{
  "best_global_step": null,
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.21333333333333335,
  "eval_steps": 500,
  "global_step": 200,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.884798833302089,
      "aux_distill/mean_u": 0.2680448395277542,
      "aux_distill/n_active_tok": 22.285714285714285,
      "calib/answer_extract_rate": 0.0546875,
      "calib/avg_num_step_conf": 0.44921875,
      "calib/ece": 0.16135714285714264,
      "calib/final_conf_rate": 0.0546875,
      "calib/format_rate": 0.05078125,
      "calib/frac_conf_gt_0.9": 0.8571428571428571,
      "calib/gap": 0.12215151515151501,
      "calib/mean_conf": 0.9256428571428571,
      "calib/mu_c": 0.9518181818181817,
      "calib/mu_w": 0.8296666666666667,
      "calib/nonempty_final_conf_rate": 0.0546875,
      "calib/nonempty_reasoning_rate": 0.08203125,
      "calib/nonempty_step_conf_rate": 0.078125,
      "calib/pce": 0.15064285714285694,
      "calib/std_conf": 0.1237495207164497,
      "calib/step_conf_rate": 0.078125,
      "calib/step_q_c": 0.8832727272727273,
      "calib/step_q_c_n": 55.0,
      "calib/step_q_gap": 0.042872727272727285,
      "calib/step_q_w": 0.8404,
      "calib/step_q_w_n": 60.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10546875,
      "completions/max_length": 2909.0,
      "completions/max_terminated_length": 2909.0,
      "completions/mean_length": 579.68359375,
      "completions/mean_terminated_length": 648.0305786132812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.0010666666666666667,
      "grad_norm": 2.8604674339294434,
      "learning_rate": 0.0,
      "loss": 0.1208,
      "num_tokens": 255983.0,
      "reward": 0.06787636131048203,
      "reward_std": 0.1474149525165558,
      "rewards/accuracy_reward_step": 0.04296875,
      "rewards/final_brier_reward_step": 0.04200273007154465,
      "rewards/format_reward_step": 0.05078125,
      "step": 1
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.3874275883038838,
      "aux_distill/mean_u": 0.329474540462036,
      "aux_distill/n_active_tok": 27.333333333333332,
      "calib/answer_extract_rate": 0.0546875,
      "calib/avg_num_step_conf": 0.359375,
      "calib/ece": 0.6484615384615385,
      "calib/final_conf_rate": 0.05078125,
      "calib/format_rate": 0.046875,
      "calib/frac_conf_gt_0.9": 0.9230769230769231,
      "calib/gap": 0.016388888888888786,
      "calib/mean_conf": 0.9561538461538462,
      "calib/mu_c": 0.9674999999999999,
      "calib/mu_w": 0.9511111111111111,
      "calib/nonempty_final_conf_rate": 0.05078125,
      "calib/nonempty_reasoning_rate": 0.07421875,
      "calib/nonempty_step_conf_rate": 0.06640625,
      "calib/pce": 0.6484615384615385,
      "calib/std_conf": 0.024663414679817527,
      "calib/step_conf_rate": 0.06640625,
      "calib/step_q_c": 0.8756521739130435,
      "calib/step_q_c_n": 23.0,
      "calib/step_q_gap": 0.015072463768115885,
      "calib/step_q_w": 0.8605797101449276,
      "calib/step_q_w_n": 69.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05078125,
      "completions/max_length": 2949.0,
      "completions/max_terminated_length": 2949.0,
      "completions/mean_length": 785.5390625,
      "completions/mean_terminated_length": 827.5637817382812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0021333333333333334,
      "grad_norm": 1.514235496520996,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.0819,
      "num_tokens": 560369.0,
      "reward": 0.040345899760723114,
      "reward_std": 0.1000000610947609,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.01819179579615593,
      "rewards/format_reward_step": 0.046875,
      "step": 2
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.5618272125720978,
      "aux_distill/mean_u": 0.10840428648371338,
      "aux_distill/n_active_tok": 31.0,
      "calib/answer_extract_rate": 0.02734375,
      "calib/avg_num_step_conf": 0.16015625,
      "calib/ece": 0.6399999999999999,
      "calib/final_conf_rate": 0.0234375,
      "calib/format_rate": 0.015625,
      "calib/frac_conf_gt_0.9": 1.0,
      "calib/gap": -0.012500000000000067,
      "calib/mean_conf": 0.9733333333333333,
      "calib/mu_c": 0.965,
      "calib/mu_w": 0.9775,
      "calib/nonempty_final_conf_rate": 0.0234375,
      "calib/nonempty_reasoning_rate": 0.03125,
      "calib/nonempty_step_conf_rate": 0.0234375,
      "calib/pce": 0.6399999999999999,
      "calib/std_conf": 0.019720265943665403,
      "calib/step_conf_rate": 0.0234375,
      "calib/step_q_c": 0.8381818181818183,
      "calib/step_q_c_n": 11.0,
      "calib/step_q_gap": -0.06198484848484842,
      "calib/step_q_w": 0.9001666666666667,
      "calib/step_q_w_n": 30.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06640625,
      "completions/max_length": 3059.0,
      "completions/max_terminated_length": 3059.0,
      "completions/mean_length": 649.97265625,
      "completions/mean_terminated_length": 696.2050170898438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0032,
      "grad_norm": 0.8159453868865967,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0354,
      "num_tokens": 832018.0,
      "reward": 0.015773242339491844,
      "reward_std": 0.038874901831150055,
      "rewards/accuracy_reward_step": 0.0078125,
      "rewards/final_brier_reward_step": 0.008108984678983688,
      "rewards/format_reward_step": 0.015625,
      "step": 3
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.4118732213974,
      "aux_distill/mean_u": 0.3019396197880862,
      "aux_distill/n_active_tok": 24.0,
      "calib/answer_extract_rate": 0.03125,
      "calib/avg_num_step_conf": 0.109375,
      "calib/ece": 0.7300000000000001,
      "calib/final_conf_rate": 0.03515625,
      "calib/format_rate": 0.01953125,
      "calib/frac_conf_gt_0.9": 0.8888888888888888,
      "calib/gap": -0.04785714285714282,
      "calib/mean_conf": 0.9522222222222223,
      "calib/mu_c": 0.915,
      "calib/mu_w": 0.9628571428571429,
      "calib/nonempty_final_conf_rate": 0.03515625,
      "calib/nonempty_reasoning_rate": 0.03125,
      "calib/nonempty_step_conf_rate": 0.01953125,
      "calib/pce": 0.7300000000000001,
      "calib/std_conf": 0.02657391276244754,
      "calib/step_conf_rate": 0.01953125,
      "calib/step_q_c": 0.892,
      "calib/step_q_c_n": 10.0,
      "calib/step_q_gap": 0.015888888888888952,
      "calib/step_q_w": 0.8761111111111111,
      "calib/step_q_w_n": 18.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08984375,
      "completions/max_length": 2858.0,
      "completions/max_terminated_length": 2858.0,
      "completions/mean_length": 685.22265625,
      "completions/mean_terminated_length": 752.8626708984375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.004266666666666667,
      "grad_norm": 0.6611509919166565,
      "learning_rate": 7.5e-07,
      "loss": 0.04,
      "num_tokens": 1113603.0,
      "reward": 0.018045702949166298,
      "reward_std": 0.04522097483277321,
      "rewards/accuracy_reward_step": 0.0078125,
      "rewards/final_brier_reward_step": 0.008747655898332596,
      "rewards/format_reward_step": 0.01953125,
      "step": 4
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.2684955596923828,
      "aux_distill/mean_u": 0.2648181885977936,
      "aux_distill/n_active_tok": 31.333333333333332,
      "calib/answer_extract_rate": 0.03125,
      "calib/avg_num_step_conf": 0.203125,
      "calib/ece": 0.6962499999999999,
      "calib/final_conf_rate": 0.03125,
      "calib/format_rate": 0.02734375,
      "calib/frac_conf_gt_0.9": 0.875,
      "calib/gap": 0.04499999999999993,
      "calib/mean_conf": 0.94625,
      "calib/mu_c": 0.98,
      "calib/mu_w": 0.935,
      "calib/nonempty_final_conf_rate": 0.03125,
      "calib/nonempty_reasoning_rate": 0.03515625,
      "calib/nonempty_step_conf_rate": 0.03125,
      "calib/pce": 0.6962499999999999,
      "calib/std_conf": 0.07532222447591415,
      "calib/step_conf_rate": 0.03125,
      "calib/step_q_c": 0.9345454545454545,
      "calib/step_q_c_n": 22.0,
      "calib/step_q_gap": 0.07754545454545458,
      "calib/step_q_w": 0.8569999999999999,
      "calib/step_q_w_n": 30.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10546875,
      "completions/max_length": 2978.0,
      "completions/max_terminated_length": 2978.0,
      "completions/mean_length": 621.35546875,
      "completions/mean_terminated_length": 694.61572265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.005333333333333333,
      "grad_norm": 1.1881181001663208,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0332,
      "num_tokens": 1379358.0,
      "reward": 0.022720899432897568,
      "reward_std": 0.06022542715072632,
      "rewards/accuracy_reward_step": 0.0078125,
      "rewards/final_brier_reward_step": 0.010285547003149986,
      "rewards/format_reward_step": 0.02734375,
      "step": 5
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 2.126640498638153,
      "aux_distill/mean_u": 0.2676678498777883,
      "aux_distill/n_active_tok": 14.0,
      "calib/answer_extract_rate": 0.04296875,
      "calib/avg_num_step_conf": 0.21484375,
      "calib/ece": 0.4745454545454545,
      "calib/final_conf_rate": 0.04296875,
      "calib/format_rate": 0.03515625,
      "calib/frac_conf_gt_0.9": 0.7272727272727273,
      "calib/gap": 0.04933333333333334,
      "calib/mean_conf": 0.929090909090909,
      "calib/mu_c": 0.9560000000000001,
      "calib/mu_w": 0.9066666666666667,
      "calib/nonempty_final_conf_rate": 0.04296875,
      "calib/nonempty_reasoning_rate": 0.046875,
      "calib/nonempty_step_conf_rate": 0.04296875,
      "calib/pce": 0.4745454545454545,
      "calib/std_conf": 0.07774720637536474,
      "calib/step_conf_rate": 0.04296875,
      "calib/step_q_c": 0.915,
      "calib/step_q_c_n": 24.0,
      "calib/step_q_gap": 0.08887096774193559,
      "calib/step_q_w": 0.8261290322580644,
      "calib/step_q_w_n": 31.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06640625,
      "completions/max_length": 2847.0,
      "completions/max_terminated_length": 2847.0,
      "completions/mean_length": 622.8046875,
      "completions/mean_terminated_length": 667.1046142578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0064,
      "grad_norm": 1.6805046796798706,
      "learning_rate": 1.25e-06,
      "loss": 0.075,
      "num_tokens": 1644748.0,
      "reward": 0.03894374892115593,
      "reward_std": 0.08292990177869797,
      "rewards/accuracy_reward_step": 0.0234375,
      "rewards/final_brier_reward_step": 0.01929374970495701,
      "rewards/format_reward_step": 0.03515625,
      "step": 6
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.2389460972377233,
      "aux_distill/mean_u": 0.28400473774278345,
      "aux_distill/n_active_tok": 34.285714285714285,
      "calib/answer_extract_rate": 0.05078125,
      "calib/avg_num_step_conf": 0.32421875,
      "calib/ece": 0.7879166666666666,
      "calib/final_conf_rate": 0.046875,
      "calib/format_rate": 0.0390625,
      "calib/frac_conf_gt_0.9": 0.9166666666666666,
      "calib/gap": 0.036499999999999866,
      "calib/mean_conf": 0.9545833333333333,
      "calib/mu_c": 0.985,
      "calib/mu_w": 0.9485000000000001,
      "calib/nonempty_final_conf_rate": 0.046875,
      "calib/nonempty_reasoning_rate": 0.0546875,
      "calib/nonempty_step_conf_rate": 0.046875,
      "calib/pce": 0.7879166666666666,
      "calib/std_conf": 0.048796274334101455,
      "calib/step_conf_rate": 0.046875,
      "calib/step_q_c": 0.869,
      "calib/step_q_c_n": 20.0,
      "calib/step_q_gap": -0.005761904761904746,
      "calib/step_q_w": 0.8747619047619047,
      "calib/step_q_w_n": 63.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.10546875,
      "completions/max_length": 3040.0,
      "completions/max_terminated_length": 3040.0,
      "completions/mean_length": 722.31640625,
      "completions/mean_terminated_length": 807.4803466796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.007466666666666667,
      "grad_norm": 1.6011173725128174,
      "learning_rate": 1.5e-06,
      "loss": 0.0944,
      "num_tokens": 1937085.0,
      "reward": 0.031165577471256256,
      "reward_std": 0.08389748632907867,
      "rewards/accuracy_reward_step": 0.01171875,
      "rewards/final_brier_reward_step": 0.011549903079867363,
      "rewards/format_reward_step": 0.0390625,
      "step": 7
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.5897511541843414,
      "aux_distill/mean_u": 0.29507590785209553,
      "aux_distill/n_active_tok": 23.0,
      "calib/answer_extract_rate": 0.04296875,
      "calib/avg_num_step_conf": 0.1875,
      "calib/ece": 0.5155555555555555,
      "calib/final_conf_rate": 0.03515625,
      "calib/format_rate": 0.0234375,
      "calib/frac_conf_gt_0.9": 0.6666666666666666,
      "calib/gap": -0.012000000000000122,
      "calib/mean_conf": 0.9266666666666666,
      "calib/mu_c": 0.9199999999999999,
      "calib/mu_w": 0.932,
      "calib/nonempty_final_conf_rate": 0.03515625,
      "calib/nonempty_reasoning_rate": 0.046875,
      "calib/nonempty_step_conf_rate": 0.03515625,
      "calib/pce": 0.4988888888888889,
      "calib/std_conf": 0.03399346342395188,
      "calib/step_conf_rate": 0.03515625,
      "calib/step_q_c": 0.7973913043478259,
      "calib/step_q_c_n": 23.0,
      "calib/step_q_gap": -0.043008695652174134,
      "calib/step_q_w": 0.8404,
      "calib/step_q_w_n": 25.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05078125,
      "completions/max_length": 2924.0,
      "completions/max_terminated_length": 2924.0,
      "completions/mean_length": 670.5859375,
      "completions/mean_terminated_length": 706.4608764648438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.008533333333333334,
      "grad_norm": 1.4276400804519653,
      "learning_rate": 1.75e-06,
      "loss": 0.0649,
      "num_tokens": 2215267.0,
      "reward": 0.024509180337190628,
      "reward_std": 0.05961586534976959,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.009955858811736107,
      "rewards/format_reward_step": 0.0234375,
      "step": 8
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.7846717834472656,
      "aux_distill/mean_u": 0.2995953277024473,
      "aux_distill/n_active_tok": 16.0,
      "calib/answer_extract_rate": 0.0390625,
      "calib/avg_num_step_conf": 0.109375,
      "calib/ece": 0.9800000000000001,
      "calib/final_conf_rate": 0.01953125,
      "calib/format_rate": 0.015625,
      "calib/frac_conf_gt_0.9": 1.0,
      "calib/mean_conf": 0.9800000000000001,
      "calib/mu_c": NaN,
      "calib/mu_w": 0.9800000000000001,
      "calib/nonempty_final_conf_rate": 0.01953125,
      "calib/nonempty_reasoning_rate": 0.05078125,
      "calib/nonempty_step_conf_rate": 0.02734375,
      "calib/pce": 0.9800000000000001,
      "calib/std_conf": 0.016733200530681527,
      "calib/step_conf_rate": 0.02734375,
      "calib/step_q_w": 0.8967857142857143,
      "calib/step_q_w_n": 28.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06640625,
      "completions/max_length": 3060.0,
      "completions/max_terminated_length": 3060.0,
      "completions/mean_length": 653.74609375,
      "completions/mean_terminated_length": 700.246826171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0096,
      "grad_norm": 2.4460411071777344,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0439,
      "num_tokens": 2490162.0,
      "reward": 0.008043359033763409,
      "reward_std": 0.02275005541741848,
      "rewards/accuracy_reward_step": 0.0,
      "rewards/final_brier_reward_step": 0.0004617187660187483,
      "rewards/format_reward_step": 0.015625,
      "step": 9
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.5699462890625,
      "aux_distill/mean_u": 0.14541823796785602,
      "aux_distill/n_active_tok": 4.0,
      "calib/answer_extract_rate": 0.03515625,
      "calib/avg_num_step_conf": 0.11328125,
      "calib/ece": 0.40714285714285725,
      "calib/final_conf_rate": 0.02734375,
      "calib/format_rate": 0.01171875,
      "calib/frac_conf_gt_0.9": 1.0,
      "calib/gap": 0.014999999999999902,
      "calib/mean_conf": 0.9785714285714286,
      "calib/mu_c": 0.985,
      "calib/mu_w": 0.9700000000000001,
      "calib/nonempty_final_conf_rate": 0.02734375,
      "calib/nonempty_reasoning_rate": 0.0546875,
      "calib/nonempty_step_conf_rate": 0.03125,
      "calib/pce": 0.40714285714285725,
      "calib/std_conf": 0.01551825784457175,
      "calib/step_conf_rate": 0.03125,
      "calib/step_q_c": 0.985,
      "calib/step_q_c_n": 2.0,
      "calib/step_q_gap": 0.33018518518518514,
      "calib/step_q_w": 0.6548148148148148,
      "calib/step_q_w_n": 27.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08203125,
      "completions/max_length": 3069.0,
      "completions/max_terminated_length": 3069.0,
      "completions/mean_length": 766.7890625,
      "completions/mean_terminated_length": 835.3106079101562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.010666666666666666,
      "grad_norm": 4.1238274574279785,
      "learning_rate": 2.25e-06,
      "loss": 0.0581,
      "num_tokens": 2793260.0,
      "reward": 0.01581699214875698,
      "reward_std": 0.04473721235990524,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.004290234297513962,
      "rewards/format_reward_step": 0.01171875,
      "step": 10
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.314156413078308,
      "aux_distill/mean_u": 0.16583724328311117,
      "aux_distill/n_active_tok": 17.6,
      "calib/answer_extract_rate": 0.05859375,
      "calib/avg_num_step_conf": 0.3203125,
      "calib/ece": 0.6253333333333335,
      "calib/final_conf_rate": 0.05859375,
      "calib/format_rate": 0.05078125,
      "calib/frac_conf_gt_0.9": 0.7333333333333333,
      "calib/gap": 0.10636363636363633,
      "calib/mean_conf": 0.8919999999999999,
      "calib/mu_c": 0.97,
      "calib/mu_w": 0.8636363636363636,
      "calib/nonempty_final_conf_rate": 0.05859375,
      "calib/nonempty_reasoning_rate": 0.078125,
      "calib/nonempty_step_conf_rate": 0.0703125,
      "calib/pce": 0.6253333333333335,
      "calib/std_conf": 0.14043741191956888,
      "calib/step_conf_rate": 0.0703125,
      "calib/step_q_c": 0.663125,
      "calib/step_q_c_n": 16.0,
      "calib/step_q_gap": -0.10748409090909095,
      "calib/step_q_w": 0.7706090909090909,
      "calib/step_q_w_n": 66.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07421875,
      "completions/max_length": 2907.0,
      "completions/max_terminated_length": 2907.0,
      "completions/mean_length": 641.875,
      "completions/mean_terminated_length": 693.3333129882812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.011733333333333333,
      "grad_norm": 1.7187901735305786,
      "learning_rate": 2.5e-06,
      "loss": 0.0682,
      "num_tokens": 3062060.0,
      "reward": 0.042952343821525574,
      "reward_std": 0.08737139403820038,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.019498437643051147,
      "rewards/format_reward_step": 0.05078125,
      "step": 11
    },
    {
      "aux_distill/lambda": 0.29999999999999993,
      "aux_distill/loss": 1.278182077407837,
      "aux_distill/mean_u": 0.24745318100648223,
      "aux_distill/n_active_tok": 21.6,
      "calib/answer_extract_rate": 0.0546875,
      "calib/avg_num_step_conf": 0.3515625,
      "calib/ece": 0.37499999999999994,
      "calib/final_conf_rate": 0.0390625,
      "calib/format_rate": 0.03125,
      "calib/frac_conf_gt_0.9": 0.7,
      "calib/gap": -0.012499999999999956,
      "calib/mean_conf": 0.945,
      "calib/mu_c": 0.9400000000000001,
      "calib/mu_w": 0.9525,
      "calib/nonempty_final_conf_rate": 0.0390625,
      "calib/nonempty_reasoning_rate": 0.078125,
      "calib/nonempty_step_conf_rate": 0.0625,
      "calib/pce": 0.35999999999999993,
      "calib/std_conf": 0.04341658669218481,
      "calib/step_conf_rate": 0.0625,
      "calib/step_q_c": 0.8353846153846152,
      "calib/step_q_c_n": 39.0,
      "calib/step_q_gap": 0.016953242835595628,
      "calib/step_q_w": 0.8184313725490195,
      "calib/step_q_w_n": 51.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08203125,
      "completions/max_length": 3059.0,
      "completions/max_terminated_length": 3059.0,
      "completions/mean_length": 605.703125,
      "completions/mean_terminated_length": 659.8297729492188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0128,
      "grad_norm": 1.3734772205352783,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.0688,
      "num_tokens": 3321296.0,
      "reward": 0.03764042630791664,
      "reward_std": 0.10646320879459381,
      "rewards/accuracy_reward_step": 0.0234375,
      "rewards/final_brier_reward_step": 0.02059336006641388,
      "rewards/format_reward_step": 0.03125,
      "step": 12
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.0337275192141533,
      "aux_distill/mean_u": 0.1903723552664163,
      "aux_distill/n_active_tok": 25.0,
      "calib/answer_extract_rate": 0.03515625,
      "calib/avg_num_step_conf": 0.30078125,
      "calib/ece": 0.43499999999999994,
      "calib/final_conf_rate": 0.0390625,
      "calib/format_rate": 0.02734375,
      "calib/frac_conf_gt_0.9": 0.4,
      "calib/gap": 0.3252380952380952,
      "calib/mean_conf": 0.679,
      "calib/mu_c": 0.9066666666666667,
      "calib/mu_w": 0.5814285714285715,
      "calib/nonempty_final_conf_rate": 0.0390625,
      "calib/nonempty_reasoning_rate": 0.0546875,
      "calib/nonempty_step_conf_rate": 0.05078125,
      "calib/pce": 0.407,
      "calib/std_conf": 0.37479194228264834,
      "calib/step_conf_rate": 0.05078125,
      "calib/step_q_c": 0.7618181818181817,
      "calib/step_q_c_n": 11.0,
      "calib/step_q_gap": 0.15924242424242419,
      "calib/step_q_w": 0.6025757575757575,
      "calib/step_q_w_n": 66.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08984375,
      "completions/max_length": 3068.0,
      "completions/max_terminated_length": 3068.0,
      "completions/mean_length": 673.48828125,
      "completions/mean_terminated_length": 739.969970703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.013866666666666666,
      "grad_norm": 1.3304049968719482,
      "learning_rate": 3e-06,
      "loss": 0.0784,
      "num_tokens": 3598301.0,
      "reward": 0.028053514659404755,
      "reward_std": 0.07934731990098953,
      "rewards/accuracy_reward_step": 0.01171875,
      "rewards/final_brier_reward_step": 0.01704453118145466,
      "rewards/format_reward_step": 0.02734375,
      "step": 13
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.1433793703715007,
      "aux_distill/mean_u": 0.4785981771689191,
      "aux_distill/n_active_tok": 9.333333333333334,
      "calib/answer_extract_rate": 0.02734375,
      "calib/avg_num_step_conf": 0.265625,
      "calib/ece": 0.6233333333333333,
      "calib/final_conf_rate": 0.01171875,
      "calib/format_rate": 0.00390625,
      "calib/frac_conf_gt_0.9": 0.3333333333333333,
      "calib/gap": -0.31499999999999995,
      "calib/mean_conf": 0.44999999999999996,
      "calib/mu_c": 0.24,
      "calib/mu_w": 0.5549999999999999,
      "calib/nonempty_final_conf_rate": 0.01171875,
      "calib/nonempty_reasoning_rate": 0.0703125,
      "calib/nonempty_step_conf_rate": 0.05078125,
      "calib/pce": 0.37,
      "calib/std_conf": 0.34765883660086455,
      "calib/step_conf_rate": 0.05078125,
      "calib/step_q_c": 0.262,
      "calib/step_q_c_n": 5.0,
      "calib/step_q_gap": -0.07657142857142857,
      "calib/step_q_w": 0.3385714285714286,
      "calib/step_q_w_n": 63.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.07421875,
      "completions/max_length": 3059.0,
      "completions/max_terminated_length": 3059.0,
      "completions/mean_length": 624.95703125,
      "completions/mean_terminated_length": 675.0590209960938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.014933333333333333,
      "grad_norm": 1.457983374595642,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.0601,
      "num_tokens": 3863690.0,
      "reward": 0.004731249995529652,
      "reward_std": 0.01338199619203806,
      "rewards/accuracy_reward_step": 0.00390625,
      "rewards/final_brier_reward_step": 0.0016499999910593033,
      "rewards/format_reward_step": 0.00390625,
      "step": 14
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.2922158042589824,
      "aux_distill/mean_u": 0.3207821715853844,
      "aux_distill/n_active_tok": 10.0,
      "calib/answer_extract_rate": 0.04296875,
      "calib/avg_num_step_conf": 0.3125,
      "calib/ece": 0.4483333333333333,
      "calib/final_conf_rate": 0.0234375,
      "calib/format_rate": 0.0234375,
      "calib/frac_conf_gt_0.9": 0.16666666666666666,
      "calib/gap": -0.14600000000000002,
      "calib/mean_conf": 0.36166666666666664,
      "calib/mu_c": 0.24,
      "calib/mu_w": 0.386,
      "calib/nonempty_final_conf_rate": 0.0234375,
      "calib/nonempty_reasoning_rate": 0.07421875,
      "calib/nonempty_step_conf_rate": 0.0625,
      "calib/pce": 0.32166666666666666,
      "calib/std_conf": 0.33012203130896245,
      "calib/step_conf_rate": 0.0625,
      "calib/step_q_c": 0.42400000000000004,
      "calib/step_q_c_n": 5.0,
      "calib/step_q_gap": -0.0019999999999999463,
      "calib/step_q_w": 0.426,
      "calib/step_q_w_n": 75.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05859375,
      "completions/max_length": 3057.0,
      "completions/max_terminated_length": 3057.0,
      "completions/mean_length": 497.453125,
      "completions/mean_terminated_length": 528.4149780273438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.016,
      "grad_norm": 1.143528938293457,
      "learning_rate": 3.5e-06,
      "loss": 0.0974,
      "num_tokens": 4098918.0,
      "reward": 0.021565038710832596,
      "reward_std": 0.06099514290690422,
      "rewards/accuracy_reward_step": 0.00390625,
      "rewards/final_brier_reward_step": 0.01578632742166519,
      "rewards/format_reward_step": 0.0234375,
      "step": 15
    },
    {
      "aux_distill/lambda": 0.29999999999999993,
      "aux_distill/loss": 0.8635341763496399,
      "aux_distill/mean_u": 0.31292070845885867,
      "aux_distill/n_active_tok": 51.2,
      "calib/answer_extract_rate": 0.0546875,
      "calib/avg_num_step_conf": 0.7578125,
      "calib/ece": 0.37249999999999994,
      "calib/final_conf_rate": 0.046875,
      "calib/format_rate": 0.0390625,
      "calib/frac_conf_gt_0.9": 0.08333333333333333,
      "calib/gap": 0.10875000000000001,
      "calib/mean_conf": 0.3475,
      "calib/mu_c": 0.42,
      "calib/mu_w": 0.31124999999999997,
      "calib/nonempty_final_conf_rate": 0.046875,
      "calib/nonempty_reasoning_rate": 0.11328125,
      "calib/nonempty_step_conf_rate": 0.09765625,
      "calib/pce": 0.1933333333333333,
      "calib/std_conf": 0.34285869295284516,
      "calib/step_conf_rate": 0.09765625,
      "calib/step_q_c": 0.42625,
      "calib/step_q_c_n": 16.0,
      "calib/step_q_gap": 0.17636797752808991,
      "calib/step_q_w": 0.2498820224719101,
      "calib/step_q_w_n": 178.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.078125,
      "completions/max_length": 3002.0,
      "completions/max_terminated_length": 3002.0,
      "completions/mean_length": 624.296875,
      "completions/mean_terminated_length": 677.203369140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.017066666666666667,
      "grad_norm": 1.7086925506591797,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0955,
      "num_tokens": 4367586.0,
      "reward": 0.04019277170300484,
      "reward_std": 0.0873384177684784,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.025698047131299973,
      "rewards/format_reward_step": 0.0390625,
      "step": 16
    },
    {
      "aux_distill/lambda": 0.3,
      "aux_distill/loss": 1.1117850542068481,
      "aux_distill/mean_u": 0.3679425643558005,
      "aux_distill/n_active_tok": 38.0,
      "calib/answer_extract_rate": 0.0546875,
      "calib/avg_num_step_conf": 0.54296875,
      "calib/ece": 0.4572727272727272,
      "calib/final_conf_rate": 0.04296875,
      "calib/format_rate": 0.0234375,
      "calib/frac_conf_gt_0.9": 0.2727272727272727,
      "calib/mean_conf": 0.4572727272727272,
      "calib/mu_c": NaN,
      "calib/mu_w": 0.4572727272727272,
      "calib/nonempty_final_conf_rate": 0.04296875,
      "calib/nonempty_reasoning_rate": 0.1015625,
      "calib/nonempty_step_conf_rate": 0.07421875,
      "calib/pce": 0.4572727272727272,
      "calib/std_conf": 0.4130395294551866,
      "calib/step_conf_rate": 0.07421875,
      "calib/step_q_w": 0.3162045755422472,
      "calib/step_q_w_n": 139.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0703125,
      "completions/max_length": 3053.0,
      "completions/max_terminated_length": 3053.0,
      "completions/mean_length": 555.11328125,
      "completions/mean_terminated_length": 597.0966796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.018133333333333335,
      "grad_norm": 0.8777731657028198,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0304,
      "num_tokens": 4613223.0,
      "reward": 0.02120097726583481,
      "reward_std": 0.05278696119785309,
      "rewards/accuracy_reward_step": 0.0,
      "rewards/final_brier_reward_step": 0.018964452669024467,
      "rewards/format_reward_step": 0.0234375,
      "step": 17
    },
    {
      "aux_distill/lambda": 0.29999999999999993,
      "aux_distill/loss": 0.9870762129624685,
      "aux_distill/mean_u": 0.3191481519811337,
      "aux_distill/n_active_tok": 22.0,
      "calib/answer_extract_rate": 0.05078125,
      "calib/avg_num_step_conf": 0.3203125,
      "calib/ece": 0.3189693576383376,
      "calib/final_conf_rate": 0.0546875,
      "calib/format_rate": 0.02734375,
      "calib/frac_conf_gt_0.9": 0.14285714285714285,
      "calib/gap": -0.1487975839113939,
      "calib/mean_conf": 0.34754078620976614,
      "calib/mu_c": 0.22,
      "calib/mu_w": 0.3687975839113939,
      "calib/nonempty_final_conf_rate": 0.0546875,
      "calib/nonempty_reasoning_rate": 0.08984375,
      "calib/nonempty_step_conf_rate": 0.06640625,
      "calib/pce": 0.26182650049548045,
      "calib/std_conf": 0.3387916578440022,
      "calib/step_conf_rate": 0.06640625,
      "calib/step_q_c": 0.10200000000000001,
      "calib/step_q_c_n": 5.0,
      "calib/step_q_gap": -0.19884221498891771,
      "calib/step_q_w": 0.3008422149889177,
      "calib/step_q_w_n": 77.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0546875,
      "completions/max_length": 3008.0,
      "completions/max_terminated_length": 3008.0,
      "completions/mean_length": 602.00390625,
      "completions/mean_terminated_length": 636.83056640625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0192,
      "grad_norm": 2.490189790725708,
      "learning_rate": 4.25e-06,
      "loss": 0.0671,
      "num_tokens": 4878056.0,
      "reward": 0.02656581997871399,
      "reward_std": 0.06706564873456955,
      "rewards/accuracy_reward_step": 0.0078125,
      "rewards/final_brier_reward_step": 0.01797538995742798,
      "rewards/format_reward_step": 0.02734375,
      "step": 18
    },
    {
      "aux_distill/lambda": 0.29999999999999993,
      "aux_distill/loss": 0.9927313059568406,
      "aux_distill/mean_u": 0.30662449710636475,
      "aux_distill/n_active_tok": 31.8,
      "calib/answer_extract_rate": 0.09765625,
      "calib/avg_num_step_conf": 1.18359375,
      "calib/ece": 0.276476284445,
      "calib/final_conf_rate": 0.078125,
      "calib/format_rate": 0.0625,
      "calib/frac_conf_gt_0.9": 0.05,
      "calib/gap": 0.02218476339803918,
      "calib/mean_conf": 0.22447628444499998,
      "calib/mu_c": 0.24333333333333332,
      "calib/mu_w": 0.22114856993529414,
      "calib/nonempty_final_conf_rate": 0.078125,
      "calib/nonempty_reasoning_rate": 0.21875,
      "calib/nonempty_step_conf_rate": 0.19921875,
      "calib/pce": 0.17547628444500002,
      "calib/std_conf": 0.246312235256632,
      "calib/step_conf_rate": 0.19921875,
      "calib/step_q_c": 0.23055555555555557,
      "calib/step_q_c_n": 18.0,
      "calib/step_q_gap": 0.02889905699921752,
      "calib/step_q_w": 0.20165649855633805,
      "calib/step_q_w_n": 284.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06640625,
      "completions/max_length": 2978.0,
      "completions/max_terminated_length": 2978.0,
      "completions/mean_length": 639.859375,
      "completions/mean_terminated_length": 685.3723754882812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.020266666666666665,
      "grad_norm": 2.305959939956665,
      "learning_rate": 4.5e-06,
      "loss": 0.1893,
      "num_tokens": 5146620.0,
      "reward": 0.06604452431201935,
      "reward_std": 0.1502785086631775,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.05396406352519989,
      "rewards/format_reward_step": 0.0625,
      "step": 19
    },
    {
      "aux_distill/lambda": 0.2999999999999999,
      "aux_distill/loss": 1.0178489529568215,
      "aux_distill/mean_u": 0.44129956033095913,
      "aux_distill/n_active_tok": 34.08695652173913,
      "calib/answer_extract_rate": 0.2421875,
      "calib/avg_num_step_conf": 1.69921875,
      "calib/ece": 0.1661569117133794,
      "calib/final_conf_rate": 0.23046875,
      "calib/format_rate": 0.1484375,
      "calib/frac_conf_gt_0.9": 0.01694915254237288,
      "calib/gap": -0.05294987261431322,
      "calib/mean_conf": 0.21577022886999955,
      "calib/mu_c": 0.17,
      "calib/mu_w": 0.22294987261431323,
      "calib/nonempty_final_conf_rate": 0.23046875,
      "calib/nonempty_reasoning_rate": 0.375,
      "calib/nonempty_step_conf_rate": 0.29296875,
      "calib/pce": 0.12316696012219798,
      "calib/std_conf": 0.20314595471244742,
      "calib/step_conf_rate": 0.29296875,
      "calib/step_q_c": 0.21828571428571428,
      "calib/step_q_c_n": 35.0,
      "calib/step_q_gap": -0.005639482913952498,
      "calib/step_q_w": 0.22392519719966678,
      "calib/step_q_w_n": 362.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0390625,
      "completions/max_length": 2923.0,
      "completions/max_terminated_length": 2923.0,
      "completions/mean_length": 573.02734375,
      "completions/mean_terminated_length": 596.3211059570312,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.021333333333333333,
      "grad_norm": 2.455202579498291,
      "learning_rate": 4.75e-06,
      "loss": 0.2093,
      "num_tokens": 5398187.0,
      "reward": 0.15179908275604248,
      "reward_std": 0.2831573188304901,
      "rewards/accuracy_reward_step": 0.03515625,
      "rewards/final_brier_reward_step": 0.12000440806150436,
      "rewards/format_reward_step": 0.1484375,
      "step": 20
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9270459283143282,
      "aux_distill/mean_u": 0.35117479873859514,
      "aux_distill/n_active_tok": 41.25,
      "calib/answer_extract_rate": 0.4296875,
      "calib/avg_num_step_conf": 2.91015625,
      "calib/ece": 0.1688007413558772,
      "calib/final_conf_rate": 0.37109375,
      "calib/format_rate": 0.24609375,
      "calib/frac_conf_gt_0.9": 0.010526315789473684,
      "calib/gap": 0.02934422203383541,
      "calib/mean_conf": 0.2818624165388597,
      "calib/mu_c": 0.30750000000000005,
      "calib/mu_w": 0.27815577796616464,
      "calib/nonempty_final_conf_rate": 0.37109375,
      "calib/nonempty_reasoning_rate": 0.62109375,
      "calib/nonempty_step_conf_rate": 0.48828125,
      "calib/pce": 0.1621736842105263,
      "calib/std_conf": 0.2205194295691477,
      "calib/step_conf_rate": 0.48828125,
      "calib/step_q_c": 0.21250000000000002,
      "calib/step_q_c_n": 64.0,
      "calib/step_q_gap": -0.010935235866823712,
      "calib/step_q_w": 0.22343523586682373,
      "calib/step_q_w_n": 681.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.02734375,
      "completions/max_length": 2797.0,
      "completions/max_terminated_length": 2797.0,
      "completions/mean_length": 480.07421875,
      "completions/mean_terminated_length": 493.57025146484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 4.0,
      "epoch": 0.0224,
      "grad_norm": 2.3599891662597656,
      "learning_rate": 5e-06,
      "loss": 0.35,
      "num_tokens": 5624046.0,
      "reward": 0.2571793496608734,
      "reward_std": 0.3987439274787903,
      "rewards/accuracy_reward_step": 0.05859375,
      "rewards/final_brier_reward_step": 0.20967116951942444,
      "rewards/format_reward_step": 0.24609375,
      "step": 21
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9610700905323029,
      "aux_distill/mean_u": 0.4553320713262558,
      "aux_distill/n_active_tok": 53.25,
      "calib/answer_extract_rate": 0.58984375,
      "calib/avg_num_step_conf": 3.44921875,
      "calib/ece": 0.17043563629820027,
      "calib/final_conf_rate": 0.56640625,
      "calib/format_rate": 0.4375,
      "calib/frac_conf_gt_0.9": 0.013793103448275862,
      "calib/gap": 0.04415089183342563,
      "calib/mean_conf": 0.2646309511539849,
      "calib/mu_c": 0.30391002044027393,
      "calib/mu_w": 0.2597591286068483,
      "calib/nonempty_final_conf_rate": 0.56640625,
      "calib/nonempty_reasoning_rate": 0.78515625,
      "calib/nonempty_step_conf_rate": 0.66796875,
      "calib/pce": 0.16236087993298912,
      "calib/std_conf": 0.21311021832354388,
      "calib/step_conf_rate": 0.66796875,
      "calib/step_q_c": 0.31052220008859993,
      "calib/step_q_c_n": 61.0,
      "calib/step_q_gap": 0.0386094263659722,
      "calib/step_q_w": 0.27191277372262773,
      "calib/step_q_w_n": 822.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03515625,
      "completions/max_length": 3012.0,
      "completions/max_terminated_length": 3012.0,
      "completions/mean_length": 442.33203125,
      "completions/mean_terminated_length": 458.44940185546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 4.0,
      "epoch": 0.023466666666666667,
      "grad_norm": 3.0896050930023193,
      "learning_rate": 4.9722222222222224e-06,
      "loss": 0.3149,
      "num_tokens": 5839099.0,
      "reward": 0.4424505829811096,
      "reward_std": 0.483396053314209,
      "rewards/accuracy_reward_step": 0.078125,
      "rewards/final_brier_reward_step": 0.36927616596221924,
      "rewards/format_reward_step": 0.4375,
      "step": 22
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9353764075785875,
      "aux_distill/mean_u": 0.35294600129183773,
      "aux_distill/n_active_tok": 72.625,
      "calib/answer_extract_rate": 0.625,
      "calib/avg_num_step_conf": 3.7578125,
      "calib/ece": 0.27355118955660707,
      "calib/final_conf_rate": 0.63671875,
      "calib/format_rate": 0.47265625,
      "calib/frac_conf_gt_0.9": 0.03680981595092025,
      "calib/gap": 0.028277245810025498,
      "calib/mean_conf": 0.35243953311488924,
      "calib/mu_c": 0.37846153846153846,
      "calib/mu_w": 0.35018429265151296,
      "calib/nonempty_final_conf_rate": 0.63671875,
      "calib/nonempty_reasoning_rate": 0.84765625,
      "calib/nonempty_step_conf_rate": 0.71484375,
      "calib/pce": 0.2731180607222512,
      "calib/std_conf": 0.2499565020687635,
      "calib/step_conf_rate": 0.71484375,
      "calib/step_q_c": 0.33299999999999996,
      "calib/step_q_c_n": 50.0,
      "calib/step_q_gap": -0.0760557458927632,
      "calib/step_q_w": 0.40905574589276317,
      "calib/step_q_w_n": 912.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03125,
