head/pred_top1_mass": 0.06525605171918869,
      "grad_norm": 0.3550872504711151,
      "learning_rate": 3e-06,
      "loss": 0.6741,
      "step": 1949,
      "step_time": 7.695300258696079
    },
    {
      "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,
      "entropy": 1.3165979385375977,
      "epoch": 0.9339080459770115,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1033894270658493,
      "learning_rate": 3e-06,
      "loss": 0.0065,
      "step": 1950,
      "step_time": 7.5347340777516365
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.05,
      "clip_ratio/low_min": 0.05,
      "clip_ratio/region_mean": 0.05,
      "entropy": 1.13466796875,
      "epoch": 0.9343869731800766,
      "fork_head/loss": 2.6988911628723145,
      "fork_head/num_positive": 12.4375,
      "fork_head/pred_top1_mass": 0.07896512746810913,
      "grad_norm": 0.33511507511138916,
      "learning_rate": 3e-06,
      "loss": 0.671,
      "step": 1951,
      "step_time": 7.6094740852713585
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.027777777777777776,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.027777777777777776,
      "entropy": 1.3468026717503865,
      "epoch": 0.9348659003831418,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.08875462412834167,
      "learning_rate": 3e-06,
      "loss": 0.0054,
      "step": 1952,
      "step_time": 7.539103619754314
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 152.521484375,
      "completions/mean_terminated_length": 152.521484375,
      "completions/min_length": 41.0,
      "completions/min_terminated_length": 41.0,
      "entropy": 1.0426728291945024,
      "epoch": 0.9353448275862069,
      "fork/depth_mean": 0.3794085681438446,
      "fork/depth_std": 0.29138559103012085,
      "fork_head/loss": 2.727698802947998,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07845080643892288,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.5625340938568115,
      "learning_rate": 3e-06,
      "loss": 0.6823,
      "num_tokens": 49028208.0,
      "reward": 0.9740825295448303,
      "reward_std": 0.5059677958488464,
      "rewards/code_format_reward/mean": 0.771484375,
      "rewards/code_format_reward/std": 0.31980812549591064,
      "rewards/code_reward/mean": 0.20259815454483032,
      "rewards/code_reward/std": 0.33722934126853943,
      "step": 1953,
      "step_time": 376.05313532054424
    },
    {
      "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,
      "entropy": 1.2195763221153846,
      "epoch": 0.935823754789272,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.16561752557754517,
      "learning_rate": 3e-06,
      "loss": 0.0022,
      "step": 1954,
      "step_time": 7.77594755589962
    },
    {
      "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,
      "entropy": 1.038204312324524,
      "epoch": 0.9363026819923371,
      "fork_head/loss": 2.7185187339782715,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.0796513482928276,
      "grad_norm": 0.543079137802124,
      "learning_rate": 3e-06,
      "loss": 0.6788,
      "step": 1955,
      "step_time": 8.187141105532646
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.07692307692307693,
      "entropy": 1.2217548076923077,
      "epoch": 0.9367816091954023,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14762653410434723,
      "learning_rate": 3e-06,
      "loss": 0.0008,
      "step": 1956,
      "step_time": 7.736348420381546
    },
    {
      "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,
      "entropy": 1.0412227673964067,
      "epoch": 0.9372605363984674,
      "fork_head/loss": 2.7017934322357178,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.08213059604167938,
      "grad_norm": 0.49447640776634216,
      "learning_rate": 3e-06,
      "loss": 0.6737,
      "step": 1957,
      "step_time": 7.419141545891762
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.038461538461538464,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.11538461538461539,
      "entropy": 1.2276141826923077,
      "epoch": 0.9377394636015326,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14429673552513123,
      "learning_rate": 3e-06,
      "loss": -0.0006,
      "step": 1958,
      "step_time": 7.655941762030125
    },
    {
      "clip_ratio/high_max": 0.09090909090909091,
      "clip_ratio/high_mean": 0.03030303120613098,
      "clip_ratio/low_mean": 0.09090909090909091,
      "clip_ratio/low_min": 0.09090909090909091,
      "clip_ratio/region_mean": 0.12121212211522189,
      "entropy": 1.0349786931818181,
      "epoch": 0.9382183908045977,
      "fork_head/loss": 2.6813817024230957,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.10538545995950699,
      "grad_norm": 0.4438776671886444,
      "learning_rate": 3e-06,
      "loss": 0.6667,
