{
  "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": 0.09999999999999999,
      "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": 1.1307653188705444,
      "learning_rate": 0.0,
      "loss": 0.0331,
      "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": 0.09999999999999999,
      "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": 0.7146468162536621,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.0274,
      "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": 0.10000000000000002,
      "aux_distill/loss": 1.3222839037577312,
      "aux_distill/mean_u": 0.11826397114259928,
      "aux_distill/n_active_tok": 32.0,
      "calib/answer_extract_rate": 0.02734375,
      "calib/avg_num_step_conf": 0.125,
      "calib/ece": 0.7699999999999998,
      "calib/final_conf_rate": 0.0234375,
      "calib/format_rate": 0.01171875,
      "calib/frac_conf_gt_0.9": 0.6666666666666666,
      "calib/mean_conf": 0.77,
      "calib/mu_c": NaN,
      "calib/mu_w": 0.77,
      "calib/nonempty_final_conf_rate": 0.0234375,
      "calib/nonempty_reasoning_rate": 0.03125,
      "calib/nonempty_step_conf_rate": 0.015625,
      "calib/pce": 0.7699999999999998,
      "calib/std_conf": 0.3528928071430945,
      "calib/step_conf_rate": 0.015625,
      "calib/step_q_w": 0.863125,
      "calib/step_q_w_n": 32.0,
      "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": 0.1015625,
      "completions/max_length": 3032.0,
      "completions/max_terminated_length": 3032.0,
      "completions/mean_length": 699.63671875,
      "completions/mean_terminated_length": 778.72607421875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0032,
      "grad_norm": 0.3191400170326233,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0012,
      "num_tokens": 844732.0,
      "reward": 0.006911718752235174,
      "reward_std": 0.0195492934435606,
      "rewards/accuracy_reward_step": 0.0,
      "rewards/final_brier_reward_step": 0.0021046875044703484,
      "rewards/format_reward_step": 0.01171875,
      "step": 3
    },
    {
      "aux_distill/lambda": 0.1,
      "aux_distill/loss": 2.1405008137226105,
      "aux_distill/mean_u": 0.22794245170372168,
      "aux_distill/n_active_tok": 12.0,
      "calib/answer_extract_rate": 0.046875,
      "calib/avg_num_step_conf": 0.1953125,
      "calib/ece": 0.361,
      "calib/final_conf_rate": 0.0390625,
      "calib/format_rate": 0.0390625,
      "calib/frac_conf_gt_0.9": 0.9,
      "calib/gap": 0.026666666666666727,
      "calib/mean_conf": 0.961,
      "calib/mu_c": 0.9716666666666667,
      "calib/mu_w": 0.945,
      "calib/nonempty_final_conf_rate": 0.0390625,
      "calib/nonempty_reasoning_rate": 0.046875,
      "calib/nonempty_step_conf_rate": 0.0390625,
      "calib/pce": 0.361,
      "calib/std_conf": 0.02913760456866692,
      "calib/step_conf_rate": 0.0390625,
      "calib/step_q_c": 0.9222222222222222,
      "calib/step_q_c_n": 27.0,
      "calib/step_q_gap": 0.027004830917874534,
      "calib/step_q_w": 0.8952173913043476,
      "calib/step_q_w_n": 23.0,
      "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": 2925.0,
      "completions/max_terminated_length": 2925.0,
      "completions/mean_length": 740.88671875,
      "completions/mean_terminated_length": 807.0935668945312,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.004266666666666667,
      "grad_norm": 0.6574472784996033,
      "learning_rate": 7.5e-07,
      "loss": 0.0335,
      "num_tokens": 1140567.0,
      "reward": 0.043781835585832596,
      "reward_std": 0.10839106887578964,
      "rewards/accuracy_reward_step": 0.0234375,
      "rewards/final_brier_reward_step": 0.02506367117166519,
      "rewards/format_reward_step": 0.0390625,
      "step": 4
    },
    {
      "aux_distill/lambda": 0.1,
      "aux_distill/loss": 1.5584832191467286,
      "aux_distill/mean_u": 0.555644570917228,
      "aux_distill/n_active_tok": 23.2,
      "calib/answer_extract_rate": 0.05078125,
      "calib/avg_num_step_conf": 0.21875,
      "calib/ece": 0.22153846153846146,
      "calib/final_conf_rate": 0.05078125,
      "calib/format_rate": 0.0390625,
      "calib/frac_conf_gt_0.9": 0.6923076923076923,
      "calib/gap": 0.1283333333333332,
      "calib/mean_conf": 0.9138461538461538,
      "calib/mu_c": 0.9533333333333331,
      "calib/mu_w": 0.825,
      "calib/nonempty_final_conf_rate": 0.05078125,
      "calib/nonempty_reasoning_rate": 0.0546875,
      "calib/nonempty_step_conf_rate": 0.04296875,
      "calib/pce": 0.22153846153846146,
      "calib/std_conf": 0.12344656031837367,
      "calib/step_conf_rate": 0.04296875,
      "calib/step_q_c": 0.8882857142857142,
      "calib/step_q_c_n": 35.0,
      "calib/step_q_gap": 0.04019047619047622,
      "calib/step_q_w": 0.848095238095238,
      "calib/step_q_w_n": 21.0,
      "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": 2855.0,
      "completions/max_terminated_length": 2855.0,
      "completions/mean_length": 686.07421875,
      "completions/mean_terminated_length": 734.8744506835938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.005333333333333333,
      "grad_norm": 0.7364365458488464,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0574,
      "num_tokens": 1422890.0,
      "reward": 0.05148027092218399,
      "reward_std": 0.1402149796485901,
      "rewards/accuracy_reward_step": 0.03515625,
      "rewards/final_brier_reward_step": 0.028741799294948578,
      "rewards/format_reward_step": 0.0390625,
      "step": 5
    },
    {
      "aux_distill/lambda": 0.1,
      "aux_distill/loss": 1.6745985746383667,
      "aux_distill/mean_u": 0.28617723428204433,
      "aux_distill/n_active_tok": 26.0,
      "calib/answer_extract_rate": 0.046875,
      "calib/avg_num_step_conf": 0.15234375,
      "calib/ece": 0.6709999999999998,
      "calib/final_conf_rate": 0.0390625,
      "calib/format_rate": 0.01953125,
      "calib/frac_conf_gt_0.9": 1.0,
      "calib/gap": -0.010952380952381047,
      "calib/mean_conf": 0.9709999999999999,
      "calib/mu_c": 0.9633333333333333,
      "calib/mu_w": 0.9742857142857143,
      "calib/nonempty_final_conf_rate": 0.0390625,
      "calib/nonempty_reasoning_rate": 0.0546875,
      "calib/nonempty_step_conf_rate": 0.02734375,
      "calib/pce": 0.6709999999999998,
      "calib/std_conf": 0.018681541692269422,
      "calib/step_conf_rate": 0.02734375,
      "calib/step_q_c": 0.8683333333333332,
      "calib/step_q_c_n": 12.0,
      "calib/step_q_gap": 0.008703703703703547,
      "calib/step_q_w": 0.8596296296296296,
      "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": 0.06640625,
      "completions/max_length": 2973.0,
      "completions/max_terminated_length": 2973.0,
      "completions/mean_length": 638.53515625,
      "completions/mean_terminated_length": 683.9539794921875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0064,
      "grad_norm": 0.46913886070251465,
      "learning_rate": 1.25e-06,
      "loss": 0.007,
      "num_tokens": 1692307.0,
      "reward": 0.02002226561307907,
      "reward_std": 0.05123838037252426,
      "rewards/accuracy_reward_step": 0.01171875,