      "completions/max_length": 3058.0,
      "completions/max_terminated_length": 3058.0,
      "completions/mean_length": 468.2890625,
      "completions/mean_terminated_length": 483.3951416015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 6.0,
      "epoch": 0.024533333333333334,
      "grad_norm": 3.023226737976074,
      "learning_rate": 4.944444444444445e-06,
      "loss": 0.2974,
      "num_tokens": 6062917.0,
      "reward": 0.45471012592315674,
      "reward_std": 0.4595857262611389,
      "rewards/accuracy_reward_step": 0.05078125,
      "rewards/final_brier_reward_step": 0.3859827518463135,
      "rewards/format_reward_step": 0.47265625,
      "step": 23
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.918319652788341,
      "aux_distill/mean_u": 0.3763038000716115,
      "aux_distill/n_active_tok": 54.5,
      "calib/answer_extract_rate": 0.7421875,
      "calib/avg_num_step_conf": 3.53125,
      "calib/ece": 0.25175623019718174,
      "calib/final_conf_rate": 0.71875,
      "calib/format_rate": 0.57421875,
      "calib/frac_conf_gt_0.9": 0.021739130434782608,
      "calib/gap": -0.017374254595942218,
      "calib/mean_conf": 0.35396275193631216,
      "calib/mu_c": 0.3385714285714286,
      "calib/mu_w": 0.3559456831673708,
      "calib/nonempty_final_conf_rate": 0.71875,
      "calib/nonempty_reasoning_rate": 0.87890625,
      "calib/nonempty_step_conf_rate": 0.75390625,
      "calib/pce": 0.2457942736754426,
      "calib/std_conf": 0.21846927693240498,
      "calib/step_conf_rate": 0.75390625,
      "calib/step_q_c": 0.4223333333333334,
      "calib/step_q_c_n": 90.0,
      "calib/step_q_gap": 0.05739322645372652,
      "calib/step_q_w": 0.3649401068796069,
      "calib/step_q_w_n": 814.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03515625,
      "completions/max_length": 2973.0,
      "completions/max_terminated_length": 2973.0,
      "completions/mean_length": 432.51171875,
      "completions/mean_terminated_length": 448.2712707519531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 15.0,
      "epoch": 0.0256,
      "grad_norm": 1.662244439125061,
      "learning_rate": 4.9166666666666665e-06,
      "loss": 0.3306,
      "num_tokens": 6278152.0,
      "reward": 0.5555780529975891,
      "reward_std": 0.4755654036998749,
      "rewards/accuracy_reward_step": 0.08203125,
      "rewards/final_brier_reward_step": 0.45490607619285583,
      "rewards/format_reward_step": 0.57421875,
      "step": 24
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.940506137907505,
      "aux_distill/mean_u": 0.37342990083807837,
      "aux_distill/n_active_tok": 66.0,
      "calib/answer_extract_rate": 0.78125,
      "calib/avg_num_step_conf": 3.96875,
      "calib/ece": 0.2819637535040166,
      "calib/final_conf_rate": 0.80078125,
      "calib/format_rate": 0.703125,
      "calib/frac_conf_gt_0.9": 0.03414634146341464,
      "calib/gap": -0.0317800534933525,
      "calib/mean_conf": 0.3949393632601142,
      "calib/mu_c": 0.36750000000000005,
      "calib/mu_w": 0.39928005349335255,
      "calib/nonempty_final_conf_rate": 0.80078125,
      "calib/nonempty_reasoning_rate": 0.91796875,
      "calib/nonempty_step_conf_rate": 0.84375,
      "calib/pce": 0.2701588754552361,
      "calib/std_conf": 0.23788335954319495,
      "calib/step_conf_rate": 0.84375,
      "calib/step_q_c": 0.34856725146198836,
      "calib/step_q_c_n": 171.0,
      "calib/step_q_gap": -0.08577948073599456,
      "calib/step_q_w": 0.4343467321979829,
      "calib/step_q_w_n": 845.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.02734375,
      "completions/max_length": 2925.0,
      "completions/max_terminated_length": 2925.0,
      "completions/mean_length": 364.23046875,
      "completions/mean_terminated_length": 374.4698791503906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 24.0,
      "epoch": 0.02666666666666667,
      "grad_norm": 1.7440599203109741,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.3463,
      "num_tokens": 6474619.0,
      "reward": 0.6718124151229858,
      "reward_std": 0.4372907280921936,
      "rewards/accuracy_reward_step": 0.11328125,
      "rewards/final_brier_reward_step": 0.5272184610366821,
      "rewards/format_reward_step": 0.703125,
      "step": 25
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9505161102861166,
      "aux_distill/mean_u": 0.43465847075130665,
      "aux_distill/n_active_tok": 70.75,
      "calib/answer_extract_rate": 0.87890625,
      "calib/avg_num_step_conf": 4.234375,
      "calib/ece": 0.35119691629955946,
      "calib/final_conf_rate": 0.88671875,
      "calib/format_rate": 0.80859375,
      "calib/frac_conf_gt_0.9": 0.048458149779735685,
      "calib/gap": -0.11122185534591206,
      "calib/mean_conf": 0.4052057268722467,
      "calib/mu_c": 0.3013333333333333,
      "calib/mu_w": 0.41255518867924534,
      "calib/nonempty_final_conf_rate": 0.88671875,
      "calib/nonempty_reasoning_rate": 0.9765625,
      "calib/nonempty_step_conf_rate": 0.9296875,
      "calib/pce": 0.34516167400881054,
      "calib/std_conf": 0.23691412078640622,
      "calib/step_conf_rate": 0.9296875,
      "calib/step_q_c": 0.39131578947368423,
      "calib/step_q_c_n": 76.0,
      "calib/step_q_gap": -0.013101670843776114,
      "calib/step_q_w": 0.40441746031746034,
      "calib/step_q_w_n": 1008.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.046875,
      "completions/max_length": 2894.0,
      "completions/max_terminated_length": 2894.0,
      "completions/mean_length": 351.65625,
      "completions/mean_terminated_length": 368.9508056640625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 25.0,
      "epoch": 0.027733333333333332,
      "grad_norm": 1.5558923482894897,
      "learning_rate": 4.861111111111111e-06,
      "loss": 0.2922,
      "num_tokens": 6669883.0,
      "reward": 0.7352690696716309,
      "reward_std": 0.3406623303890228,
      "rewards/accuracy_reward_step": 0.05859375,
      "rewards/final_brier_reward_step": 0.6033506393432617,
      "rewards/format_reward_step": 0.80859375,
      "step": 26
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8715469744056463,
      "aux_distill/mean_u": 0.4131292154875933,
      "aux_distill/n_active_tok": 68.875,
      "calib/answer_extract_rate": 0.9296875,
      "calib/avg_num_step_conf": 4.6640625,
      "calib/ece": 0.32931550512155855,
      "calib/final_conf_rate": 0.94140625,
      "calib/format_rate": 0.87890625,
      "calib/frac_conf_gt_0.9": 0.04564315352697095,
      "calib/gap": 0.014844166011933668,
      "calib/mean_conf": 0.42060181217550047,
      "calib/mu_c": 0.434090909090909,
      "calib/mu_w": 0.41924674307897536,
      "calib/nonempty_final_conf_rate": 0.94140625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.96484375,
      "calib/pce": 0.32931550512155855,
      "calib/std_conf": 0.23038737837096063,
      "calib/step_conf_rate": 0.96484375,
      "calib/step_q_c": 0.423968253968254,
      "calib/step_q_c_n": 126.0,
      "calib/step_q_gap": -0.006461334046727274,
      "calib/step_q_w": 0.43042958801498127,
      "calib/step_q_w_n": 1068.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2791.0,
      "completions/max_terminated_length": 2791.0,
      "completions/mean_length": 354.55859375,
      "completions/mean_terminated_length": 358.7628479003906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 24.0,
      "epoch": 0.0288,
      "grad_norm": 1.6147425174713135,
      "learning_rate": 4.833333333333333e-06,
      "loss": 0.2992,
      "num_tokens": 6865866.0,
      "reward": 0.8195099830627441,
      "reward_std": 0.28160202503204346,
      "rewards/accuracy_reward_step": 0.08984375,
      "rewards/final_brier_reward_step": 0.6702699661254883,
      "rewards/format_reward_step": 0.87890625,
      "step": 27
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8409523535519838,
      "aux_distill/mean_u": 0.39175565836695353,
      "aux_distill/n_active_tok": 96.0,
      "calib/answer_extract_rate": 0.9296875,
      "calib/avg_num_step_conf": 5.765625,
      "calib/ece": 0.3076613089487777,
      "calib/final_conf_rate": 0.94921875,
      "calib/format_rate": 0.91015625,
      "calib/frac_conf_gt_0.9": 0.024691358024691357,
      "calib/gap": -0.04317698626550687,
      "calib/mean_conf": 0.4128464941339629,
      "calib/mu_c": 0.37500000000000006,
      "calib/mu_w": 0.4181769862655069,
      "calib/nonempty_final_conf_rate": 0.94921875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.2985255064796419,
      "calib/std_conf": 0.22269757326949044,
      "calib/step_conf_rate": 0.98828125,
      "calib/step_q_c": 0.3717714285714286,
      "calib/step_q_c_n": 175.0,
      "calib/step_q_gap": -0.047476346667450486,
      "calib/step_q_w": 0.4192477752388791,
      "calib/step_q_w_n": 1301.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 2873.0,
      "completions/max_terminated_length": 2873.0,
      "completions/mean_length": 360.7265625,
      "completions/mean_terminated_length": 366.452392578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 50.0,
      "epoch": 0.029866666666666666,
      "grad_norm": 1.035765290260315,
      "learning_rate": 4.805555555555556e-06,
      "loss": 0.2905,
      "num_tokens": 7065156.0,
      "reward": 0.8525996804237366,
      "reward_std": 0.2693730294704437,
      "rewards/accuracy_reward_step": 0.1171875,
      "rewards/final_brier_reward_step": 0.6778556108474731,
      "rewards/format_reward_step": 0.91015625,
      "step": 28
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8522766791284084,
      "aux_distill/mean_u": 0.37731298414794345,
      "aux_distill/n_active_tok": 85.25,
      "calib/answer_extract_rate": 0.94140625,
      "calib/avg_num_step_conf": 5.11328125,
      "calib/ece": 0.3199395161290322,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.9140625,
      "calib/frac_conf_gt_0.9": 0.012096774193548387,
      "calib/gap": 0.011023706896551744,
      "calib/mean_conf": 0.384375,
      "calib/mu_c": 0.39468749999999997,
      "calib/mu_w": 0.3836637931034482,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9765625,
      "calib/pce": 0.3198991935483871,
      "calib/std_conf": 0.21358195266462066,
      "calib/step_conf_rate": 0.9765625,
      "calib/step_q_c": 0.4183333333333333,
      "calib/step_q_c_n": 84.0,
      "calib/step_q_gap": 0.05758231292517002,
      "calib/step_q_w": 0.36075102040816326,
      "calib/step_q_w_n": 1225.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 2043.0,
      "completions/max_terminated_length": 2043.0,
      "completions/mean_length": 308.078125,
      "completions/mean_terminated_length": 314.21514892578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 27.0,
      "epoch": 0.030933333333333334,
      "grad_norm": 1.1891764402389526,
      "learning_rate": 4.777777777777778e-06,
      "loss": 0.2311,
      "num_tokens": 7251152.0,
      "reward": 0.8489661812782288,
      "reward_std": 0.25030654668807983,
      "rewards/accuracy_reward_step": 0.0625,
      "rewards/final_brier_reward_step": 0.7213698029518127,
      "rewards/format_reward_step": 0.9140625,
      "step": 29
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9309793021529913,
      "aux_distill/mean_u": 0.3487368012615765,
      "aux_distill/n_active_tok": 76.625,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 5.078125,
      "calib/ece": 0.28369047619047616,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.007936507936507936,
      "calib/gap": -0.023942428035043872,
      "calib/mean_conf": 0.3511507936507937,
      "calib/mu_c": 0.3288235294117647,
      "calib/mu_w": 0.35276595744680855,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.28369047619047616,
      "calib/std_conf": 0.195815006942304,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.35017094017094014,
      "calib/step_q_c_n": 117.0,
      "calib/step_q_gap": -0.006929651545036164,
      "calib/step_q_w": 0.3571005917159763,
      "calib/step_q_w_n": 1183.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2379.0,
      "completions/max_terminated_length": 2379.0,
      "completions/mean_length": 272.4453125,
      "completions/mean_terminated_length": 275.6759033203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 59.0,
      "epoch": 0.032,
      "grad_norm": 1.1468839645385742,
      "learning_rate": 4.75e-06,
      "loss": 0.206,
      "num_tokens": 7427882.0,
      "reward": 0.9133496284484863,
      "reward_std": 0.16822901368141174,
      "rewards/accuracy_reward_step": 0.06640625,
      "rewards/final_brier_reward_step": 0.7915430068969727,
      "rewards/format_reward_step": 0.96875,
      "step": 30
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9184452090412378,
      "aux_distill/mean_u": 0.3828412534240243,
      "aux_distill/n_active_tok": 76.625,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 4.96484375,
      "calib/ece": 0.25460396083361886,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.01568627450980392,
      "calib/gap": -0.0025642798847438364,
      "calib/mean_conf": 0.3598588627944031,
      "calib/mu_c": 0.3575862068965517,
      "calib/mu_w": 0.3601504867812955,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.2503686667159718,
      "calib/std_conf": 0.18381201605366868,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3186231884057971,
      "calib/step_q_c_n": 138.0,
      "calib/step_q_gap": -0.037029460320658014,
      "calib/step_q_w": 0.35565264872645513,
      "calib/step_q_w_n": 1133.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1761.0,
      "completions/max_terminated_length": 1761.0,
      "completions/mean_length": 247.328125,
      "completions/mean_terminated_length": 249.2755889892578,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 51.0,
      "epoch": 0.03306666666666667,
      "grad_norm": 1.1739736795425415,
      "learning_rate": 4.722222222222222e-06,
      "loss": 0.2572,
      "num_tokens": 7597110.0,
      "reward": 0.9391323328018188,
      "reward_std": 0.15756180882453918,
      "rewards/accuracy_reward_step": 0.1171875,
      "rewards/final_brier_reward_step": 0.7845146656036377,
      "rewards/format_reward_step": 0.9765625,
      "step": 31
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8915242683142424,
      "aux_distill/mean_u": 0.3488786209027339,
      "aux_distill/n_active_tok": 91.375,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 5.765625,
      "calib/ece": 0.22366141732283462,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.007874015748031496,
      "calib/gap": -0.025626875407697258,
      "calib/mean_conf": 0.33295275590551177,
      "calib/mu_c": 0.3108571428571429,
      "calib/mu_w": 0.33648401826484015,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.20940944881889761,
      "calib/std_conf": 0.19328373978999924,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.310981308411215,
      "calib/step_q_c_n": 214.0,
      "calib/step_q_gap": 0.002391767999170624,
      "calib/step_q_w": 0.3085895404120444,
      "calib/step_q_w_n": 1262.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 904.0,
      "completions/max_terminated_length": 904.0,
      "completions/mean_length": 256.578125,
      "completions/mean_terminated_length": 258.5984191894531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 61.0,
      "epoch": 0.034133333333333335,
      "grad_norm": 1.5438235998153687,
      "learning_rate": 4.694444444444445e-06,
      "loss": 0.2457,
      "num_tokens": 7769498.0,
      "reward": 0.9572702646255493,
      "reward_std": 0.1368747055530548,
      "rewards/accuracy_reward_step": 0.140625,
      "rewards/final_brier_reward_step": 0.7895406484603882,
      "rewards/format_reward_step": 0.984375,
      "step": 32
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.917936485260725,
      "aux_distill/mean_u": 0.3906123472827693,
      "aux_distill/n_active_tok": 113.0,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 7.078125,
      "calib/ece": 0.20196850393700788,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.007874015748031496,
      "calib/gap": -0.00590214067278283,
      "calib/mean_conf": 0.32173228346456695,
      "calib/mu_c": 0.3166666666666667,
      "calib/mu_w": 0.32256880733944954,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.19098425196850397,
      "calib/std_conf": 0.1856955195088237,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.33299107142857143,
      "calib/step_q_c_n": 224.0,
      "calib/step_q_gap": 0.03755026538323136,
      "calib/step_q_w": 0.29544080604534007,
      "calib/step_q_w_n": 1588.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1418.0,
      "completions/max_terminated_length": 1418.0,
      "completions/mean_length": 311.609375,
      "completions/mean_terminated_length": 314.06298828125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 86.0,
      "epoch": 0.0352,
      "grad_norm": 4.340875625610352,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.2648,
      "num_tokens": 7956142.0,
      "reward": 0.9643667936325073,
      "reward_std": 0.1367204785346985,
      "rewards/accuracy_reward_step": 0.140625,
      "rewards/final_brier_reward_step": 0.7998273372650146,
      "rewards/format_reward_step": 0.98828125,
      "step": 33
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.851872069761157,
      "aux_distill/mean_u": 0.32349913069650194,
      "aux_distill/n_active_tok": 118.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 7.109375,
      "calib/ece": 0.20184476940382454,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.015748031496062992,
      "calib/gap": 0.00206108374384234,
      "calib/mean_conf": 0.30886389201349834,
      "calib/mu_c": 0.31068965517241376,
      "calib/mu_w": 0.3086285714285714,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1982677165354331,
      "calib/std_conf": 0.1947032285675119,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3370466321243524,
      "calib/step_q_c_n": 193.0,
      "calib/step_q_gap": 0.028676819146917942,
      "calib/step_q_w": 0.30836981297743443,
      "calib/step_q_w_n": 1627.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1659.0,
      "completions/max_terminated_length": 1659.0,
      "completions/mean_length": 294.09375,
      "completions/mean_terminated_length": 296.4094543457031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.03626666666666667,
      "grad_norm": 1.6347882747650146,
      "learning_rate": 4.638888888888889e-06,
      "loss": 0.2868,
      "num_tokens": 8136542.0,
      "reward": 0.9612504243850708,
      "reward_std": 0.13310450315475464,
      "rewards/accuracy_reward_step": 0.11328125,
      "rewards/final_brier_reward_step": 0.8170320987701416,
      "rewards/format_reward_step": 0.9921875,
      "step": 34
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8687707763165236,
      "aux_distill/mean_u": 0.3417219541422614,
      "aux_distill/n_active_tok": 135.125,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.8671875,
      "calib/ece": 0.18003937007874016,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.003937007874015748,
      "calib/gap": -0.012003913894324814,
      "calib/mean_conf": 0.2749212598425197,
      "calib/mu_c": 0.2645714285714286,
      "calib/mu_w": 0.27657534246575344,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.15858267716535435,
      "calib/std_conf": 0.1757632223629866,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2934931506849316,
      "calib/step_q_c_n": 292.0,
      "calib/step_q_gap": -0.002841834148233191,
      "calib/step_q_w": 0.2963349848331648,
      "calib/step_q_w_n": 1978.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1794.0,
      "completions/max_terminated_length": 1794.0,
      "completions/mean_length": 339.9765625,
      "completions/mean_terminated_length": 342.6535339355469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.037333333333333336,
      "grad_norm": 1.1217002868652344,
      "learning_rate": 4.611111111111112e-06,
      "loss": 0.2413,
      "num_tokens": 8332832.0,
      "reward": 0.9755380749702454,
      "reward_std": 0.11975622177124023,
      "rewards/accuracy_reward_step": 0.13671875,
      "rewards/final_brier_reward_step": 0.8221698999404907,
      "rewards/format_reward_step": 0.9921875,
      "step": 35
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9129628445953131,
      "aux_distill/mean_u": 0.421250388575979,
      "aux_distill/n_active_tok": 130.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.28515625,
      "calib/ece": 0.131640625,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": 0.005971291866028627,
      "calib/mean_conf": 0.28375,
      "calib/mu_c": 0.28818181818181815,
      "calib/mu_w": 0.2822105263157895,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0787890625,
      "calib/std_conf": 0.1903717580157309,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.32663507109004736,
      "calib/step_q_c_n": 422.0,
      "calib/step_q_gap": 0.001790927476156834,
      "calib/step_q_w": 0.3248441436138905,
      "calib/step_q_w_n": 1699.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1530.0,
      "completions/max_terminated_length": 1530.0,
      "completions/mean_length": 312.66796875,
      "completions/mean_terminated_length": 315.1299133300781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 135.0,
      "epoch": 0.0384,
      "grad_norm": 1.9868746995925903,
      "learning_rate": 4.583333333333333e-06,
      "loss": 0.2626,
      "num_tokens": 8515587.0,
      "reward": 1.0159192085266113,
      "reward_std": 0.12932753562927246,
      "rewards/accuracy_reward_step": 0.2578125,
      "rewards/final_brier_reward_step": 0.7740257978439331,
      "rewards/format_reward_step": 1.0,
      "step": 36
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8150140605866909,
      "aux_distill/mean_u": 0.2819595940367653,
      "aux_distill/n_active_tok": 142.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 9.19140625,
      "calib/ece": 0.18827624015779124,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0078125,
      "calib/gap": -0.02756114637071816,
      "calib/mean_conf": 0.31054186515779125,
      "calib/mu_c": 0.28717948717948716,
      "calib/mu_w": 0.3147406335502053,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.17323717765779123,
      "calib/std_conf": 0.19026432766581608,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28545090909090914,
      "calib/step_q_c_n": 275.0,
      "calib/step_q_gap": -0.08267099806038813,
      "calib/step_q_w": 0.3681219071512973,
      "calib/step_q_w_n": 2078.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1464.0,
      "completions/max_terminated_length": 1464.0,
      "completions/mean_length": 338.0078125,
      "completions/mean_terminated_length": 340.6692810058594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 137.0,
      "epoch": 0.039466666666666664,
      "grad_norm": 0.9691076874732971,
      "learning_rate": 4.555555555555556e-06,
      "loss": 0.2546,
      "num_tokens": 8709213.0,
      "reward": 0.9745661616325378,
      "reward_std": 0.12055601924657822,
      "rewards/accuracy_reward_step": 0.15234375,
      "rewards/final_brier_reward_step": 0.8006948828697205,
      "rewards/format_reward_step": 0.99609375,
      "step": 37
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9269427638500929,
      "aux_distill/mean_u": 0.39672384793919174,
      "aux_distill/n_active_tok": 149.75,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 9.75,
      "calib/ece": 0.13189723320158103,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.003952569169960474,
      "calib/gap": 0.017112944162436555,
      "calib/mean_conf": 0.25292490118577077,
      "calib/mu_c": 0.26625,
      "calib/mu_w": 0.24913705583756343,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.08173913043478262,
      "calib/std_conf": 0.17077636687404305,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.3071501272264631,
      "calib/step_q_c_n": 393.0,
      "calib/step_q_gap": 0.03841698409759958,
      "calib/step_q_w": 0.26873314312886354,
      "calib/step_q_w_n": 2103.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1516.0,
      "completions/max_terminated_length": 1516.0,
      "completions/mean_length": 348.8671875,
      "completions/mean_terminated_length": 353.00396728515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 100.0,
      "epoch": 0.04053333333333333,
      "grad_norm": 1.204365611076355,
      "learning_rate": 4.527777777777778e-06,
      "loss": 0.2647,
      "num_tokens": 8905411.0,
      "reward": 0.9932834506034851,
      "reward_std": 0.14956696331501007,
      "rewards/accuracy_reward_step": 0.21875,
      "rewards/final_brier_reward_step": 0.7873480916023254,
      "rewards/format_reward_step": 0.98046875,
      "step": 38
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8835952132940292,
      "aux_distill/mean_u": 0.28569774736326614,
      "aux_distill/n_active_tok": 114.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.1328125,
      "calib/ece": 0.1694797277269374,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": -0.0029947900175748554,
      "calib/mean_conf": 0.2791672277269374,
      "calib/mu_c": 0.27679245283018866,
      "calib/mu_w": 0.2797872428477635,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12080785272693742,
      "calib/std_conf": 0.19193509897197172,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.30876010781671165,
      "calib/step_q_c_n": 371.0,
      "calib/step_q_gap": 0.0038731106546076033,
      "calib/step_q_w": 0.30488699716210405,
      "calib/step_q_w_n": 1711.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 811.0,
      "completions/max_terminated_length": 811.0,
      "completions/mean_length": 332.22265625,
      "completions/mean_terminated_length": 334.8385925292969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.0416,
      "grad_norm": 1.4351422786712646,
      "learning_rate": 4.5e-06,
      "loss": 0.2705,
      "num_tokens": 9096548.0,
      "reward": 0.9999179840087891,
      "reward_std": 0.11247202754020691,
      "rewards/accuracy_reward_step": 0.20703125,
      "rewards/final_brier_reward_step": 0.7928047180175781,
      "rewards/format_reward_step": 1.0,
      "step": 39
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.900733919814229,
      "aux_distill/mean_u": 0.333621742383105,
      "aux_distill/n_active_tok": 122.625,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.61328125,
      "calib/ece": 0.16886274509803922,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00392156862745098,
      "calib/gap": 0.016845509893455124,
      "calib/mean_conf": 0.251921568627451,
      "calib/mu_c": 0.26638888888888895,
      "calib/mu_w": 0.24954337899543383,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.13980392156862748,
      "calib/std_conf": 0.19011408784605996,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2860714285714286,
      "calib/step_q_c_n": 252.0,
      "calib/step_q_gap": -0.0015418842805939303,
      "calib/step_q_w": 0.2876133128520225,
      "calib/step_q_w_n": 1953.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1043.0,
      "completions/max_terminated_length": 1043.0,
      "completions/mean_length": 333.73828125,
      "completions/mean_terminated_length": 336.36614990234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.042666666666666665,
      "grad_norm": 1.1747071743011475,
      "learning_rate": 4.472222222222223e-06,
      "loss": 0.2479,
      "num_tokens": 9288745.0,
      "reward": 0.9839453101158142,
      "reward_std": 0.10935330390930176,
      "rewards/accuracy_reward_step": 0.140625,
      "rewards/final_brier_reward_step": 0.8311718702316284,
      "rewards/format_reward_step": 0.99609375,
      "step": 40
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.885868264362216,
      "aux_distill/mean_u": 0.34837704053660135,
      "aux_distill/n_active_tok": 120.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.34375,
      "calib/ece": 0.18067968750000002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0078125,
      "calib/gap": -0.016945614035087653,
      "calib/mean_conf": 0.2678359375,
      "calib/mu_c": 0.255921052631579,
      "calib/mu_w": 0.27286666666666665,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.07582031250000001,
      "calib/std_conf": 0.2042344423903473,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.33281376518218625,
      "calib/step_q_c_n": 494.0,
      "calib/step_q_gap": 0.044120475138896165,
      "calib/step_q_w": 0.2886932900432901,
      "calib/step_q_w_n": 1386.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 876.0,
      "completions/max_terminated_length": 876.0,
      "completions/mean_length": 297.26171875,
      "completions/mean_terminated_length": 299.60235595703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 116.0,
      "epoch": 0.04373333333333333,
      "grad_norm": 2.003614664077759,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.2752,
      "num_tokens": 9472092.0,
      "reward": 1.0167245864868164,
      "reward_std": 0.14265429973602295,
      "rewards/accuracy_reward_step": 0.296875,
      "rewards/final_brier_reward_step": 0.7404803037643433,
      "rewards/format_reward_step": 0.99609375,
      "step": 41
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9000848159193993,
      "aux_distill/mean_u": 0.3127236620497125,
      "aux_distill/n_active_tok": 118.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.03125,
      "calib/ece": 0.15384313725490198,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.01568627450980392,
      "calib/gap": 0.031094871346052766,
      "calib/mean_conf": 0.2975294117647058,
      "calib/mu_c": 0.3215517241379309,
      "calib/mu_w": 0.29045685279187816,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1119607843137255,
      "calib/std_conf": 0.2243818832417291,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3662303664921467,
      "calib/step_q_c_n": 382.0,
      "calib/step_q_gap": 0.07238765442524109,
      "calib/step_q_w": 0.2938427120669056,
      "calib/step_q_w_n": 1674.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1343.0,
      "completions/max_terminated_length": 1343.0,
      "completions/mean_length": 299.609375,
      "completions/mean_terminated_length": 301.968505859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.0448,
      "grad_norm": 3.954033613204956,
      "learning_rate": 4.416666666666667e-06,
      "loss": 0.2571,
      "num_tokens": 9653160.0,
      "reward": 0.9958779215812683,
      "reward_std": 0.16326522827148438,
      "rewards/accuracy_reward_step": 0.2265625,
      "rewards/final_brier_reward_step": 0.7730058431625366,
      "rewards/format_reward_step": 0.9921875,
      "step": 42
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8822179082781076,
      "aux_distill/mean_u": 0.2652459974117178,
      "aux_distill/n_active_tok": 110.625,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.265625,
      "calib/ece": 0.10752070312499999,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.032956915113871665,
      "calib/mean_conf": 0.253182421875,
      "calib/mu_c": 0.28021739130434786,
      "calib/mu_w": 0.2472604761904762,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.09050781249999999,
      "calib/std_conf": 0.20561714812989704,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3243166666666667,
      "calib/step_q_c_n": 336.0,
      "calib/step_q_gap": 0.03327598425196859,
      "calib/step_q_w": 0.2910406824146981,
      "calib/step_q_w_n": 1524.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 974.0,
      "completions/max_terminated_length": 974.0,
      "completions/mean_length": 286.375,
      "completions/mean_terminated_length": 288.6299133300781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 98.0,
      "epoch": 0.04586666666666667,
      "grad_norm": 4.156473636627197,
      "learning_rate": 4.388888888888889e-06,
      "loss": 0.2712,
      "num_tokens": 9831696.0,
      "reward": 0.9935085773468018,
      "reward_std": 0.12804695963859558,
      "rewards/accuracy_reward_step": 0.1796875,
      "rewards/final_brier_reward_step": 0.8112359046936035,
      "rewards/format_reward_step": 0.99609375,
      "step": 43
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8823073115199804,
      "aux_distill/mean_u": 0.2861128941515574,
      "aux_distill/n_active_tok": 130.25,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 7.52734375,
      "calib/ece": 0.18015810276679844,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.019762845849802372,
      "calib/gap": 0.010590909090909095,
      "calib/mean_conf": 0.271699604743083,
      "calib/mu_c": 0.2809090909090909,
      "calib/mu_w": 0.2703181818181818,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.16071146245059292,
      "calib/std_conf": 0.2293293984725821,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.32342342342342345,
      "calib/step_q_c_n": 222.0,
      "calib/step_q_gap": 0.01449292488969911,
      "calib/step_q_w": 0.30893049853372434,
      "calib/step_q_w_n": 1705.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1593.0,
      "completions/max_terminated_length": 1593.0,
      "completions/mean_length": 303.41796875,
      "completions/mean_terminated_length": 304.60784912109375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.046933333333333334,
      "grad_norm": 3.0850069522857666,
      "learning_rate": 4.361111111111112e-06,
      "loss": 0.2627,
      "num_tokens": 10015691.0,
      "reward": 0.9543378353118896,
      "reward_std": 0.15263614058494568,
      "rewards/accuracy_reward_step": 0.12890625,
      "rewards/final_brier_reward_step": 0.7993007898330688,
      "rewards/format_reward_step": 0.98046875,
      "step": 44
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8355578556656837,
      "aux_distill/mean_u": 0.26338739376699555,
      "aux_distill/n_active_tok": 115.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.21484375,
      "calib/ece": 0.21674509803921563,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.01568627450980392,
      "calib/gap": -0.055726495726495684,
      "calib/mean_conf": 0.265921568627451,
      "calib/mu_c": 0.2187179487179487,
      "calib/mu_w": 0.2744444444444444,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1648627450980392,
      "calib/std_conf": 0.21819184247590376,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25475471698113206,
      "calib/step_q_c_n": 265.0,
      "calib/step_q_gap": -0.03257777353719915,
      "calib/step_q_w": 0.2873324905183312,
      "calib/step_q_w_n": 1582.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 801.0,
      "completions/max_terminated_length": 801.0,
      "completions/mean_length": 296.41796875,
      "completions/mean_terminated_length": 298.751953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 98.0,
      "epoch": 0.048,
      "grad_norm": 3.737006902694702,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.2522,
      "num_tokens": 10196622.0,
      "reward": 0.9704841375350952,
      "reward_std": 0.11979597806930542,
      "rewards/accuracy_reward_step": 0.15234375,
      "rewards/final_brier_reward_step": 0.7925308346748352,
      "rewards/format_reward_step": 0.99609375,
      "step": 45
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9158951379358768,
      "aux_distill/mean_u": 0.3747160606423709,
      "aux_distill/n_active_tok": 109.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.03515625,
      "calib/ece": 0.1684313725490196,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.011764705882352941,
      "calib/gap": -0.008765096618357532,
      "calib/mean_conf": 0.2248235294117647,
      "calib/mu_c": 0.2177083333333333,
      "calib/mu_w": 0.22647342995169084,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10250980392156862,
      "calib/std_conf": 0.2021353367136176,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.31444210526315797,
      "calib/step_q_c_n": 285.0,
      "calib/step_q_gap": 0.040425746424107845,
      "calib/step_q_w": 0.2740163588390501,
      "calib/step_q_w_n": 1516.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 839.0,
      "completions/max_terminated_length": 839.0,
      "completions/mean_length": 285.61328125,
      "completions/mean_terminated_length": 287.8622131347656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 92.0,
      "epoch": 0.04906666666666667,
      "grad_norm": 1.399847149848938,
      "learning_rate": 4.305555555555556e-06,
      "loss": 0.2632,
      "num_tokens": 10374507.0,
      "reward": 0.9875702857971191,
      "reward_std": 0.10994257032871246,
      "rewards/accuracy_reward_step": 0.1875,
      "rewards/final_brier_reward_step": 0.7954530715942383,
      "rewards/format_reward_step": 0.9921875,
      "step": 46
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8687598332762718,
      "aux_distill/mean_u": 0.30231973079135227,
      "aux_distill/n_active_tok": 139.625,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 9.08203125,
      "calib/ece": 0.15363636363636363,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.007905138339920948,
      "calib/gap": -0.013971698113207559,
      "calib/mean_conf": 0.2055731225296443,
      "calib/mu_c": 0.19452830188679246,
      "calib/mu_w": 0.20850000000000002,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.07486166007905139,
      "calib/std_conf": 0.2029918697843591,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26025,
      "calib/step_q_c_n": 360.0,
      "calib/step_q_gap": 0.023171119592875294,
      "calib/step_q_w": 0.2370788804071247,
      "calib/step_q_w_n": 1965.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 869.0,
      "completions/max_terminated_length": 869.0,
      "completions/mean_length": 315.52734375,
      "completions/mean_terminated_length": 319.268798828125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 115.0,
      "epoch": 0.050133333333333335,
      "grad_norm": 3.7428154945373535,
      "learning_rate": 4.277777777777778e-06,
      "loss": 0.2504,
      "num_tokens": 10561258.0,
      "reward": 0.9834374785423279,
      "reward_std": 0.13472025096416473,
      "rewards/accuracy_reward_step": 0.20703125,
      "rewards/final_brier_reward_step": 0.7754687070846558,
      "rewards/format_reward_step": 0.984375,
      "step": 47
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8738913396373391,
      "aux_distill/mean_u": 0.2939748861844822,
      "aux_distill/n_active_tok": 111.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.14453125,
      "calib/ece": 0.19413944223107568,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.02494813725490197,
      "calib/mean_conf": 0.2220358565737052,
      "calib/mu_c": 0.202156862745098,
      "calib/mu_w": 0.22710499999999997,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.10649402390438249,
      "calib/std_conf": 0.196638085229977,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.21553977272727273,
      "calib/step_q_c_n": 352.0,
      "calib/step_q_gap": -0.07732170323752077,
      "calib/step_q_w": 0.2928614759647935,