      "step": 1959,
      "step_time": 7.628019578754902
    },
    {
      "clip_ratio/high_max": 0.07692307692307693,
      "clip_ratio/high_mean": 0.07692307692307693,
      "clip_ratio/low_mean": 0.057692307692307696,
      "clip_ratio/low_min": 0.038461538461538464,
      "clip_ratio/region_mean": 0.1346153846153846,
      "entropy": 1.2264873798076923,
      "epoch": 0.9386973180076629,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.1160198301076889,
      "learning_rate": 3e-06,
      "loss": -0.0014,
      "step": 1960,
      "step_time": 7.531622804701328
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 159.2265625,
      "completions/mean_terminated_length": 159.2265625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.20079209903876,
      "epoch": 0.939176245210728,
      "fork/depth_mean": 0.33118435740470886,
      "fork/depth_std": 0.27628859877586365,
      "fork_head/loss": 2.7131447792053223,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.07041054219007492,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.7099963426589966,
      "learning_rate": 3e-06,
      "loss": 0.6787,
      "num_tokens": 49209444.0,
      "reward": 0.9623632431030273,
      "reward_std": 0.5540130138397217,
      "rewards/code_format_reward/mean": 0.767578125,
      "rewards/code_format_reward/std": 0.3107399046421051,
      "rewards/code_reward/mean": 0.19478510320186615,
      "rewards/code_reward/std": 0.3576781451702118,
      "step": 1961,
      "step_time": 350.7146336585283
    },
    {
      "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,
      "entropy": 1.3727032790581386,
      "epoch": 0.9396551724137931,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.14267563819885254,
      "learning_rate": 3e-06,
      "loss": -0.0035,
      "step": 1962,
      "step_time": 8.593812622129917
    },
    {
      "clip_ratio/high_max": 0.08333333333333333,
      "clip_ratio/high_mean": 0.041666666666666664,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.1825629323720932,
      "epoch": 0.9401340996168582,
      "fork_head/loss": 2.699500560760498,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.07020226120948792,
      "grad_norm": 0.615203320980072,
      "learning_rate": 3e-06,
      "loss": 0.6745,
      "step": 1963,
      "step_time": 8.068834595382214
    },
    {
      "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,
      "entropy": 1.3696922014156978,
      "epoch": 0.9406130268199234,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.10964889079332352,
      "learning_rate": 3e-06,
      "loss": -0.0042,
      "step": 1964,
      "step_time": 7.933951422572136
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.125,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.125,
      "entropy": 1.1878255208333333,
      "epoch": 0.9410919540229885,
      "fork_head/loss": 2.6796350479125977,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.1007409542798996,
      "grad_norm": 0.4893675148487091,
      "learning_rate": 3e-06,
      "loss": 0.6699,
      "step": 1965,
      "step_time": 7.640578433871269
    },
    {
      "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,
      "entropy": 1.3461733261744182,
      "epoch": 0.9415708812260536,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.11688773334026337,
      "learning_rate": 3e-06,
      "loss": -0.0043,
      "step": 1966,
      "step_time": 8.250076651573181
    },
    {
      "clip_ratio/high_max": 0.25,
      "clip_ratio/high_mean": 0.16666666666666666,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.16666666666666666,
      "entropy": 1.1682400157054265,
      "epoch": 0.9420498084291188,
      "fork_head/loss": 2.6595816612243652,
      "fork_head/num_positive": 11.3125,
      "fork_head/pred_top1_mass": 0.11255902051925659,
      "grad_norm": 0.4053972661495209,
      "learning_rate": 3e-06,
      "loss": 0.664,
      "step": 1967,
      "step_time": 7.459468826651573
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.041666666666666664,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.041666666666666664,
      "entropy": 1.3725314686695735,
      "epoch": 0.9425287356321839,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.13966289162635803,
      "learning_rate": 3e-06,
      "loss": -0.0048,
      "step": 1968,
      "step_time": 8.29489654302597
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 159.67578125,
      "completions/mean_terminated_length": 159.67578125,
      "completions/min_length": 52.0,
      "completions/min_terminated_length": 52.0,
      "entropy": 1.2944742813706398,
      "epoch": 0.9430076628352491,
      "fork/depth_mean": 0.32926759123802185,
      "fork/depth_std": 0.28303906321525574,