      "rewards/final_brier_reward_step": 0.008794531226158142,
      "rewards/format_reward_step": 0.01953125,
      "step": 6
    },
    {
      "aux_distill/lambda": 0.10000000000000002,
      "aux_distill/loss": 1.4568543831507366,
      "aux_distill/mean_u": 0.3033000773576174,
      "aux_distill/n_active_tok": 17.333333333333332,
      "calib/answer_extract_rate": 0.0625,
      "calib/avg_num_step_conf": 0.30859375,
      "calib/ece": 0.6199999999999999,
      "calib/final_conf_rate": 0.05859375,
      "calib/format_rate": 0.0390625,
      "calib/frac_conf_gt_0.9": 0.9333333333333333,
      "calib/gap": 0.006999999999999895,
      "calib/mean_conf": 0.9533333333333333,
      "calib/mu_c": 0.9579999999999999,
      "calib/mu_w": 0.951,
      "calib/nonempty_final_conf_rate": 0.05859375,
      "calib/nonempty_reasoning_rate": 0.07421875,
      "calib/nonempty_step_conf_rate": 0.0546875,
      "calib/pce": 0.6199999999999999,
      "calib/std_conf": 0.035150470203904174,
      "calib/step_conf_rate": 0.0546875,
      "calib/step_q_c": 0.89,
      "calib/step_q_c_n": 25.0,
      "calib/step_q_gap": 0.026296296296296262,
      "calib/step_q_w": 0.8637037037037038,
      "calib/step_q_w_n": 54.0,
      "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": 2881.0,
      "completions/max_terminated_length": 2881.0,
      "completions/mean_length": 764.57421875,
      "completions/mean_terminated_length": 808.8057250976562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.007466666666666667,
      "grad_norm": 0.7070335149765015,
      "learning_rate": 1.5e-06,
      "loss": 0.0243,
      "num_tokens": 1995462.0,
      "reward": 0.0416390635073185,
      "reward_std": 0.09732498973608017,
      "rewards/accuracy_reward_step": 0.0234375,
      "rewards/final_brier_reward_step": 0.020778125151991844,
      "rewards/format_reward_step": 0.0390625,
      "step": 7
    },
    {
      "aux_distill/lambda": 0.1,
      "aux_distill/loss": 1.481102204322815,
      "aux_distill/mean_u": 0.2925329127684373,
      "aux_distill/n_active_tok": 59.2,
      "calib/answer_extract_rate": 0.0546875,
      "calib/avg_num_step_conf": 0.3984375,
      "calib/ece": 0.537857142857143,
      "calib/final_conf_rate": 0.0546875,
      "calib/format_rate": 0.03125,
      "calib/frac_conf_gt_0.9": 0.9285714285714286,
      "calib/gap": -0.028749999999999942,
      "calib/mean_conf": 0.9664285714285716,
      "calib/mu_c": 0.9500000000000001,
      "calib/mu_w": 0.97875,
      "calib/nonempty_final_conf_rate": 0.0546875,
      "calib/nonempty_reasoning_rate": 0.05859375,
      "calib/nonempty_step_conf_rate": 0.04296875,
      "calib/pce": 0.537857142857143,
      "calib/std_conf": 0.024671308627386493,
      "calib/step_conf_rate": 0.04296875,
      "calib/step_q_c": 0.8899999999999999,
      "calib/step_q_c_n": 23.0,
      "calib/step_q_gap": -0.01430379746835464,
      "calib/step_q_w": 0.9043037974683545,
      "calib/step_q_w_n": 79.0,
      "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": 3053.0,
      "completions/max_terminated_length": 3053.0,
      "completions/mean_length": 646.93359375,
      "completions/mean_terminated_length": 698.7974243164062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.008533333333333334,
      "grad_norm": 0.6346028447151184,
      "learning_rate": 1.75e-06,
      "loss": 0.0211,
      "num_tokens": 2267589.0,
      "reward": 0.03550625219941139,
      "reward_std": 0.09643617272377014,
      "rewards/accuracy_reward_step": 0.0234375,
      "rewards/final_brier_reward_step": 0.016325000673532486,
      "rewards/format_reward_step": 0.03125,
      "step": 8
    },
    {
      "aux_distill/lambda": 0.1,
      "aux_distill/loss": 1.111382430791855,
      "aux_distill/mean_u": 0.26760089571160023,
      "aux_distill/n_active_tok": 18.4,
      "calib/answer_extract_rate": 0.0625,
      "calib/avg_num_step_conf": 0.33984375,
      "calib/ece": 0.4546111111111111,
      "calib/final_conf_rate": 0.0703125,
      "calib/format_rate": 0.05078125,
      "calib/frac_conf_gt_0.9": 0.7777777777777778,
      "calib/gap": 0.07819999999999983,
      "calib/mean_conf": 0.8990555555555555,
      "calib/mu_c": 0.9424999999999999,
      "calib/mu_w": 0.8643000000000001,
      "calib/nonempty_final_conf_rate": 0.0703125,
      "calib/nonempty_reasoning_rate": 0.0703125,
      "calib/nonempty_step_conf_rate": 0.05859375,
      "calib/pce": 0.4546111111111111,
      "calib/std_conf": 0.2067651572362074,
      "calib/step_conf_rate": 0.05859375,
      "calib/step_q_c": 0.8628571428571428,
      "calib/step_q_c_n": 35.0,
      "calib/step_q_gap": 0.020260989010988828,
      "calib/step_q_w": 0.8425961538461539,
      "calib/step_q_w_n": 52.0,
      "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": 2991.0,
      "completions/max_terminated_length": 2991.0,
      "completions/mean_length": 629.04296875,
      "completions/mean_terminated_length": 668.195068359375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.0096,
      "grad_norm": 0.7474179863929749,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0211,
      "num_tokens": 2536160.0,
      "reward": 0.05555380508303642,
      "reward_std": 0.10311586409807205,
      "rewards/accuracy_reward_step": 0.03125,
      "rewards/final_brier_reward_step": 0.02907637134194374,
      "rewards/format_reward_step": 0.05078125,
      "step": 9
    },
    {
      "aux_distill/lambda": 0.09999999999999999,
      "aux_distill/loss": 1.1272154109818595,
      "aux_distill/mean_u": 0.17926448979161314,
      "aux_distill/n_active_tok": 26.285714285714285,
      "calib/answer_extract_rate": 0.06640625,
      "calib/avg_num_step_conf": 0.28515625,
      "calib/ece": 0.7218749999999999,
      "calib/final_conf_rate": 0.0625,
      "calib/format_rate": 0.0546875,
      "calib/frac_conf_gt_0.9": 0.75,
      "calib/gap": 0.0541025641025642,
      "calib/mean_conf": 0.909375,
      "calib/mu_c": 0.9533333333333333,
      "calib/mu_w": 0.8992307692307691,
      "calib/nonempty_final_conf_rate": 0.0625,
      "calib/nonempty_reasoning_rate": 0.0703125,
      "calib/nonempty_step_conf_rate": 0.05859375,
      "calib/pce": 0.7218749999999999,
      "calib/std_conf": 0.15469602249249975,
      "calib/step_conf_rate": 0.05859375,
      "calib/step_q_c": 0.6654545454545455,
      "calib/step_q_c_n": 11.0,
      "calib/step_q_gap": -0.14260997067448677,
      "calib/step_q_w": 0.8080645161290323,
      "calib/step_q_w_n": 62.0,
      "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": 3028.0,
      "completions/max_terminated_length": 3028.0,
      "completions/mean_length": 657.84765625,
      "completions/mean_terminated_length": 690.2008056640625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.010666666666666666,
      "grad_norm": 0.6247566938400269,
      "learning_rate": 2.25e-06,
      "loss": 0.0382,
      "num_tokens": 2811369.0,
      "reward": 0.04085097461938858,
      "reward_std": 0.08594364672899246,
      "rewards/accuracy_reward_step": 0.01171875,
      "rewards/final_brier_reward_step": 0.015295703895390034,