      "calib/step_q_w_n": 1477.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2449.0,
      "completions/max_terminated_length": 2449.0,
      "completions/mean_length": 298.18359375,
      "completions/mean_terminated_length": 300.531494140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 122.0,
      "epoch": 0.0512,
      "grad_norm": 4.377185344696045,
      "learning_rate": 4.25e-06,
      "loss": 0.3166,
      "num_tokens": 10741281.0,
      "reward": 0.9776182174682617,
      "reward_std": 0.1417025923728943,
      "rewards/accuracy_reward_step": 0.20703125,
      "rewards/final_brier_reward_step": 0.7716426253318787,
      "rewards/format_reward_step": 0.9765625,
      "step": 48
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.877561567351222,
      "aux_distill/mean_u": 0.30501746020629955,
      "aux_distill/n_active_tok": 108.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.66796875,
      "calib/ece": 0.16074803149606298,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.007874015748031496,
      "calib/gap": 0.01150063251106892,
      "calib/mean_conf": 0.19996062992125985,
      "calib/mu_c": 0.20838235294117644,
      "calib/mu_w": 0.19688172043010752,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.04649606299212598,
      "calib/std_conf": 0.19759478545092854,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.26375000000000004,
      "calib/step_q_c_n": 488.0,
      "calib/step_q_gap": 0.028097796610169534,
      "calib/step_q_w": 0.2356522033898305,
      "calib/step_q_w_n": 1475.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1379.0,
      "completions/max_terminated_length": 1379.0,
      "completions/mean_length": 301.2578125,
      "completions/mean_terminated_length": 302.4392395019531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 136.0,
      "epoch": 0.05226666666666667,
      "grad_norm": 2.8171491622924805,
      "learning_rate": 4.222222222222223e-06,
      "loss": 0.2598,
      "num_tokens": 10922939.0,
      "reward": 1.0083338022232056,
      "reward_std": 0.1236845999956131,
      "rewards/accuracy_reward_step": 0.265625,
      "rewards/final_brier_reward_step": 0.7588551044464111,
      "rewards/format_reward_step": 0.9921875,
      "step": 49
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9258559513837099,
      "aux_distill/mean_u": 0.32496536010658356,
      "aux_distill/n_active_tok": 125.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.70703125,
      "calib/ece": 0.1795275590551181,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.01968503937007874,
      "calib/gap": -0.0027361005331302257,
      "calib/mean_conf": 0.2192913385826772,
      "calib/mu_c": 0.2171153846153846,
      "calib/mu_w": 0.21985148514851482,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.09704724409448819,
      "calib/std_conf": 0.21140021045424237,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2707202216066482,
      "calib/step_q_c_n": 361.0,
      "calib/step_q_gap": 0.0019423059738938875,
      "calib/step_q_w": 0.2687779156327543,
      "calib/step_q_w_n": 1612.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 833.0,
      "completions/max_terminated_length": 833.0,
      "completions/mean_length": 307.43359375,
      "completions/mean_terminated_length": 309.8543395996094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.05333333333333334,
      "grad_norm": 0.9597369432449341,
      "learning_rate": 4.194444444444445e-06,
      "loss": 0.2235,
      "num_tokens": 11107002.0,
      "reward": 0.9902620911598206,
      "reward_std": 0.11881484091281891,
      "rewards/accuracy_reward_step": 0.203125,
      "rewards/final_brier_reward_step": 0.7852116823196411,
      "rewards/format_reward_step": 0.9921875,
      "step": 50
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9441087357699871,
      "aux_distill/mean_u": 0.3196395898677443,
      "aux_distill/n_active_tok": 126.625,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.15234375,
      "calib/ece": 0.229809918419611,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.007874015748031496,
      "calib/gap": -0.039765872397339,
      "calib/mean_conf": 0.21152866425755446,
      "calib/mu_c": 0.18397435897435896,
      "calib/mu_w": 0.22374023137169796,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0671259842519685,
      "calib/std_conf": 0.19343077812634754,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.23406779661016952,
      "calib/step_q_c_n": 590.0,
      "calib/step_q_gap": -0.005498214038365784,
      "calib/step_q_w": 0.2395660106485353,
      "calib/step_q_w_n": 1497.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1077.0,
      "completions/max_terminated_length": 1077.0,
      "completions/mean_length": 313.10546875,
      "completions/mean_terminated_length": 316.8182067871094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 108.0,
      "epoch": 0.0544,
      "grad_norm": 2.159965753555298,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.2578,
      "num_tokens": 11296453.0,
      "reward": 1.0037527084350586,
      "reward_std": 0.12112631648778915,
      "rewards/accuracy_reward_step": 0.3046875,
      "rewards/final_brier_reward_step": 0.7145366668701172,
      "rewards/format_reward_step": 0.98828125,
      "step": 51
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9033014588057995,
      "aux_distill/mean_u": 0.3339367217015873,
      "aux_distill/n_active_tok": 109.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.2890625,
      "calib/ece": 0.17832031250000002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": 0.015150966183574893,
      "calib/mean_conf": 0.2298046875,
      "calib/mu_c": 0.24069444444444446,
      "calib/mu_w": 0.22554347826086957,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.06343750000000001,
      "calib/std_conf": 0.214132426848498,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.29648148148148146,
      "calib/step_q_c_n": 486.0,
      "calib/step_q_gap": 0.017575684380032153,
      "calib/step_q_w": 0.2789057971014493,
      "calib/step_q_w_n": 1380.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 668.0,
      "completions/max_terminated_length": 668.0,
      "completions/mean_length": 280.90234375,
      "completions/mean_terminated_length": 283.1141662597656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 85.0,
      "epoch": 0.055466666666666664,
      "grad_norm": 1.1030815839767456,
      "learning_rate": 4.138888888888889e-06,
      "loss": 0.2568,
      "num_tokens": 11476316.0,
      "reward": 1.0183639526367188,
      "reward_std": 0.1275119185447693,
      "rewards/accuracy_reward_step": 0.28125,
      "rewards/final_brier_reward_step": 0.7554777264595032,
      "rewards/format_reward_step": 1.0,
      "step": 52
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8972780760377645,
      "aux_distill/mean_u": 0.33077345545734277,
      "aux_distill/n_active_tok": 127.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.09765625,
      "calib/ece": 0.21405078125000002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": 0.008478112175102581,
      "calib/mean_conf": 0.21173046875,
      "calib/mu_c": 0.21736046511627904,
      "calib/mu_w": 0.20888235294117646,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04492187499999999,
      "calib/std_conf": 0.19493206350971426,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25831831932773114,
      "calib/step_q_c_n": 595.0,
      "calib/step_q_gap": -0.013571058209481346,
      "calib/step_q_w": 0.2718893775372125,
      "calib/step_q_w_n": 1478.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 758.0,
      "completions/max_terminated_length": 758.0,
      "completions/mean_length": 320.73046875,
      "completions/mean_terminated_length": 323.2558898925781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 122.0,
      "epoch": 0.05653333333333333,
      "grad_norm": 0.7430399656295776,
      "learning_rate": 4.111111111111111e-06,
      "loss": 0.299,
      "num_tokens": 11664247.0,
      "reward": 1.0316053628921509,
      "reward_std": 0.11501534283161163,
      "rewards/accuracy_reward_step": 0.3359375,
      "rewards/final_brier_reward_step": 0.7272732257843018,
      "rewards/format_reward_step": 1.0,
      "step": 53
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8624539021402597,
      "aux_distill/mean_u": 0.2733415144863499,
      "aux_distill/n_active_tok": 115.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.9453125,
      "calib/ece": 0.28133359375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": -0.01467832984326417,
      "calib/mean_conf": 0.20022890625,
      "calib/mu_c": 0.19145631067961166,
      "calib/mu_w": 0.20613464052287583,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.039609375,
      "calib/std_conf": 0.19904292178146832,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2401891891891892,
      "calib/step_q_c_n": 740.0,
      "calib/step_q_gap": -0.017947982371861793,
      "calib/step_q_w": 0.258137171561051,
      "calib/step_q_w_n": 1294.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 867.0,
      "completions/max_terminated_length": 867.0,
      "completions/mean_length": 304.75,
      "completions/mean_terminated_length": 307.14959716796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 100.0,
      "epoch": 0.0576,
      "grad_norm": 2.56850004196167,
      "learning_rate": 4.083333333333334e-06,
      "loss": 0.2742,
      "num_tokens": 11848495.0,
      "reward": 1.0347485542297363,
      "reward_std": 0.11770911514759064,
      "rewards/accuracy_reward_step": 0.40234375,
      "rewards/final_brier_reward_step": 0.6710594296455383,
      "rewards/format_reward_step": 0.99609375,
      "step": 54
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8497700281441212,
      "aux_distill/mean_u": 0.2503387007848704,
      "aux_distill/n_active_tok": 123.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.36328125,
      "calib/ece": 0.15098549019607843,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.019738239538239555,
      "calib/mean_conf": 0.18736745098039215,
      "calib/mu_c": 0.2019969696969697,
      "calib/mu_w": 0.18225873015873015,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03976470588235294,
      "calib/std_conf": 0.16331167867467933,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27153822894168467,
      "calib/step_q_c_n": 463.0,
      "calib/step_q_gap": -0.015342581546992307,
      "calib/step_q_w": 0.286880810488677,
      "calib/step_q_w_n": 1678.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 881.0,
      "completions/max_terminated_length": 881.0,
      "completions/mean_length": 311.5,
      "completions/mean_terminated_length": 313.9527587890625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 105.0,
      "epoch": 0.058666666666666666,
      "grad_norm": 1.1233389377593994,
      "learning_rate": 4.055555555555556e-06,
      "loss": 0.2892,
      "num_tokens": 12036063.0,
      "reward": 1.0174031257629395,
      "reward_std": 0.09957180917263031,
      "rewards/accuracy_reward_step": 0.2578125,
      "rewards/final_brier_reward_step": 0.7809000015258789,
      "rewards/format_reward_step": 0.99609375,
      "step": 55
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8872428350150585,
      "aux_distill/mean_u": 0.32575370195465403,
      "aux_distill/n_active_tok": 117.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.10546875,
      "calib/ece": 0.19675787401574807,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.007874015748031496,
      "calib/gap": -0.010327287296037324,
      "calib/mean_conf": 0.1839507874015748,
      "calib/mu_c": 0.17679487179487177,
      "calib/mu_w": 0.1871221590909091,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.036811023622047245,
      "calib/std_conf": 0.18785905835616928,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.29798865478119935,
      "calib/step_q_c_n": 617.0,
      "calib/step_q_gap": -0.007017655232518072,
      "calib/step_q_w": 0.3050063100137174,
      "calib/step_q_w_n": 1458.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1047.0,
      "completions/max_terminated_length": 1047.0,
      "completions/mean_length": 310.671875,
      "completions/mean_terminated_length": 313.11810302734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.05973333333333333,
      "grad_norm": 0.5889172554016113,
      "learning_rate": 4.027777777777779e-06,
      "loss": 0.2527,
      "num_tokens": 12222435.0,
      "reward": 1.0117602348327637,
      "reward_std": 0.1229933425784111,
      "rewards/accuracy_reward_step": 0.3046875,
      "rewards/final_brier_reward_step": 0.7266455292701721,
      "rewards/format_reward_step": 0.9921875,
      "step": 56
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.935738505795598,
      "aux_distill/mean_u": 0.31780845640126715,
      "aux_distill/n_active_tok": 125.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.62890625,
      "calib/ece": 0.21763294117647058,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.011764705882352941,
      "calib/gap": -0.006428235294117651,
      "calib/mean_conf": 0.18785725490196079,
      "calib/mu_c": 0.18357176470588232,
      "calib/mu_w": 0.18999999999999997,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03607843137254902,
      "calib/std_conf": 0.19378627973123452,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.290451226993865,
      "calib/step_q_c_n": 652.0,
      "calib/step_q_gap": 0.011611727957256124,
      "calib/step_q_w": 0.2788394990366089,
      "calib/step_q_w_n": 1557.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 925.0,
      "completions/max_terminated_length": 925.0,
      "completions/mean_length": 307.55859375,
      "completions/mean_terminated_length": 309.9803161621094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.0608,
      "grad_norm": 0.4905681610107422,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.2566,
      "num_tokens": 12407962.0,
      "reward": 1.0207659006118774,
      "reward_std": 0.1187235563993454,
      "rewards/accuracy_reward_step": 0.33203125,
      "rewards/final_brier_reward_step": 0.7134066820144653,
      "rewards/format_reward_step": 0.99609375,
      "step": 57
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8835944719612598,
      "aux_distill/mean_u": 0.31737898162086964,
      "aux_distill/n_active_tok": 131.5,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 7.71875,
      "calib/ece": 0.12244212598425194,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.03459161626344087,
      "calib/mean_conf": 0.1974003937007874,
      "calib/mu_c": 0.2235483870967742,
      "calib/mu_w": 0.18895677083333332,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.037874015748031495,
      "calib/std_conf": 0.18803480296431632,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2734507042253521,
      "calib/step_q_c_n": 426.0,
      "calib/step_q_gap": -0.01306226351658335,
      "calib/step_q_w": 0.28651296774193546,
      "calib/step_q_w_n": 1550.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 838.0,
      "completions/max_terminated_length": 838.0,
      "completions/mean_length": 296.234375,
      "completions/mean_terminated_length": 298.5669250488281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 117.0,
      "epoch": 0.06186666666666667,
      "grad_norm": 0.6383081078529358,
      "learning_rate": 3.972222222222223e-06,
      "loss": 0.237,
      "num_tokens": 12590118.0,
      "reward": 1.0094563961029053,
      "reward_std": 0.11579343676567078,
      "rewards/accuracy_reward_step": 0.2421875,
      "rewards/final_brier_reward_step": 0.7845379114151001,
      "rewards/format_reward_step": 0.9921875,
      "step": 58
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8786263391375542,
      "aux_distill/mean_u": 0.2848541231171143,
      "aux_distill/n_active_tok": 115.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.18359375,
      "calib/ece": 0.19727568627450984,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.03612934017146355,
      "calib/mean_conf": 0.18790078431372548,
      "calib/mu_c": 0.16183098591549297,
      "calib/mu_w": 0.19796032608695652,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05337254901960784,
      "calib/std_conf": 0.17178196828199355,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.263248031496063,
      "calib/step_q_c_n": 508.0,
      "calib/step_q_gap": -0.022587885328133606,
      "calib/step_q_w": 0.2858359168241966,
      "calib/step_q_w_n": 1587.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 888.0,
      "completions/max_terminated_length": 888.0,
      "completions/mean_length": 303.1171875,
      "completions/mean_terminated_length": 305.5039367675781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.06293333333333333,
      "grad_norm": 1.045523762702942,
      "learning_rate": 3.944444444444445e-06,
      "loss": 0.2482,
      "num_tokens": 12773964.0,
      "reward": 1.0048048496246338,
      "reward_std": 0.1106722503900528,
      "rewards/accuracy_reward_step": 0.27734375,
      "rewards/final_brier_reward_step": 0.7400784492492676,
      "rewards/format_reward_step": 0.9921875,
      "step": 59
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9114331156015396,
      "aux_distill/mean_u": 0.3258913867277279,
      "aux_distill/n_active_tok": 117.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.6015625,
      "calib/ece": 0.15221093749999998,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": -0.0014270833333333288,
      "calib/mean_conf": 0.1487265625,
      "calib/mu_c": 0.14765625000000002,
      "calib/mu_w": 0.14908333333333335,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.02546875,
      "calib/std_conf": 0.15708154803137636,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.23213580246913576,
      "calib/step_q_c_n": 405.0,
      "calib/step_q_gap": 0.006057866064203887,
      "calib/step_q_w": 0.22607793640493187,
      "calib/step_q_w_n": 1541.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 733.0,
      "completions/max_terminated_length": 733.0,
      "completions/mean_length": 281.43359375,
      "completions/mean_terminated_length": 283.64959716796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 72.0,
      "epoch": 0.064,
      "grad_norm": 1.4852534532546997,
      "learning_rate": 3.916666666666667e-06,
      "loss": 0.2697,
      "num_tokens": 12954867.0,
      "reward": 1.009638786315918,
      "reward_std": 0.09001928567886353,
      "rewards/accuracy_reward_step": 0.25,
      "rewards/final_brier_reward_step": 0.7731839418411255,
      "rewards/format_reward_step": 0.99609375,
      "step": 60
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8951994087547064,
      "aux_distill/mean_u": 0.2844116796439261,
      "aux_distill/n_active_tok": 131.875,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 9.3125,
      "calib/ece": 0.3061264822134387,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.007905138339920948,
      "calib/gap": -0.01790540540540539,
      "calib/mean_conf": 0.17047430830039526,
      "calib/mu_c": 0.16,
      "calib/mu_w": 0.1779054054054054,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.030790513833992093,
      "calib/std_conf": 0.17423512217240988,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2416358839050132,
      "calib/step_q_c_n": 758.0,
      "calib/step_q_gap": -0.061912701580841695,
      "calib/step_q_w": 0.3035485854858549,
      "calib/step_q_w_n": 1626.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1082.0,
      "completions/max_terminated_length": 1082.0,
      "completions/mean_length": 301.35546875,
      "completions/mean_terminated_length": 304.9288635253906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 0.06506666666666666,
      "grad_norm": 0.9578151702880859,
      "learning_rate": 3.88888888888889e-06,
      "loss": 0.2356,
      "num_tokens": 13136078.0,
      "reward": 1.0245447158813477,
      "reward_std": 0.1300700306892395,
      "rewards/accuracy_reward_step": 0.41015625,
      "rewards/final_brier_reward_step": 0.6506519317626953,
      "rewards/format_reward_step": 0.98828125,
      "step": 61
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9100477192550898,
      "aux_distill/mean_u": 0.28706747487226597,
      "aux_distill/n_active_tok": 125.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.265625,
      "calib/ece": 0.18208267716535437,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.011811023622047244,
      "calib/gap": 0.001003481419813823,
      "calib/mean_conf": 0.14453149606299212,
      "calib/mu_c": 0.14524657534246574,
      "calib/mu_w": 0.14424309392265192,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.019606299212598426,
      "calib/std_conf": 0.14875648927857302,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2428534136546185,
      "calib/step_q_c_n": 498.0,
      "calib/step_q_gap": -0.006505053588892001,
      "calib/step_q_w": 0.2493584672435105,
      "calib/step_q_w_n": 1618.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 978.0,
      "completions/max_terminated_length": 978.0,
      "completions/mean_length": 317.453125,
      "completions/mean_terminated_length": 319.9527587890625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.06613333333333334,
      "grad_norm": 0.9756151437759399,
      "learning_rate": 3.861111111111112e-06,
      "loss": 0.286,
      "num_tokens": 13324426.0,
      "reward": 1.012264609336853,
      "reward_std": 0.09289586544036865,
      "rewards/accuracy_reward_step": 0.28515625,
      "rewards/final_brier_reward_step": 0.7471854090690613,
      "rewards/format_reward_step": 0.9921875,
      "step": 62
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.921884486451745,
      "aux_distill/mean_u": 0.37658791087950677,
      "aux_distill/n_active_tok": 159.625,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.546875,
      "calib/ece": 0.2384722222222222,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.015873015873015872,
      "calib/gap": -0.018688953488372073,
      "calib/mean_conf": 0.1461309523809524,
      "calib/mu_c": 0.13337500000000002,
      "calib/mu_w": 0.1520639534883721,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03357142857142857,
      "calib/std_conf": 0.16492485877498914,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2725210084033613,
      "calib/step_q_c_n": 595.0,
      "calib/step_q_gap": 0.01504974949599791,
      "calib/step_q_w": 0.2574712589073634,
      "calib/step_q_w_n": 2105.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1623.0,
      "completions/max_terminated_length": 1623.0,
      "completions/mean_length": 342.53125,
      "completions/mean_terminated_length": 347.96826171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 106.0,
      "epoch": 0.0672,
      "grad_norm": 0.6014112830162048,
      "learning_rate": 3.833333333333334e-06,
      "loss": 0.2515,
      "num_tokens": 13520754.0,
      "reward": 1.0021567344665527,
      "reward_std": 0.11314384639263153,
      "rewards/accuracy_reward_step": 0.3125,
      "rewards/final_brier_reward_step": 0.707438588142395,
      "rewards/format_reward_step": 0.984375,
      "step": 63
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8175699971616268,
      "aux_distill/mean_u": 0.24582131315860167,
      "aux_distill/n_active_tok": 137.5,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 10.3125,
      "calib/ece": 0.24190476190476193,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.003968253968253968,
      "calib/gap": -0.01925925925925928,
      "calib/mean_conf": 0.14269841269841274,
      "calib/mu_c": 0.12962962962962962,
      "calib/mu_w": 0.1488888888888889,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03158730158730158,
      "calib/std_conf": 0.16538753287121283,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25657575757575757,
      "calib/step_q_c_n": 660.0,
      "calib/step_q_gap": 0.033606060606060584,
      "calib/step_q_w": 0.22296969696969698,
      "calib/step_q_w_n": 1980.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 2310.0,
      "completions/max_terminated_length": 2310.0,
      "completions/mean_length": 331.27734375,
      "completions/mean_terminated_length": 336.5357360839844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 69.0,
      "epoch": 0.06826666666666667,
      "grad_norm": 0.8397740721702576,
      "learning_rate": 3.8055555555555556e-06,
      "loss": 0.2273,
      "num_tokens": 13709337.0,
      "reward": 1.0019054412841797,
      "reward_std": 0.11516478657722473,
      "rewards/accuracy_reward_step": 0.31640625,
      "rewards/final_brier_reward_step": 0.7030296325683594,
      "rewards/format_reward_step": 0.984375,
      "step": 64
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8168295957148075,
      "aux_distill/mean_u": 0.2424435817139507,
      "aux_distill/n_active_tok": 123.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.08203125,
      "calib/ece": 0.2042647058823529,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.00392156862745098,
      "calib/gap": 0.010540230317627547,
      "calib/mean_conf": 0.14083333333333334,
      "calib/mu_c": 0.14835616438356164,
      "calib/mu_w": 0.1378159340659341,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.029411764705882353,
      "calib/std_conf": 0.1520072130211508,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27874517374517377,
      "calib/step_q_c_n": 518.0,
      "calib/step_q_gap": 0.01918327819391652,
      "calib/step_q_w": 0.25956189555125725,
      "calib/step_q_w_n": 1551.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 901.0,
      "completions/max_terminated_length": 901.0,
      "completions/mean_length": 292.21875,
      "completions/mean_terminated_length": 294.5196838378906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 107.0,
      "epoch": 0.06933333333333333,
      "grad_norm": 1.7245993614196777,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.2277,
      "num_tokens": 13889169.0,
      "reward": 1.0131059885025024,
      "reward_std": 0.09085900336503983,
      "rewards/accuracy_reward_step": 0.28515625,
      "rewards/final_brier_reward_step": 0.7488681674003601,
      "rewards/format_reward_step": 0.9921875,
      "step": 65
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8621831871569157,
      "aux_distill/mean_u": 0.2643594781882684,
      "aux_distill/n_active_tok": 153.5,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 12.0234375,
      "calib/ece": 0.2113147410358566,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.025648954038784533,
      "calib/mean_conf": 0.141195219123506,
      "calib/mu_c": 0.1231081081081081,
      "calib/mu_w": 0.14875706214689263,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.02884462151394422,
      "calib/std_conf": 0.14315235015754815,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24553872053872058,
      "calib/step_q_c_n": 594.0,
      "calib/step_q_gap": -0.024567036305079698,
      "calib/step_q_w": 0.2701057568438003,
      "calib/step_q_w_n": 2484.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 2387.0,
      "completions/max_terminated_length": 2387.0,
      "completions/mean_length": 373.921875,
      "completions/mean_terminated_length": 381.37054443359375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 78.0,
      "epoch": 0.0704,
      "grad_norm": 5.546261310577393,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.1916,
      "num_tokens": 14091245.0,
      "reward": 0.9962350726127625,
      "reward_std": 0.11140823364257812,
      "rewards/accuracy_reward_step": 0.2890625,
      "rewards/final_brier_reward_step": 0.722939133644104,
      "rewards/format_reward_step": 0.98046875,
      "step": 66
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8063938282430172,
      "aux_distill/mean_u": 0.24485005814775876,
      "aux_distill/n_active_tok": 176.875,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 14.796875,
      "calib/ece": 0.2592798353909465,
      "calib/final_conf_rate": 0.94921875,
      "calib/format_rate": 0.94921875,
      "calib/frac_conf_gt_0.9": 0.00823045267489712,
      "calib/gap": -0.030675828313253023,
      "calib/mean_conf": 0.15405349794238685,
      "calib/mu_c": 0.133855421686747,
      "calib/mu_w": 0.16453125000000002,
      "calib/nonempty_final_conf_rate": 0.94921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.035884773662551446,
      "calib/std_conf": 0.166563329748725,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2717168674698795,
      "calib/step_q_c_n": 996.0,
      "calib/step_q_gap": 0.04492449640970758,
      "calib/step_q_w": 0.22679237106017192,
      "calib/step_q_w_n": 2792.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.05078125,
      "completions/max_length": 1735.0,
      "completions/max_terminated_length": 1735.0,
      "completions/mean_length": 383.7421875,
      "completions/mean_terminated_length": 404.2716064453125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.07146666666666666,
      "grad_norm": 0.7894091010093689,
      "learning_rate": 3.7222222222222225e-06,
      "loss": 0.1327,
      "num_tokens": 14294491.0,
      "reward": 0.9720925092697144,
      "reward_std": 0.17863138020038605,
      "rewards/accuracy_reward_step": 0.33203125,
      "rewards/final_brier_reward_step": 0.6629350781440735,
      "rewards/format_reward_step": 0.94921875,
      "step": 67
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7980727730318904,
      "aux_distill/mean_u": 0.2929847676459035,
      "aux_distill/n_active_tok": 199.875,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 17.66015625,
      "calib/ece": 0.21976569037656904,
      "calib/final_conf_rate": 0.93359375,
      "calib/format_rate": 0.93359375,
      "calib/frac_conf_gt_0.9": 0.02092050209205021,
      "calib/gap": -0.004586732186732195,
      "calib/mean_conf": 0.1737071129707113,
      "calib/mu_c": 0.17054054054054055,
      "calib/mu_w": 0.17512727272727274,
      "calib/nonempty_final_conf_rate": 0.93359375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04192468619246862,
      "calib/std_conf": 0.18586754065481934,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2742739726027397,
      "calib/step_q_c_n": 730.0,
      "calib/step_q_gap": -0.031045995743343102,
      "calib/step_q_w": 0.3053199683460828,
      "calib/step_q_w_n": 3791.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.06640625,
      "completions/max_length": 1614.0,
      "completions/max_terminated_length": 1614.0,
      "completions/mean_length": 403.64453125,
      "completions/mean_terminated_length": 432.35565185546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 107.0,
      "epoch": 0.07253333333333334,
      "grad_norm": 1.416663408279419,
      "learning_rate": 3.694444444444445e-06,
      "loss": 0.0853,
      "num_tokens": 14501912.0,
      "reward": 0.9546322226524353,
      "reward_std": 0.20958805084228516,
      "rewards/accuracy_reward_step": 0.29296875,
      "rewards/final_brier_reward_step": 0.6827020049095154,
      "rewards/format_reward_step": 0.93359375,
      "step": 68
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7827791813760996,
      "aux_distill/mean_u": 0.2817591572783052,
      "aux_distill/n_active_tok": 236.25,
      "calib/answer_extract_rate": 0.9375,
      "calib/avg_num_step_conf": 23.88671875,
      "calib/ece": 0.27657333333333334,
      "calib/final_conf_rate": 0.87890625,
      "calib/format_rate": 0.87890625,
      "calib/frac_conf_gt_0.9": 0.017777777777777778,
      "calib/gap": -0.0463560344827586,
      "calib/mean_conf": 0.15924888888888888,
      "calib/mu_c": 0.12937500000000002,
      "calib/mu_w": 0.17573103448275862,
      "calib/nonempty_final_conf_rate": 0.87890625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04013333333333333,
      "calib/std_conf": 0.1781372013330327,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.29911104868913857,
      "calib/step_q_c_n": 1068.0,
      "calib/step_q_gap": 0.05638503323441299,
      "calib/step_q_w": 0.24272601545472558,
      "calib/step_q_w_n": 5047.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.12109375,
      "completions/max_length": 2415.0,
      "completions/max_terminated_length": 2415.0,
      "completions/mean_length": 432.390625,
      "completions/mean_terminated_length": 491.9644470214844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 137.0,
      "epoch": 0.0736,
      "grad_norm": 1.3896609544754028,
      "learning_rate": 3.6666666666666666e-06,
      "loss": -0.0331,
      "num_tokens": 14717100.0,
      "reward": 0.9001055955886841,
      "reward_std": 0.2826094627380371,
      "rewards/accuracy_reward_step": 0.32421875,
      "rewards/final_brier_reward_step": 0.5970861911773682,
      "rewards/format_reward_step": 0.87890625,
      "step": 69
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8019177615642548,
      "aux_distill/mean_u": 0.3130857905734622,
      "aux_distill/n_active_tok": 260.625,
      "calib/answer_extract_rate": 0.84765625,
      "calib/avg_num_step_conf": 29.87890625,
      "calib/ece": 0.17838095238095236,
      "calib/final_conf_rate": 0.8203125,
      "calib/format_rate": 0.8125,
      "calib/frac_conf_gt_0.9": 0.004761904761904762,
      "calib/gap": -0.02826281389748886,
      "calib/mean_conf": 0.16304761904761902,
      "calib/mu_c": 0.14245614035087717,
      "calib/mu_w": 0.17071895424836603,
      "calib/nonempty_final_conf_rate": 0.8203125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.035,
      "calib/std_conf": 0.16869879503633695,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2659308203991131,
      "calib/step_q_c_n": 902.0,
      "calib/step_q_gap": 0.02679896920599023,
      "calib/step_q_w": 0.23913185119312286,
      "calib/step_q_w_n": 6747.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1796875,
      "completions/max_length": 3065.0,
      "completions/max_terminated_length": 3065.0,
      "completions/mean_length": 443.58984375,
      "completions/mean_terminated_length": 540.7571411132812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 147.0,
      "epoch": 0.07466666666666667,
      "grad_norm": 2.984570264816284,
      "learning_rate": 3.638888888888889e-06,
      "loss": -0.1281,
      "num_tokens": 14937651.0,
      "reward": 0.8255972862243652,
      "reward_std": 0.34680598974227905,
      "rewards/accuracy_reward_step": 0.23046875,
      "rewards/final_brier_reward_step": 0.6082257628440857,
      "rewards/format_reward_step": 0.8125,
      "step": 70
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8115239245817065,
      "aux_distill/mean_u": 0.36577361168431416,
      "aux_distill/n_active_tok": 248.875,
      "calib/answer_extract_rate": 0.89453125,
      "calib/avg_num_step_conf": 26.7578125,
      "calib/ece": 0.21968468468468466,
      "calib/final_conf_rate": 0.8671875,
      "calib/format_rate": 0.86328125,
      "calib/frac_conf_gt_0.9": 0.0045045045045045045,
      "calib/gap": 0.0011015763454901295,
      "calib/mean_conf": 0.159954954954955,
      "calib/mu_c": 0.16070422535211265,
      "calib/mu_w": 0.15960264900662252,
      "calib/nonempty_final_conf_rate": 0.8671875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.029909909909909906,
      "calib/std_conf": 0.155692286453228,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2715031982942431,
      "calib/step_q_c_n": 938.0,
      "calib/step_q_gap": 0.0020297544512119514,
      "calib/step_q_w": 0.2694734438430311,
      "calib/step_q_w_n": 5912.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1328125,
      "completions/max_length": 2533.0,
      "completions/max_terminated_length": 2533.0,
      "completions/mean_length": 462.32421875,
      "completions/mean_terminated_length": 533.130615234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 110.0,
      "epoch": 0.07573333333333333,
      "grad_norm": 1.4650132656097412,
      "learning_rate": 3.6111111111111115e-06,
      "loss": 0.0103,
      "num_tokens": 15160414.0,
      "reward": 0.888353705406189,
      "reward_std": 0.30983930826187134,
      "rewards/accuracy_reward_step": 0.28125,
      "rewards/final_brier_reward_step": 0.6321761608123779,
      "rewards/format_reward_step": 0.86328125,
      "step": 71
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7690277192741632,
      "aux_distill/mean_u": 0.3222443399292448,
      "aux_distill/n_active_tok": 245.125,
      "calib/answer_extract_rate": 0.9375,
      "calib/avg_num_step_conf": 22.91015625,
      "calib/ece": 0.16442553191489365,
      "calib/final_conf_rate": 0.91796875,
      "calib/format_rate": 0.91796875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.014400000000000024,
      "calib/mean_conf": 0.17872340425531916,
      "calib/mu_c": 0.16799999999999998,
      "calib/mu_w": 0.1824,
      "calib/nonempty_final_conf_rate": 0.91796875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.04391489361702128,
      "calib/std_conf": 0.15519948756874982,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.25215622732769044,
      "calib/step_q_c_n": 827.0,
      "calib/step_q_gap": -0.0312737051852115,
      "calib/step_q_w": 0.28342993251290194,
      "calib/step_q_w_n": 5038.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08203125,
      "completions/max_length": 2636.0,
      "completions/max_terminated_length": 2636.0,
      "completions/mean_length": 511.328125,
      "completions/mean_terminated_length": 557.021240234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 110.0,
      "epoch": 0.0768,
      "grad_norm": 0.9580546021461487,
      "learning_rate": 3.5833333333333335e-06,
      "loss": 0.0729,
      "num_tokens": 15395722.0,
      "reward": 0.9316273331642151,
      "reward_std": 0.2351001799106598,
      "rewards/accuracy_reward_step": 0.234375,
      "rewards/final_brier_reward_step": 0.7109109163284302,
      "rewards/format_reward_step": 0.91796875,
      "step": 72
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7873094780370593,
      "aux_distill/mean_u": 0.2826349133400705,
      "aux_distill/n_active_tok": 212.0,
      "calib/answer_extract_rate": 0.921875,
      "calib/avg_num_step_conf": 19.82421875,
      "calib/ece": 0.2566312352930599,
      "calib/final_conf_rate": 0.91796875,
      "calib/format_rate": 0.91796875,
      "calib/frac_conf_gt_0.9": 0.03829787234042553,
      "calib/gap": -0.031687061508404446,
      "calib/mean_conf": 0.2211559987494933,
      "calib/mu_c": 0.2003908605695176,
      "calib/mu_w": 0.23207792207792205,
      "calib/nonempty_final_conf_rate": 0.91796875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0665531914893617,
      "calib/std_conf": 0.2149404265793188,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3638418079096045,
      "calib/step_q_c_n": 885.0,
      "calib/step_q_gap": 0.03793607998597681,
      "calib/step_q_w": 0.3259057279236277,
      "calib/step_q_w_n": 4190.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08203125,
      "completions/max_length": 2226.0,
      "completions/max_terminated_length": 2226.0,
      "completions/mean_length": 423.15234375,