      "fork_head/loss": 2.74293851852417,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07233453541994095,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6299628019332886,
      "learning_rate": 3e-06,
      "loss": 0.6884,
      "num_tokens": 49382254.0,
      "reward": 0.9283899664878845,
      "reward_std": 0.5447753667831421,
      "rewards/code_format_reward/mean": 0.7470703125,
      "rewards/code_format_reward/std": 0.34407368302345276,
      "rewards/code_reward/mean": 0.18131965398788452,
      "rewards/code_reward/std": 0.29995134472846985,
      "step": 1969,
      "step_time": 383.8746586665511
    },
    {
      "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,
      "entropy": 1.24658203125,
      "epoch": 0.9434865900383141,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06380510330200195,
      "learning_rate": 3e-06,
      "loss": -0.0024,
      "step": 1970,
      "step_time": 7.570208512246609
    },
    {
      "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,
      "entropy": 1.295654296875,
      "epoch": 0.9439655172413793,
      "fork_head/loss": 2.734373092651367,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.0716598778963089,
      "grad_norm": 0.5049958825111389,
      "learning_rate": 3e-06,
      "loss": 0.685,
      "step": 1971,
      "step_time": 8.873059995472431
    },
    {
      "clip_ratio/high_max": 0.16666666666666666,
      "clip_ratio/high_mean": 0.16666666666666666,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.16666666666666666,
      "entropy": 1.21728515625,
      "epoch": 0.9444444444444444,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.021909773349761963,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 1972,
      "step_time": 7.293678469955921
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.3326822966337204,
      "epoch": 0.9449233716475096,
      "fork_head/loss": 2.7191872596740723,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07823405414819717,
      "grad_norm": 0.4468959867954254,
      "learning_rate": 3e-06,
      "loss": 0.6807,
      "step": 1973,
      "step_time": 7.566475011408329
    },
    {
      "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,
      "entropy": 1.2498372395833333,
      "epoch": 0.9454022988505747,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.07047799229621887,
      "learning_rate": 3e-06,
      "loss": -0.0024,
      "step": 1974,
      "step_time": 7.562328509986401
    },
    {
      "clip_ratio/high_max": 0.0416666679084301,
      "clip_ratio/high_mean": 0.02083333395421505,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.08333333395421505,
      "entropy": 1.3409016877412796,
      "epoch": 0.9458812260536399,
      "fork_head/loss": 2.700735330581665,
      "fork_head/num_positive": 11.8125,
      "fork_head/pred_top1_mass": 0.07988199591636658,
      "grad_norm": 0.4034043252468109,
      "learning_rate": 3e-06,
      "loss": 0.6756,
      "step": 1975,
      "step_time": 7.775803335011005
    },
    {
      "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,
      "entropy": 1.2291666666666667,
      "epoch": 0.946360153256705,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06389022618532181,
      "learning_rate": 3e-06,
      "loss": -0.0025,
      "step": 1976,
      "step_time": 7.7087241262197495
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 156.56640625,
      "completions/mean_terminated_length": 156.56640625,
      "completions/min_length": 32.0,
      "completions/min_terminated_length": 32.0,
      "entropy": 1.522705078125,
      "epoch": 0.9468390804597702,
      "fork/depth_mean": 0.3517395555973053,
      "fork/depth_std": 0.29796767234802246,
      "fork_head/loss": 2.7224042415618896,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.06535390764474869,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.41159915924072266,
      "learning_rate": 3e-06,
      "loss": 0.6827,
      "num_tokens": 49555136.0,
      "reward": 1.0071886777877808,
      "reward_std": 0.5234795808792114,
      "rewards/code_format_reward/mean": 0.7880859375,
      "rewards/code_format_reward/std": 0.30889400839805603,
      "rewards/code_reward/mean": 0.219102680683136,
      "rewards/code_reward/std": 0.3478129804134369,
      "step": 1977,
      "step_time": 381.71982277184725
    },
    {
      "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,
      "entropy": 1.2537841796875,
      "epoch": 0.9473180076628352,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09904736280441284,
      "learning_rate": 3e-06,
      "loss": -0.0044,
      "step": 1978,
      "step_time": 7.6998604238033295
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.484375,
      "epoch": 0.9477969348659003,
      "fork_head/loss": 2.716715097427368,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.0706479400396347,
      "grad_norm": 0.39702656865119934,