      "rewards/format_reward_step": 0.0546875,
      "step": 10
    },
    {
      "aux_distill/lambda": 0.09999999999999999,
      "aux_distill/loss": 1.1807828417846136,
      "aux_distill/mean_u": 0.3193847890032968,
      "aux_distill/n_active_tok": 25.714285714285715,
      "calib/answer_extract_rate": 0.08984375,
      "calib/avg_num_step_conf": 0.43359375,
      "calib/ece": 0.5734782608695652,
      "calib/final_conf_rate": 0.08984375,
      "calib/format_rate": 0.06640625,
      "calib/frac_conf_gt_0.9": 0.6956521739130435,
      "calib/gap": 0.012333333333333307,
      "calib/mean_conf": 0.8569565217391302,
      "calib/mu_c": 0.865,
      "calib/mu_w": 0.8526666666666667,
      "calib/nonempty_final_conf_rate": 0.08984375,
      "calib/nonempty_reasoning_rate": 0.09765625,
      "calib/nonempty_step_conf_rate": 0.078125,
      "calib/pce": 0.5413043478260869,
      "calib/std_conf": 0.23582297676543393,
      "calib/step_conf_rate": 0.078125,
      "calib/step_q_c": 0.8226470588235294,
      "calib/step_q_c_n": 34.0,
      "calib/step_q_gap": -0.006573720397249794,
      "calib/step_q_w": 0.8292207792207792,
      "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": 0.07421875,
      "completions/max_length": 3061.0,
      "completions/max_terminated_length": 3061.0,
      "completions/mean_length": 704.5546875,
      "completions/mean_terminated_length": 761.0379638671875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.011733333333333333,
      "grad_norm": 0.8381440043449402,
      "learning_rate": 2.5e-06,
      "loss": 0.0418,
      "num_tokens": 3096215.0,
      "reward": 0.06206425651907921,
      "reward_std": 0.14835157990455627,
      "rewards/accuracy_reward_step": 0.03125,
      "rewards/final_brier_reward_step": 0.026472264900803566,
      "rewards/format_reward_step": 0.06640625,
      "step": 11
    },
    {
      "aux_distill/lambda": 0.10000000000000003,
      "aux_distill/loss": 1.1680403964387045,
      "aux_distill/mean_u": 0.39082487489789425,
      "aux_distill/n_active_tok": 38.22222222222222,
      "calib/answer_extract_rate": 0.109375,
      "calib/avg_num_step_conf": 0.9609375,
      "calib/ece": 0.45625000000000004,
      "calib/final_conf_rate": 0.09375,
      "calib/format_rate": 0.0703125,
      "calib/frac_conf_gt_0.9": 0.6666666666666666,
      "calib/gap": 0.07741258741258716,
      "calib/mean_conf": 0.8562500000000001,
      "calib/mu_c": 0.8981818181818181,
      "calib/mu_w": 0.8207692307692309,
      "calib/nonempty_final_conf_rate": 0.09375,
      "calib/nonempty_reasoning_rate": 0.16796875,
      "calib/nonempty_step_conf_rate": 0.1328125,
      "calib/pce": 0.42708333333333337,
      "calib/std_conf": 0.22052612279123154,
      "calib/step_conf_rate": 0.1328125,
      "calib/step_q_c": 0.8590789473684209,
      "calib/step_q_c_n": 76.0,
      "calib/step_q_gap": 0.005726006191950406,
      "calib/step_q_w": 0.8533529411764705,
      "calib/step_q_w_n": 170.0,
      "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": 3066.0,
      "completions/max_terminated_length": 3066.0,
      "completions/mean_length": 582.1171875,
      "completions/mean_terminated_length": 631.4491577148438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0128,
      "grad_norm": 1.0448517799377441,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.0591,
      "num_tokens": 3349413.0,
      "reward": 0.07992050796747208,
      "reward_std": 0.18436968326568604,
      "rewards/accuracy_reward_step": 0.04296875,
      "rewards/final_brier_reward_step": 0.04655976593494415,
      "rewards/format_reward_step": 0.0703125,
      "step": 12
    },
    {
      "aux_distill/lambda": 0.10000000000000003,
      "aux_distill/loss": 1.087590326865514,
      "aux_distill/mean_u": 0.2842842258965096,
      "aux_distill/n_active_tok": 26.88888888888889,
      "calib/answer_extract_rate": 0.09375,
      "calib/avg_num_step_conf": 0.71484375,
      "calib/ece": 0.4649230769230769,
      "calib/final_conf_rate": 0.1015625,
      "calib/format_rate": 0.0859375,
      "calib/frac_conf_gt_0.9": 0.5769230769230769,
      "calib/gap": 0.12395151515151503,
      "calib/mean_conf": 0.8603076923076922,
      "calib/mu_c": 0.9318181818181818,
      "calib/mu_w": 0.8078666666666667,
      "calib/nonempty_final_conf_rate": 0.1015625,
      "calib/nonempty_reasoning_rate": 0.1484375,
      "calib/nonempty_step_conf_rate": 0.140625,
      "calib/pce": 0.45107692307692304,
      "calib/std_conf": 0.15624826626257063,
      "calib/step_conf_rate": 0.140625,
      "calib/step_q_c": 0.8850000000000001,
      "calib/step_q_c_n": 48.0,
      "calib/step_q_gap": 0.1614444444444445,
      "calib/step_q_w": 0.7235555555555556,
      "calib/step_q_w_n": 135.0,
      "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": 0.04296875,
      "completions/max_length": 3017.0,
      "completions/max_terminated_length": 3017.0,
      "completions/mean_length": 657.3046875,
      "completions/mean_terminated_length": 686.8162841796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.013866666666666666,
      "grad_norm": 1.8521384000778198,
      "learning_rate": 3e-06,
      "loss": 0.0762,
      "num_tokens": 3622275.0,
      "reward": 0.0911479964852333,
      "reward_std": 0.21495230495929718,
      "rewards/accuracy_reward_step": 0.04296875,
      "rewards/final_brier_reward_step": 0.053389742970466614,
      "rewards/format_reward_step": 0.0859375,
      "step": 13
    },
    {
      "aux_distill/lambda": 0.10000000000000002,
      "aux_distill/loss": 0.9630577005445957,
      "aux_distill/mean_u": 0.2799823846927252,
      "aux_distill/n_active_tok": 30.5,
      "calib/answer_extract_rate": 0.140625,
      "calib/avg_num_step_conf": 1.26953125,
      "calib/ece": 0.4797850574712644,
      "calib/final_conf_rate": 0.11328125,
      "calib/format_rate": 0.08203125,
      "calib/frac_conf_gt_0.9": 0.3103448275862069,
      "calib/gap": -0.0819315359477123,
      "calib/mean_conf": 0.6202149425287355,
      "calib/mu_c": 0.5721861111111112,
      "calib/mu_w": 0.6541176470588235,
      "calib/nonempty_final_conf_rate": 0.11328125,
      "calib/nonempty_reasoning_rate": 0.21875,
      "calib/nonempty_step_conf_rate": 0.1640625,
      "calib/pce": 0.3431034482758621,
      "calib/std_conf": 0.36152277783771075,
      "calib/step_conf_rate": 0.1640625,
      "calib/step_q_c": 0.5964516129032258,
      "calib/step_q_c_n": 62.0,
      "calib/step_q_gap": 0.09865338730119794,
      "calib/step_q_w": 0.49779822560202786,
      "calib/step_q_w_n": 263.0,
      "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": 2949.0,
      "completions/max_terminated_length": 2949.0,
      "completions/mean_length": 585.1875,
      "completions/mean_terminated_length": 619.0413208007812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.014933333333333333,
      "grad_norm": 1.0055991411209106,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.0355,
      "num_tokens": 3877483.0,
      "reward": 0.09005921334028244,
      "reward_std": 0.18428009748458862,
      "rewards/accuracy_reward_step": 0.05078125,