      "completions/mean_terminated_length": 460.9659423828125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 107.0,
      "epoch": 0.07786666666666667,
      "grad_norm": 0.8632502555847168,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0653,
      "num_tokens": 15611081.0,
      "reward": 0.937720000743866,
      "reward_std": 0.2524094581604004,
      "rewards/accuracy_reward_step": 0.31640625,
      "rewards/final_brier_reward_step": 0.6410648822784424,
      "rewards/format_reward_step": 0.91796875,
      "step": 73
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8297929093241692,
      "aux_distill/mean_u": 0.3083530604700027,
      "aux_distill/n_active_tok": 203.375,
      "calib/answer_extract_rate": 0.9609375,
      "calib/avg_num_step_conf": 15.5703125,
      "calib/ece": 0.12463332343236579,
      "calib/final_conf_rate": 0.9609375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.012195121951219513,
      "calib/gap": 0.07146305939212541,
      "calib/mean_conf": 0.19877966489578042,
      "calib/mu_c": 0.2536842105263158,
      "calib/mu_w": 0.18222115113419038,
      "calib/nonempty_final_conf_rate": 0.9609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.04585283562748774,
      "calib/std_conf": 0.18025502827472875,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3164716006884682,
      "calib/step_q_c_n": 581.0,
      "calib/step_q_gap": -0.012599678947922632,
      "calib/step_q_w": 0.3290712796363908,
      "calib/step_q_w_n": 3405.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0390625,
      "completions/max_length": 2231.0,
      "completions/max_terminated_length": 2231.0,
      "completions/mean_length": 400.96875,
      "completions/mean_terminated_length": 417.2682800292969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 96.0,
      "epoch": 0.07893333333333333,
      "grad_norm": 0.44695550203323364,
      "learning_rate": 3.5277777777777784e-06,
      "loss": 0.1973,
      "num_tokens": 15817657.0,
      "reward": 0.9808335900306702,
      "reward_std": 0.17156293988227844,
      "rewards/accuracy_reward_step": 0.22265625,
      "rewards/final_brier_reward_step": 0.7819797992706299,
      "rewards/format_reward_step": 0.95703125,
      "step": 74
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8280173083767295,
      "aux_distill/mean_u": 0.3092613757817121,
      "aux_distill/n_active_tok": 160.375,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 12.41796875,
      "calib/ece": 0.29354824224200393,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.008032128514056224,
      "calib/gap": -0.035276279308424696,
      "calib/mean_conf": 0.17400195856120884,
      "calib/mu_c": 0.15303453150238613,
      "calib/mu_w": 0.18831081081081083,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03096385542168674,
      "calib/std_conf": 0.1677046066909969,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2633288827668051,
      "calib/step_q_c_n": 867.0,
      "calib/step_q_gap": 0.0049077322477739105,
      "calib/step_q_w": 0.2584211505190312,
      "calib/step_q_w_n": 2312.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.02734375,
      "completions/max_length": 3068.0,
      "completions/max_terminated_length": 3068.0,
      "completions/mean_length": 390.31640625,
      "completions/mean_terminated_length": 401.2891540527344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 86.0,
      "epoch": 0.08,
      "grad_norm": 0.5844130516052246,
      "learning_rate": 3.5e-06,
      "loss": 0.2072,
      "num_tokens": 16022330.0,
      "reward": 1.0046308040618896,
      "reward_std": 0.15581795573234558,
      "rewards/accuracy_reward_step": 0.39453125,
      "rewards/final_brier_reward_step": 0.6420742273330688,
      "rewards/format_reward_step": 0.97265625,
      "step": 75
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8842153791338205,
      "aux_distill/mean_u": 0.3634484035747906,
      "aux_distill/n_active_tok": 141.5,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.1015625,
      "calib/ece": 0.23534384920634918,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.011904761904761904,
      "calib/gap": -0.017342894461859937,
      "calib/mean_conf": 0.19783075396825398,
      "calib/mu_c": 0.18647528735632185,
      "calib/mu_w": 0.2038181818181818,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04396825396825396,
      "calib/std_conf": 0.19580172393485226,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2866044303797468,
      "calib/step_q_c_n": 711.0,
      "calib/step_q_gap": 0.00703109704641347,
      "calib/step_q_w": 0.27957333333333334,
      "calib/step_q_w_n": 1875.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1357.0,
      "completions/max_terminated_length": 1357.0,
      "completions/mean_length": 348.10546875,
      "completions/mean_terminated_length": 353.6309814453125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 75.0,
      "epoch": 0.08106666666666666,
      "grad_norm": 0.5333477854728699,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 0.2624,
      "num_tokens": 16214501.0,
      "reward": 1.0057470798492432,
      "reward_std": 0.15070591866970062,
      "rewards/accuracy_reward_step": 0.33984375,
      "rewards/final_brier_reward_step": 0.6911816596984863,
      "rewards/format_reward_step": 0.98046875,
      "step": 76
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8543671872466803,
      "aux_distill/mean_u": 0.33429203086375125,
      "aux_distill/n_active_tok": 122.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 8.23828125,
      "calib/ece": 0.17608695652173917,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.011857707509881422,
      "calib/gap": 0.04061777965624119,
      "calib/mean_conf": 0.19727272727272727,
      "calib/mu_c": 0.2244047619047619,
      "calib/mu_w": 0.18378698224852072,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.020671936758893283,
      "calib/std_conf": 0.18031174345926798,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3276148409893993,
      "calib/step_q_c_n": 566.0,
      "calib/step_q_gap": 0.04176941001078621,
      "calib/step_q_w": 0.2858454309786131,
      "calib/step_q_w_n": 1543.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 874.0,
      "completions/max_terminated_length": 874.0,
      "completions/mean_length": 292.85546875,
      "completions/mean_terminated_length": 296.32806396484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 94.0,
      "epoch": 0.08213333333333334,
      "grad_norm": 0.6401911973953247,
      "learning_rate": 3.444444444444445e-06,
      "loss": 0.2291,
      "num_tokens": 16394136.0,
      "reward": 1.019188404083252,
      "reward_std": 0.14684052765369415,
      "rewards/accuracy_reward_step": 0.328125,
      "rewards/final_brier_reward_step": 0.7297831773757935,
      "rewards/format_reward_step": 0.98046875,
      "step": 77
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.856964997947216,
      "aux_distill/mean_u": 0.3375044024240291,
      "aux_distill/n_active_tok": 111.125,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 8.05078125,
      "calib/ece": 0.2119047619047619,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.015873015873015872,
      "calib/gap": 0.0001949317738791645,
      "calib/mean_conf": 0.20912698412698413,
      "calib/mu_c": 0.20925925925925928,
      "calib/mu_w": 0.20906432748538012,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.049801587301587305,
      "calib/std_conf": 0.202887958138405,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.3684361842105263,
      "calib/step_q_c_n": 608.0,
      "calib/step_q_gap": 0.04266309405223312,
      "calib/step_q_w": 0.3257730901582932,
      "calib/step_q_w_n": 1453.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 2738.0,
      "completions/max_terminated_length": 2738.0,
      "completions/mean_length": 286.5078125,
      "completions/mean_terminated_length": 291.0555725097656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 101.0,
      "epoch": 0.0832,
      "grad_norm": 0.9656935930252075,
      "learning_rate": 3.416666666666667e-06,
      "loss": 0.2081,
      "num_tokens": 16575506.0,
      "reward": 1.0049442052841187,
      "reward_std": 0.146946519613266,
      "rewards/accuracy_reward_step": 0.31640625,
      "rewards/final_brier_reward_step": 0.7130132913589478,
      "rewards/format_reward_step": 0.98046875,
      "step": 78
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8454724475741386,
      "aux_distill/mean_u": 0.36820621793473934,
      "aux_distill/n_active_tok": 148.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.87109375,
      "calib/ece": 0.16129370078740157,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.023622047244094488,
      "calib/gap": 0.057363827133318585,
      "calib/mean_conf": 0.21405196850393704,
      "calib/mu_c": 0.254025974025974,
      "calib/mu_w": 0.1966621468926554,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.036098031496062985,
      "calib/std_conf": 0.2157638236920991,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3824413145539906,
      "calib/step_q_c_n": 639.0,
      "calib/step_q_gap": 0.06627826308340234,
      "calib/step_q_w": 0.31616305147058826,
      "calib/step_q_w_n": 1632.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1231.0,
      "completions/max_terminated_length": 1231.0,
      "completions/mean_length": 341.9609375,
      "completions/mean_terminated_length": 344.6535339355469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 80.0,
      "epoch": 0.08426666666666667,
      "grad_norm": 0.8393629193305969,
      "learning_rate": 3.3888888888888893e-06,
      "loss": 0.2185,
      "num_tokens": 16769424.0,
      "reward": 1.022768497467041,
      "reward_std": 0.1480945348739624,
      "rewards/accuracy_reward_step": 0.30078125,
      "rewards/final_brier_reward_step": 0.7525681257247925,
      "rewards/format_reward_step": 0.9921875,
      "step": 79
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.831517806276679,
      "aux_distill/mean_u": 0.2914098852630286,
      "aux_distill/n_active_tok": 171.125,
      "calib/answer_extract_rate": 0.953125,
      "calib/avg_num_step_conf": 12.109375,
      "calib/ece": 0.20660023795359905,
      "calib/final_conf_rate": 0.9609375,
      "calib/format_rate": 0.953125,
      "calib/frac_conf_gt_0.9": 0.024390243902439025,
      "calib/gap": 0.009875955039670925,
      "calib/mean_conf": 0.22096073765615706,
      "calib/mu_c": 0.22762500000000002,
      "calib/mu_w": 0.2177490449603291,
      "calib/nonempty_final_conf_rate": 0.9609375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.05117886178861789,
      "calib/std_conf": 0.20067466003042428,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3260610263522885,
      "calib/step_q_c_n": 721.0,
      "calib/step_q_gap": 0.028705904320342257,
      "calib/step_q_w": 0.29735512203194625,
      "calib/step_q_w_n": 2379.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03515625,
      "completions/max_length": 2370.0,
      "completions/max_terminated_length": 2370.0,
      "completions/mean_length": 429.87890625,
      "completions/mean_terminated_length": 445.5425109863281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.08533333333333333,
      "grad_norm": 2.0035974979400635,
      "learning_rate": 3.3611111111111117e-06,
      "loss": 0.2209,
      "num_tokens": 16981633.0,
      "reward": 0.9816856980323792,
      "reward_std": 0.21914240717887878,
      "rewards/accuracy_reward_step": 0.3125,
      "rewards/final_brier_reward_step": 0.6977464556694031,
      "rewards/format_reward_step": 0.953125,
      "step": 80
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8161282241344452,
      "aux_distill/mean_u": 0.2957661482389787,
      "aux_distill/n_active_tok": 191.625,
      "calib/answer_extract_rate": 0.69140625,
      "calib/avg_num_step_conf": 26.234375,
      "calib/ece": 0.20500000000000002,
      "calib/final_conf_rate": 0.6953125,
      "calib/format_rate": 0.6796875,
      "calib/frac_conf_gt_0.9": 0.02247191011235955,
      "calib/gap": 0.010498504486540439,
      "calib/mean_conf": 0.22230337078651682,
      "calib/mu_c": 0.22932203389830508,
      "calib/mu_w": 0.21882352941176464,
      "calib/nonempty_final_conf_rate": 0.6953125,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.047921348314606746,
      "calib/std_conf": 0.21159845927172033,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.34729551451187335,
      "calib/step_q_c_n": 758.0,
      "calib/step_q_gap": -0.0039771776667100744,
      "calib/step_q_w": 0.3512726921785834,
      "calib/step_q_w_n": 5958.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.27734375,
      "completions/max_length": 3030.0,
      "completions/max_terminated_length": 3030.0,
      "completions/mean_length": 394.12890625,
      "completions/mean_terminated_length": 545.3892211914062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.0864,
      "grad_norm": 3.689892053604126,
      "learning_rate": 3.3333333333333333e-06,
      "loss": -0.1709,
      "num_tokens": 17188778.0,
      "reward": 0.701918363571167,
      "reward_std": 0.4659685492515564,
      "rewards/accuracy_reward_step": 0.234375,
      "rewards/final_brier_reward_step": 0.489774227142334,
      "rewards/format_reward_step": 0.6796875,
      "step": 81
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7570554517930553,
      "aux_distill/mean_u": 0.2875513874466876,
      "aux_distill/n_active_tok": 217.5483870967742,
      "calib/answer_extract_rate": 0.51953125,
      "calib/avg_num_step_conf": 29.359375,
      "calib/ece": 0.21593984962406013,
      "calib/final_conf_rate": 0.51953125,
      "calib/format_rate": 0.51171875,
      "calib/frac_conf_gt_0.9": 0.015037593984962405,
      "calib/gap": -0.006004497751124438,
      "calib/mean_conf": 0.242406015037594,
      "calib/mu_c": 0.23847826086956522,
      "calib/mu_w": 0.24448275862068966,
      "calib/nonempty_final_conf_rate": 0.51953125,
      "calib/nonempty_reasoning_rate": 0.96484375,
      "calib/nonempty_step_conf_rate": 0.96484375,
      "calib/pce": 0.0562406015037594,
      "calib/std_conf": 0.20432964070120244,
      "calib/step_conf_rate": 0.96484375,
      "calib/step_q_c": 0.3298311444652908,
      "calib/step_q_c_n": 533.0,
      "calib/step_q_gap": -0.07547177691520468,
      "calib/step_q_w": 0.4053029213804955,
      "calib/step_q_w_n": 6983.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.421875,
      "completions/max_length": 3071.0,
      "completions/max_terminated_length": 3071.0,
      "completions/mean_length": 385.625,
      "completions/mean_terminated_length": 667.0270385742188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 8.0,
      "epoch": 0.08746666666666666,
      "grad_norm": 8.525459289550781,
      "learning_rate": 3.3055555555555558e-06,
      "loss": -0.366,
      "num_tokens": 17393050.0,
      "reward": 0.5286312699317932,
      "reward_std": 0.5281660556793213,
      "rewards/accuracy_reward_step": 0.1796875,
      "rewards/final_brier_reward_step": 0.36585623025894165,
      "rewards/format_reward_step": 0.51171875,
      "step": 82
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7571941297501326,
      "aux_distill/mean_u": 0.2952067848573532,
      "aux_distill/n_active_tok": 255.875,
      "calib/answer_extract_rate": 0.62109375,
      "calib/avg_num_step_conf": 27.77734375,
      "calib/ece": 0.2588588957055215,
      "calib/final_conf_rate": 0.63671875,
      "calib/format_rate": 0.61328125,
      "calib/frac_conf_gt_0.9": 0.012269938650306749,
      "calib/gap": -0.07034216785592931,
      "calib/mean_conf": 0.23574233128834357,
      "calib/mu_c": 0.18870370370370373,
      "calib/mu_w": 0.25904587155963305,
      "calib/nonempty_final_conf_rate": 0.63671875,
      "calib/nonempty_reasoning_rate": 0.98828125,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.0816564417177914,
      "calib/std_conf": 0.20480982017762447,
      "calib/step_conf_rate": 0.98828125,
      "calib/step_q_c": 0.30772997032640953,
      "calib/step_q_c_n": 674.0,
      "calib/step_q_gap": -0.06138778639255893,
      "calib/step_q_w": 0.36911775671896846,
      "calib/step_q_w_n": 6437.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.30078125,
      "completions/max_length": 2857.0,
      "completions/max_terminated_length": 2857.0,
      "completions/mean_length": 454.484375,
      "completions/mean_terminated_length": 649.9888305664062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 111.0,
      "epoch": 0.08853333333333334,
      "grad_norm": 3.0912773609161377,
      "learning_rate": 3.277777777777778e-06,
      "loss": -0.194,
      "num_tokens": 17616662.0,
      "reward": 0.6224585771560669,
      "reward_std": 0.4793688654899597,
      "rewards/accuracy_reward_step": 0.2109375,
      "rewards/final_brier_reward_step": 0.420698344707489,
      "rewards/format_reward_step": 0.61328125,
      "step": 83
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7821133900433779,
      "aux_distill/mean_u": 0.31811979441848154,
      "aux_distill/n_active_tok": 229.125,
      "calib/answer_extract_rate": 0.72265625,
      "calib/avg_num_step_conf": 24.9453125,
      "calib/ece": 0.19843243243243244,
      "calib/final_conf_rate": 0.72265625,
      "calib/format_rate": 0.71484375,
      "calib/frac_conf_gt_0.9": 0.02702702702702703,
      "calib/gap": -0.043775894538606475,
      "calib/mean_conf": 0.2565945945945946,
      "calib/mu_c": 0.22677966101694913,
      "calib/mu_w": 0.2705555555555556,
      "calib/nonempty_final_conf_rate": 0.72265625,
      "calib/nonempty_reasoning_rate": 0.98828125,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.06805405405405407,
      "calib/std_conf": 0.21356123996180776,
      "calib/step_conf_rate": 0.98828125,
      "calib/step_q_c": 0.3681395348837209,
      "calib/step_q_c_n": 774.0,
      "calib/step_q_gap": -0.03123756775348513,
      "calib/step_q_w": 0.39937710263720605,
      "calib/step_q_w_n": 5612.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.19921875,
      "completions/max_length": 2697.0,
      "completions/max_terminated_length": 2697.0,
      "completions/mean_length": 455.98046875,
      "completions/mean_terminated_length": 569.4194946289062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 41.0,
      "epoch": 0.0896,
      "grad_norm": 1.6149286031723022,
      "learning_rate": 3.2500000000000002e-06,
      "loss": -0.0904,
      "num_tokens": 17839313.0,
      "reward": 0.7286218404769897,
      "reward_std": 0.47204819321632385,
      "rewards/accuracy_reward_step": 0.23046875,
      "rewards/final_brier_reward_step": 0.5119312405586243,
      "rewards/format_reward_step": 0.71484375,
      "step": 84
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8418309669941664,
      "aux_distill/mean_u": 0.30520798490763335,
      "aux_distill/n_active_tok": 183.625,
      "calib/answer_extract_rate": 0.9375,
      "calib/avg_num_step_conf": 15.20703125,
      "calib/ece": 0.21260330578512396,
      "calib/final_conf_rate": 0.9453125,
      "calib/format_rate": 0.9375,
      "calib/frac_conf_gt_0.9": 0.049586776859504134,
      "calib/gap": 0.029399350649350675,
      "calib/mean_conf": 0.27004132231404954,
      "calib/mu_c": 0.28875,
      "calib/mu_w": 0.25935064935064933,
      "calib/nonempty_final_conf_rate": 0.9453125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05950413223140496,
      "calib/std_conf": 0.2362106111882379,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3968901734104046,
      "calib/step_q_c_n": 865.0,
      "calib/step_q_gap": 0.025058931666679385,
      "calib/step_q_w": 0.3718312417437252,
      "calib/step_q_w_n": 3028.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.046875,
      "completions/max_length": 2467.0,
      "completions/max_terminated_length": 2467.0,
      "completions/mean_length": 410.4453125,
      "completions/mean_terminated_length": 430.6311340332031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 0.09066666666666667,
      "grad_norm": 0.6718313097953796,
      "learning_rate": 3.2222222222222227e-06,
      "loss": 0.1399,
      "num_tokens": 18052211.0,
      "reward": 0.9761437177658081,
      "reward_std": 0.251767098903656,
      "rewards/accuracy_reward_step": 0.34375,
      "rewards/final_brier_reward_step": 0.671037495136261,
      "rewards/format_reward_step": 0.9375,
      "step": 85
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8239959748461843,
      "aux_distill/mean_u": 0.319219554660902,
      "aux_distill/n_active_tok": 175.25,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 12.54296875,
      "calib/ece": 0.2043265306122449,
      "calib/final_conf_rate": 0.95703125,
      "calib/format_rate": 0.9453125,
      "calib/frac_conf_gt_0.9": 0.02857142857142857,
      "calib/gap": -0.009132947976878636,
      "calib/mean_conf": 0.20644897959183672,
      "calib/mu_c": 0.2,
      "calib/mu_w": 0.20913294797687865,
      "calib/nonempty_final_conf_rate": 0.95703125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.058448979591836744,
      "calib/std_conf": 0.2032512330182736,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3402790697674418,
      "calib/step_q_c_n": 645.0,
      "calib/step_q_gap": -0.013185466475738206,
      "calib/step_q_w": 0.35346453624318,
      "calib/step_q_w_n": 2566.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03125,
      "completions/max_length": 2073.0,
      "completions/max_terminated_length": 2073.0,
      "completions/mean_length": 374.6640625,
      "completions/mean_terminated_length": 386.75,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 108.0,
      "epoch": 0.09173333333333333,
      "grad_norm": 0.5359158515930176,
      "learning_rate": 3.1944444444444443e-06,
      "loss": 0.1646,
      "num_tokens": 18253637.0,
      "reward": 0.9615132808685303,
      "reward_std": 0.20205360651016235,
      "rewards/accuracy_reward_step": 0.28125,
      "rewards/final_brier_reward_step": 0.6964640617370605,
      "rewards/format_reward_step": 0.9453125,
      "step": 86
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.803918688558042,
      "aux_distill/mean_u": 0.2932792815270562,
      "aux_distill/n_active_tok": 159.0,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.21484375,
      "calib/ece": 0.30135999999999996,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.02,
      "calib/gap": -0.022649350649350697,
      "calib/mean_conf": 0.20632000000000003,
      "calib/mu_c": 0.1936363636363636,
      "calib/mu_w": 0.2162857142857143,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.03384000000000001,
      "calib/std_conf": 0.18657560826646125,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3034076433121019,
      "calib/step_q_c_n": 942.0,
      "calib/step_q_gap": -0.05141004945538169,
      "calib/step_q_w": 0.3548176927674836,
      "calib/step_q_w_n": 1673.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 3022.0,
      "completions/max_terminated_length": 3022.0,
      "completions/mean_length": 347.71875,
      "completions/mean_terminated_length": 354.6454162597656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 92.0,
      "epoch": 0.0928,
      "grad_norm": 0.5403491854667664,
      "learning_rate": 3.1666666666666667e-06,
      "loss": 0.1776,
      "num_tokens": 18448149.0,
      "reward": 1.0203272104263306,
      "reward_std": 0.17660395801067352,
      "rewards/accuracy_reward_step": 0.43359375,
      "rewards/final_brier_reward_step": 0.6344043016433716,
      "rewards/format_reward_step": 0.97265625,
      "step": 87
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.876606872305274,
      "aux_distill/mean_u": 0.2893717978872386,
      "aux_distill/n_active_tok": 136.125,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.7890625,
      "calib/ece": 0.2384285714285714,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.003968253968253968,
      "calib/gap": -0.003660256410256385,
      "calib/mean_conf": 0.18934920634920635,
      "calib/mu_c": 0.18708333333333335,
      "calib/mu_w": 0.19074358974358974,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.023412698412698413,
      "calib/std_conf": 0.1634352120294649,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2629210836277974,
      "calib/step_q_c_n": 849.0,
      "calib/step_q_gap": -0.02230522898535886,
      "calib/step_q_w": 0.2852263126131563,
      "calib/step_q_w_n": 1657.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1649.0,
      "completions/max_terminated_length": 1649.0,
      "completions/mean_length": 345.0546875,
      "completions/mean_terminated_length": 350.5317687988281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 36.0,
      "epoch": 0.09386666666666667,
      "grad_norm": 0.5013801455497742,
      "learning_rate": 3.138888888888889e-06,
      "loss": 0.233,
      "num_tokens": 18646331.0,
      "reward": 1.019582748413086,
      "reward_std": 0.14026647806167603,
      "rewards/accuracy_reward_step": 0.37890625,
      "rewards/final_brier_reward_step": 0.6836965084075928,
      "rewards/format_reward_step": 0.9765625,
      "step": 88
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8640087898820639,
      "aux_distill/mean_u": 0.3608727952693905,
      "aux_distill/n_active_tok": 140.125,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.90234375,
      "calib/ece": 0.18127058823529407,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0196078431372549,
      "calib/gap": 0.037335528159148146,
      "calib/mean_conf": 0.19276862745098042,
      "calib/mu_c": 0.2179518072289156,
      "calib/mu_w": 0.18061627906976746,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.024274509803921575,
      "calib/std_conf": 0.18715120602797072,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3426186830015313,
      "calib/step_q_c_n": 653.0,
      "calib/step_q_gap": 0.04391204093511064,
      "calib/step_q_w": 0.29870664206642067,
      "calib/step_q_w_n": 1626.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1065.0,
      "completions/max_terminated_length": 1065.0,
      "completions/mean_length": 346.13671875,
      "completions/mean_terminated_length": 348.8622131347656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 92.0,
      "epoch": 0.09493333333333333,
      "grad_norm": 0.5498507618904114,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.266,
      "num_tokens": 18843830.0,
      "reward": 1.0269007682800293,
      "reward_std": 0.13228027522563934,
      "rewards/accuracy_reward_step": 0.32421875,
      "rewards/final_brier_reward_step": 0.7373951077461243,
      "rewards/format_reward_step": 0.9921875,
      "step": 89
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8565371185541153,
      "aux_distill/mean_u": 0.307992225866492,
      "aux_distill/n_active_tok": 141.0,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.86328125,
      "calib/ece": 0.23880478087649398,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.01195219123505976,
      "calib/gap": 0.007009454796411352,
      "calib/mean_conf": 0.2145816733067729,
      "calib/mu_c": 0.21907777777777782,
      "calib/mu_w": 0.21206832298136646,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04741035856573704,
      "calib/std_conf": 0.198886632863703,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.33993487698986974,
      "calib/step_q_c_n": 691.0,
      "calib/step_q_gap": 0.002947417884090031,
      "calib/step_q_w": 0.3369874591057797,
      "calib/step_q_w_n": 1834.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 2610.0,
      "completions/max_terminated_length": 2610.0,
      "completions/mean_length": 351.59765625,
      "completions/mean_terminated_length": 358.6015930175781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.096,
      "grad_norm": 0.4637305438518524,
      "learning_rate": 3.0833333333333336e-06,
      "loss": 0.2443,
      "num_tokens": 19037159.0,
      "reward": 1.011687994003296,
      "reward_std": 0.16643886268138885,
      "rewards/accuracy_reward_step": 0.3515625,
      "rewards/final_brier_reward_step": 0.6952507495880127,
      "rewards/format_reward_step": 0.9765625,
      "step": 90
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8760084416717291,
      "aux_distill/mean_u": 0.28022362415155183,
      "aux_distill/n_active_tok": 129.5,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.640625,
      "calib/ece": 0.258492125984252,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.015748031496062992,
      "calib/gap": -0.027924675324675324,
      "calib/mean_conf": 0.20683070866141734,
      "calib/mu_c": 0.1899,
      "calib/mu_w": 0.21782467532467534,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.03581102362204725,
      "calib/std_conf": 0.17850015682377415,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2720631313131313,
      "calib/step_q_c_n": 792.0,
      "calib/step_q_gap": -0.08380024896855881,
      "calib/step_q_w": 0.3558633802816901,
      "calib/step_q_w_n": 1420.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1145.0,
      "completions/max_terminated_length": 1145.0,
      "completions/mean_length": 339.00390625,
      "completions/mean_terminated_length": 341.6732177734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.09706666666666666,
      "grad_norm": 0.5233395099639893,
      "learning_rate": 3.055555555555556e-06,
      "loss": 0.2388,
      "num_tokens": 19231656.0,
      "reward": 1.0215727090835571,
      "reward_std": 0.14010092616081238,
      "rewards/accuracy_reward_step": 0.390625,
      "rewards/final_brier_reward_step": 0.6681454181671143,
      "rewards/format_reward_step": 0.984375,
      "step": 91
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8196988068521023,
      "aux_distill/mean_u": 0.24410908331532602,
      "aux_distill/n_active_tok": 123.625,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.5,
      "calib/ece": 0.21699604743083004,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.019762845849802372,
      "calib/gap": 0.008241670012043234,
      "calib/mean_conf": 0.21928853754940714,
      "calib/mu_c": 0.22446808510638291,
      "calib/mu_w": 0.21622641509433968,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.032371541501976284,
      "calib/std_conf": 0.20788290958580805,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.29255714285714285,
      "calib/step_q_c_n": 700.0,
      "calib/step_q_gap": -0.015381365803365243,
      "calib/step_q_w": 0.3079385086605081,
      "calib/step_q_w_n": 1732.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1419.0,
      "completions/max_terminated_length": 1419.0,
      "completions/mean_length": 337.625,
      "completions/mean_terminated_length": 342.9841613769531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.09813333333333334,
      "grad_norm": 0.5248092412948608,
      "learning_rate": 3.0277777777777776e-06,
      "loss": 0.2386,
      "num_tokens": 19424808.0,
      "reward": 1.0217368602752686,
      "reward_std": 0.15499769151210785,
      "rewards/accuracy_reward_step": 0.3671875,
      "rewards/final_brier_reward_step": 0.6919113397598267,
      "rewards/format_reward_step": 0.984375,
      "step": 92
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8715116884559393,
      "aux_distill/mean_u": 0.292919875969081,
      "aux_distill/n_active_tok": 134.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.38671875,
      "calib/ece": 0.2876086956521739,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.015810276679841896,
      "calib/gap": -0.03291698595146875,
      "calib/mean_conf": 0.2075691699604743,
      "calib/mu_c": 0.1887037037037037,
      "calib/mu_w": 0.22162068965517245,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0341501976284585,
      "calib/std_conf": 0.18972447298727482,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28146012269938653,
      "calib/step_q_c_n": 815.0,
      "calib/step_q_gap": -0.07955750954242713,
      "calib/step_q_w": 0.36101763224181366,
      "calib/step_q_w_n": 1588.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2593.0,
      "completions/max_terminated_length": 2593.0,
      "completions/mean_length": 341.03125,
      "completions/mean_terminated_length": 345.0751037597656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.0992,
      "grad_norm": 0.4833468794822693,
      "learning_rate": 3e-06,
      "loss": 0.263,
      "num_tokens": 19617888.0,
      "reward": 1.0288138389587402,
      "reward_std": 0.14168192446231842,
      "rewards/accuracy_reward_step": 0.421875,
      "rewards/final_brier_reward_step": 0.6474713683128357,
      "rewards/format_reward_step": 0.98828125,
      "step": 93
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.828835241496563,
      "aux_distill/mean_u": 0.27052133899454756,
      "aux_distill/n_active_tok": 121.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.53515625,
      "calib/ece": 0.34593800392156865,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.011764705882352941,
      "calib/gap": -0.030389300000000008,
      "calib/mean_conf": 0.20253258431372548,
      "calib/mu_c": 0.18704,
      "calib/mu_w": 0.21742930000000002,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.029137254901960792,
      "calib/std_conf": 0.2013929581370898,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2530616996507567,
      "calib/step_q_c_n": 859.0,
      "calib/step_q_gap": -0.032922238667000336,
      "calib/step_q_w": 0.28598393831775704,
      "calib/step_q_w_n": 1070.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 777.0,
      "completions/max_terminated_length": 777.0,
      "completions/mean_length": 299.28125,
      "completions/mean_terminated_length": 301.6377868652344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 111.0,
      "epoch": 0.10026666666666667,
      "grad_norm": 0.5058930516242981,
      "learning_rate": 2.9722222222222225e-06,
      "loss": 0.2347,
      "num_tokens": 19803184.0,
      "reward": 1.046791911125183,
      "reward_std": 0.12935060262680054,
      "rewards/accuracy_reward_step": 0.48828125,
      "rewards/final_brier_reward_step": 0.6092088222503662,
      "rewards/format_reward_step": 0.99609375,
      "step": 94
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9062896221876144,
      "aux_distill/mean_u": 0.3493818523568767,
      "aux_distill/n_active_tok": 113.75,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.93359375,
      "calib/ece": 0.28477968749999993,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.0078125,
      "calib/gap": 0.006114682539682503,
      "calib/mean_conf": 0.18225156250000002,
      "calib/mu_c": 0.1856910714285714,
      "calib/mu_w": 0.1795763888888889,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.014765625000000001,
      "calib/std_conf": 0.1670195296329402,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2656130909090909,
      "calib/step_q_c_n": 825.0,
      "calib/step_q_gap": 0.026708447459671325,
      "calib/step_q_w": 0.23890464344941958,
      "calib/step_q_w_n": 1206.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 934.0,
      "completions/max_terminated_length": 934.0,
      "completions/mean_length": 314.15234375,
      "completions/mean_terminated_length": 316.6259765625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 89.0,
      "epoch": 0.10133333333333333,
      "grad_norm": 0.5349423885345459,
      "learning_rate": 2.944444444444445e-06,
      "loss": 0.2295,
      "num_tokens": 19989735.0,
      "reward": 1.0411256551742554,
      "reward_std": 0.11611491441726685,
      "rewards/accuracy_reward_step": 0.4375,
      "rewards/final_brier_reward_step": 0.6564700603485107,
      "rewards/format_reward_step": 0.98828125,
      "step": 95
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8619091520085931,
      "aux_distill/mean_u": 0.3177986059032521,
      "aux_distill/n_active_tok": 143.75,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 9.99609375,
      "calib/ece": 0.3648896,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.004,
      "calib/gap": 0.0050328205128205294,
      "calib/mean_conf": 0.16543040000000003,
      "calib/mu_c": 0.16784615384615384,
      "calib/mu_w": 0.1628133333333333,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0051600000000000005,
      "calib/std_conf": 0.1503400339092685,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2535573122529644,
      "calib/step_q_c_n": 1012.0,
      "calib/step_q_gap": 0.02521956176815507,
      "calib/step_q_w": 0.22833775048480934,
      "calib/step_q_w_n": 1547.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 1337.0,
      "completions/max_terminated_length": 1337.0,
      "completions/mean_length": 348.24609375,
      "completions/mean_terminated_length": 355.18328857421875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 74.0,
      "epoch": 0.1024,
      "grad_norm": 0.5326002240180969,
      "learning_rate": 2.916666666666667e-06,
      "loss": 0.2346,
      "num_tokens": 20184702.0,
      "reward": 1.0335478782653809,
      "reward_std": 0.15459981560707092,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.5866272449493408,
      "rewards/format_reward_step": 0.97265625,
      "step": 96
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8531998693943024,
      "aux_distill/mean_u": 0.2890356585287972,
      "aux_distill/n_active_tok": 144.625,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 10.1796875,
      "calib/ece": 0.20959055118110237,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.003937007874015748,
      "calib/gap": 0.013189355742296932,
      "calib/mean_conf": 0.16426771653543307,
      "calib/mu_c": 0.1730952380952381,
      "calib/mu_w": 0.15990588235294118,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0215748031496063,
      "calib/std_conf": 0.15347099286409566,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25576441441441444,
      "calib/step_q_c_n": 666.0,
      "calib/step_q_gap": 0.04668194018761032,
      "calib/step_q_w": 0.20908247422680412,
      "calib/step_q_w_n": 1940.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1587.0,
      "completions/max_terminated_length": 1587.0,
      "completions/mean_length": 362.73046875,
      "completions/mean_terminated_length": 365.58660888671875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 88.0,
      "epoch": 0.10346666666666667,
      "grad_norm": 0.5830482840538025,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.277,
      "num_tokens": 20382633.0,
      "reward": 1.0239131450653076,
      "reward_std": 0.10469962656497955,
      "rewards/accuracy_reward_step": 0.328125,
      "rewards/final_brier_reward_step": 0.7275139093399048,
      "rewards/format_reward_step": 0.9921875,
      "step": 97
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8644020799547434,
      "aux_distill/mean_u": 0.296143036334745,
      "aux_distill/n_active_tok": 152.25,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 10.77734375,
      "calib/ece": 0.32207928286852594,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.00796812749003984,
      "calib/gap": 0.006815569053708392,
      "calib/mean_conf": 0.163498406374502,