      "learning_rate": 3e-06,
      "loss": 0.6814,
      "step": 1979,
      "step_time": 8.363416604697704
    },
    {
      "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,
      "entropy": 1.23486328125,
      "epoch": 0.9482758620689655,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.3166596591472626,
      "learning_rate": 3e-06,
      "loss": -0.0045,
      "step": 1980,
      "step_time": 7.607493683695793
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.25,
      "entropy": 1.49609375,
      "epoch": 0.9487547892720306,
      "fork_head/loss": 2.7069692611694336,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.07307548820972443,
      "grad_norm": 0.3981577754020691,
      "learning_rate": 3e-06,
      "loss": 0.6782,
      "step": 1981,
      "step_time": 7.363748036324978
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0625,
      "clip_ratio/low_min": 0.0625,
      "clip_ratio/region_mean": 0.0625,
      "entropy": 1.2418212890625,
      "epoch": 0.9492337164750958,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09390246123075485,
      "learning_rate": 3e-06,
      "loss": -0.0047,
      "step": 1982,
      "step_time": 7.462623484432697
    },
    {
      "clip_ratio/high_max": 0.0625,
      "clip_ratio/high_mean": 0.0625,
      "clip_ratio/low_mean": 0.25,
      "clip_ratio/low_min": 0.25,
      "clip_ratio/region_mean": 0.3125,
      "entropy": 1.566650390625,
      "epoch": 0.9497126436781609,
      "fork_head/loss": 2.6946215629577637,
      "fork_head/num_positive": 11.65625,
      "fork_head/pred_top1_mass": 0.07269571721553802,
      "grad_norm": 0.35555464029312134,
      "learning_rate": 3e-06,
      "loss": 0.6742,
      "step": 1983,
      "step_time": 7.636840604245663
    },
    {
      "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,
      "entropy": 1.2210693359375,
      "epoch": 0.9501915708812261,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09117218106985092,
      "learning_rate": 3e-06,
      "loss": -0.0053,
      "step": 1984,
      "step_time": 7.377876952290535
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 157.87109375,
      "completions/mean_terminated_length": 157.87109375,
      "completions/min_length": 2.0,
      "completions/min_terminated_length": 2.0,
      "entropy": 1.3151041666666667,
      "epoch": 0.9506704980842912,
      "fork/depth_mean": 0.3787665069103241,
      "fork/depth_std": 0.31904175877571106,
      "fork_head/loss": 2.730168581008911,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.0712750256061554,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.6399468183517456,
      "learning_rate": 3e-06,
      "loss": 0.6798,
      "num_tokens": 49730462.0,
      "reward": 0.9858189821243286,
      "reward_std": 0.5946834683418274,
      "rewards/code_format_reward/mean": 0.75,
      "rewards/code_format_reward/std": 0.33251693844795227,
      "rewards/code_reward/mean": 0.2358189821243286,
      "rewards/code_reward/std": 0.3791508674621582,
      "step": 1985,
      "step_time": 381.01982209831476
    },
    {
      "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,
      "entropy": 1.5548095703125,
      "epoch": 0.9511494252873564,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.0663454681634903,
      "learning_rate": 3e-06,
      "loss": 0.0028,
      "step": 1986,
      "step_time": 7.581993393599987
    },
    {
      "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,
      "entropy": 1.3287760416666667,
      "epoch": 0.9516283524904214,
      "fork_head/loss": 2.719486951828003,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.07992683351039886,
      "grad_norm": 0.5454797744750977,
      "learning_rate": 3e-06,
      "loss": 0.6772,
      "step": 1987,
      "step_time": 7.794211506843567
    },
    {
      "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,
      "entropy": 1.5677490234375,
      "epoch": 0.9521072796934866,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05947815254330635,
      "learning_rate": 3e-06,
      "loss": 0.0026,
      "step": 1988,
      "step_time": 7.646610975265503
    },
    {
      "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,
      "entropy": 1.3326822916666667,
      "epoch": 0.9525862068965517,
      "fork_head/loss": 2.7038018703460693,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.0788998007774353,
      "grad_norm": 0.45884162187576294,
      "learning_rate": 3e-06,
      "loss": 0.6733,
      "step": 1989,
      "step_time": 7.267617389559746
    },
    {
      "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,
      "entropy": 1.558349609375,
      "epoch": 0.9530651340996169,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.05711139738559723,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 1990,