      "rewards/final_brier_reward_step": 0.04730593040585518,
      "rewards/format_reward_step": 0.08203125,
      "step": 14
    },
    {
      "aux_distill/lambda": 0.10000000000000003,
      "aux_distill/loss": 0.9747249976448391,
      "aux_distill/mean_u": 0.39136178132639804,
      "aux_distill/n_active_tok": 33.04347826086956,
      "calib/answer_extract_rate": 0.14453125,
      "calib/avg_num_step_conf": 1.41796875,
      "calib/ece": 0.526875,
      "calib/final_conf_rate": 0.125,
      "calib/format_rate": 0.109375,
      "calib/frac_conf_gt_0.9": 0.3125,
      "calib/gap": -0.17142857142857143,
      "calib/mean_conf": 0.5725,
      "calib/mu_c": 0.43857142857142856,
      "calib/mu_w": 0.61,
      "calib/nonempty_final_conf_rate": 0.125,
      "calib/nonempty_reasoning_rate": 0.24609375,
      "calib/nonempty_step_conf_rate": 0.2109375,
      "calib/pce": 0.4403125,
      "calib/std_conf": 0.37339657202497184,
      "calib/step_conf_rate": 0.2109375,
      "calib/step_q_c": 0.42551724137931035,
      "calib/step_q_c_n": 58.0,
      "calib/step_q_gap": 0.011806287248435299,
      "calib/step_q_w": 0.41371095413087505,
      "calib/step_q_w_n": 305.0,
      "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": 3066.0,
      "completions/max_terminated_length": 3066.0,
      "completions/mean_length": 540.75,
      "completions/mean_terminated_length": 555.9517822265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.016,
      "grad_norm": 1.5617997646331787,
      "learning_rate": 3.5e-06,
      "loss": 0.0336,
      "num_tokens": 4123795.0,
      "reward": 0.0969601571559906,
      "reward_std": 0.2092396765947342,
      "rewards/accuracy_reward_step": 0.02734375,
      "rewards/final_brier_reward_step": 0.0572015605866909,
      "rewards/format_reward_step": 0.109375,
      "step": 15
    },
    {
      "aux_distill/lambda": 0.10000000000000003,
      "aux_distill/loss": 0.9529272094368935,
      "aux_distill/mean_u": 0.37173205164412104,
      "aux_distill/n_active_tok": 56.0,
      "calib/answer_extract_rate": 0.16015625,
      "calib/avg_num_step_conf": 1.6328125,
      "calib/ece": 0.42747368421052634,
      "calib/final_conf_rate": 0.1484375,
      "calib/format_rate": 0.10546875,
      "calib/frac_conf_gt_0.9": 0.2894736842105263,
      "calib/gap": 0.060500000000000054,
      "calib/mean_conf": 0.4624210526315789,
      "calib/mu_c": 0.507,
      "calib/mu_w": 0.44649999999999995,
      "calib/nonempty_final_conf_rate": 0.1484375,
      "calib/nonempty_reasoning_rate": 0.265625,
      "calib/nonempty_step_conf_rate": 0.2265625,
      "calib/pce": 0.3133684210526316,
      "calib/std_conf": 0.4219147227118263,
      "calib/step_conf_rate": 0.2265625,
      "calib/step_q_c": 0.4690909090909091,
      "calib/step_q_c_n": 44.0,
      "calib/step_q_gap": 0.07918449197860966,
      "calib/step_q_w": 0.38990641711229945,
      "calib/step_q_w_n": 374.0,
      "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": 3072.0,
      "completions/max_terminated_length": 3072.0,
      "completions/mean_length": 575.58984375,
      "completions/mean_terminated_length": 608.888427734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.017066666666666667,
      "grad_norm": 0.7805458307266235,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0193,
      "num_tokens": 4379994.0,
      "reward": 0.10980231314897537,
      "reward_std": 0.21534091234207153,
      "rewards/accuracy_reward_step": 0.04296875,
      "rewards/final_brier_reward_step": 0.07116712629795074,
      "rewards/format_reward_step": 0.10546875,
      "step": 16
    },
    {
      "aux_distill/lambda": 0.10000000000000003,
      "aux_distill/loss": 0.9404510505821394,
      "aux_distill/mean_u": 0.3841350244727926,
      "aux_distill/n_active_tok": 52.69565217391305,
      "calib/answer_extract_rate": 0.25390625,
      "calib/avg_num_step_conf": 2.19921875,
      "calib/ece": 0.27664382713964286,
      "calib/final_conf_rate": 0.21875,
      "calib/format_rate": 0.16015625,
      "calib/frac_conf_gt_0.9": 0.16071428571428573,
      "calib/gap": 0.21649394476619044,
      "calib/mean_conf": 0.3804866842825,
      "calib/mu_c": 0.5428571428571428,
      "calib/mu_w": 0.3263631980909524,
      "calib/nonempty_final_conf_rate": 0.21875,
      "calib/nonempty_reasoning_rate": 0.3984375,
      "calib/nonempty_step_conf_rate": 0.32421875,
      "calib/pce": 0.20356525571107142,
      "calib/std_conf": 0.36673813544202716,
      "calib/step_conf_rate": 0.32421875,
      "calib/step_q_c": 0.27522388059701497,
      "calib/step_q_c_n": 67.0,
      "calib/step_q_gap": -0.06842597302122227,
      "calib/step_q_w": 0.34364985361823724,
      "calib/step_q_w_n": 496.0,
      "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": 2893.0,
      "completions/max_terminated_length": 2893.0,
      "completions/mean_length": 498.5,
      "completions/mean_terminated_length": 514.5806274414062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.018133333333333335,
      "grad_norm": 1.1857376098632812,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0915,
      "num_tokens": 4611138.0,
      "reward": 0.1654895544052124,
      "reward_std": 0.3047766089439392,
      "rewards/accuracy_reward_step": 0.0546875,
      "rewards/final_brier_reward_step": 0.11613535135984421,
      "rewards/format_reward_step": 0.16015625,
      "step": 17
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9438614919781685,
      "aux_distill/mean_u": 0.3519507888561258,
      "aux_distill/n_active_tok": 69.125,
      "calib/answer_extract_rate": 0.3359375,
      "calib/avg_num_step_conf": 3.0390625,
      "calib/ece": 0.3567121111111111,
      "calib/final_conf_rate": 0.328125,
      "calib/format_rate": 0.25,
      "calib/frac_conf_gt_0.9": 0.14285714285714285,
      "calib/gap": -0.17255727187591455,
      "calib/mean_conf": 0.33294084920634925,
      "calib/mu_c": 0.1953058823529412,
      "calib/mu_w": 0.36786315422885574,
      "calib/nonempty_final_conf_rate": 0.328125,
      "calib/nonempty_reasoning_rate": 0.5078125,
      "calib/nonempty_step_conf_rate": 0.453125,
      "calib/pce": 0.24363600396825394,
      "calib/std_conf": 0.34057311000666146,
      "calib/step_conf_rate": 0.453125,
      "calib/step_q_c": 0.19010934579439256,
      "calib/step_q_c_n": 107.0,
      "calib/step_q_gap": -0.0462484400259055,
      "calib/step_q_w": 0.23635778582029807,
      "calib/step_q_w_n": 671.0,
      "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": 0.0234375,
      "completions/max_length": 2714.0,
      "completions/max_terminated_length": 2714.0,
      "completions/mean_length": 514.56640625,
      "completions/mean_terminated_length": 526.916015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.0192,
      "grad_norm": 1.1566716432571411,
      "learning_rate": 4.25e-06,
      "loss": 0.0994,
      "num_tokens": 4853587.0,
      "reward": 0.24126005172729492,
      "reward_std": 0.35394883155822754,
      "rewards/accuracy_reward_step": 0.0703125,
      "rewards/final_brier_reward_step": 0.16220760345458984,