      "calib/mu_c": 0.16719130434782606,
      "calib/mu_w": 0.16037573529411767,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.013705179282868525,
      "calib/std_conf": 0.15263408655975402,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28418463073852296,
      "calib/step_q_c_n": 1002.0,
      "calib/step_q_gap": 0.02861587718132319,
      "calib/step_q_w": 0.25556875355719977,
      "calib/step_q_w_n": 1757.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1548.0,
      "completions/max_terminated_length": 1548.0,
      "completions/mean_length": 362.5625,
      "completions/mean_terminated_length": 368.3174743652344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 0.10453333333333334,
      "grad_norm": 0.49543944001197815,
      "learning_rate": 2.861111111111111e-06,
      "loss": 0.2354,
      "num_tokens": 20581633.0,
      "reward": 1.03104829788208,
      "reward_std": 0.1321575939655304,
      "rewards/accuracy_reward_step": 0.44921875,
      "rewards/final_brier_reward_step": 0.6324091553688049,
      "rewards/format_reward_step": 0.98046875,
      "step": 98
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8613427989184856,
      "aux_distill/mean_u": 0.36263934351551574,
      "aux_distill/n_active_tok": 190.0,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 13.87890625,
      "calib/ece": 0.16048192771084335,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.008032128514056224,
      "calib/gap": 0.010117528019925298,
      "calib/mean_conf": 0.18120481927710846,
      "calib/mu_c": 0.18835616438356168,
      "calib/mu_w": 0.17823863636363638,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.024257028112449796,
      "calib/std_conf": 0.1701622017856304,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2812059392265193,
      "calib/step_q_c_n": 724.0,
      "calib/step_q_gap": -0.06427320534753506,
      "calib/step_q_w": 0.3454791445740544,
      "calib/step_q_w_n": 2829.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.02734375,
      "completions/max_length": 2363.0,
      "completions/max_terminated_length": 2363.0,
      "completions/mean_length": 428.9375,
      "completions/mean_terminated_length": 440.9959716796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.1056,
      "grad_norm": 0.39491331577301025,
      "learning_rate": 2.8333333333333335e-06,
      "loss": 0.187,
      "num_tokens": 20797241.0,
      "reward": 0.9963167905807495,
      "reward_std": 0.16188617050647736,
      "rewards/accuracy_reward_step": 0.28515625,
      "rewards/final_brier_reward_step": 0.734821081161499,
      "rewards/format_reward_step": 0.97265625,
      "step": 99
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.820311427116394,
      "aux_distill/mean_u": 0.3183899844738844,
      "aux_distill/n_active_tok": 192.75,
      "calib/answer_extract_rate": 0.9296875,
      "calib/avg_num_step_conf": 18.1953125,
      "calib/ece": 0.3286554621848739,
      "calib/final_conf_rate": 0.9296875,
      "calib/format_rate": 0.9296875,
      "calib/frac_conf_gt_0.9": 0.012605042016806723,
      "calib/gap": -0.01797679683078668,
      "calib/mean_conf": 0.19445378151260503,
      "calib/mu_c": 0.1850877192982456,
      "calib/mu_w": 0.20306451612903229,
      "calib/nonempty_final_conf_rate": 0.9296875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.02205882352941177,
      "calib/std_conf": 0.17215245090107312,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3088545150501672,
      "calib/step_q_c_n": 1196.0,
      "calib/step_q_gap": -0.022706836769590122,
      "calib/step_q_w": 0.3315613518197573,
      "calib/step_q_w_n": 3462.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0703125,
      "completions/max_length": 1488.0,
      "completions/max_terminated_length": 1488.0,
      "completions/mean_length": 402.67578125,
      "completions/mean_terminated_length": 433.1302795410156,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 133.0,
      "epoch": 0.10666666666666667,
      "grad_norm": 0.4126844108104706,
      "learning_rate": 2.805555555555556e-06,
      "loss": 0.0821,
      "num_tokens": 21007734.0,
      "reward": 0.9807562232017517,
      "reward_std": 0.2439180612564087,
      "rewards/accuracy_reward_step": 0.4453125,
      "rewards/final_brier_reward_step": 0.5865125060081482,
      "rewards/format_reward_step": 0.9296875,
      "step": 100
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8029415346682072,
      "aux_distill/mean_u": 0.2606578303927161,
      "aux_distill/n_active_tok": 277.625,
      "calib/answer_extract_rate": 0.8125,
      "calib/avg_num_step_conf": 31.69921875,
      "calib/ece": 0.14891980676328506,
      "calib/final_conf_rate": 0.80859375,
      "calib/format_rate": 0.8046875,
      "calib/frac_conf_gt_0.9": 0.00966183574879227,
      "calib/gap": 0.07122306390181968,
      "calib/mean_conf": 0.21310917874396135,
      "calib/mu_c": 0.26093529411764704,
      "calib/mu_w": 0.18971223021582737,
      "calib/nonempty_final_conf_rate": 0.80859375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.01676328502415459,
      "calib/std_conf": 0.17152507024488853,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3626553168635876,
      "calib/step_q_c_n": 931.0,
      "calib/step_q_gap": -0.057379482690978145,
      "calib/step_q_w": 0.42003479955456574,
      "calib/step_q_w_n": 7184.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.19140625,
      "completions/max_length": 2237.0,
      "completions/max_terminated_length": 2237.0,
      "completions/mean_length": 435.09375,
      "completions/mean_terminated_length": 538.0869750976562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 209.0,
      "epoch": 0.10773333333333333,
      "grad_norm": 0.3700943887233734,
      "learning_rate": 2.7777777777777783e-06,
      "loss": -0.1194,
      "num_tokens": 21226110.0,
      "reward": 0.8445423245429993,
      "reward_std": 0.388075590133667,
      "rewards/accuracy_reward_step": 0.265625,
      "rewards/final_brier_reward_step": 0.6187721490859985,
      "rewards/format_reward_step": 0.8046875,
      "step": 101
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.753003110177815,
      "aux_distill/mean_u": 0.29103300838633506,
      "aux_distill/n_active_tok": 254.5,
      "calib/answer_extract_rate": 0.796875,
      "calib/avg_num_step_conf": 32.359375,
      "calib/ece": 0.26816666666666666,
      "calib/final_conf_rate": 0.796875,
      "calib/format_rate": 0.79296875,
      "calib/frac_conf_gt_0.9": 0.00980392156862745,
      "calib/gap": -0.007465838509316758,
      "calib/mean_conf": 0.24399019607843137,
      "calib/mu_c": 0.2398913043478261,
      "calib/mu_w": 0.24735714285714286,
      "calib/nonempty_final_conf_rate": 0.796875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03058823529411765,
      "calib/std_conf": 0.2015103264564598,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.42965867992766726,
      "calib/step_q_c_n": 1106.0,
      "calib/step_q_gap": 0.023690713950568654,
      "calib/step_q_w": 0.4059679659770986,
      "calib/step_q_w_n": 7178.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.19921875,
      "completions/max_length": 2892.0,
      "completions/max_terminated_length": 2892.0,
      "completions/mean_length": 419.28125,
      "completions/mean_terminated_length": 523.5902709960938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 145.0,
      "epoch": 0.1088,
      "grad_norm": 0.4960228204727173,
      "learning_rate": 2.7500000000000004e-06,
      "loss": -0.0723,
      "num_tokens": 21440142.0,
      "reward": 0.8393516540527344,
      "reward_std": 0.39622962474823,
      "rewards/accuracy_reward_step": 0.359375,
      "rewards/final_brier_reward_step": 0.5263594388961792,
      "rewards/format_reward_step": 0.79296875,
      "step": 102
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7848681882023811,
      "aux_distill/mean_u": 0.29963380616280844,
      "aux_distill/n_active_tok": 223.0,
      "calib/answer_extract_rate": 0.9140625,
      "calib/avg_num_step_conf": 22.140625,
      "calib/ece": 0.2553653846153846,
      "calib/final_conf_rate": 0.9140625,
      "calib/format_rate": 0.9140625,
      "calib/frac_conf_gt_0.9": 0.01282051282051282,
      "calib/gap": 0.0185940487660268,
      "calib/mean_conf": 0.21856623931623936,
      "calib/mu_c": 0.22897572815533979,
      "calib/mu_w": 0.210381679389313,
      "calib/nonempty_final_conf_rate": 0.9140625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.016880341880341883,
      "calib/std_conf": 0.18379481805063905,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3806787834902245,
      "calib/step_q_c_n": 1381.0,
      "calib/step_q_gap": -0.018342770043715317,
      "calib/step_q_w": 0.3990215535339398,
      "calib/step_q_w_n": 4287.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.08203125,
      "completions/max_length": 1995.0,
      "completions/max_terminated_length": 1995.0,
      "completions/mean_length": 493.2265625,
      "completions/mean_terminated_length": 537.3021240234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 151.0,
      "epoch": 0.10986666666666667,
      "grad_norm": 0.3963656425476074,
      "learning_rate": 2.7222222222222224e-06,
      "loss": 0.0785,
      "num_tokens": 21670960.0,
      "reward": 0.968917727470398,
      "reward_std": 0.26493752002716064,
      "rewards/accuracy_reward_step": 0.40234375,
      "rewards/final_brier_reward_step": 0.6214292645454407,
      "rewards/format_reward_step": 0.9140625,
      "step": 103
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.75134018342942,
      "aux_distill/mean_u": 0.30651122907575223,
      "aux_distill/n_active_tok": 190.125,
      "calib/answer_extract_rate": 0.9609375,
      "calib/avg_num_step_conf": 14.66796875,
      "calib/ece": 0.2352226720647774,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.024291497975708502,
      "calib/gap": -0.015760073260073365,
      "calib/mean_conf": 0.22720647773279354,
      "calib/mu_c": 0.2172527472527472,
      "calib/mu_w": 0.23301282051282057,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.04700404858299595,
      "calib/std_conf": 0.20290573856710922,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2972996746203905,
      "calib/step_q_c_n": 922.0,
      "calib/step_q_gap": -0.10208966530195324,
      "calib/step_q_w": 0.3993893399223437,
      "calib/step_q_w_n": 2833.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03515625,
      "completions/max_length": 1175.0,
      "completions/max_terminated_length": 1175.0,
      "completions/mean_length": 418.80859375,
      "completions/mean_terminated_length": 434.06884765625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 148.0,
      "epoch": 0.11093333333333333,
      "grad_norm": 0.4534111022949219,
      "learning_rate": 2.6944444444444444e-06,
      "loss": 0.167,
      "num_tokens": 21884855.0,
      "reward": 0.9896644353866577,
      "reward_std": 0.20764201879501343,
      "rewards/accuracy_reward_step": 0.35546875,
      "rewards/final_brier_reward_step": 0.6668288707733154,
      "rewards/format_reward_step": 0.95703125,
      "step": 104
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.829353941604495,
      "aux_distill/mean_u": 0.2884675067906213,
      "aux_distill/n_active_tok": 160.625,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 10.84375,
      "calib/ece": 0.33407086614173226,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.003937007874015748,
      "calib/gap": -0.006714950660956981,
      "calib/mean_conf": 0.19175590551181104,
      "calib/mu_c": 0.18829268292682927,
      "calib/mu_w": 0.19500763358778625,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.02078740157480315,
      "calib/std_conf": 0.1637721297920406,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3003345388788427,
      "calib/step_q_c_n": 1106.0,
      "calib/step_q_gap": -0.05909935333672617,
      "calib/step_q_w": 0.3594338922155689,
      "calib/step_q_w_n": 1670.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1180.0,
      "completions/max_terminated_length": 1180.0,
      "completions/mean_length": 382.421875,
      "completions/mean_terminated_length": 386.95654296875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 116.0,
      "epoch": 0.112,
      "grad_norm": 0.4169822931289673,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.2266,
      "num_tokens": 22088515.0,
      "reward": 1.0472151041030884,
      "reward_std": 0.13726508617401123,
      "rewards/accuracy_reward_step": 0.48046875,
      "rewards/final_brier_reward_step": 0.625680148601532,
      "rewards/format_reward_step": 0.98828125,
      "step": 105
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8308898760005832,
      "aux_distill/mean_u": 0.2872108335961295,
      "aux_distill/n_active_tok": 125.625,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.94140625,
      "calib/ece": 0.26890625,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": -0.022763157894736846,
      "calib/mean_conf": 0.208515625,
      "calib/mu_c": 0.195,
      "calib/mu_w": 0.21776315789473685,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0355859375,
      "calib/std_conf": 0.17521901403917148,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27391780821917805,
      "calib/step_q_c_n": 730.0,
      "calib/step_q_gap": -0.046343128081666185,
      "calib/step_q_w": 0.32026093630084423,
      "calib/step_q_w_n": 1303.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 813.0,
      "completions/max_terminated_length": 813.0,
      "completions/mean_length": 315.41015625,
      "completions/mean_terminated_length": 317.8937072753906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 121.0,
      "epoch": 0.11306666666666666,
      "grad_norm": 0.4877546429634094,
      "learning_rate": 2.6388888888888893e-06,
      "loss": 0.2546,
      "num_tokens": 22273844.0,
      "reward": 1.0382380485534668,
      "reward_std": 0.11429660767316818,
      "rewards/accuracy_reward_step": 0.40625,
      "rewards/final_brier_reward_step": 0.6741324663162231,
      "rewards/format_reward_step": 0.99609375,
      "step": 106
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9066454619169235,
      "aux_distill/mean_u": 0.3440224827633614,
      "aux_distill/n_active_tok": 122.25,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.82421875,
      "calib/ece": 0.27308593750000004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.02734375,
      "calib/gap": 0.010836730853666238,
      "calib/mean_conf": 0.2028515625,
      "calib/mu_c": 0.20915887850467296,
      "calib/mu_w": 0.19832214765100672,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.028984375,
      "calib/std_conf": 0.1833160091653443,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.32453738910012675,
      "calib/step_q_c_n": 789.0,
      "calib/step_q_gap": 0.001094226002927412,
      "calib/step_q_w": 0.32344316309719934,
      "calib/step_q_w_n": 1214.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 802.0,
      "completions/max_terminated_length": 802.0,
      "completions/mean_length": 307.30859375,
      "completions/mean_terminated_length": 309.72833251953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 118.0,
      "epoch": 0.11413333333333334,
      "grad_norm": 0.46674540638923645,
      "learning_rate": 2.6111111111111113e-06,
      "loss": 0.2812,
      "num_tokens": 22457131.0,
      "reward": 1.046167016029358,
      "reward_std": 0.123149573802948,
      "rewards/accuracy_reward_step": 0.41796875,
      "rewards/final_brier_reward_step": 0.678271472454071,
      "rewards/format_reward_step": 0.99609375,
      "step": 107
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8467370923608541,
      "aux_distill/mean_u": 0.29868711003865983,
      "aux_distill/n_active_tok": 125.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.4140625,
      "calib/ece": 0.4441796875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.03470141700404858,
      "calib/mean_conf": 0.18699218750000002,
      "calib/mu_c": 0.17289473684210527,
      "calib/mu_w": 0.20759615384615385,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.018710937499999997,
      "calib/std_conf": 0.14887053919081789,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2749911190053286,
      "calib/step_q_c_n": 1126.0,
      "calib/step_q_gap": -0.021511471668246573,
      "calib/step_q_w": 0.29650259067357515,
      "calib/step_q_w_n": 772.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 666.0,
      "completions/max_terminated_length": 666.0,
      "completions/mean_length": 294.11328125,
      "completions/mean_terminated_length": 296.42913818359375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 102.0,
      "epoch": 0.1152,
      "grad_norm": 0.4774184226989746,
      "learning_rate": 2.5833333333333337e-06,
      "loss": 0.2502,
      "num_tokens": 22635656.0,
      "reward": 1.0740920305252075,
      "reward_std": 0.10364996641874313,
      "rewards/accuracy_reward_step": 0.59375,
      "rewards/final_brier_reward_step": 0.5544339418411255,
      "rewards/format_reward_step": 1.0,
      "step": 108
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.9085110891610384,
      "aux_distill/mean_u": 0.30277771671408027,
      "aux_distill/n_active_tok": 127.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.1953125,
      "calib/ece": 0.22762500000000005,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": 0.0026041626142774,
      "calib/mean_conf": 0.202609375,
      "calib/mu_c": 0.20422680412371136,
      "calib/mu_w": 0.20162264150943396,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0256640625,
      "calib/std_conf": 0.17991144672340717,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2743208092485549,
      "calib/step_q_c_n": 692.0,
      "calib/step_q_gap": -0.007309347223706797,
      "calib/step_q_w": 0.2816301564722617,
      "calib/step_q_w_n": 1406.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 858.0,
      "completions/max_terminated_length": 858.0,
      "completions/mean_length": 318.0546875,
      "completions/mean_terminated_length": 320.5590515136719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.11626666666666667,
      "grad_norm": 0.45826196670532227,
      "learning_rate": 2.5555555555555557e-06,
      "loss": 0.2573,
      "num_tokens": 22821678.0,
      "reward": 1.0406734943389893,
      "reward_std": 0.10282213985919952,
      "rewards/accuracy_reward_step": 0.37890625,
      "rewards/final_brier_reward_step": 0.7024407386779785,
      "rewards/format_reward_step": 1.0,
      "step": 109
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8161964118480682,
      "aux_distill/mean_u": 0.20091813851221327,
      "aux_distill/n_active_tok": 124.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.921875,
      "calib/ece": 0.250546875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": 0.015232793522267224,
      "calib/mean_conf": 0.19210937499999997,
      "calib/mu_c": 0.20115384615384618,
      "calib/mu_w": 0.18592105263157896,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.018203125,
      "calib/std_conf": 0.16482543959325385,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.339047619047619,
      "calib/step_q_c_n": 756.0,
      "calib/step_q_gap": 0.07994699011680134,
      "calib/step_q_w": 0.25910062893081764,
      "calib/step_q_w_n": 1272.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 881.0,
      "completions/max_terminated_length": 881.0,
      "completions/mean_length": 309.60546875,
      "completions/mean_terminated_length": 312.0433044433594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 55.0,
      "epoch": 0.11733333333333333,
      "grad_norm": 0.524115800857544,
      "learning_rate": 2.5277777777777778e-06,
      "loss": 0.219,
      "num_tokens": 23005857.0,
      "reward": 1.0496820211410522,
      "reward_std": 0.09998592734336853,
      "rewards/accuracy_reward_step": 0.40625,
      "rewards/final_brier_reward_step": 0.6931140422821045,
      "rewards/format_reward_step": 1.0,
      "step": 110
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8629253953695297,
      "aux_distill/mean_u": 0.27772488076746643,
      "aux_distill/n_active_tok": 129.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.34765625,
      "calib/ece": 0.29621093750000005,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.03939433068929471,
      "calib/mean_conf": 0.1716015625,
      "calib/mu_c": 0.19299145299145298,
      "calib/mu_w": 0.15359712230215827,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.005390625,
      "calib/std_conf": 0.11243128733834988,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2669621621621622,
      "calib/step_q_c_n": 925.0,
      "calib/step_q_gap": -0.011140148068860933,
      "calib/step_q_w": 0.2781023102310231,
      "calib/step_q_w_n": 1212.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1113.0,
      "completions/max_terminated_length": 1113.0,
      "completions/mean_length": 321.51953125,
      "completions/mean_terminated_length": 324.0511779785156,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 119.0,
      "epoch": 0.1184,
      "grad_norm": 0.45644694566726685,
      "learning_rate": 2.5e-06,
      "loss": 0.2514,
      "num_tokens": 23195574.0,
      "reward": 1.0671591758728027,
      "reward_std": 0.07571589201688766,
      "rewards/accuracy_reward_step": 0.45703125,
      "rewards/final_brier_reward_step": 0.6772871017456055,
      "rewards/format_reward_step": 1.0,
      "step": 111
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8283669967204332,
      "aux_distill/mean_u": 0.2970689657933608,
      "aux_distill/n_active_tok": 144.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.5234375,
      "calib/ece": 0.25892578125000004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": 0.007412488174077636,
      "calib/mean_conf": 0.17919921875,
      "calib/mu_c": 0.18357142857142864,
      "calib/mu_w": 0.176158940397351,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.013984374999999999,
      "calib/std_conf": 0.13212152372206296,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24368415051311287,
      "calib/step_q_c_n": 877.0,
      "calib/step_q_gap": -0.0294882632799906,
      "calib/step_q_w": 0.2731724137931035,
      "calib/step_q_w_n": 1305.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 951.0,
      "completions/max_terminated_length": 951.0,
      "completions/mean_length": 334.32421875,
      "completions/mean_terminated_length": 336.9566955566406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 134.0,
      "epoch": 0.11946666666666667,
      "grad_norm": 0.42700129747390747,
      "learning_rate": 2.4722222222222226e-06,
      "loss": 0.25,
      "num_tokens": 23389081.0,
      "reward": 1.0481845140457153,
      "reward_std": 0.0933864563703537,
      "rewards/accuracy_reward_step": 0.41015625,
      "rewards/final_brier_reward_step": 0.6901190280914307,
      "rewards/format_reward_step": 0.99609375,
      "step": 112
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.825203481130302,
      "aux_distill/mean_u": 0.23171280180171644,
      "aux_distill/n_active_tok": 133.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.5,
      "calib/ece": 0.33029296875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": 0.04800671550671551,
      "calib/mean_conf": 0.18291015625,
      "calib/mu_c": 0.20653846153846156,
      "calib/mu_w": 0.15853174603174605,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0026953125,
      "calib/std_conf": 0.12331135854758307,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2763565891472869,
      "calib/step_q_c_n": 1032.0,
      "calib/step_q_gap": -0.012678375887678173,
      "calib/step_q_w": 0.28903496503496506,
      "calib/step_q_w_n": 1144.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 789.0,
      "completions/max_terminated_length": 789.0,
      "completions/mean_length": 313.15625,
      "completions/mean_terminated_length": 315.6220397949219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.12053333333333334,
      "grad_norm": 0.474639892578125,
      "learning_rate": 2.4444444444444447e-06,
      "loss": 0.2397,
      "num_tokens": 23574449.0,
      "reward": 1.0805518627166748,
      "reward_std": 0.09097933769226074,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.6532913446426392,
      "rewards/format_reward_step": 1.0,
      "step": 113
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8264994230121374,
      "aux_distill/mean_u": 0.260318213715089,
      "aux_distill/n_active_tok": 130.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.26953125,
      "calib/ece": 0.3955468749999999,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.02640873015873016,
      "calib/mean_conf": 0.17601562499999998,
      "calib/mu_c": 0.18756944444444446,
      "calib/mu_w": 0.1611607142857143,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.00453125,
      "calib/std_conf": 0.12373636189843054,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2543778801843318,
      "calib/step_q_c_n": 1085.0,
      "calib/step_q_gap": 0.0020522987889829425,
      "calib/step_q_w": 0.25232558139534883,
      "calib/step_q_w_n": 1032.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 770.0,
      "completions/max_terminated_length": 770.0,
      "completions/mean_length": 321.21484375,
      "completions/mean_terminated_length": 323.74407958984375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 94.0,
      "epoch": 0.1216,
      "grad_norm": 0.40753239393234253,
      "learning_rate": 2.4166666666666667e-06,
      "loss": 0.2685,
      "num_tokens": 23761704.0,
      "reward": 1.0823616981506348,
      "reward_std": 0.08417695015668869,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.6022234559059143,
      "rewards/format_reward_step": 1.0,
      "step": 114
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8344746474176645,
      "aux_distill/mean_u": 0.26471749218578394,
      "aux_distill/n_active_tok": 153.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 9.45703125,
      "calib/ece": 0.29492156862745095,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00392156862745098,
      "calib/gap": 0.03482805374090964,
      "calib/mean_conf": 0.1933921568627451,
      "calib/mu_c": 0.21155737704918032,
      "calib/mu_w": 0.17672932330827068,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.004941176470588235,
      "calib/std_conf": 0.14304244241008351,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24150134048257374,
      "calib/step_q_c_n": 1119.0,
      "calib/step_q_gap": -0.033216785477487715,
      "calib/step_q_w": 0.27471812596006145,
      "calib/step_q_w_n": 1302.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1701.0,
      "completions/max_terminated_length": 1701.0,
      "completions/mean_length": 359.45703125,
      "completions/mean_terminated_length": 362.28741455078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 80.0,
      "epoch": 0.12266666666666666,
      "grad_norm": 0.4399702250957489,
      "learning_rate": 2.388888888888889e-06,
      "loss": 0.2428,
      "num_tokens": 23958989.0,
      "reward": 1.0680962800979614,
      "reward_std": 0.11737754195928574,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.6635362505912781,
      "rewards/format_reward_step": 0.99609375,
      "step": 115
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8296884559094906,
      "aux_distill/mean_u": 0.2383707263061892,
      "aux_distill/n_active_tok": 157.25,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.125,
      "calib/ece": 0.2138117647058823,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00392156862745098,
      "calib/gap": 0.019737956487956454,
      "calib/mean_conf": 0.20187450980392158,
      "calib/mu_c": 0.21394949494949492,
      "calib/mu_w": 0.19421153846153846,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.013725490196078431,
      "calib/std_conf": 0.14172868286387316,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28197294250281846,
      "calib/step_q_c_n": 887.0,
      "calib/step_q_gap": -0.003939667467856012,
      "calib/step_q_w": 0.28591260997067447,
      "calib/step_q_w_n": 1705.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1067.0,
      "completions/max_terminated_length": 1067.0,
      "completions/mean_length": 408.76953125,
      "completions/mean_terminated_length": 410.37255859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 126.0,
      "epoch": 0.12373333333333333,
      "grad_norm": 0.4059218168258667,
      "learning_rate": 2.361111111111111e-06,
      "loss": 0.2438,
      "num_tokens": 24168154.0,
      "reward": 1.0485306978225708,
      "reward_std": 0.09994710981845856,
      "rewards/accuracy_reward_step": 0.38671875,
      "rewards/final_brier_reward_step": 0.7142488956451416,
      "rewards/format_reward_step": 0.99609375,
      "step": 116
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8448361586779356,
      "aux_distill/mean_u": 0.25940913176972474,
      "aux_distill/n_active_tok": 156.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.4375,
      "calib/ece": 0.22378515625,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": -0.007329716352560495,
      "calib/mean_conf": 0.23301171875,
      "calib/mu_c": 0.22863106796116503,
      "calib/mu_w": 0.23596078431372552,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.027226562499999996,
      "calib/std_conf": 0.15947716045384336,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2669742574257426,
      "calib/step_q_c_n": 1010.0,
      "calib/step_q_gap": -0.025353059060418626,
      "calib/step_q_w": 0.2923273164861612,
      "calib/step_q_w_n": 1662.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1321.0,
      "completions/max_terminated_length": 1321.0,
      "completions/mean_length": 397.28515625,
      "completions/mean_terminated_length": 400.41339111328125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 105.0,
      "epoch": 0.1248,
      "grad_norm": 0.3755255937576294,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.2472,
      "num_tokens": 24376459.0,
      "reward": 1.0521246194839478,
      "reward_std": 0.1031055748462677,
      "rewards/accuracy_reward_step": 0.40234375,
      "rewards/final_brier_reward_step": 0.7019054293632507,
      "rewards/format_reward_step": 1.0,
      "step": 117
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8845678055658937,
      "aux_distill/mean_u": 0.3274318211216818,
      "aux_distill/n_active_tok": 165.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.44140625,
      "calib/ece": 0.25635039370078744,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.003937007874015748,
      "calib/gap": 0.015536764705882306,
      "calib/mean_conf": 0.23168110236220477,
      "calib/mu_c": 0.23999999999999996,
      "calib/mu_w": 0.22446323529411766,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.011732283464566926,
      "calib/std_conf": 0.14912145830804718,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2450046425255339,
      "calib/step_q_c_n": 1077.0,
      "calib/step_q_gap": -0.023299868752661657,
      "calib/step_q_w": 0.26830451127819555,
      "calib/step_q_w_n": 1596.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1194.0,
      "completions/max_terminated_length": 1194.0,
      "completions/mean_length": 393.8046875,
      "completions/mean_terminated_length": 396.905517578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 123.0,
      "epoch": 0.12586666666666665,
      "grad_norm": 0.35954874753952026,
      "learning_rate": 2.305555555555556e-06,
      "loss": 0.2548,
      "num_tokens": 24581281.0,
      "reward": 1.0651524066925049,
      "reward_std": 0.12032070755958557,
      "rewards/accuracy_reward_step": 0.4609375,
      "rewards/final_brier_reward_step": 0.6771796941757202,
      "rewards/format_reward_step": 0.9921875,
      "step": 118
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8137836419045925,
      "aux_distill/mean_u": 0.26442520070357983,
      "aux_distill/n_active_tok": 150.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.61328125,
      "calib/ece": 0.2552968750000001,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.043520966855887216,
      "calib/mean_conf": 0.252828125,
      "calib/mu_c": 0.2744186046511628,
      "calib/mu_w": 0.23089763779527558,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0021093750000000006,
      "calib/std_conf": 0.14796788130024832,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27225104602510464,
      "calib/step_q_c_n": 1195.0,
      "calib/step_q_gap": -0.018978500624041217,
      "calib/step_q_w": 0.29122954664914585,
      "calib/step_q_w_n": 1522.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1890.0,
      "completions/max_terminated_length": 1890.0,
      "completions/mean_length": 415.8203125,
      "completions/mean_terminated_length": 419.094482421875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.12693333333333334,
      "grad_norm": 0.3934442698955536,
      "learning_rate": 2.277777777777778e-06,
      "loss": 0.2799,
      "num_tokens": 24792795.0,
      "reward": 1.0953730344772339,
      "reward_std": 0.10872112214565277,
      "rewards/accuracy_reward_step": 0.50390625,
      "rewards/final_brier_reward_step": 0.6868396997451782,
      "rewards/format_reward_step": 1.0,
      "step": 119
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8701824210584164,
      "aux_distill/mean_u": 0.2888762191205573,
      "aux_distill/n_active_tok": 162.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.875,
      "calib/ece": 0.31291015625,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": 0.026631944444444444,
      "calib/mean_conf": 0.27279296875000003,
      "calib/mu_c": 0.2844444444444445,
      "calib/mu_w": 0.25781250000000006,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.011601562499999997,
      "calib/std_conf": 0.1702758335225863,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23827830188679247,
      "calib/step_q_c_n": 1272.0,
      "calib/step_q_gap": -0.05679892741257056,
      "calib/step_q_w": 0.29507722929936303,
      "calib/step_q_w_n": 1256.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 972.0,
      "completions/max_terminated_length": 972.0,
      "completions/mean_length": 387.85546875,
      "completions/mean_terminated_length": 390.9094543457031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 165.0,
      "epoch": 0.128,
      "grad_norm": 0.7766105532646179,
      "learning_rate": 2.25e-06,
      "loss": 0.2532,
      "num_tokens": 24998774.0,
      "reward": 1.10829496383667,
      "reward_std": 0.12710832059383392,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.654090166091919,
      "rewards/format_reward_step": 1.0,
      "step": 120
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8371859528124332,
      "aux_distill/mean_u": 0.2782328386251819,
      "aux_distill/n_active_tok": 159.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.3515625,
      "calib/ece": 0.3069925490196078,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00784313725490196,
      "calib/gap": 0.020572110739502047,
      "calib/mean_conf": 0.2653211764705883,
      "calib/mu_c": 0.27476014492753625,
      "calib/mu_w": 0.2541880341880342,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.015568627450980392,
      "calib/std_conf": 0.1585131689419768,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.277452566096423,
      "calib/step_q_c_n": 1286.0,
      "calib/step_q_gap": -0.02520139284785855,
      "calib/step_q_w": 0.30265395894428154,
      "calib/step_q_w_n": 1364.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1525.0,
      "completions/max_terminated_length": 1525.0,
      "completions/mean_length": 413.54296875,
      "completions/mean_terminated_length": 416.7992248535156,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.12906666666666666,
      "grad_norm": 0.36996495723724365,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.2268,
      "num_tokens": 25209697.0,
      "reward": 1.0985853672027588,
      "reward_std": 0.12847691774368286,
      "rewards/accuracy_reward_step": 0.54296875,
      "rewards/final_brier_reward_step": 0.6581084132194519,
      "rewards/format_reward_step": 0.99609375,
      "step": 121
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8512360025197268,
      "aux_distill/mean_u": 0.2924762686230811,
      "aux_distill/n_active_tok": 160.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.0546875,
      "calib/ece": 0.2778117647058823,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.011764705882352941,
      "calib/gap": 0.07618210380796786,
      "calib/mean_conf": 0.30673725490196074,
      "calib/mu_c": 0.3393013698630137,
      "calib/mu_w": 0.2631192660550458,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.006,
      "calib/std_conf": 0.171524986987587,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23481716475095785,
      "calib/step_q_c_n": 1305.0,
      "calib/step_q_gap": -0.06490702752642588,
      "calib/step_q_w": 0.2997241922773837,
      "calib/step_q_w_n": 1269.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1217.0,
      "completions/max_terminated_length": 1217.0,
      "completions/mean_length": 394.52734375,
      "completions/mean_terminated_length": 397.63385009765625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 140.0,
      "epoch": 0.13013333333333332,
      "grad_norm": 0.39588937163352966,
      "learning_rate": 2.1944444444444445e-06,
      "loss": 0.2451,
      "num_tokens": 25418040.0,
      "reward": 1.1280884742736816,
      "reward_std": 0.12784995138645172,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6897708177566528,
      "rewards/format_reward_step": 0.99609375,
      "step": 122
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8117791870608926,
      "aux_distill/mean_u": 0.22482812837851487,
      "aux_distill/n_active_tok": 182.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.44140625,
      "calib/ece": 0.22187109374999997,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": 0.04839890109890105,
      "calib/mean_conf": 0.30586328125,
      "calib/mu_c": 0.3296846153846154,
      "calib/mu_w": 0.28128571428571436,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0099609375,
      "calib/std_conf": 0.1724087645371702,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2834858528698464,
      "calib/step_q_c_n": 1237.0,
      "calib/step_q_gap": -0.042104676378064476,
      "calib/step_q_w": 0.3255905292479109,
      "calib/step_q_w_n": 1436.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2279.0,
      "completions/max_terminated_length": 2279.0,
      "completions/mean_length": 436.1875,
      "completions/mean_terminated_length": 439.6220397949219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 148.0,
      "epoch": 0.1312,
      "grad_norm": 0.3385443389415741,
      "learning_rate": 2.166666666666667e-06,
      "loss": 0.2469,
      "num_tokens": 25634992.0,
      "reward": 1.1057794094085693,