      "step_time": 7.733510971069336
    },
    {
      "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,
      "entropy": 1.3255208333333333,
      "epoch": 0.953544061302682,
      "fork_head/loss": 2.686600923538208,
      "fork_head/num_positive": 11.21875,
      "fork_head/pred_top1_mass": 0.08445923775434494,
      "grad_norm": 0.4192532002925873,
      "learning_rate": 3e-06,
      "loss": 0.669,
      "step": 1991,
      "step_time": 7.353450790047646
    },
    {
      "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,
      "entropy": 1.5396728515625,
      "epoch": 0.9540229885057471,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06536472588777542,
      "learning_rate": 3e-06,
      "loss": 0.0025,
      "step": 1992,
      "step_time": 7.323419667780399
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0,
      "completions/max_length": 256.0,
      "completions/max_terminated_length": 256.0,
      "completions/mean_length": 164.744140625,
      "completions/mean_terminated_length": 164.744140625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "entropy": 1.37880859375,
      "epoch": 0.9545019157088123,
      "fork/depth_mean": 0.2969675064086914,
      "fork/depth_std": 0.2506706416606903,
      "fork_head/loss": 2.722534418106079,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.06630251556634903,
      "frac_reward_zero_std": 0.0,
      "grad_norm": 0.45662060379981995,
      "learning_rate": 3e-06,
      "loss": 0.683,
      "num_tokens": 49903979.0,
      "reward": 0.9859260320663452,
      "reward_std": 0.5082409381866455,
      "rewards/code_format_reward/mean": 0.8046875,
      "rewards/code_format_reward/std": 0.29499658942222595,
      "rewards/code_reward/mean": 0.18123850226402283,
      "rewards/code_reward/std": 0.34059739112854004,
      "step": 1993,
      "step_time": 438.7314872890711
    },
    {
      "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,
      "entropy": 1.7078993055555556,
      "epoch": 0.9549808429118773,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09477447718381882,
      "learning_rate": 3e-06,
      "loss": -0.0024,
      "step": 1994,
      "step_time": 7.469042636454105
    },
    {
      "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,
      "entropy": 1.3919921875,
      "epoch": 0.9554597701149425,
      "fork_head/loss": 2.714840888977051,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.07498389482498169,
      "grad_norm": 0.4351447522640228,
      "learning_rate": 3e-06,
      "loss": 0.6808,
      "step": 1995,
      "step_time": 7.886467829346657
    },
    {
      "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,
      "entropy": 1.7081163194444444,
      "epoch": 0.9559386973180076,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.09058965742588043,
      "learning_rate": 3e-06,
      "loss": -0.0026,
      "step": 1996,
      "step_time": 7.349710524082184
    },
    {
      "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,
      "entropy": 1.374609375,
      "epoch": 0.9564176245210728,
      "fork_head/loss": 2.7022721767425537,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.08135698735713959,
      "grad_norm": 0.4036998450756073,
      "learning_rate": 3e-06,
      "loss": 0.6776,
      "step": 1997,
      "step_time": 7.2395289093256
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.6997612847222223,
      "epoch": 0.9568965517241379,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06605171412229538,
      "learning_rate": 3e-06,
      "loss": -0.003,
      "step": 1998,
      "step_time": 7.824833035469055
    },
    {
      "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,
      "entropy": 1.42099609375,
      "epoch": 0.9573754789272031,
      "fork_head/loss": 2.6861302852630615,
      "fork_head/num_positive": 11.28125,
      "fork_head/pred_top1_mass": 0.08745279163122177,
      "grad_norm": 0.3689590394496918,
      "learning_rate": 3e-06,
      "loss": 0.6732,
      "step": 1999,
      "step_time": 7.253084123134613
    },
    {
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.1111111111111111,
      "clip_ratio/low_min": 0.1111111111111111,
      "clip_ratio/region_mean": 0.1111111111111111,
      "entropy": 1.7191840277777777,
      "epoch": 0.9578544061302682,
      "fork_head/loss": null,
      "fork_head/num_positive": null,
      "fork_head/pred_top1_mass": null,
      "grad_norm": 0.06431084871292114,
      "learning_rate": 3e-06,
      "loss": -0.0032,
      "step": 2000,
      "step_time": 7.835825972259045
    }
  ],
  "logging_steps": 1,
  "max_steps": 2000,
  "num_input_tokens_seen": 49903979,
  "num_train_epochs": 1,
  "save_steps": 100,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 0.0,
  "train_batch_size": 2,
  "trial_name": null,
  "trial_params": null
}