      "rewards/format_reward_step": 0.25,
      "step": 18
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9123309293100911,
      "aux_distill/mean_u": 0.35264843125580947,
      "aux_distill/n_active_tok": 64.51612903225806,
      "calib/answer_extract_rate": 0.65625,
      "calib/avg_num_step_conf": 3.90234375,
      "calib/ece": 0.1993193862581379,
      "calib/final_conf_rate": 0.66015625,
      "calib/format_rate": 0.53515625,
      "calib/frac_conf_gt_0.9": 0.05325443786982249,
      "calib/gap": -0.06653554882625726,
      "calib/mean_conf": 0.21027278051141096,
      "calib/mu_c": 0.15239866407673747,
      "calib/mu_w": 0.21893421290299472,
      "calib/nonempty_final_conf_rate": 0.66015625,
      "calib/nonempty_reasoning_rate": 0.81640625,
      "calib/nonempty_step_conf_rate": 0.72265625,
      "calib/pce": 0.13970732598832475,
      "calib/std_conf": 0.2727893687990524,
      "calib/step_conf_rate": 0.72265625,
      "calib/step_q_c": 0.16477064220183493,
      "calib/step_q_c_n": 109.0,
      "calib/step_q_gap": -0.04536888836948741,
      "calib/step_q_w": 0.21013953057132234,
      "calib/step_q_w_n": 890.0,
      "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": 2972.0,
      "completions/max_terminated_length": 2972.0,
      "completions/mean_length": 407.1015625,
      "completions/mean_terminated_length": 408.69805908203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 1.0,
      "epoch": 0.020266666666666665,
      "grad_norm": 1.1330541372299194,
      "learning_rate": 4.5e-06,
      "loss": 0.128,
      "num_tokens": 5062565.0,
      "reward": 0.5311158299446106,
      "reward_std": 0.45769965648651123,
      "rewards/accuracy_reward_step": 0.09765625,
      "rewards/final_brier_reward_step": 0.4294191896915436,
      "rewards/format_reward_step": 0.53515625,
      "step": 19
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9665737859904766,
      "aux_distill/mean_u": 0.446485798432125,
      "aux_distill/n_active_tok": 63.875,
      "calib/answer_extract_rate": 0.78515625,
      "calib/avg_num_step_conf": 3.921875,
      "calib/ece": 0.1426651822136941,
      "calib/final_conf_rate": 0.76171875,
      "calib/format_rate": 0.6640625,
      "calib/frac_conf_gt_0.9": 0.010256410256410256,
      "calib/gap": 0.020352225624694464,
      "calib/mean_conf": 0.1951296895811777,
      "calib/mu_c": 0.2115158404687522,
      "calib/mu_w": 0.19116361484405772,
      "calib/nonempty_final_conf_rate": 0.76171875,
      "calib/nonempty_reasoning_rate": 0.90625,
      "calib/nonempty_step_conf_rate": 0.81640625,
      "calib/pce": 0.07146153846153845,
      "calib/std_conf": 0.19727081252457762,
      "calib/step_conf_rate": 0.81640625,
      "calib/step_q_c": 0.23115456434267534,
      "calib/step_q_c_n": 149.0,
      "calib/step_q_gap": 0.031229714981662104,
      "calib/step_q_w": 0.19992484936101323,
      "calib/step_q_w_n": 855.0,
      "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": 2996.0,
      "completions/max_terminated_length": 2996.0,
      "completions/mean_length": 338.4453125,
      "completions/mean_terminated_length": 342.4585266113281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 0.021333333333333333,
      "grad_norm": 1.3762656450271606,
      "learning_rate": 4.75e-06,
      "loss": 0.1177,
      "num_tokens": 5254079.0,
      "reward": 0.6778804063796997,
      "reward_std": 0.46077030897140503,
      "rewards/accuracy_reward_step": 0.15234375,
      "rewards/final_brier_reward_step": 0.5393545627593994,
      "rewards/format_reward_step": 0.6640625,
      "step": 20
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9324986916035414,
      "aux_distill/mean_u": 0.32273819262569403,
      "aux_distill/n_active_tok": 61.75,
      "calib/answer_extract_rate": 0.890625,
      "calib/avg_num_step_conf": 3.81640625,
      "calib/ece": 0.1712819804076931,
      "calib/final_conf_rate": 0.8984375,
      "calib/format_rate": 0.78125,
      "calib/frac_conf_gt_0.9": 0.030434782608695653,
      "calib/gap": -0.01407527168420239,
      "calib/mean_conf": 0.24356141418077143,
      "calib/mu_c": 0.23223999999999997,
      "calib/mu_w": 0.24631527168420236,
      "calib/nonempty_final_conf_rate": 0.8984375,
      "calib/nonempty_reasoning_rate": 0.97265625,
      "calib/nonempty_step_conf_rate": 0.90234375,
      "calib/pce": 0.10959561033771054,
      "calib/std_conf": 0.2399938362844077,
      "calib/step_conf_rate": 0.90234375,
      "calib/step_q_c": 0.2613429411764706,
      "calib/step_q_c_n": 170.0,
      "calib/step_q_gap": 0.012324789999271113,
      "calib/step_q_w": 0.2490181511771995,
      "calib/step_q_w_n": 807.0,
      "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": 2985.0,
      "completions/max_terminated_length": 2985.0,
      "completions/mean_length": 297.56640625,
      "completions/mean_terminated_length": 298.73333740234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0224,
      "grad_norm": 0.912643313407898,
      "learning_rate": 5e-06,
      "loss": 0.1296,
      "num_tokens": 5433216.0,
      "reward": 0.7877582907676697,
      "reward_std": 0.39034831523895264,
      "rewards/accuracy_reward_step": 0.18359375,
      "rewards/final_brier_reward_step": 0.6106728911399841,
      "rewards/format_reward_step": 0.78125,
      "step": 21
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9481630343943834,
      "aux_distill/mean_u": 0.3599154632144565,
      "aux_distill/n_active_tok": 57.625,
      "calib/answer_extract_rate": 0.91015625,
      "calib/avg_num_step_conf": 3.5390625,
      "calib/ece": 0.19719833333333334,
      "calib/final_conf_rate": 0.9375,
      "calib/format_rate": 0.8515625,
      "calib/frac_conf_gt_0.9": 0.025,
      "calib/gap": -0.0323800114220445,
      "calib/mean_conf": 0.2848016666666667,
      "calib/mu_c": 0.2570088235294118,
      "calib/mu_w": 0.2893888349514563,
      "calib/nonempty_final_conf_rate": 0.9375,
      "calib/nonempty_reasoning_rate": 0.96484375,
      "calib/nonempty_step_conf_rate": 0.921875,
      "calib/pce": 0.1701666666666667,
      "calib/std_conf": 0.24047366504579154,
      "calib/step_conf_rate": 0.921875,
      "calib/step_q_c": 0.3154104575163399,
      "calib/step_q_c_n": 153.0,
      "calib/step_q_gap": -0.01267838710517405,
      "calib/step_q_w": 0.32808884462151394,
      "calib/step_q_w_n": 753.0,
      "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": 2532.0,
      "completions/max_terminated_length": 2532.0,
      "completions/mean_length": 292.546875,
      "completions/mean_terminated_length": 293.6941223144531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 23.0,
      "epoch": 0.023466666666666667,
      "grad_norm": 1.1575355529785156,
      "learning_rate": 4.9722222222222224e-06,
      "loss": 0.0719,
      "num_tokens": 5609924.0,
      "reward": 0.8310613632202148,
      "reward_std": 0.34224647283554077,
      "rewards/accuracy_reward_step": 0.13671875,
      "rewards/final_brier_reward_step": 0.6738414764404297,
      "rewards/format_reward_step": 0.8515625,
      "step": 22