      "reward_std": 0.12787926197052002,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.7037463188171387,
      "rewards/format_reward_step": 1.0,
      "step": 123
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.879656570032239,
      "aux_distill/mean_u": 0.2719786785542341,
      "aux_distill/n_active_tok": 141.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.8359375,
      "calib/ece": 0.28291796875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": 0.026609239164921583,
      "calib/mean_conf": 0.34137890625,
      "calib/mu_c": 0.35208496732026145,
      "calib/mu_w": 0.32547572815533987,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0133203125,
      "calib/std_conf": 0.16329816295293623,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2698769968051118,
      "calib/step_q_c_n": 1252.0,
      "calib/step_q_gap": -0.0455032012146902,
      "calib/step_q_w": 0.315380198019802,
      "calib/step_q_w_n": 1010.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 949.0,
      "completions/max_terminated_length": 949.0,
      "completions/mean_length": 377.21875,
      "completions/mean_terminated_length": 380.18896484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 118.0,
      "epoch": 0.13226666666666667,
      "grad_norm": 0.40829819440841675,
      "learning_rate": 2.138888888888889e-06,
      "loss": 0.2655,
      "num_tokens": 25838376.0,
      "reward": 1.1388227939605713,
      "reward_std": 0.13053259253501892,
      "rewards/accuracy_reward_step": 0.59765625,
      "rewards/final_brier_reward_step": 0.6799894571304321,
      "rewards/format_reward_step": 1.0,
      "step": 124
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8070436110720038,
      "aux_distill/mean_u": 0.218702439653177,
      "aux_distill/n_active_tok": 135.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.32421875,
      "calib/ece": 0.2130560342555995,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.019762845849802372,
      "calib/gap": -0.008654143619905685,
      "calib/mean_conf": 0.40035635046113305,
      "calib/mu_c": 0.39618320610687024,
      "calib/mu_w": 0.40483734972677593,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04781291172595521,
      "calib/std_conf": 0.1918678271576223,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26378531073446326,
      "calib/step_q_c_n": 1062.0,
      "calib/step_q_gap": -0.01954426159258077,
      "calib/step_q_w": 0.28332957232704403,
      "calib/step_q_w_n": 1325.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1575.0,
      "completions/max_terminated_length": 1575.0,
      "completions/mean_length": 378.16796875,
      "completions/mean_terminated_length": 382.6521911621094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 120.0,
      "epoch": 0.13333333333333333,
      "grad_norm": 0.5166134238243103,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 0.1924,
      "num_tokens": 26039995.0,
      "reward": 1.0936212539672852,
      "reward_std": 0.16634529829025269,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.6872425079345703,
      "rewards/format_reward_step": 0.98828125,
      "step": 125
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7905949633568525,
      "aux_distill/mean_u": 0.22697852024378237,
      "aux_distill/n_active_tok": 147.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.375,
      "calib/ece": 0.10710690234375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0078125,
      "calib/gap": 0.06187749438339435,
      "calib/mean_conf": 0.43086184765625,
      "calib/mu_c": 0.4613171769230769,
      "calib/mu_w": 0.39943968253968254,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.015078125000000012,
      "calib/std_conf": 0.18609255587625964,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.25488984684684685,
      "calib/step_q_c_n": 1110.0,
      "calib/step_q_gap": -0.0629840291221454,
      "calib/step_q_w": 0.31787387596899225,
      "calib/step_q_w_n": 1290.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1404.0,
      "completions/max_terminated_length": 1404.0,
      "completions/mean_length": 413.5078125,
      "completions/mean_terminated_length": 416.7637634277344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 0.1344,
      "grad_norm": 0.4073919951915741,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.2482,
      "num_tokens": 26251317.0,
      "reward": 1.1208109855651855,
      "reward_std": 0.1425214558839798,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.7377157807350159,
      "rewards/format_reward_step": 0.99609375,
      "step": 126
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7611914118751884,
      "aux_distill/mean_u": 0.21427045188004712,
      "aux_distill/n_active_tok": 145.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 9.41015625,
      "calib/ece": 0.13895176470588233,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0196078431372549,
      "calib/gap": 0.08786426725729052,
      "calib/mean_conf": 0.4333854901960784,
      "calib/mu_c": 0.47680077519379843,
      "calib/mu_w": 0.3889365079365079,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03322745098039215,
      "calib/std_conf": 0.18705611191605545,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2536829690346084,
      "calib/step_q_c_n": 1098.0,
      "calib/step_q_gap": -0.017251432185834026,
      "calib/step_q_w": 0.27093440122044243,
      "calib/step_q_w_n": 1311.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2870.0,
      "completions/max_terminated_length": 2870.0,
      "completions/mean_length": 388.625,
      "completions/mean_terminated_length": 391.6850280761719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 100.0,
      "epoch": 0.13546666666666668,
      "grad_norm": 0.4389866292476654,
      "learning_rate": 2.0555555555555555e-06,
      "loss": 0.2181,
      "num_tokens": 26454477.0,
      "reward": 1.1253852844238281,
      "reward_std": 0.15292197465896606,
      "rewards/accuracy_reward_step": 0.50390625,
      "rewards/final_brier_reward_step": 0.7507706880569458,
      "rewards/format_reward_step": 0.99609375,
      "step": 127
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8607869111001492,
      "aux_distill/mean_u": 0.2650464180056679,
      "aux_distill/n_active_tok": 145.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.3203125,
      "calib/ece": 0.08300393700787398,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.027559055118110236,
      "calib/gap": 0.09334362766420179,
      "calib/mean_conf": 0.5029015748031496,
      "calib/mu_c": 0.5473684210526315,
      "calib/mu_w": 0.4540247933884297,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.031141732283464563,
      "calib/std_conf": 0.20164805759386392,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25387522935779816,
      "calib/step_q_c_n": 1090.0,
      "calib/step_q_gap": -0.05599745582738702,
      "calib/step_q_w": 0.3098726851851852,
      "calib/step_q_w_n": 1296.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1275.0,
      "completions/max_terminated_length": 1275.0,
      "completions/mean_length": 385.42578125,
      "completions/mean_terminated_length": 389.9960632324219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.13653333333333334,
      "grad_norm": 0.43388524651527405,
      "learning_rate": 2.027777777777778e-06,
      "loss": 0.2457,
      "num_tokens": 26659810.0,
      "reward": 1.1309232711791992,
      "reward_std": 0.20366542041301727,
      "rewards/accuracy_reward_step": 0.51953125,
      "rewards/final_brier_reward_step": 0.7501278519630432,
      "rewards/format_reward_step": 0.9921875,
      "step": 128
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8683709055185318,
      "aux_distill/mean_u": 0.31790983151919006,
      "aux_distill/n_active_tok": 144.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.734375,
      "calib/ece": 0.10494927843137251,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0196078431372549,
      "calib/gap": 0.052860777350533295,
      "calib/mean_conf": 0.5346585647058825,
      "calib/mu_c": 0.558705035971223,
      "calib/mu_w": 0.5058442586206897,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0472549019607843,
      "calib/std_conf": 0.18598623563819291,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23410098302055407,
      "calib/step_q_c_n": 1119.0,
      "calib/step_q_gap": -0.02256104115133492,
      "calib/step_q_w": 0.256662024171889,
      "calib/step_q_w_n": 1117.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1457.0,
      "completions/max_terminated_length": 1457.0,
      "completions/mean_length": 368.296875,
      "completions/mean_terminated_length": 371.19683837890625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 134.0,
      "epoch": 0.1376,
      "grad_norm": 0.5401434898376465,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.2675,
      "num_tokens": 26856478.0,
      "reward": 1.1418068408966064,
      "reward_std": 0.1939820498228073,
      "rewards/accuracy_reward_step": 0.546875,
      "rewards/final_brier_reward_step": 0.7406448721885681,
      "rewards/format_reward_step": 0.99609375,
      "step": 129
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8692066632211208,
      "aux_distill/mean_u": 0.25088071420496283,
      "aux_distill/n_active_tok": 125.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.48828125,
      "calib/ece": 0.04881510416666664,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.05859375,
      "calib/gap": 0.1445388415851992,
      "calib/mean_conf": 0.5760807291666666,
      "calib/mu_c": 0.635364238410596,
      "calib/mu_w": 0.4908253968253968,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.017526041666666662,
      "calib/std_conf": 0.20678414523099387,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27217905405405407,
      "calib/step_q_c_n": 1184.0,
      "calib/step_q_gap": 0.030604062480073635,
      "calib/step_q_w": 0.24157499157398044,
      "calib/step_q_w_n": 989.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1276.0,
      "completions/max_terminated_length": 1276.0,
      "completions/mean_length": 352.51171875,
      "completions/mean_terminated_length": 355.28741455078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 115.0,
      "epoch": 0.13866666666666666,
      "grad_norm": 0.40447065234184265,
      "learning_rate": 1.9722222222222224e-06,
      "loss": 0.2612,
      "num_tokens": 27052009.0,
      "reward": 1.1874513626098633,
      "reward_std": 0.17112727463245392,
      "rewards/accuracy_reward_step": 0.58984375,
      "rewards/final_brier_reward_step": 0.7850587964057922,
      "rewards/format_reward_step": 1.0,
      "step": 130
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8476680275052786,
      "aux_distill/mean_u": 0.23583052530939752,
      "aux_distill/n_active_tok": 144.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.87890625,
      "calib/ece": 0.17316406249999994,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0703125,
      "calib/gap": 0.10731275303643739,
      "calib/mean_conf": 0.5771484375,
      "calib/mu_c": 0.6408653846153847,
      "calib/mu_w": 0.5335526315789473,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1720312499999999,
      "calib/std_conf": 0.20614598096084383,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23984172661870506,
      "calib/step_q_c_n": 834.0,
      "calib/step_q_gap": -0.055043610420905775,
      "calib/step_q_w": 0.29488533703961084,
      "calib/step_q_w_n": 1439.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1479.0,
      "completions/max_terminated_length": 1479.0,
      "completions/mean_length": 371.578125,
      "completions/mean_terminated_length": 374.5039367675781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 150.0,
      "epoch": 0.13973333333333332,
      "grad_norm": 0.6352536678314209,
      "learning_rate": 1.944444444444445e-06,
      "loss": 0.239,
      "num_tokens": 27253341.0,
      "reward": 1.0725533962249756,
      "reward_std": 0.1738070249557495,
      "rewards/accuracy_reward_step": 0.40625,
      "rewards/final_brier_reward_step": 0.7388566732406616,
      "rewards/format_reward_step": 1.0,
      "step": 131
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8875786550343037,
      "aux_distill/mean_u": 0.2981896139011602,
      "aux_distill/n_active_tok": 133.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.28515625,
      "calib/ece": 0.10589843750000008,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.078125,
      "calib/gap": 0.09960147695441823,
      "calib/mean_conf": 0.6212890624999999,
      "calib/mu_c": 0.660974025974026,
      "calib/mu_w": 0.5613725490196078,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.06281250000000008,
      "calib/std_conf": 0.21616756565884507,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2856017328519856,
      "calib/step_q_c_n": 1385.0,
      "calib/step_q_gap": -0.010592319567369224,
      "calib/step_q_w": 0.2961940524193548,
      "calib/step_q_w_n": 992.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1142.0,
      "completions/max_terminated_length": 1142.0,
      "completions/mean_length": 395.390625,
      "completions/mean_terminated_length": 398.5039367675781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 116.0,
      "epoch": 0.1408,
      "grad_norm": 0.4326506555080414,
      "learning_rate": 1.916666666666667e-06,
      "loss": 0.2447,
      "num_tokens": 27460153.0,
      "reward": 1.1812529563903809,
      "reward_std": 0.21404148638248444,
      "rewards/accuracy_reward_step": 0.6015625,
      "rewards/final_brier_reward_step": 0.7609433531761169,
      "rewards/format_reward_step": 1.0,
      "step": 132
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8566825911402702,
      "aux_distill/mean_u": 0.30097786155730477,
      "aux_distill/n_active_tok": 153.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.39453125,
      "calib/ece": 0.25664062500000007,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.10546875,
      "calib/gap": 0.048932330827067605,
      "calib/mean_conf": 0.601015625,
      "calib/mu_c": 0.6317894736842105,
      "calib/mu_w": 0.5828571428571429,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24328125000000003,
      "calib/std_conf": 0.22125649652803278,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25251977401129944,
      "calib/step_q_c_n": 885.0,
      "calib/step_q_gap": -0.027565752304490043,
      "calib/step_q_w": 0.2800855263157895,
      "calib/step_q_w_n": 1520.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1081.0,
      "completions/max_terminated_length": 1081.0,
      "completions/mean_length": 416.375,
      "completions/mean_terminated_length": 419.6535339355469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 102.0,
      "epoch": 0.14186666666666667,
      "grad_norm": 0.3714953660964966,
      "learning_rate": 1.888888888888889e-06,
      "loss": 0.2421,
      "num_tokens": 27673089.0,
      "reward": 1.0293660163879395,
      "reward_std": 0.2450329065322876,
      "rewards/accuracy_reward_step": 0.37109375,
      "rewards/final_brier_reward_step": 0.6876382827758789,
      "rewards/format_reward_step": 1.0,
      "step": 133
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8316728174686432,
      "aux_distill/mean_u": 0.24845326043078172,
      "aux_distill/n_active_tok": 168.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.4765625,
      "calib/ece": 0.13919140625000004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.11328125,
      "calib/gap": 0.18020545265603027,
      "calib/mean_conf": 0.56841015625,
      "calib/mu_c": 0.6620325203252032,
      "calib/mu_w": 0.48182706766917294,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11356640625000003,
      "calib/std_conf": 0.2657971643059808,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2906495589414595,
      "calib/step_q_c_n": 1247.0,
      "calib/step_q_gap": -0.010627792974916783,
      "calib/step_q_w": 0.3012773519163763,
      "calib/step_q_w_n": 1435.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1473.0,
      "completions/max_terminated_length": 1473.0,
      "completions/mean_length": 450.24609375,
      "completions/mean_terminated_length": 455.5849914550781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 173.0,
      "epoch": 0.14293333333333333,
      "grad_norm": 0.38375750184059143,
      "learning_rate": 1.8611111111111113e-06,
      "loss": 0.2399,
      "num_tokens": 27897304.0,
      "reward": 1.1212167739868164,
      "reward_std": 0.20416364073753357,
      "rewards/accuracy_reward_step": 0.48046875,
      "rewards/final_brier_reward_step": 0.7619649171829224,
      "rewards/format_reward_step": 1.0,
      "step": 134
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8126184120774269,
      "aux_distill/mean_u": 0.2249086348603118,
      "aux_distill/n_active_tok": 148.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.84375,
      "calib/ece": 0.167433881854416,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.12109375,
      "calib/gap": 0.12185181360830022,
      "calib/mean_conf": 0.6109254931455841,
      "calib/mu_c": 0.6708994326559193,
      "calib/mu_w": 0.5490476190476191,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1352734375000001,
      "calib/std_conf": 0.24969082446086727,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2639072319720659,
      "calib/step_q_c_n": 1275.0,
      "calib/step_q_gap": -0.01553051903195013,
      "calib/step_q_w": 0.279437751004016,
      "calib/step_q_w_n": 1245.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1691.0,
      "completions/max_terminated_length": 1691.0,
      "completions/mean_length": 426.47265625,
      "completions/mean_terminated_length": 429.8307189941406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 131.0,
      "epoch": 0.144,
      "grad_norm": 0.3621184527873993,
      "learning_rate": 1.8333333333333333e-06,
      "loss": 0.2554,
      "num_tokens": 28112361.0,
      "reward": 1.1229034662246704,
      "reward_std": 0.20501597225666046,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.7379943132400513,
      "rewards/format_reward_step": 1.0,
      "step": 135
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8527163006365299,
      "aux_distill/mean_u": 0.24802126757150933,
      "aux_distill/n_active_tok": 156.625,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 10.765625,
      "calib/ece": 0.08708661417322835,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.09448818897637795,
      "calib/gap": 0.2711115248557066,
      "calib/mean_conf": 0.5421259842519685,
      "calib/mu_c": 0.6819512195121952,
      "calib/mu_w": 0.41083969465648856,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.07248031496062994,
      "calib/std_conf": 0.2681876470854467,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2807493061979648,
      "calib/step_q_c_n": 1081.0,
      "calib/step_q_gap": -0.02145964902591585,
      "calib/step_q_w": 0.30220895522388064,
      "calib/step_q_w_n": 1675.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1400.0,
      "completions/max_terminated_length": 1400.0,
      "completions/mean_length": 441.3359375,
      "completions/mean_terminated_length": 446.5691833496094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 126.0,
      "epoch": 0.14506666666666668,
      "grad_norm": 0.44931545853614807,
      "learning_rate": 1.8055555555555557e-06,
      "loss": 0.207,
      "num_tokens": 28333831.0,
      "reward": 1.1383601427078247,
      "reward_std": 0.1638520359992981,
      "rewards/accuracy_reward_step": 0.48046875,
      "rewards/final_brier_reward_step": 0.8040640354156494,
      "rewards/format_reward_step": 0.9921875,
      "step": 136
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8319171629846096,
      "aux_distill/mean_u": 0.2752806769768428,
      "aux_distill/n_active_tok": 145.25,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 9.7421875,
      "calib/ece": 0.09984313725490195,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.11764705882352941,
      "calib/gap": 0.2407757135826773,
      "calib/mean_conf": 0.5584313725490195,
      "calib/mu_c": 0.6792913385826773,
      "calib/mu_w": 0.438515625,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08011764705882352,
      "calib/std_conf": 0.28072296405028024,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2751356080489939,
      "calib/step_q_c_n": 1143.0,
      "calib/step_q_gap": 0.03524504032138473,
      "calib/step_q_w": 0.23989056772760917,
      "calib/step_q_w_n": 1351.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2805.0,
      "completions/max_terminated_length": 2805.0,
      "completions/mean_length": 425.0390625,
      "completions/mean_terminated_length": 426.7059020996094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 136.0,
      "epoch": 0.14613333333333334,
      "grad_norm": 0.35590553283691406,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.2827,
      "num_tokens": 28549625.0,
      "reward": 1.1385234594345093,
      "reward_std": 0.17667999863624573,
      "rewards/accuracy_reward_step": 0.49609375,
      "rewards/final_brier_reward_step": 0.7848593592643738,
      "rewards/format_reward_step": 0.99609375,
      "step": 137
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8382508996874094,
      "aux_distill/mean_u": 0.2673697069633504,
      "aux_distill/n_active_tok": 147.75,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.53515625,
      "calib/ece": 0.12353333333333334,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.11372549019607843,
      "calib/gap": 0.20276334453154166,
      "calib/mean_conf": 0.5598705882352941,
      "calib/mu_c": 0.6505177304964539,
      "calib/mu_w": 0.44775438596491224,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0652313725490196,
      "calib/std_conf": 0.26972782744769586,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21309735849056602,
      "calib/step_q_c_n": 1325.0,
      "calib/step_q_gap": -0.04360307882721823,
      "calib/step_q_w": 0.25670043731778425,
      "calib/step_q_w_n": 1372.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1641.0,
      "completions/max_terminated_length": 1641.0,
      "completions/mean_length": 444.98828125,
      "completions/mean_terminated_length": 448.49212646484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.1472,
      "grad_norm": 0.394085168838501,
      "learning_rate": 1.75e-06,
      "loss": 0.2411,
      "num_tokens": 28767878.0,
      "reward": 1.1620368957519531,
      "reward_std": 0.20082634687423706,
      "rewards/accuracy_reward_step": 0.55078125,
      "rewards/final_brier_reward_step": 0.7771989107131958,
      "rewards/format_reward_step": 0.99609375,
      "step": 138
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8436331208795309,
      "aux_distill/mean_u": 0.2814290154348513,
      "aux_distill/n_active_tok": 138.875,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.296875,
      "calib/ece": 0.11386718749999998,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.12109375,
      "calib/gap": 0.23992398594698178,
      "calib/mean_conf": 0.5486328125,
      "calib/mu_c": 0.6432903225806451,
      "calib/mu_w": 0.40336633663366334,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.028515624999999992,
      "calib/std_conf": 0.2902290413162333,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25776119402985076,
      "calib/step_q_c_n": 1407.0,
      "calib/step_q_gap": -0.02217714101639795,
      "calib/step_q_w": 0.2799383350462487,
      "calib/step_q_w_n": 973.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1872.0,
      "completions/max_terminated_length": 1872.0,
      "completions/mean_length": 397.0625,
      "completions/mean_terminated_length": 400.18896484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.14826666666666666,
      "grad_norm": 0.4404184818267822,
      "learning_rate": 1.7222222222222224e-06,
      "loss": 0.2486,
      "num_tokens": 28972622.0,
      "reward": 1.1968767642974854,
      "reward_std": 0.18953311443328857,
      "rewards/accuracy_reward_step": 0.60546875,
      "rewards/final_brier_reward_step": 0.7882847785949707,
      "rewards/format_reward_step": 1.0,
      "step": 139
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8466813545674086,
      "aux_distill/mean_u": 0.23530338947091445,
      "aux_distill/n_active_tok": 169.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.015625,
      "calib/ece": 0.17171874999999995,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.171875,
      "calib/gap": 0.21049717744677948,
      "calib/mean_conf": 0.569765625,
      "calib/mu_c": 0.641301775147929,
      "calib/mu_w": 0.4308045977011495,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.040664062499999945,
      "calib/std_conf": 0.309708695903682,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2294391934467549,
      "calib/step_q_c_n": 1587.0,
      "calib/step_q_gap": -0.00437861617453475,
      "calib/step_q_w": 0.23381780962128965,
      "calib/step_q_w_n": 977.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 994.0,
      "completions/max_terminated_length": 994.0,
      "completions/mean_length": 428.6171875,
      "completions/mean_terminated_length": 431.99212646484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 135.0,
      "epoch": 0.14933333333333335,
      "grad_norm": 0.6040653586387634,
      "learning_rate": 1.6944444444444446e-06,
      "loss": 0.2489,
      "num_tokens": 29187364.0,
      "reward": 1.2130831480026245,
      "reward_std": 0.1731255054473877,
      "rewards/accuracy_reward_step": 0.66015625,
      "rewards/final_brier_reward_step": 0.7660101652145386,
      "rewards/format_reward_step": 1.0,
      "step": 140
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8504192354157567,
      "aux_distill/mean_u": 0.27125072911758774,
      "aux_distill/n_active_tok": 163.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.58203125,
      "calib/ece": 0.17248046874999998,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.296875,
      "calib/gap": 0.1967826293953614,
      "calib/mean_conf": 0.65916015625,
      "calib/mu_c": 0.7260355029585798,
      "calib/mu_w": 0.5292528735632184,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08574218749999997,
      "calib/std_conf": 0.3202229718395068,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24454718176060797,
      "calib/step_q_c_n": 1579.0,
      "calib/step_q_gap": 0.021320637138182302,
      "calib/step_q_w": 0.22322654462242567,
      "calib/step_q_w_n": 874.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1278.0,
      "completions/max_terminated_length": 1278.0,
      "completions/mean_length": 444.90234375,
      "completions/mean_terminated_length": 448.405517578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 142.0,
      "epoch": 0.1504,
      "grad_norm": 0.4739407002925873,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.2614,
      "num_tokens": 29408355.0,
      "reward": 1.2107794284820557,
      "reward_std": 0.21473020315170288,
      "rewards/accuracy_reward_step": 0.66015625,
      "rewards/final_brier_reward_step": 0.7614026069641113,
      "rewards/format_reward_step": 1.0,
      "step": 141
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8319859988987446,
      "aux_distill/mean_u": 0.23112263651612833,
      "aux_distill/n_active_tok": 160.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 9.91015625,
      "calib/ece": 0.13419607843137257,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.27450980392156865,
      "calib/gap": 0.26719765684051405,
      "calib/mean_conf": 0.6069019607843138,
      "calib/mu_c": 0.7200680272108844,
      "calib/mu_w": 0.4528703703703703,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08231372549019614,
      "calib/std_conf": 0.3279362237675192,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2621822358346095,
      "calib/step_q_c_n": 1306.0,
      "calib/step_q_gap": -0.07119713053419635,
      "calib/step_q_w": 0.33337936636880583,
      "calib/step_q_w_n": 1231.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1303.0,
      "completions/max_terminated_length": 1303.0,
      "completions/mean_length": 446.37109375,
      "completions/mean_terminated_length": 449.8858337402344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 143.0,
      "epoch": 0.15146666666666667,
      "grad_norm": 0.3541559875011444,
      "learning_rate": 1.638888888888889e-06,
      "loss": 0.1921,
      "num_tokens": 29627786.0,
      "reward": 1.1725636720657349,
      "reward_std": 0.19649362564086914,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.7748148441314697,
      "rewards/format_reward_step": 0.99609375,
      "step": 142
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.855469910427928,
      "aux_distill/mean_u": 0.2705275509708212,
      "aux_distill/n_active_tok": 142.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.390625,
      "calib/ece": 0.15089843749999995,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.3515625,
      "calib/gap": 0.3437423538047467,
      "calib/mean_conf": 0.6123046875,
      "calib/mu_c": 0.7761194029850746,
      "calib/mu_w": 0.43237704918032793,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11988281249999996,
      "calib/std_conf": 0.36266989679394035,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2441018998272884,
      "calib/step_q_c_n": 1158.0,
      "calib/step_q_gap": -0.03865893484365859,
      "calib/step_q_w": 0.282760834670947,
      "calib/step_q_w_n": 1246.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1415.0,
      "completions/max_terminated_length": 1415.0,
      "completions/mean_length": 422.3984375,
      "completions/mean_terminated_length": 425.7243957519531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 115.0,
      "epoch": 0.15253333333333333,
      "grad_norm": 0.3848201632499695,
      "learning_rate": 1.6111111111111113e-06,
      "loss": 0.2682,
      "num_tokens": 29843256.0,
      "reward": 1.153026819229126,
      "reward_std": 0.15764334797859192,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.7826160192489624,
      "rewards/format_reward_step": 1.0,
      "step": 143
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8302166722714901,
      "aux_distill/mean_u": 0.24965398126664595,
      "aux_distill/n_active_tok": 149.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.83203125,
      "calib/ece": 0.19162499999999993,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.484375,
      "calib/gap": 0.31800591675266615,
      "calib/mean_conf": 0.66540625,
      "calib/mu_c": 0.7709941520467838,
      "calib/mu_w": 0.45298823529411764,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.09453124999999996,
      "calib/std_conf": 0.36710423923040914,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2574477244772448,
      "calib/step_q_c_n": 1626.0,
      "calib/step_q_gap": 0.05942303311922009,
      "calib/step_q_w": 0.1980246913580247,
      "calib/step_q_w_n": 891.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1286.0,
      "completions/max_terminated_length": 1286.0,
      "completions/mean_length": 444.6015625,
      "completions/mean_terminated_length": 448.10235595703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 90.0,
      "epoch": 0.1536,
      "grad_norm": 0.4171602725982666,
      "learning_rate": 1.5833333333333333e-06,
      "loss": 0.2536,
      "num_tokens": 30061202.0,
      "reward": 1.2262345552444458,
      "reward_std": 0.14458578824996948,
      "rewards/accuracy_reward_step": 0.66796875,
      "rewards/final_brier_reward_step": 0.784500241279602,
      "rewards/format_reward_step": 1.0,
      "step": 144
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7345545515418053,
      "aux_distill/mean_u": 0.25175217095796415,
      "aux_distill/n_active_tok": 182.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 11.49609375,
      "calib/ece": 0.18062055335968383,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.5335968379446641,
      "calib/gap": 0.322297159979156,
      "calib/mean_conf": 0.6942964426877472,
      "calib/mu_c": 0.8229605263157896,
      "calib/mu_w": 0.5006633663366337,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.13706324110671939,
      "calib/std_conf": 0.365968198616236,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24804450438300743,
      "calib/step_q_c_n": 1483.0,
      "calib/step_q_gap": -0.055667824384115855,
      "calib/step_q_w": 0.3037123287671233,
      "calib/step_q_w_n": 1460.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1904.0,
      "completions/max_terminated_length": 1904.0,
      "completions/mean_length": 460.94140625,
      "completions/mean_terminated_length": 466.4071350097656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 155.0,
      "epoch": 0.15466666666666667,
      "grad_norm": 0.983404278755188,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 0.2223,
      "num_tokens": 30281907.0,
      "reward": 1.1744862794876099,
      "reward_std": 0.23245766758918762,
      "rewards/accuracy_reward_step": 0.59765625,
      "rewards/final_brier_reward_step": 0.7630351185798645,
      "rewards/format_reward_step": 0.98828125,
      "step": 145
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8586379196494818,
      "aux_distill/mean_u": 0.2799588618828064,
      "aux_distill/n_active_tok": 202.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 11.796875,
      "calib/ece": 0.25566929133858274,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.5118110236220472,
      "calib/gap": 0.30277665995975855,
      "calib/mean_conf": 0.6478740157480315,
      "calib/mu_c": 0.8171428571428573,
      "calib/mu_w": 0.5143661971830987,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.23129921259842523,
      "calib/std_conf": 0.39229915854773334,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27074539363484085,
      "calib/step_q_c_n": 1194.0,
      "calib/step_q_gap": -0.0038603018744691187,
      "calib/step_q_w": 0.27460569550930997,
      "calib/step_q_w_n": 1826.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1861.0,
      "completions/max_terminated_length": 1861.0,
      "completions/mean_length": 506.44921875,
      "completions/mean_terminated_length": 508.4353332519531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 151.0,
      "epoch": 0.15573333333333333,
      "grad_norm": 0.4449126720428467,
      "learning_rate": 1.527777777777778e-06,
      "loss": 0.2461,
      "num_tokens": 30518774.0,
      "reward": 1.0651085376739502,
      "reward_std": 0.24594148993492126,
      "rewards/accuracy_reward_step": 0.4375,
      "rewards/final_brier_reward_step": 0.7005296945571899,
      "rewards/format_reward_step": 0.9921875,
      "step": 146
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8714845888316631,
      "aux_distill/mean_u": 0.3000520868029459,
      "aux_distill/n_active_tok": 202.375,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.421875,
      "calib/ece": 0.26945098039215687,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.596078431372549,
      "calib/gap": 0.2942093252744541,
      "calib/mean_conf": 0.7012156862745098,
      "calib/mu_c": 0.840820895522388,
      "calib/mu_w": 0.5466115702479339,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.22258823529411764,
      "calib/std_conf": 0.39100936821059745,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2534572784810127,
      "calib/step_q_c_n": 1264.0,
      "calib/step_q_gap": -0.14323787821414397,
      "calib/step_q_w": 0.39669515669515665,
      "calib/step_q_w_n": 1404.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2762.0,
      "completions/max_terminated_length": 2762.0,
      "completions/mean_length": 458.91015625,
      "completions/mean_terminated_length": 460.7098388671875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 161.0,
      "epoch": 0.1568,
      "grad_norm": 0.3406878113746643,
      "learning_rate": 1.5e-06,
      "loss": 0.2536,
      "num_tokens": 30739935.0,
      "reward": 1.1151740550994873,
      "reward_std": 0.1806105673313141,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.7108168005943298,
      "rewards/format_reward_step": 0.99609375,
      "step": 147
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8154330048710108,
      "aux_distill/mean_u": 0.2554237223511033,
      "aux_distill/n_active_tok": 159.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.0546875,
      "calib/ece": 0.17080468749999994,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.69140625,
      "calib/gap": 0.3563581395348837,
      "calib/mean_conf": 0.7840859375,
      "calib/mu_c": 0.9038,
      "calib/mu_w": 0.5474418604651163,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.14541406249999994,
      "calib/std_conf": 0.34397198576300664,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2468400263331139,
      "calib/step_q_c_n": 1519.0,
      "calib/step_q_gap": -0.04597513954366342,
      "calib/step_q_w": 0.2928151658767773,
      "calib/step_q_w_n": 1055.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1552.0,
      "completions/max_terminated_length": 1552.0,
      "completions/mean_length": 434.2109375,
      "completions/mean_terminated_length": 437.6299133300781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 93.0,
      "epoch": 0.15786666666666666,
      "grad_norm": 0.3257892429828644,
      "learning_rate": 1.4722222222222225e-06,
      "loss": 0.2332,
      "num_tokens": 30956205.0,
      "reward": 1.233625888824463,
      "reward_std": 0.23119288682937622,
      "rewards/accuracy_reward_step": 0.6640625,
      "rewards/final_brier_reward_step": 0.8031893968582153,
      "rewards/format_reward_step": 1.0,
      "step": 148
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7958731511607766,
      "aux_distill/mean_u": 0.2204305141091541,
      "aux_distill/n_active_tok": 161.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.69921875,
      "calib/ece": 0.22457031249999992,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.64453125,
      "calib/gap": 0.3587686450727079,
      "calib/mean_conf": 0.7226171875,
      "calib/mu_c": 0.875374149659864,
      "calib/mu_w": 0.5166055045871562,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.18648437499999992,
      "calib/std_conf": 0.3883586579498258,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.30053857350800584,
      "calib/step_q_c_n": 1374.0,
      "calib/step_q_gap": -0.06163725066781833,
      "calib/step_q_w": 0.36217582417582417,
      "calib/step_q_w_n": 1365.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1747.0,
      "completions/max_terminated_length": 1747.0,
      "completions/mean_length": 504.1796875,
      "completions/mean_terminated_length": 508.14959716796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 102.0,
      "epoch": 0.15893333333333334,
      "grad_norm": 0.587183952331543,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 0.243,
      "num_tokens": 31189731.0,
      "reward": 1.1661572456359863,
      "reward_std": 0.2592434883117676,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.7580957412719727,