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9651179648935795,
      "aux_distill/mean_u": 0.3030497133401629,
      "aux_distill/n_active_tok": 66.0,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 4.62109375,
      "calib/ece": 0.2240234610788449,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.93359375,
      "calib/frac_conf_gt_0.9": 0.028112449799196786,
      "calib/gap": -0.005876867276822795,
      "calib/mean_conf": 0.3204090032475196,
      "calib/mu_c": 0.31540540540540546,
      "calib/mu_w": 0.32128227268222825,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.96875,
      "calib/pce": 0.1979190434081622,
      "calib/std_conf": 0.24102069592003547,
      "calib/step_conf_rate": 0.96875,
      "calib/step_q_c": 0.4652601156069365,
      "calib/step_q_c_n": 173.0,
      "calib/step_q_gap": 0.13476971956733252,
      "calib/step_q_w": 0.33049039603960395,
      "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": 0.00390625,
      "completions/max_length": 2817.0,
      "completions/max_terminated_length": 2817.0,
      "completions/mean_length": 330.66796875,
      "completions/mean_terminated_length": 331.9647216796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 25.0,
      "epoch": 0.024533333333333334,
      "grad_norm": 1.0878534317016602,
      "learning_rate": 4.944444444444445e-06,
      "loss": 0.0965,
      "num_tokens": 5798511.0,
      "reward": 0.90683913230896,
      "reward_std": 0.24767723679542542,
      "rewards/accuracy_reward_step": 0.14453125,
      "rewards/final_brier_reward_step": 0.7355532646179199,
      "rewards/format_reward_step": 0.93359375,
      "step": 23
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.96570317260921,
      "aux_distill/mean_u": 0.3328278768185367,
      "aux_distill/n_active_tok": 78.375,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 5.0546875,
      "calib/ece": 0.21902016129032253,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.91796875,
      "calib/frac_conf_gt_0.9": 0.036290322580645164,
      "calib/gap": -0.04096695273472528,
      "calib/mean_conf": 0.3345846774193549,
      "calib/mu_c": 0.2997297297297297,
      "calib/mu_w": 0.340696682464455,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.96484375,
      "calib/pce": 0.20220564516129028,
      "calib/std_conf": 0.23942838701197067,
      "calib/step_conf_rate": 0.96484375,
      "calib/step_q_c": 0.36037037037037034,
      "calib/step_q_c_n": 189.0,
      "calib/step_q_gap": -0.026124652254064074,
      "calib/step_q_w": 0.3864950226244344,
      "calib/step_q_w_n": 1105.0,
      "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": 3041.0,
      "completions/max_terminated_length": 3041.0,
      "completions/mean_length": 345.12109375,
      "completions/mean_terminated_length": 346.4745178222656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 61.0,
      "epoch": 0.0256,
      "grad_norm": 1.195889949798584,
      "learning_rate": 4.9166666666666665e-06,
      "loss": 0.1758,
      "num_tokens": 5991374.0,
      "reward": 0.8862032294273376,
      "reward_std": 0.26511549949645996,
      "rewards/accuracy_reward_step": 0.14453125,
      "rewards/final_brier_reward_step": 0.7099063396453857,
      "rewards/format_reward_step": 0.91796875,
      "step": 24
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8784007225185633,
      "aux_distill/mean_u": 0.3437077893096381,
      "aux_distill/n_active_tok": 96.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 6.0546875,
      "calib/ece": 0.22140242914979763,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.94140625,
      "calib/frac_conf_gt_0.9": 0.016194331983805668,
      "calib/gap": -0.028569543147208054,
      "calib/mean_conf": 0.3309862348178138,
      "calib/mu_c": 0.30820000000000003,
      "calib/mu_w": 0.3367695431472081,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.984375,
      "calib/pce": 0.17497975708502028,
      "calib/std_conf": 0.21689816182725485,
      "calib/step_conf_rate": 0.984375,
      "calib/step_q_c": 0.3235438596491228,
      "calib/step_q_c_n": 285.0,
      "calib/step_q_gap": -0.04858870952083766,
      "calib/step_q_w": 0.37213256916996046,
      "calib/step_q_w_n": 1265.0,
      "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": 0.0,
      "completions/max_length": 2073.0,
      "completions/max_terminated_length": 2073.0,
      "completions/mean_length": 333.0390625,
      "completions/mean_terminated_length": 333.0390625,
      "completions/min_length": 77.0,
      "completions/min_terminated_length": 77.0,
      "epoch": 0.02666666666666667,
      "grad_norm": 0.9269714951515198,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.1322,
      "num_tokens": 6179856.0,
      "reward": 0.9290375113487244,
      "reward_std": 0.24123907089233398,
      "rewards/accuracy_reward_step": 0.1953125,
      "rewards/final_brier_reward_step": 0.7213562726974487,
      "rewards/format_reward_step": 0.94140625,
      "step": 25
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9190471321344376,
      "aux_distill/mean_u": 0.3904257065951485,
      "aux_distill/n_active_tok": 92.0,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 5.9453125,
      "calib/ece": 0.22666126482213436,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.023715415019762844,
      "calib/gap": -0.01109248826291076,
      "calib/mean_conf": 0.3333387351778656,
      "calib/mu_c": 0.324,
      "calib/mu_w": 0.33509248826291077,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.20094861660079053,
      "calib/std_conf": 0.22792036210925268,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.26669642857142856,
      "calib/step_q_c_n": 224.0,
      "calib/step_q_gap": -0.1110853896103896,
      "calib/step_q_w": 0.37778181818181816,
      "calib/step_q_w_n": 1298.0,
      "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": 1246.0,
      "completions/max_terminated_length": 1246.0,
      "completions/mean_length": 313.1171875,
      "completions/mean_terminated_length": 315.5826721191406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.027733333333333332,
      "grad_norm": 0.9309002161026001,
      "learning_rate": 4.861111111111111e-06,
      "loss": 0.1212,
      "num_tokens": 6365254.0,
      "reward": 0.9508861303329468,
      "reward_std": 0.17603799700737,
      "rewards/accuracy_reward_step": 0.15625,
      "rewards/final_brier_reward_step": 0.7650535702705383,
      "rewards/format_reward_step": 0.98046875,
      "step": 26
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.880526764318347,
      "aux_distill/mean_u": 0.38630980740941306,
      "aux_distill/n_active_tok": 89.625,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 6.5234375,
      "calib/ece": 0.16436765534052672,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.011904761904761904,
      "calib/gap": 0.0424597766099421,
      "calib/mean_conf": 0.26168194105481246,
      "calib/mu_c": 0.29875,
      "calib/mu_w": 0.2562902233900579,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1495327347056061,
      "calib/std_conf": 0.18758624185278192,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.35888888888888887,
      "calib/step_q_c_n": 162.0,
      "calib/step_q_gap": 0.06739253610374296,
      "calib/step_q_w": 0.2914963527851459,
      "calib/step_q_w_n": 1508.0,
      "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": 2837.0,
      "completions/max_terminated_length": 2837.0,
      "completions/mean_length": 310.01953125,
      "completions/mean_terminated_length": 313.6956787109375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 107.0,
      "epoch": 0.0288,
      "grad_norm": 0.6328495740890503,
      "learning_rate": 4.833333333333333e-06,
      "loss": 0.0679,
      "num_tokens": 6549835.0,
      "reward": 0.9706956148147583,
      "reward_std": 0.1332382708787918,
      "rewards/accuracy_reward_step": 0.125,
      "rewards/final_brier_reward_step": 0.8320162296295166,
      "rewards/format_reward_step": 0.984375,
      "step": 27
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8589735962450504,
      "aux_distill/mean_u": 0.3424635989604109,
      "aux_distill/n_active_tok": 99.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 6.57421875,
      "calib/ece": 0.11967999992745071,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.011764705882352941,
      "calib/gap": 0.029057711345396253,
      "calib/mean_conf": 0.2874447058098037,
      "calib/mu_c": 0.31057692307692303,
      "calib/mu_w": 0.2815192117315268,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.10160156855490168,
      "calib/std_conf": 0.196537932612208,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.34105919003115265,
      "calib/step_q_c_n": 321.0,
      "calib/step_q_gap": 0.032916616935873055,
      "calib/step_q_w": 0.3081425730952796,
      "calib/step_q_w_n": 1362.0,
      "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": 1546.0,
      "completions/max_terminated_length": 1546.0,
      "completions/mean_length": 316.87890625,
      "completions/mean_terminated_length": 319.3740234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 108.0,
      "epoch": 0.029866666666666666,
      "grad_norm": 43.48025894165039,
      "learning_rate": 4.805555555555556e-06,
      "loss": 0.0589,
      "num_tokens": 6737900.0,
      "reward": 0.9931570887565613,
      "reward_std": 0.14155761897563934,
      "rewards/accuracy_reward_step": 0.203125,
      "rewards/final_brier_reward_step": 0.7949079275131226,
      "rewards/format_reward_step": 0.98828125,
      "step": 28
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8502343278378248,
      "aux_distill/mean_u": 0.3553671577059875,
      "aux_distill/n_active_tok": 101.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 6.75,
      "calib/ece": 0.15306274509803927,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00392156862745098,
      "calib/gap": 0.057138248847926315,
      "calib/mean_conf": 0.2413764705882353,
      "calib/mu_c": 0.29000000000000004,
      "calib/mu_w": 0.23286175115207372,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12270980392156865,
      "calib/std_conf": 0.17569065794601613,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3440454545454546,
      "calib/step_q_c_n": 220.0,
      "calib/step_q_gap": 0.05622582589823971,
      "calib/step_q_w": 0.2878196286472149,
      "calib/step_q_w_n": 1508.0,
      "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": 764.0,
      "completions/max_terminated_length": 764.0,
      "completions/mean_length": 304.12109375,
      "completions/mean_terminated_length": 306.5157470703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.030933333333333334,
      "grad_norm": 10.140074729919434,
      "learning_rate": 4.777777777777778e-06,
      "loss": 0.0734,
      "num_tokens": 6922883.0,
      "reward": 0.9947497844696045,
      "reward_std": 0.09830287843942642,
      "rewards/accuracy_reward_step": 0.1484375,
      "rewards/final_brier_reward_step": 0.8449683785438538,
      "rewards/format_reward_step": 0.99609375,
      "step": 29
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8902030065655708,
      "aux_distill/mean_u": 0.298180368346847,
      "aux_distill/n_active_tok": 103.125,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 7.24609375,
      "calib/ece": 0.15161501976284586,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.007905138339920948,
      "calib/gap": 0.04919317754499064,
      "calib/mean_conf": 0.2650648221343873,
      "calib/mu_c": 0.30764705882352944,
      "calib/mu_w": 0.2584538812785388,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.14114624505928855,
      "calib/std_conf": 0.20693506549564705,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.3071291866028708,
      "calib/step_q_c_n": 209.0,
      "calib/step_q_gap": 0.021537023783915754,
      "calib/step_q_w": 0.28559216281895505,
      "calib/step_q_w_n": 1646.0,
      "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": 1770.0,
      "completions/max_terminated_length": 1770.0,
      "completions/mean_length": 324.484375,
      "completions/mean_terminated_length": 328.33203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 110.0,
      "epoch": 0.032,
      "grad_norm": 4.368605613708496,
      "learning_rate": 4.75e-06,
      "loss": 0.0494,
      "num_tokens": 7112935.0,
      "reward": 0.9643255472183228,
      "reward_std": 0.15291661024093628,
      "rewards/accuracy_reward_step": 0.1328125,
      "rewards/final_brier_reward_step": 0.8192760944366455,
      "rewards/format_reward_step": 0.9765625,
      "step": 30
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8743925169110298,
      "aux_distill/mean_u": 0.3390971488901282,
      "aux_distill/n_active_tok": 103.625,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 7.328125,
      "calib/ece": 0.1774110671936759,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.007905138339920948,
      "calib/gap": -0.05309655172413791,
      "calib/mean_conf": 0.27140316205533593,
      "calib/mu_c": 0.2288,
      "calib/mu_w": 0.2818965517241379,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12559288537549407,
      "calib/std_conf": 0.19619026497013756,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.30663663663663665,
      "calib/step_q_c_n": 333.0,
      "calib/step_q_gap": 0.03359710326009746,
      "calib/step_q_w": 0.2730395333765392,
      "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": 0.0078125,
      "completions/max_length": 861.0,
      "completions/max_terminated_length": 861.0,
      "completions/mean_length": 291.5625,
      "completions/mean_terminated_length": 293.8582763671875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 129.0,
      "epoch": 0.03306666666666667,
      "grad_norm": 24.946491241455078,
      "learning_rate": 4.722222222222222e-06,
      "loss": 0.059,
      "num_tokens": 7293487.0,
      "reward": 0.9699900150299072,
      "reward_std": 0.15348196029663086,
      "rewards/accuracy_reward_step": 0.1953125,
      "rewards/final_brier_reward_step": 0.7641987204551697,
      "rewards/format_reward_step": 0.98046875,
      "step": 31