      "rewards/format_reward_step": 1.0,
      "step": 149
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8890952225774527,
      "aux_distill/mean_u": 0.311144694626446,
      "aux_distill/n_active_tok": 176.75,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 9.375,
      "calib/ece": 0.21964705882352953,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.788235294117647,
      "calib/gap": 0.34338901213623896,
      "calib/mean_conf": 0.8215294117647061,
      "calib/mu_c": 0.9521518987341772,
      "calib/mu_w": 0.6087628865979382,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2107843137254903,
      "calib/std_conf": 0.3387757863966573,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23309803921568625,
      "calib/step_q_c_n": 1275.0,
      "calib/step_q_gap": -0.10549751633986934,
      "calib/step_q_w": 0.3385955555555556,
      "calib/step_q_w_n": 1125.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2533.0,
      "completions/max_terminated_length": 2533.0,
      "completions/mean_length": 424.16015625,
      "completions/mean_terminated_length": 427.5,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.16,
      "grad_norm": 0.4133373498916626,
      "learning_rate": 1.4166666666666667e-06,
      "loss": 0.2409,
      "num_tokens": 31403276.0,
      "reward": 1.1904525756835938,
      "reward_std": 0.18449309468269348,
      "rewards/accuracy_reward_step": 0.6171875,
      "rewards/final_brier_reward_step": 0.767623782157898,
      "rewards/format_reward_step": 0.99609375,
      "step": 150
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8351477207615972,
      "aux_distill/mean_u": 0.2853782974883022,
      "aux_distill/n_active_tok": 207.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 11.62890625,
      "calib/ece": 0.2806274509803922,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.6705882352941176,
      "calib/gap": 0.26302917403384696,
      "calib/mean_conf": 0.7217254901960785,
      "calib/mu_c": 0.8320945945945948,
      "calib/mu_w": 0.5690654205607478,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.21098039215686276,
      "calib/std_conf": 0.4011681021878422,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2626766040181465,
      "calib/step_q_c_n": 1543.0,
      "calib/step_q_gap": -0.049359658185479716,
      "calib/step_q_w": 0.31203626220362624,
      "calib/step_q_w_n": 1434.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2941.0,
      "completions/max_terminated_length": 2941.0,
      "completions/mean_length": 494.9296875,
      "completions/mean_terminated_length": 498.8267822265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.16106666666666666,
      "grad_norm": 0.3437005579471588,
      "learning_rate": 1.3888888888888892e-06,
      "loss": 0.2287,
      "num_tokens": 31637002.0,
      "reward": 1.1375683546066284,
      "reward_std": 0.19518016278743744,
      "rewards/accuracy_reward_step": 0.578125,
      "rewards/final_brier_reward_step": 0.7009179592132568,
      "rewards/format_reward_step": 0.99609375,
      "step": 151
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8661780599504709,
      "aux_distill/mean_u": 0.29071673239059204,
      "aux_distill/n_active_tok": 143.25,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.77734375,
      "calib/ece": 0.25152941176470583,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.6823529411764706,
      "calib/gap": 0.3766925351071694,
      "calib/mean_conf": 0.7257254901960786,
      "calib/mu_c": 0.9074242424242426,
      "calib/mu_w": 0.5307317073170732,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2298039215686274,
      "calib/std_conf": 0.4019698592949756,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2951077943615257,
      "calib/step_q_c_n": 1206.0,
      "calib/step_q_gap": -0.012941143178718972,
      "calib/step_q_w": 0.30804893754024465,
      "calib/step_q_w_n": 1553.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2099.0,
      "completions/max_terminated_length": 2099.0,
      "completions/mean_length": 473.08203125,
      "completions/mean_terminated_length": 476.8070983886719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 179.0,
      "epoch": 0.16213333333333332,
      "grad_norm": 0.31601613759994507,
      "learning_rate": 1.3611111111111112e-06,
      "loss": 0.1938,
      "num_tokens": 31863503.0,
      "reward": 1.1211999654769897,
      "reward_std": 0.2283134013414383,
      "rewards/accuracy_reward_step": 0.515625,
      "rewards/final_brier_reward_step": 0.730681300163269,
      "rewards/format_reward_step": 0.99609375,
      "step": 152
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8924239929765463,
      "aux_distill/mean_u": 0.3081032026315535,
      "aux_distill/n_active_tok": 148.875,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 10.80859375,
      "calib/ece": 0.27490234375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.6640625,
      "calib/gap": 0.27321370680153023,
      "calib/mean_conf": 0.70943359375,
      "calib/mu_c": 0.8129559748427673,
      "calib/mu_w": 0.539742268041237,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.18162109375000002,
      "calib/std_conf": 0.40331460110589845,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3348767288033674,
      "calib/step_q_c_n": 1663.0,
      "calib/step_q_gap": 0.051987236049744145,
      "calib/step_q_w": 0.28288949275362324,
      "calib/step_q_w_n": 1104.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1936.0,
      "completions/max_terminated_length": 1936.0,
      "completions/mean_length": 495.67578125,
      "completions/mean_terminated_length": 499.5787353515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 156.0,
      "epoch": 0.1632,
      "grad_norm": 0.3540187180042267,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.2643,
      "num_tokens": 32097716.0,
      "reward": 1.1719425916671753,
      "reward_std": 0.20725688338279724,
      "rewards/accuracy_reward_step": 0.62109375,
      "rewards/final_brier_reward_step": 0.7227912545204163,
      "rewards/format_reward_step": 1.0,
      "step": 153
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8481863886117935,
      "aux_distill/mean_u": 0.2958641144448045,
      "aux_distill/n_active_tok": 167.125,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 11.3515625,
      "calib/ece": 0.25767068273092375,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.6746987951807228,
      "calib/gap": 0.42878292871900814,
      "calib/mean_conf": 0.7087550200803213,
      "calib/mu_c": 0.9291735537190082,
      "calib/mu_w": 0.5003906250000001,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2402409638554217,
      "calib/std_conf": 0.41155261930734266,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28757575757575754,
      "calib/step_q_c_n": 1089.0,
      "calib/step_q_gap": -0.09181097880288852,
      "calib/step_q_w": 0.37938673637864606,
      "calib/step_q_w_n": 1817.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 3017.0,
      "completions/max_terminated_length": 3017.0,
      "completions/mean_length": 480.69921875,
      "completions/mean_terminated_length": 488.32940673828125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 174.0,
      "epoch": 0.16426666666666667,
      "grad_norm": 0.3609575629234314,
      "learning_rate": 1.3055555555555556e-06,
      "loss": 0.2392,
      "num_tokens": 32325215.0,
      "reward": 1.083282232284546,
      "reward_std": 0.2492285668849945,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.7212519645690918,
      "rewards/format_reward_step": 0.96875,
      "step": 154
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8189228344708681,
      "aux_distill/mean_u": 0.20074483546096195,
      "aux_distill/n_active_tok": 151.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.73828125,
      "calib/ece": 0.33277343750000005,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.7265625,
      "calib/gap": 0.23265598238316632,
      "calib/mean_conf": 0.7462890625,
      "calib/mu_c": 0.8571641791044777,
      "calib/mu_w": 0.6245081967213114,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2778125000000001,
      "calib/std_conf": 0.39972536768120065,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.302288595810706,
      "calib/step_q_c_n": 1289.0,
      "calib/step_q_gap": -0.01769479289361292,
      "calib/step_q_w": 0.3199833887043189,
      "calib/step_q_w_n": 1204.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1084.0,
      "completions/max_terminated_length": 1084.0,
      "completions/mean_length": 423.73828125,
      "completions/mean_terminated_length": 427.0747985839844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 162.0,
      "epoch": 0.16533333333333333,
      "grad_norm": 0.4874245822429657,
      "learning_rate": 1.2777777777777779e-06,
      "loss": 0.2194,
      "num_tokens": 32540908.0,
      "reward": 1.0903079509735107,
      "reward_std": 0.25045233964920044,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.657178521156311,
      "rewards/format_reward_step": 1.0,
      "step": 155
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8304382590577006,
      "aux_distill/mean_u": 0.3006649706126327,
      "aux_distill/n_active_tok": 165.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 10.59375,
      "calib/ece": 0.23736220472440941,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.594488188976378,
      "calib/gap": 0.3713015557762331,
      "calib/mean_conf": 0.632007874015748,
      "calib/mu_c": 0.7708805031446541,
      "calib/mu_w": 0.399578947368421,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12169291338582677,
      "calib/std_conf": 0.43995022975605924,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3129776021080369,
      "calib/step_q_c_n": 1518.0,
      "calib/step_q_gap": 0.0053394111532630295,
      "calib/step_q_w": 0.30763819095477385,
      "calib/step_q_w_n": 1194.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2464.0,
      "completions/max_terminated_length": 2464.0,
      "completions/mean_length": 504.10546875,
      "completions/mean_terminated_length": 506.0823974609375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 183.0,
      "epoch": 0.1664,
      "grad_norm": 0.4381815195083618,
      "learning_rate": 1.25e-06,
      "loss": 0.2404,
      "num_tokens": 32774719.0,
      "reward": 1.176797866821289,
      "reward_std": 0.20549890398979187,
      "rewards/accuracy_reward_step": 0.62109375,
      "rewards/final_brier_reward_step": 0.7403144836425781,
      "rewards/format_reward_step": 0.9921875,
      "step": 156
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8249999564141035,
      "aux_distill/mean_u": 0.2358457818520905,
      "aux_distill/n_active_tok": 198.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 11.9296875,
      "calib/ece": 0.2089330708661416,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.6259842519685039,
      "calib/gap": 0.4298190476190476,
      "calib/mean_conf": 0.6625551181102363,
      "calib/mu_c": 0.8046999999999999,
      "calib/mu_w": 0.37488095238095226,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10109842519685026,
      "calib/std_conf": 0.43245846702562246,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.258047619047619,
      "calib/step_q_c_n": 1680.0,
      "calib/step_q_gap": -0.06967836348513212,
      "calib/step_q_w": 0.32772598253275115,
      "calib/step_q_w_n": 1374.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2192.0,
      "completions/max_terminated_length": 2192.0,
      "completions/mean_length": 506.83203125,
      "completions/mean_terminated_length": 512.8419189453125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 146.0,
      "epoch": 0.16746666666666668,
      "grad_norm": 0.37992429733276367,
      "learning_rate": 1.2222222222222223e-06,
      "loss": 0.2279,
      "num_tokens": 33008196.0,
      "reward": 1.2160041332244873,
      "reward_std": 0.17098858952522278,
      "rewards/accuracy_reward_step": 0.6640625,
      "rewards/final_brier_reward_step": 0.7757582068443298,
      "rewards/format_reward_step": 0.9921875,
      "step": 157
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.746043186634779,
      "aux_distill/mean_u": 0.24562522883955865,
      "aux_distill/n_active_tok": 183.0,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 11.19921875,
      "calib/ece": 0.18522529644268768,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.6245059288537549,
      "calib/gap": 0.4845342349957736,
      "calib/mean_conf": 0.6513754940711463,
      "calib/mu_c": 0.8122485207100593,
      "calib/mu_w": 0.3277142857142857,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08430830039525684,
      "calib/std_conf": 0.43960899500620837,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26825465838509316,
      "calib/step_q_c_n": 1610.0,
      "calib/step_q_gap": -0.0423897330230214,
      "calib/step_q_w": 0.31064439140811456,
      "calib/step_q_w_n": 1257.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 2494.0,
      "completions/max_terminated_length": 2494.0,
      "completions/mean_length": 484.640625,
      "completions/mean_terminated_length": 492.3333740234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 118.0,
      "epoch": 0.16853333333333334,
      "grad_norm": 0.5074396133422852,
      "learning_rate": 1.1944444444444446e-06,
      "loss": 0.2143,
      "num_tokens": 33237504.0,
      "reward": 1.2193374633789062,
      "reward_std": 0.18944784998893738,
      "rewards/accuracy_reward_step": 0.66015625,
      "rewards/final_brier_reward_step": 0.7902376651763916,
      "rewards/format_reward_step": 0.98828125,
      "step": 158
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8241626480594277,
      "aux_distill/mean_u": 0.26721943700812645,
      "aux_distill/n_active_tok": 166.125,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 11.44921875,
      "calib/ece": 0.2765357142857143,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.6031746031746031,
      "calib/gap": 0.3270807702248644,
      "calib/mean_conf": 0.6278293650793652,
      "calib/mu_c": 0.7654109589041097,
      "calib/mu_w": 0.4383301886792453,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1625,
      "calib/std_conf": 0.44762274038426675,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20295438596491228,
      "calib/step_q_c_n": 1425.0,
      "calib/step_q_gap": -0.07211666317187393,
      "calib/step_q_w": 0.2750710491367862,
      "calib/step_q_w_n": 1506.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 2296.0,
      "completions/max_terminated_length": 2296.0,
      "completions/mean_length": 473.73828125,
      "completions/mean_terminated_length": 481.2579650878906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 145.0,
      "epoch": 0.1696,
      "grad_norm": 0.4542931914329529,
      "learning_rate": 1.1666666666666668e-06,
      "loss": 0.1962,
      "num_tokens": 33463565.0,
      "reward": 1.1282752752304077,
      "reward_std": 0.21358579397201538,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.701863169670105,
      "rewards/format_reward_step": 0.984375,
      "step": 159
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.879276018589735,
      "aux_distill/mean_u": 0.3238322491555653,
      "aux_distill/n_active_tok": 163.25,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 11.703125,
      "calib/ece": 0.30257028112449796,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.5502008032128514,
      "calib/gap": 0.311876102292769,
      "calib/mean_conf": 0.5762248995983936,
      "calib/mu_c": 0.6776785714285715,
      "calib/mu_w": 0.36580246913580244,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1020481927710843,
      "calib/std_conf": 0.45937294175971427,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2581012658227848,
      "calib/step_q_c_n": 1659.0,
      "calib/step_q_gap": 0.027802088560256732,
      "calib/step_q_w": 0.23029917726252808,
      "calib/step_q_w_n": 1337.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.02734375,
      "completions/max_length": 2121.0,
      "completions/max_terminated_length": 2121.0,
      "completions/mean_length": 468.46484375,
      "completions/mean_terminated_length": 481.634521484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 141.0,
      "epoch": 0.17066666666666666,
      "grad_norm": 0.559697151184082,
      "learning_rate": 1.138888888888889e-06,
      "loss": 0.1897,
      "num_tokens": 33688332.0,
      "reward": 1.153278112411499,
      "reward_std": 0.2089913785457611,
      "rewards/accuracy_reward_step": 0.65625,
      "rewards/final_brier_reward_step": 0.677649974822998,
      "rewards/format_reward_step": 0.97265625,
      "step": 160
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8093566577881575,
      "aux_distill/mean_u": 0.26456430254127333,
      "aux_distill/n_active_tok": 152.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.7578125,
      "calib/ece": 0.24238281250000004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.59375,
      "calib/gap": 0.4298599162542825,
      "calib/mean_conf": 0.6154296875,
      "calib/mu_c": 0.7346486486486487,
      "calib/mu_w": 0.30478873239436616,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.06757812500000006,
      "calib/std_conf": 0.45531390585908127,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.21960448642266825,
      "calib/step_q_c_n": 1694.0,
      "calib/step_q_gap": 0.014156725228638395,
      "calib/step_q_w": 0.20544776119402985,
      "calib/step_q_w_n": 804.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1067.0,
      "completions/max_terminated_length": 1067.0,
      "completions/mean_length": 451.3671875,
      "completions/mean_terminated_length": 454.9212646484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 139.0,
      "epoch": 0.17173333333333332,
      "grad_norm": 0.5177509784698486,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.2397,
      "num_tokens": 33907802.0,
      "reward": 1.2339601516723633,
      "reward_std": 0.19392341375350952,
      "rewards/accuracy_reward_step": 0.72265625,
      "rewards/final_brier_reward_step": 0.7491703033447266,
      "rewards/format_reward_step": 0.99609375,
      "step": 161
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8228137148544192,
      "aux_distill/mean_u": 0.23076060475157925,
      "aux_distill/n_active_tok": 158.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.84375,
      "calib/ece": 0.20257812500000003,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.66796875,
      "calib/gap": 0.4412045454545454,
      "calib/mean_conf": 0.690703125,
      "calib/mu_c": 0.8285795454545454,
      "calib/mu_w": 0.38737499999999997,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10289062500000003,
      "calib/std_conf": 0.42986132137613214,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.19520210896309315,
      "calib/step_q_c_n": 1707.0,
      "calib/step_q_gap": -0.018709330151297998,
      "calib/step_q_w": 0.21391143911439114,
      "calib/step_q_w_n": 813.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1096.0,
      "completions/max_terminated_length": 1096.0,
      "completions/mean_length": 449.9765625,
      "completions/mean_terminated_length": 453.5196838378906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 142.0,
      "epoch": 0.1728,
      "grad_norm": 0.39314666390419006,
      "learning_rate": 1.0833333333333335e-06,
      "loss": 0.2273,
      "num_tokens": 34127140.0,
      "reward": 1.2387226819992065,
      "reward_std": 0.182912677526474,
      "rewards/accuracy_reward_step": 0.6875,
      "rewards/final_brier_reward_step": 0.7899453043937683,
      "rewards/format_reward_step": 1.0,
      "step": 162
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8334611281752586,
      "aux_distill/mean_u": 0.29235313919252837,
      "aux_distill/n_active_tok": 174.625,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 11.72265625,
      "calib/ece": 0.19699604743082988,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.5928853754940712,
      "calib/gap": 0.5048886731391586,
      "calib/mean_conf": 0.621185770750988,
      "calib/mu_c": 0.8267333333333334,
      "calib/mu_w": 0.32184466019417474,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11264822134387333,
      "calib/std_conf": 0.453629387954209,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23267145656433702,
      "calib/step_q_c_n": 1531.0,
      "calib/step_q_gap": 0.02569186472760232,
      "calib/step_q_w": 0.2069795918367347,
      "calib/step_q_w_n": 1470.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2867.0,
      "completions/max_terminated_length": 2867.0,
      "completions/mean_length": 539.78125,
      "completions/mean_terminated_length": 546.1818237304688,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 167.0,
      "epoch": 0.17386666666666667,
      "grad_norm": 0.5653132200241089,
      "learning_rate": 1.0555555555555557e-06,
      "loss": 0.2166,
      "num_tokens": 34370156.0,
      "reward": 1.1803362369537354,
      "reward_std": 0.18647001683712006,
      "rewards/accuracy_reward_step": 0.5859375,
      "rewards/final_brier_reward_step": 0.786453902721405,
      "rewards/format_reward_step": 0.98828125,
      "step": 163
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8464186768978834,
      "aux_distill/mean_u": 0.26922085797948403,
      "aux_distill/n_active_tok": 195.375,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 13.1640625,
      "calib/ece": 0.18371999999999988,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.652,
      "calib/gap": 0.5143531559728742,
      "calib/mean_conf": 0.67956,
      "calib/mu_c": 0.9017605633802815,
      "calib/mu_w": 0.38740740740740737,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.14763999999999985,
      "calib/std_conf": 0.4355210745761909,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2321144781144781,
      "calib/step_q_c_n": 1485.0,
      "calib/step_q_gap": 0.023690074931454225,
      "calib/step_q_w": 0.20842440318302388,
      "calib/step_q_w_n": 1885.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2886.0,
      "completions/max_terminated_length": 2886.0,
      "completions/mean_length": 601.07421875,
      "completions/mean_terminated_length": 608.2015991210938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 77.0,
      "epoch": 0.17493333333333333,
      "grad_norm": 0.3060588836669922,
      "learning_rate": 1.0277777777777777e-06,
      "loss": 0.2373,
      "num_tokens": 34630167.0,
      "reward": 1.1586520671844482,
      "reward_std": 0.1987255960702896,
      "rewards/accuracy_reward_step": 0.5546875,
      "rewards/final_brier_reward_step": 0.786054253578186,
      "rewards/format_reward_step": 0.9765625,
      "step": 164
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8290663491934538,
      "aux_distill/mean_u": 0.21056666403523405,
      "aux_distill/n_active_tok": 169.375,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 12.0703125,
      "calib/ece": 0.2945849802371541,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.7747035573122529,
      "calib/gap": 0.2901091758172408,
      "calib/mean_conf": 0.7915810276679842,
      "calib/mu_c": 0.9223021582733812,
      "calib/mu_w": 0.6321929824561404,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2683794466403162,
      "calib/std_conf": 0.37803197257410914,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2831420765027322,
      "calib/step_q_c_n": 1464.0,
      "calib/step_q_gap": -0.04385054342346706,
      "calib/step_q_w": 0.32699261992619927,
      "calib/step_q_w_n": 1626.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1866.0,
      "completions/max_terminated_length": 1866.0,
      "completions/mean_length": 513.33203125,
      "completions/mean_terminated_length": 521.4801635742188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 175.0,
      "epoch": 0.176,
      "grad_norm": 0.3841126263141632,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.2575,
      "num_tokens": 34867156.0,
      "reward": 1.1088169813156128,
      "reward_std": 0.19576653838157654,
      "rewards/accuracy_reward_step": 0.54296875,
      "rewards/final_brier_reward_step": 0.6863839626312256,
      "rewards/format_reward_step": 0.98828125,
      "step": 165
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8489821460098028,
      "aux_distill/mean_u": 0.29878565550623887,
      "aux_distill/n_active_tok": 201.375,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 12.6171875,
      "calib/ece": 0.2149802371541502,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.83399209486166,
      "calib/gap": 0.29267097829319033,
      "calib/mean_conf": 0.8430434782608697,
      "calib/mu_c": 0.930960451977401,
      "calib/mu_w": 0.6382894736842106,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.17920948616600793,
      "calib/std_conf": 0.33328023865394385,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2746310504634397,
      "calib/step_q_c_n": 1942.0,
      "calib/step_q_gap": 0.02766676474915397,
      "calib/step_q_w": 0.24696428571428575,
      "calib/step_q_w_n": 1288.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1995.0,
      "completions/max_terminated_length": 1995.0,
      "completions/mean_length": 547.95703125,
      "completions/mean_terminated_length": 554.45458984375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 40.0,
      "epoch": 0.17706666666666668,
      "grad_norm": 0.31519395112991333,
      "learning_rate": 9.722222222222224e-07,
      "loss": 0.2368,
      "num_tokens": 35113617.0,
      "reward": 1.2258692979812622,
      "reward_std": 0.23709669709205627,
      "rewards/accuracy_reward_step": 0.69140625,
      "rewards/final_brier_reward_step": 0.7720511555671692,
      "rewards/format_reward_step": 0.98828125,
      "step": 166
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8104597348719835,
      "aux_distill/mean_u": 0.2658721026737349,
      "aux_distill/n_active_tok": 182.875,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 11.82421875,
      "calib/ece": 0.17719367588932802,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.8814229249011858,
      "calib/gap": 0.26956763925729454,
      "calib/mean_conf": 0.8803557312252965,
      "calib/mu_c": 0.9421538461538463,
      "calib/mu_w": 0.6725862068965518,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.14339920948616597,
      "calib/std_conf": 0.29882290804470024,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.275175079581628,
      "calib/step_q_c_n": 2199.0,
      "calib/step_q_gap": -0.04958337452465217,
      "calib/step_q_w": 0.3247584541062802,
      "calib/step_q_w_n": 828.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3055.0,
      "completions/max_terminated_length": 3055.0,
      "completions/mean_length": 530.625,
      "completions/mean_terminated_length": 534.8031616210938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 187.0,
      "epoch": 0.17813333333333334,
      "grad_norm": 0.3957613408565521,
      "learning_rate": 9.444444444444445e-07,
      "loss": 0.2567,
      "num_tokens": 35355065.0,
      "reward": 1.2788411378860474,
      "reward_std": 0.20663800835609436,
      "rewards/accuracy_reward_step": 0.76171875,
      "rewards/final_brier_reward_step": 0.807682454586029,
      "rewards/format_reward_step": 0.98828125,
      "step": 167
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8572558909654617,
      "aux_distill/mean_u": 0.32651245974212556,
      "aux_distill/n_active_tok": 207.875,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 13.33984375,
      "calib/ece": 0.21231075697211158,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.896414342629482,
      "calib/gap": 0.26695864661654123,
      "calib/mean_conf": 0.8923107569721117,
      "calib/mu_c": 0.9731428571428571,
      "calib/mu_w": 0.7061842105263159,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.20370517928286855,
      "calib/std_conf": 0.2860310890588503,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.31775862068965516,
      "calib/step_q_c_n": 2030.0,
      "calib/step_q_gap": 0.00037233910120748615,
      "calib/step_q_w": 0.31738628158844767,
      "calib/step_q_w_n": 1385.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2769.0,
      "completions/max_terminated_length": 2769.0,
      "completions/mean_length": 595.70703125,
      "completions/mean_terminated_length": 602.770751953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 196.0,
      "epoch": 0.1792,
      "grad_norm": 0.2700546383857727,
      "learning_rate": 9.166666666666666e-07,
      "loss": 0.2194,
      "num_tokens": 35612238.0,
      "reward": 1.215261459350586,
      "reward_std": 0.27505505084991455,
      "rewards/accuracy_reward_step": 0.68359375,
      "rewards/final_brier_reward_step": 0.7664605379104614,
      "rewards/format_reward_step": 0.98046875,
      "step": 168
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8547635991126299,
      "aux_distill/mean_u": 0.2882160263843028,
      "aux_distill/n_active_tok": 185.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 12.34765625,
      "calib/ece": 0.28507936507936515,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.8412698412698413,
      "calib/gap": 0.2736870748299318,
      "calib/mean_conf": 0.8474603174603176,
      "calib/mu_c": 0.9614965986394556,
      "calib/mu_w": 0.6878095238095238,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2746031746031747,
      "calib/std_conf": 0.3311450468844275,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3273376623376623,
      "calib/step_q_c_n": 1694.0,
      "calib/step_q_gap": 0.03185027310794181,
      "calib/step_q_w": 0.2954873892297205,
      "calib/step_q_w_n": 1467.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2432.0,
      "completions/max_terminated_length": 2432.0,
      "completions/mean_length": 571.69921875,
      "completions/mean_terminated_length": 578.478271484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 140.0,
      "epoch": 0.18026666666666666,
      "grad_norm": 0.2821076214313507,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.2208,
      "num_tokens": 35862777.0,
      "reward": 1.1290287971496582,
      "reward_std": 0.2493588924407959,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.6994640231132507,
      "rewards/format_reward_step": 0.984375,
      "step": 169
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.829416710883379,
      "aux_distill/mean_u": 0.25372244946181366,
      "aux_distill/n_active_tok": 203.0,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 13.0859375,
      "calib/ece": 0.22000000000000017,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.8605577689243028,
      "calib/gap": 0.24231843575418976,
      "calib/mean_conf": 0.8678087649402392,
      "calib/mu_c": 0.9373184357541898,
      "calib/mu_w": 0.6950000000000001,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.18733067729083683,
      "calib/std_conf": 0.31213628919231023,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3354214647627822,
      "calib/step_q_c_n": 2171.0,
      "calib/step_q_gap": 0.040764976213163884,
      "calib/step_q_w": 0.2946564885496183,
      "calib/step_q_w_n": 1179.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 1930.0,
      "completions/max_terminated_length": 1930.0,
      "completions/mean_length": 547.20703125,
      "completions/mean_terminated_length": 558.1076049804688,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 146.0,
      "epoch": 0.18133333333333335,
      "grad_norm": 0.3414306044578552,
      "learning_rate": 8.611111111111112e-07,
      "loss": 0.2157,
      "num_tokens": 36107014.0,
      "reward": 1.2150764465332031,
      "reward_std": 0.2328571081161499,
      "rewards/accuracy_reward_step": 0.69921875,
      "rewards/final_brier_reward_step": 0.7543718814849854,
      "rewards/format_reward_step": 0.9765625,
      "step": 170
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8333851397037506,
      "aux_distill/mean_u": 0.31075516679071946,
      "aux_distill/n_active_tok": 157.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 11.734375,
      "calib/ece": 0.37089791514463505,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.8549019607843137,
      "calib/gap": 0.18891153961358642,
      "calib/mean_conf": 0.8657998759289488,
      "calib/mu_c": 0.9554402535495133,
      "calib/mu_w": 0.7665287139359269,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.3556037974975762,
      "calib/std_conf": 0.3119447808665385,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2976370106761566,
      "calib/step_q_c_n": 1405.0,
      "calib/step_q_gap": -0.040511832350735244,
      "calib/step_q_w": 0.33814884302689185,
      "calib/step_q_w_n": 1599.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1384.0,
      "completions/max_terminated_length": 1384.0,
      "completions/mean_length": 539.51171875,
      "completions/mean_terminated_length": 543.7598266601562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 192.0,
      "epoch": 0.1824,
      "grad_norm": 0.4528205394744873,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.2122,
      "num_tokens": 36352025.0,
      "reward": 1.074401617050171,
      "reward_std": 0.23831433057785034,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.6292720437049866,
      "rewards/format_reward_step": 0.99609375,
      "step": 171
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8144433349370956,
      "aux_distill/mean_u": 0.22704867808540513,
      "aux_distill/n_active_tok": 169.875,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 10.875,
      "calib/ece": 0.26695219123505975,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.9203187250996016,
      "calib/gap": 0.15199018617405124,
      "calib/mean_conf": 0.9280677290836654,
      "calib/mu_c": 0.9777218934911244,
      "calib/mu_w": 0.8257317073170731,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.26085657370517923,
      "calib/std_conf": 0.22707904312620714,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.29474042027194064,
      "calib/step_q_c_n": 1618.0,
      "calib/step_q_gap": -0.04203487989958593,
      "calib/step_q_w": 0.33677530017152657,
      "calib/step_q_w_n": 1166.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 2308.0,
      "completions/max_terminated_length": 2308.0,
      "completions/mean_length": 483.25390625,
      "completions/mean_terminated_length": 492.8804931640625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.18346666666666667,
      "grad_norm": 0.39163994789123535,
      "learning_rate": 8.055555555555557e-07,
      "loss": 0.2099,
      "num_tokens": 36579090.0,
      "reward": 1.1783955097198486,
      "reward_std": 0.21855786442756653,
      "rewards/accuracy_reward_step": 0.66015625,
      "rewards/final_brier_reward_step": 0.7161659002304077,
      "rewards/format_reward_step": 0.98046875,
      "step": 172
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7717863172292709,
      "aux_distill/mean_u": 0.23909250423805128,
      "aux_distill/n_active_tok": 183.125,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 11.98046875,
      "calib/ece": 0.3257193675889329,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.83399209486166,
      "calib/gap": 0.1694787360105332,
      "calib/mean_conf": 0.8410553359683796,
      "calib/mu_c": 0.9067032258064516,
      "calib/mu_w": 0.7372244897959184,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2770632411067194,
      "calib/std_conf": 0.34101605461781503,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27821873308067135,
      "calib/step_q_c_n": 1847.0,
      "calib/step_q_gap": -0.02184274232916472,
      "calib/step_q_w": 0.3000614754098361,
      "calib/step_q_w_n": 1220.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2850.0,
      "completions/max_terminated_length": 2850.0,
      "completions/mean_length": 575.21484375,
      "completions/mean_terminated_length": 582.0355834960938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.18453333333333333,
      "grad_norm": 0.43757736682891846,
      "learning_rate": 7.777777777777779e-07,
      "loss": 0.2275,
      "num_tokens": 36829505.0,
      "reward": 1.130254864692688,
      "reward_std": 0.20538073778152466,
      "rewards/accuracy_reward_step": 0.60546875,
      "rewards/final_brier_reward_step": 0.666759729385376,
      "rewards/format_reward_step": 0.98828125,
      "step": 173
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8352727349847555,
      "aux_distill/mean_u": 0.27053089226250693,
      "aux_distill/n_active_tok": 186.125,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 12.53515625,
      "calib/ece": 0.2823904382470118,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.7808764940239044,
      "calib/gap": 0.30275407697325507,
      "calib/mean_conf": 0.7901992031872511,
      "calib/mu_c": 0.9168493150684932,
      "calib/mu_w": 0.6140952380952381,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24545816733067716,
      "calib/std_conf": 0.3834145709626288,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.31455794504181595,
      "calib/step_q_c_n": 1674.0,
      "calib/step_q_gap": -0.002034888834405557,
      "calib/step_q_w": 0.3165928338762215,
      "calib/step_q_w_n": 1535.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 2609.0,
      "completions/max_terminated_length": 2609.0,
      "completions/mean_length": 592.8515625,
      "completions/mean_terminated_length": 602.261962890625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.1856,
      "grad_norm": 0.30176031589508057,
      "learning_rate": 7.5e-07,
      "loss": 0.2294,
      "num_tokens": 37085507.0,
      "reward": 1.1251819133758545,
      "reward_std": 0.304109126329422,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6995828151702881,
      "rewards/format_reward_step": 0.98046875,
      "step": 174
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8153031952679157,
      "aux_distill/mean_u": 0.2494548199217007,
      "aux_distill/n_active_tok": 179.625,
      "calib/answer_extract_rate": 0.96484375,
      "calib/avg_num_step_conf": 15.0703125,
      "calib/ece": 0.3794979757085021,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.8097165991902834,
      "calib/gap": 0.1968481510621557,
      "calib/mean_conf": 0.8256518218623481,
      "calib/mu_c": 0.9236774193548386,
      "calib/mu_w": 0.7268292682926829,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.3515627530364373,
      "calib/std_conf": 0.35116507103079797,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2560912547528517,
      "calib/step_q_c_n": 1315.0,
      "calib/step_q_gap": -0.01895396742567762,
      "calib/step_q_w": 0.27504522217852934,
      "calib/step_q_w_n": 2543.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.03125,
      "completions/max_length": 3001.0,
      "completions/max_terminated_length": 3001.0,
      "completions/mean_length": 579.5859375,
      "completions/mean_terminated_length": 598.2822265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 154.0,
      "epoch": 0.18666666666666668,
      "grad_norm": 0.2807024419307709,
      "learning_rate": 7.222222222222222e-07,
      "loss": 0.1579,
      "num_tokens": 37339705.0,
      "reward": 1.0238916873931885,
      "reward_std": 0.3080098628997803,
      "rewards/accuracy_reward_step": 0.484375,
      "rewards/final_brier_reward_step": 0.5985648036003113,
      "rewards/format_reward_step": 0.96484375,
      "step": 175
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8529551476240158,
      "aux_distill/mean_u": 0.3246562601008102,
      "aux_distill/n_active_tok": 154.875,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 12.21484375,
      "calib/ece": 0.2678740157480316,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8188976377952756,
      "calib/gap": 0.3169576923076922,
      "calib/mean_conf": 0.8336220472440946,
      "calib/mu_c": 0.9633999999999999,
      "calib/mu_w": 0.6464423076923077,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.255472440944882,
      "calib/std_conf": 0.34766072993072755,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.277580547112462,
      "calib/step_q_c_n": 1645.0,
      "calib/step_q_gap": 0.022081221876294688,
      "calib/step_q_w": 0.2554993252361673,
      "calib/step_q_w_n": 1482.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2024.0,
      "completions/max_terminated_length": 2024.0,
      "completions/mean_length": 560.88671875,
      "completions/mean_terminated_length": 567.53759765625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 142.0,
      "epoch": 0.18773333333333334,
      "grad_norm": 0.3367729187011719,
      "learning_rate": 6.944444444444446e-07,
      "loss": 0.2451,
      "num_tokens": 37587356.0,
      "reward": 1.1519694328308105,
      "reward_std": 0.2565736174583435,
      "rewards/accuracy_reward_step": 0.5859375,
      "rewards/final_brier_reward_step": 0.7258139848709106,
      "rewards/format_reward_step": 0.9921875,
      "step": 176
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.846255837008357,
      "aux_distill/mean_u": 0.2547409636449559,
      "aux_distill/n_active_tok": 172.75,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 12.03125,
      "calib/ece": 0.26491600000000004,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.832,
      "calib/gap": 0.33505267295597463,
      "calib/mean_conf": 0.839396,
      "calib/mu_c": 0.9814583333333332,
      "calib/mu_w": 0.6464056603773586,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.26415600000000006,
      "calib/std_conf": 0.34353509163402796,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24766943291839558,
      "calib/step_q_c_n": 1446.0,
      "calib/step_q_gap": -0.033187360227259294,
      "calib/step_q_w": 0.28085679314565487,
      "calib/step_q_w_n": 1634.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 2398.0,
      "completions/max_terminated_length": 2398.0,
      "completions/mean_length": 557.25390625,
      "completions/mean_terminated_length": 568.3546142578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 187.0,
      "epoch": 0.1888,
      "grad_norm": 0.3005785048007965,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.203,
      "num_tokens": 37833845.0,
      "reward": 1.1269714832305908,
      "reward_std": 0.2106054425239563,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.7148808240890503,
      "rewards/format_reward_step": 0.9765625,
      "step": 177
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8563280925154686,
      "aux_distill/mean_u": 0.2752489762335201,
      "aux_distill/n_active_tok": 221.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 12.66796875,
      "calib/ece": 0.2809055118110236,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8464566929133859,
      "calib/gap": 0.2337249472573839,
      "calib/mean_conf": 0.857992125984252,
      "calib/mu_c": 0.9463291139240505,
      "calib/mu_w": 0.7126041666666666,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.25842519685039367,
      "calib/std_conf": 0.31873443135826335,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2823424054206663,
      "calib/step_q_c_n": 1771.0,
      "calib/step_q_gap": -0.0351100402315076,
      "calib/step_q_w": 0.3174524456521739,
      "calib/step_q_w_n": 1472.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1868.0,
      "completions/max_terminated_length": 1868.0,
      "completions/mean_length": 548.85546875,
      "completions/mean_terminated_length": 555.3636474609375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 77.0,
      "epoch": 0.18986666666666666,
      "grad_norm": 0.3231167197227478,
      "learning_rate": 6.388888888888889e-07,
      "loss": 0.218,
      "num_tokens": 38080424.0,
      "reward": 1.1606513261795044,
      "reward_std": 0.2605002820491791,
      "rewards/accuracy_reward_step": 0.6171875,
      "rewards/final_brier_reward_step": 0.7119277119636536,
      "rewards/format_reward_step": 0.9921875,
      "step": 178
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8314931150525808,
      "aux_distill/mean_u": 0.2916405868492405,
      "aux_distill/n_active_tok": 216.75,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 12.07421875,
      "calib/ece": 0.19565177865612648,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.8260869565217391,
      "calib/gap": 0.38206014211886286,
      "calib/mean_conf": 0.8252960474308301,
      "calib/mu_c": 0.9476156976744184,
      "calib/mu_w": 0.5655555555555556,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.17055296442687748,
      "calib/std_conf": 0.35870026592531407,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26839024390243904,
      "calib/step_q_c_n": 1845.0,
      "calib/step_q_gap": 0.007780292056532134,
      "calib/step_q_w": 0.2606099518459069,
      "calib/step_q_w_n": 1246.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2932.0,
      "completions/max_terminated_length": 2932.0,
      "completions/mean_length": 575.703125,
      "completions/mean_terminated_length": 580.2362060546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 201.0,
      "epoch": 0.19093333333333334,
      "grad_norm": 0.3155559301376343,
      "learning_rate": 6.111111111111112e-07,
      "loss": 0.2778,
      "num_tokens": 38334068.0,
      "reward": 1.2248156070709229,
      "reward_std": 0.2330620288848877,
      "rewards/accuracy_reward_step": 0.671875,
      "rewards/final_brier_reward_step": 0.7894749641418457,
      "rewards/format_reward_step": 0.98828125,
      "step": 179
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.7971970979124308,
      "aux_distill/mean_u": 0.24041332149219932,
      "aux_distill/n_active_tok": 194.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 13.73828125,
      "calib/ece": 0.25646666666666673,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.796078431372549,
      "calib/gap": 0.28167070707070696,
      "calib/mean_conf": 0.8167019607843138,
      "calib/mu_c": 0.9161151515151514,
      "calib/mu_w": 0.6344444444444445,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.21305490196078436,
      "calib/std_conf": 0.35877278488289205,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2661774744027304,
      "calib/step_q_c_n": 2051.0,
      "calib/step_q_gap": -0.015460179076123681,
      "calib/step_q_w": 0.28163765347885406,
      "calib/step_q_w_n": 1466.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2452.0,
      "completions/max_terminated_length": 2452.0,
      "completions/mean_length": 682.99609375,
      "completions/mean_terminated_length": 688.3740234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 185.0,
      "epoch": 0.192,
      "grad_norm": 0.2555125951766968,
      "learning_rate": 5.833333333333334e-07,
      "loss": 0.2443,
      "num_tokens": 38612771.0,
      "reward": 1.190252661705017,
      "reward_std": 0.223234623670578,
      "rewards/accuracy_reward_step": 0.64453125,
      "rewards/final_brier_reward_step": 0.7398804426193237,
      "rewards/format_reward_step": 0.99609375,
      "step": 180
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8345296196639538,
      "aux_distill/mean_u": 0.26767888136809903,
      "aux_distill/n_active_tok": 186.0,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 11.2890625,
      "calib/ece": 0.2867193675889328,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.8458498023715415,
      "calib/gap": 0.23232245169886734,
      "calib/mean_conf": 0.8459288537549408,
      "calib/mu_c": 0.9331645569620252,
      "calib/mu_w": 0.7008421052631578,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.2540711462450593,
      "calib/std_conf": 0.33912761084905496,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.24617365269461078,
      "calib/step_q_c_n": 1670.0,
      "calib/step_q_gap": -0.021949298125061356,
      "calib/step_q_w": 0.26812295081967213,
      "calib/step_q_w_n": 1220.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1694.0,
      "completions/max_terminated_length": 1694.0,
      "completions/mean_length": 505.734375,
      "completions/mean_terminated_length": 513.761962890625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 198.0,
      "epoch": 0.19306666666666666,
      "grad_norm": 0.38826414942741394,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.2464,
      "num_tokens": 38848503.0,
      "reward": 1.1498777866363525,
      "reward_std": 0.21765771508216858,
      "rewards/accuracy_reward_step": 0.6171875,
      "rewards/final_brier_reward_step": 0.698193371295929,
      "rewards/format_reward_step": 0.984375,
      "step": 181
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8593632094562054,
      "aux_distill/mean_u": 0.26944253590385014,
      "aux_distill/n_active_tok": 210.625,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 11.94921875,
      "calib/ece": 0.294392156862745,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.8745098039215686,
      "calib/gap": 0.16662650602409612,
      "calib/mean_conf": 0.8784705882352943,
      "calib/mu_c": 0.9366265060240961,
      "calib/mu_w": 0.77,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2609411764705881,
      "calib/std_conf": 0.3012034914349353,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25794397759103643,
      "calib/step_q_c_n": 1785.0,
      "calib/step_q_gap": -0.014685535752762657,
      "calib/step_q_w": 0.2726295133437991,
      "calib/step_q_w_n": 1274.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2724.0,
      "completions/max_terminated_length": 2724.0,
      "completions/mean_length": 572.48828125,
      "completions/mean_terminated_length": 576.9960327148438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 225.0,
      "epoch": 0.19413333333333332,
      "grad_norm": 0.28626206517219543,
      "learning_rate": 5.277777777777779e-07,
      "loss": 0.2553,
      "num_tokens": 39101220.0,
      "reward": 1.1739048957824707,
      "reward_std": 0.24584300816059113,
      "rewards/accuracy_reward_step": 0.6484375,
      "rewards/final_brier_reward_step": 0.7032785415649414,
      "rewards/format_reward_step": 0.99609375,
      "step": 182
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8672371339052916,
      "aux_distill/mean_u": 0.30996374739638244,
      "aux_distill/n_active_tok": 220.875,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 13.296875,
      "calib/ece": 0.25592000000000004,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.804,
      "calib/gap": 0.2876544071463466,
      "calib/mean_conf": 0.8148000000000001,
      "calib/mu_c": 0.9172049689440995,
      "calib/mu_w": 0.6295505617977529,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.21336000000000005,
      "calib/std_conf": 0.3630352049044279,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28253883235136584,
      "calib/step_q_c_n": 1867.0,
      "calib/step_q_gap": 0.026566158311027543,
      "calib/step_q_w": 0.2559726740403383,
      "calib/step_q_w_n": 1537.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2908.0,
      "completions/max_terminated_length": 2908.0,
      "completions/mean_length": 625.296875,
      "completions/mean_terminated_length": 632.7114868164062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 187.0,
      "epoch": 0.1952,
      "grad_norm": 0.29089388251304626,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.2574,
      "num_tokens": 39367976.0,
      "reward": 1.164876103401184,
      "reward_std": 0.2800893187522888,
      "rewards/accuracy_reward_step": 0.62890625,
      "rewards/final_brier_reward_step": 0.7242835760116577,
      "rewards/format_reward_step": 0.9765625,
      "step": 183
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8836637884378433,
      "aux_distill/mean_u": 0.3054510764293644,
      "aux_distill/n_active_tok": 187.375,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 11.67578125,
      "calib/ece": 0.21167098039215682,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.8549019607843137,
      "calib/gap": 0.31682057142857123,
      "calib/mean_conf": 0.8529258823529412,
      "calib/mu_c": 0.9523205714285713,
      "calib/mu_w": 0.6355000000000001,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1891611764705882,
      "calib/std_conf": 0.33352521267059637,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22977933801404213,
      "calib/step_q_c_n": 1994.0,
      "calib/step_q_gap": -0.10504478258897296,
      "calib/step_q_w": 0.3348241206030151,
      "calib/step_q_w_n": 995.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1591.0,
      "completions/max_terminated_length": 1591.0,
      "completions/mean_length": 542.71484375,
      "completions/mean_terminated_length": 546.9881591796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 169.0,
      "epoch": 0.19626666666666667,
      "grad_norm": 0.28528183698654175,
      "learning_rate": 4.7222222222222226e-07,
      "loss": 0.2563,
      "num_tokens": 39612191.0,
      "reward": 1.229371428489685,
      "reward_std": 0.21543928980827332,
      "rewards/accuracy_reward_step": 0.68359375,
      "rewards/final_brier_reward_step": 0.7790553569793701,
      "rewards/format_reward_step": 0.99609375,
      "step": 184
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8230397878214717,
      "aux_distill/mean_u": 0.2919940917335714,
      "aux_distill/n_active_tok": 176.375,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 12.4375,
      "calib/ece": 0.25213095238095246,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.8571428571428571,
      "calib/gap": 0.23040740740740717,
      "calib/mean_conf": 0.8614960317460318,
      "calib/mu_c": 0.9355555555555555,
      "calib/mu_w": 0.7051481481481483,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.21752777777777788,
      "calib/std_conf": 0.32164138190799013,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.25592944718117133,
      "calib/step_q_c_n": 1827.0,
      "calib/step_q_gap": -0.06152817993747273,
      "calib/step_q_w": 0.31745762711864406,
      "calib/step_q_w_n": 1357.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1982.0,
      "completions/max_terminated_length": 1982.0,
      "completions/mean_length": 553.1875,
      "completions/mean_terminated_length": 561.96826171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 154.0,
      "epoch": 0.19733333333333333,
      "grad_norm": 0.39035874605178833,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.1929,
      "num_tokens": 39860727.0,
      "reward": 1.189182996749878,
      "reward_std": 0.25200313329696655,
      "rewards/accuracy_reward_step": 0.66796875,
      "rewards/final_brier_reward_step": 0.7299284934997559,
      "rewards/format_reward_step": 0.98046875,
      "step": 185
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8578335829079151,
      "aux_distill/mean_u": 0.2921642115088832,
      "aux_distill/n_active_tok": 171.25,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 12.21484375,
      "calib/ece": 0.19909447449562465,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.7795275590551181,
      "calib/gap": 0.42372533501685417,
      "calib/mean_conf": 0.7847637658342075,
      "calib/mu_c": 0.9315662441077632,
      "calib/mu_w": 0.5078409090909091,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1651574666216089,
      "calib/std_conf": 0.39025296827853495,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2744632469592808,
      "calib/step_q_c_n": 1891.0,
      "calib/step_q_gap": 0.0016477129786983125,
      "calib/step_q_w": 0.2728155339805825,
      "calib/step_q_w_n": 1236.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1581.0,
      "completions/max_terminated_length": 1581.0,
      "completions/mean_length": 566.46875,
      "completions/mean_terminated_length": 570.9291381835938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 157.0,
      "epoch": 0.1984,
      "grad_norm": 0.33431562781333923,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 0.2537,
      "num_tokens": 40110783.0,
      "reward": 1.215174674987793,
      "reward_std": 0.23402857780456543,
      "rewards/accuracy_reward_step": 0.6484375,
      "rewards/final_brier_reward_step": 0.789724588394165,
      "rewards/format_reward_step": 0.9921875,
      "step": 186
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.824182590469718,
      "aux_distill/mean_u": 0.2650358004125993,
      "aux_distill/n_active_tok": 204.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 13.7734375,
      "calib/ece": 0.2811732283464566,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.8070866141732284,
      "calib/gap": 0.2934272930648768,
      "calib/mean_conf": 0.8147952755905513,
      "calib/mu_c": 0.9360939597315435,
      "calib/mu_w": 0.6426666666666667,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.2546771653543306,
      "calib/std_conf": 0.36384323955490216,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.28421627525942106,
      "calib/step_q_c_n": 1831.0,
      "calib/step_q_gap": -0.027907618545888635,
      "calib/step_q_w": 0.3121238938053097,
      "calib/step_q_w_n": 1695.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2513.0,
      "completions/max_terminated_length": 2513.0,
      "completions/mean_length": 604.75390625,
      "completions/mean_terminated_length": 611.9249267578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.19946666666666665,
      "grad_norm": 0.2580687403678894,
      "learning_rate": 3.8888888888888895e-07,
      "loss": 0.2251,
      "num_tokens": 40367144.0,
      "reward": 1.138094186782837,
      "reward_std": 0.2702537178993225,
      "rewards/accuracy_reward_step": 0.58203125,
      "rewards/final_brier_reward_step": 0.7058757543563843,
      "rewards/format_reward_step": 0.98828125,
      "step": 187
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8105640029534698,
      "aux_distill/mean_u": 0.29351264664163546,
      "aux_distill/n_active_tok": 219.875,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 12.7734375,
      "calib/ece": 0.1787007874015748,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.84251968503937,
      "calib/gap": 0.35303642773207977,
      "calib/mean_conf": 0.8453937007874016,
      "calib/mu_c": 0.9412972972972973,
      "calib/mu_w": 0.5882608695652175,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1478740157480315,
      "calib/std_conf": 0.33680007989694366,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2589574366331899,
      "calib/step_q_c_n": 2091.0,
      "calib/step_q_gap": -0.03526537931252682,
      "calib/step_q_w": 0.2942228159457167,
      "calib/step_q_w_n": 1179.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3009.0,
      "completions/max_terminated_length": 3009.0,
      "completions/mean_length": 630.16796875,
      "completions/mean_terminated_length": 635.1299438476562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 175.0,
      "epoch": 0.20053333333333334,
      "grad_norm": 0.253006249666214,
      "learning_rate": 3.611111111111111e-07,
      "loss": 0.2441,
      "num_tokens": 40632539.0,
      "reward": 1.261594295501709,
      "reward_std": 0.16998478770256042,
      "rewards/accuracy_reward_step": 0.72265625,
      "rewards/final_brier_reward_step": 0.8083449006080627,
      "rewards/format_reward_step": 0.9921875,
      "step": 188
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8325030775740743,
      "aux_distill/mean_u": 0.2645952570453017,
      "aux_distill/n_active_tok": 164.625,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.9921875,
      "calib/ece": 0.23508235294117644,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.7686274509803922,
      "calib/gap": 0.3517131313131311,
      "calib/mean_conf": 0.7800235294117649,
      "calib/mu_c": 0.9041575757575756,
      "calib/mu_w": 0.5524444444444445,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.18402352941176467,
      "calib/std_conf": 0.392159210659638,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2832378826530612,
      "calib/step_q_c_n": 1568.0,
      "calib/step_q_gap": -0.033977205629442786,
      "calib/step_q_w": 0.317215088282504,
      "calib/step_q_w_n": 1246.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2109.0,
      "completions/max_terminated_length": 2109.0,
      "completions/mean_length": 517.3984375,
      "completions/mean_terminated_length": 521.472412109375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 139.0,
      "epoch": 0.2016,
      "grad_norm": 0.33088845014572144,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.2758,
      "num_tokens": 40872761.0,
      "reward": 1.1992274522781372,
      "reward_std": 0.21433404088020325,
      "rewards/accuracy_reward_step": 0.64453125,
      "rewards/final_brier_reward_step": 0.757830023765564,
      "rewards/format_reward_step": 0.99609375,
      "step": 189
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8428498767316341,
      "aux_distill/mean_u": 0.3225249213968002,
      "aux_distill/n_active_tok": 222.875,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 13.54296875,
      "calib/ece": 0.22940711462450605,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.7905138339920948,
      "calib/gap": 0.39989902912621333,
      "calib/mean_conf": 0.8003952569169961,
      "calib/mu_c": 0.9631999999999997,
      "calib/mu_w": 0.5633009708737864,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.21845849802371556,
      "calib/std_conf": 0.37837472524572235,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2910631040363843,
      "calib/step_q_c_n": 1759.0,
      "calib/step_q_gap": -0.0024757718418358388,
      "calib/step_q_w": 0.29353887587822014,
      "calib/step_q_w_n": 1708.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 1863.0,
      "completions/max_terminated_length": 1863.0,
      "completions/mean_length": 627.0234375,
      "completions/mean_terminated_length": 636.9762573242188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 179.0,
      "epoch": 0.20266666666666666,
      "grad_norm": 0.3278692960739136,
      "learning_rate": 3.055555555555556e-07,
      "loss": 0.2014,
      "num_tokens": 41138887.0,
      "reward": 1.165348768234253,
      "reward_std": 0.2440982311964035,
      "rewards/accuracy_reward_step": 0.5859375,
      "rewards/final_brier_reward_step": 0.7564789056777954,
      "rewards/format_reward_step": 0.98828125,
      "step": 190
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8193467408418655,
      "aux_distill/mean_u": 0.20845514856220498,
      "aux_distill/n_active_tok": 222.625,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 12.49609375,
      "calib/ece": 0.3277254901960783,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.7725490196078432,
      "calib/gap": 0.24052509990009974,
      "calib/mean_conf": 0.785686274509804,
      "calib/mu_c": 0.8913286713286712,
      "calib/mu_w": 0.6508035714285715,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2763137254901959,
      "calib/std_conf": 0.38520814287995353,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28625617206982545,
      "calib/step_q_c_n": 1604.0,
      "calib/step_q_gap": -0.016665458024218427,
      "calib/step_q_w": 0.3029216300940439,
      "calib/step_q_w_n": 1595.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2570.0,
      "completions/max_terminated_length": 2570.0,
      "completions/mean_length": 581.796875,
      "completions/mean_terminated_length": 584.0784912109375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 206.0,
      "epoch": 0.20373333333333332,
      "grad_norm": 0.2885873019695282,
      "learning_rate": 2.7777777777777776e-07,
      "loss": 0.2257,
      "num_tokens": 41391995.0,
      "reward": 1.112635612487793,
      "reward_std": 0.20652508735656738,
      "rewards/accuracy_reward_step": 0.55859375,
      "rewards/final_brier_reward_step": 0.670583963394165,
      "rewards/format_reward_step": 0.99609375,
      "step": 191
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8118951702490449,
      "aux_distill/mean_u": 0.24394687888891278,
      "aux_distill/n_active_tok": 194.5,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 12.828125,
      "calib/ece": 0.15258933522084617,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.782608695652174,
      "calib/gap": 0.4859580254252286,
      "calib/mean_conf": 0.796888925648719,
      "calib/mu_c": 0.9447891942563974,
      "calib/mu_w": 0.4588311688311688,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12691304347826085,
      "calib/std_conf": 0.37765356366640646,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2691260053619303,
      "calib/step_q_c_n": 1865.0,
      "calib/step_q_gap": -0.053601267365342375,
      "calib/step_q_w": 0.3227272727272727,
      "calib/step_q_w_n": 1419.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 2180.0,
      "completions/max_terminated_length": 2180.0,
      "completions/mean_length": 609.125,
      "completions/mean_terminated_length": 618.793701171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 165.0,
      "epoch": 0.2048,
      "grad_norm": 0.44815996289253235,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.2364,
      "num_tokens": 41652907.0,
      "reward": 1.2535533905029297,
      "reward_std": 0.2129342257976532,
      "rewards/accuracy_reward_step": 0.6875,
      "rewards/final_brier_reward_step": 0.831325352191925,
      "rewards/format_reward_step": 0.98828125,
      "step": 192
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8178987912833691,
      "aux_distill/mean_u": 0.23051178048996848,
      "aux_distill/n_active_tok": 182.125,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 13.16796875,
      "calib/ece": 0.23338645418326698,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.8446215139442231,
      "calib/gap": 0.3356832720838435,
      "calib/mean_conf": 0.8408764940239045,
      "calib/mu_c": 0.965253164556962,
      "calib/mu_w": 0.6295698924731185,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.222390438247012,
      "calib/std_conf": 0.3451739394729761,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2731526548672566,
      "calib/step_q_c_n": 1808.0,
      "calib/step_q_gap": 0.010663851284403114,
      "calib/step_q_w": 0.2624888035828535,
      "calib/step_q_w_n": 1563.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.015625,
      "completions/max_length": 3006.0,
      "completions/max_terminated_length": 3006.0,
      "completions/mean_length": 580.14453125,
      "completions/mean_terminated_length": 589.3532104492188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 183.0,
      "epoch": 0.20586666666666667,
      "grad_norm": 0.25259193778038025,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.2709,
      "num_tokens": 41907136.0,
      "reward": 1.1711702346801758,
      "reward_std": 0.28969404101371765,
      "rewards/accuracy_reward_step": 0.6171875,
      "rewards/final_brier_reward_step": 0.7446843385696411,
      "rewards/format_reward_step": 0.98046875,
      "step": 193
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8706032708287239,
      "aux_distill/mean_u": 0.2951731910868477,
      "aux_distill/n_active_tok": 167.5,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 11.984375,
      "calib/ece": 0.22929133858267708,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8385826771653543,
      "calib/gap": 0.32819611848825314,
      "calib/mean_conf": 0.836456692913386,
      "calib/mu_c": 0.9514545454545453,
      "calib/mu_w": 0.6232584269662922,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2080708661417322,
      "calib/std_conf": 0.3495160378622438,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27585446527012125,
      "calib/step_q_c_n": 1814.0,
      "calib/step_q_gap": -0.013284290710739999,
      "calib/step_q_w": 0.28913875598086125,
      "calib/step_q_w_n": 1254.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2842.0,
      "completions/max_terminated_length": 2842.0,
      "completions/mean_length": 527.45703125,
      "completions/mean_terminated_length": 533.7114868164062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 209.0,
      "epoch": 0.20693333333333333,
      "grad_norm": 0.358541339635849,
      "learning_rate": 1.9444444444444447e-07,
      "loss": 0.2455,
      "num_tokens": 42148109.0,
      "reward": 1.197729229927063,
      "reward_std": 0.22360165417194366,
      "rewards/accuracy_reward_step": 0.64453125,
      "rewards/final_brier_reward_step": 0.7587398290634155,
      "rewards/format_reward_step": 0.9921875,
      "step": 194
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8249854417517781,
      "aux_distill/mean_u": 0.2865456123927324,
      "aux_distill/n_active_tok": 215.625,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 13.7578125,
      "calib/ece": 0.31261043666280663,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.7349397590361446,
      "calib/gap": 0.26040804686107377,
      "calib/mean_conf": 0.7554216816427264,
      "calib/mu_c": 0.8673239347115411,
      "calib/mu_w": 0.6069158878504674,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24887549690377048,
      "calib/std_conf": 0.40102669492810117,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25076524741081707,
      "calib/step_q_c_n": 1738.0,
      "calib/step_q_gap": -0.04811367635599911,
      "calib/step_q_w": 0.2988789237668162,
      "calib/step_q_w_n": 1784.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.01953125,
      "completions/max_length": 2582.0,
      "completions/max_terminated_length": 2582.0,
      "completions/mean_length": 605.25390625,
      "completions/mean_terminated_length": 617.310791015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 201.0,
      "epoch": 0.208,
      "grad_norm": 0.3838996887207031,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 0.2191,
      "num_tokens": 42409038.0,
      "reward": 1.0980085134506226,
      "reward_std": 0.29674944281578064,
      "rewards/accuracy_reward_step": 0.5546875,
      "rewards/final_brier_reward_step": 0.6686733961105347,
      "rewards/format_reward_step": 0.97265625,
      "step": 195
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.87274937517941,
      "aux_distill/mean_u": 0.2900594268034635,
      "aux_distill/n_active_tok": 171.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 10.60546875,
      "calib/ece": 0.24312941166743263,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.9176470588235294,
      "calib/gap": 0.19994928557254454,
      "calib/mean_conf": 0.9103450979419426,
      "calib/mu_c": 0.9730742855725446,
      "calib/mu_w": 0.7731250000000001,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.23359999990272676,
      "calib/std_conf": 0.26470785692410465,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27979405034324945,
      "calib/step_q_c_n": 1748.0,
      "calib/step_q_gap": -0.004900882438549969,
      "calib/step_q_w": 0.2846949327817994,
      "calib/step_q_w_n": 967.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1804.0,
      "completions/max_terminated_length": 1804.0,
      "completions/mean_length": 482.55859375,
      "completions/mean_terminated_length": 486.3582763671875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 172.0,
      "epoch": 0.20906666666666668,
      "grad_norm": 0.2778487205505371,
      "learning_rate": 1.3888888888888888e-07,
      "loss": 0.2494,
      "num_tokens": 42635117.0,
      "reward": 1.213637351989746,
      "reward_std": 0.17422911524772644,
      "rewards/accuracy_reward_step": 0.68359375,
      "rewards/final_brier_reward_step": 0.7475874423980713,
      "rewards/format_reward_step": 0.99609375,
      "step": 196
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8694142308086157,
      "aux_distill/mean_u": 0.3005259901536683,
      "aux_distill/n_active_tok": 215.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 13.5234375,
      "calib/ece": 0.3585039370078741,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8149606299212598,
      "calib/gap": 0.13569102564102553,
      "calib/mean_conf": 0.8215748031496063,
      "calib/mu_c": 0.8771333333333333,
      "calib/mu_w": 0.7414423076923078,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2947637795275591,
      "calib/std_conf": 0.36048507389207807,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.255830258302583,
      "calib/step_q_c_n": 1897.0,
      "calib/step_q_gap": -0.03370967779965345,
      "calib/step_q_w": 0.28953993610223644,
      "calib/step_q_w_n": 1565.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2515.0,
      "completions/max_terminated_length": 2515.0,
      "completions/mean_length": 597.7734375,
      "completions/mean_terminated_length": 604.8616943359375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 152.0,
      "epoch": 0.21013333333333334,
      "grad_norm": 0.3032035529613495,
      "learning_rate": 1.1111111111111112e-07,
      "loss": 0.2319,
      "num_tokens": 42893203.0,
      "reward": 1.1068097352981567,
      "reward_std": 0.26105594635009766,
      "rewards/accuracy_reward_step": 0.5859375,
      "rewards/final_brier_reward_step": 0.6354945302009583,
      "rewards/format_reward_step": 0.9921875,
      "step": 197
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8194463439285755,
      "aux_distill/mean_u": 0.2501906911000831,
      "aux_distill/n_active_tok": 180.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 11.80859375,
      "calib/ece": 0.20453725490196056,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.803921568627451,
      "calib/gap": 0.3732018186944873,
      "calib/mean_conf": 0.8054235294117648,
      "calib/mu_c": 0.9254335260115605,
      "calib/mu_w": 0.5522317073170732,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.16576470588235273,
      "calib/std_conf": 0.37370990247927954,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2524092247301276,
      "calib/step_q_c_n": 2038.0,
      "calib/step_q_gap": 0.04279501153215809,
      "calib/step_q_w": 0.20961421319796952,
      "calib/step_q_w_n": 985.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2297.0,
      "completions/max_terminated_length": 2297.0,
      "completions/mean_length": 542.8359375,
      "completions/mean_terminated_length": 547.1102294921875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.2112,
      "grad_norm": 0.3990510404109955,
      "learning_rate": 8.333333333333334e-08,
      "loss": 0.2275,
      "num_tokens": 43137553.0,
      "reward": 1.2288410663604736,
      "reward_std": 0.2174731194972992,
      "rewards/accuracy_reward_step": 0.67578125,
      "rewards/final_brier_reward_step": 0.7858071327209473,
      "rewards/format_reward_step": 0.99609375,
      "step": 198
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8574781194329262,
      "aux_distill/mean_u": 0.2731382857877819,
      "aux_distill/n_active_tok": 191.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 12.76171875,
      "calib/ece": 0.2103906250000001,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.8515625,
      "calib/gap": 0.23776595744680862,
      "calib/mean_conf": 0.8671093750000001,
      "calib/mu_c": 0.9302659574468086,
      "calib/mu_w": 0.6925,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1715625000000001,
      "calib/std_conf": 0.30916382273336795,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26340939008336994,
      "calib/step_q_c_n": 2279.0,
      "calib/step_q_gap": -0.03318980020003082,
      "calib/step_q_w": 0.29659919028340076,
      "calib/step_q_w_n": 988.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2269.0,
      "completions/max_terminated_length": 2269.0,
      "completions/mean_length": 618.19921875,
      "completions/mean_terminated_length": 623.0669555664062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 108.0,
      "epoch": 0.21226666666666666,
      "grad_norm": 0.28748640418052673,
      "learning_rate": 5.555555555555556e-08,
      "loss": 0.2313,
      "num_tokens": 43400012.0,
      "reward": 1.2594335079193115,
      "reward_std": 0.19422554969787598,
      "rewards/accuracy_reward_step": 0.734375,
      "rewards/final_brier_reward_step": 0.7844921946525574,
      "rewards/format_reward_step": 1.0,
      "step": 199
    },
    {
      "aux_distill/lambda": 0.30000000000000004,
      "aux_distill/loss": 0.8403654135763645,
      "aux_distill/mean_u": 0.25869545011362055,
      "aux_distill/n_active_tok": 207.5,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 13.1484375,
      "calib/ece": 0.23880478087649398,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.7808764940239044,
      "calib/gap": 0.3254771813011512,
      "calib/mean_conf": 0.7949003984063744,
      "calib/mu_c": 0.9103086419753086,
      "calib/mu_w": 0.5848314606741574,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.19414342629482068,
      "calib/std_conf": 0.37575566630034124,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3023505275498241,
      "calib/step_q_c_n": 1706.0,
      "calib/step_q_gap": 0.0009710094775349232,
      "calib/step_q_w": 0.3013795180722892,
      "calib/step_q_w_n": 1660.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3017.0,
      "completions/max_terminated_length": 3017.0,
      "completions/mean_length": 630.12109375,
      "completions/mean_terminated_length": 637.5928955078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 119.0,
      "epoch": 0.21333333333333335,
      "grad_norm": 0.2310328185558319,
      "learning_rate": 2.777777777777778e-08,
      "loss": 0.236,
      "num_tokens": 43669371.0,
      "reward": 1.1727749109268188,
      "reward_std": 0.21215327084064484,
      "rewards/accuracy_reward_step": 0.6328125,
      "rewards/final_brier_reward_step": 0.740081250667572,
      "rewards/format_reward_step": 0.97265625,
      "step": 200
    },
    {
      "epoch": 0.21333333333333335,
      "step": 200,
      "total_flos": 0.0,
      "train_loss": 0.2056847586715594,
      "train_runtime": 17418.2127,
      "train_samples_per_second": 2.939,
      "train_steps_per_second": 0.011
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 43669371,
  "num_train_epochs": 1,
  "save_steps": 20,
  "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": 4,
  "trial_name": null,
  "trial_params": null
}