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8389797694981098,
      "aux_distill/mean_u": 0.305542944566539,
      "aux_distill/n_active_tok": 96.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 6.23046875,
      "calib/ece": 0.15460937500000002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": -0.014638684061259205,
      "calib/mean_conf": 0.24156249999999999,
      "calib/mu_c": 0.22926829268292684,
      "calib/mu_w": 0.24390697674418604,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11800781249999998,
      "calib/std_conf": 0.17999104795447463,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24836909871244636,
      "calib/step_q_c_n": 233.0,
      "calib/step_q_gap": -0.01399507162529226,
      "calib/step_q_w": 0.2623641703377386,
      "calib/step_q_w_n": 1362.0,
      "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": 261.515625,
      "completions/mean_terminated_length": 263.5747985839844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 136.0,
      "epoch": 0.034133333333333335,
      "grad_norm": 10.89183235168457,
      "learning_rate": 4.694444444444445e-06,
      "loss": 0.1066,
      "num_tokens": 7467139.0,
      "reward": 0.9876136779785156,
      "reward_std": 0.10019958019256592,
      "rewards/accuracy_reward_step": 0.16015625,
      "rewards/final_brier_reward_step": 0.8189772963523865,
      "rewards/format_reward_step": 0.99609375,
      "step": 32
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.875149991363287,
      "aux_distill/mean_u": 0.32872740744453877,
      "aux_distill/n_active_tok": 108.75,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 6.609375,
      "calib/ece": 0.15937007874015743,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.003937007874015748,
      "calib/gap": -0.0203460038986355,
      "calib/mean_conf": 0.2351968503937008,
      "calib/mu_c": 0.21789473684210525,
      "calib/mu_w": 0.23824074074074075,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12248031496062992,
      "calib/std_conf": 0.17148655461952939,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24656387665198237,
      "calib/step_q_c_n": 227.0,
      "calib/step_q_gap": -0.02186615747771048,
      "calib/step_q_w": 0.26843003412969285,
      "calib/step_q_w_n": 1465.0,
      "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": 1117.0,
      "completions/max_terminated_length": 1117.0,
      "completions/mean_length": 268.30078125,
      "completions/mean_terminated_length": 270.41339111328125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 101.0,
      "epoch": 0.0352,
      "grad_norm": 2.0930428504943848,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.0706,
      "num_tokens": 7642696.0,
      "reward": 0.9824995994567871,
      "reward_std": 0.09917615354061127,
      "rewards/accuracy_reward_step": 0.1484375,
      "rewards/final_brier_reward_step": 0.8243741989135742,
      "rewards/format_reward_step": 0.9921875,
      "step": 33
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8333712462335825,
      "aux_distill/mean_u": 0.3157383508748938,
      "aux_distill/n_active_tok": 89.625,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.6796875,
      "calib/ece": 0.15984313725490196,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.03704761904761905,
      "calib/mean_conf": 0.1971764705882353,
      "calib/mu_c": 0.16666666666666666,
      "calib/mu_w": 0.2037142857142857,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.09027450980392157,
      "calib/std_conf": 0.14941148679918573,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.22977973568281937,
      "calib/step_q_c_n": 227.0,
      "calib/step_q_gap": -0.008932570755648434,
      "calib/step_q_w": 0.2387123064384678,
      "calib/step_q_w_n": 1227.0,
      "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": 654.0,
      "completions/max_terminated_length": 654.0,
      "completions/mean_length": 232.14453125,
      "completions/mean_terminated_length": 233.97244262695312,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.03626666666666667,
      "grad_norm": 4.866479396820068,
      "learning_rate": 4.638888888888889e-06,
      "loss": 0.0829,
      "num_tokens": 7807237.0,
      "reward": 0.991145133972168,
      "reward_std": 0.08340153098106384,
      "rewards/accuracy_reward_step": 0.17578125,
      "rewards/final_brier_reward_step": 0.8143214583396912,
      "rewards/format_reward_step": 0.9921875,
      "step": 34
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8673066068440676,
      "aux_distill/mean_u": 0.3331934416142454,
      "aux_distill/n_active_tok": 95.375,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 6.0078125,
      "calib/ece": 0.13542890625,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.008839642684693394,
      "calib/mean_conf": 0.21121171875,
      "calib/mu_c": 0.2036842105263158,
      "calib/mu_w": 0.2125238532110092,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0991015625,
      "calib/std_conf": 0.17462509132115264,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21622727272727277,
      "calib/step_q_c_n": 220.0,
      "calib/step_q_gap": -0.017437446544350893,
      "calib/step_q_w": 0.23366471927162366,
      "calib/step_q_w_n": 1318.0,
      "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": 831.0,
      "completions/max_terminated_length": 831.0,
      "completions/mean_length": 257.53125,
      "completions/mean_terminated_length": 259.5590515136719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 90.0,
      "epoch": 0.037333333333333336,
      "grad_norm": 20.04242706298828,
      "learning_rate": 4.611111111111112e-06,
      "loss": 0.0487,
      "num_tokens": 7982421.0,
      "reward": 0.9849402904510498,
      "reward_std": 0.1031317412853241,
      "rewards/accuracy_reward_step": 0.1484375,
      "rewards/final_brier_reward_step": 0.8292554616928101,
      "rewards/format_reward_step": 0.9921875,
      "step": 35
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9580904059112072,
      "aux_distill/mean_u": 0.3945864164823561,
      "aux_distill/n_active_tok": 106.25,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 6.58203125,
      "calib/ece": 0.13430830039525696,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.003952569169960474,
      "calib/gap": 0.016415094339622627,
      "calib/mean_conf": 0.1934387351778656,
      "calib/mu_c": 0.20641509433962263,
      "calib/mu_w": 0.19,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.059130434782608696,
      "calib/std_conf": 0.16373250502921927,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22345323741007192,
      "calib/step_q_c_n": 278.0,
      "calib/step_q_gap": 0.03788820542712951,
      "calib/step_q_w": 0.1855650319829424,
      "calib/step_q_w_n": 1407.0,
      "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": 2971.0,
      "completions/max_terminated_length": 2971.0,
      "completions/mean_length": 262.90234375,
      "completions/mean_terminated_length": 266.019775390625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 94.0,
      "epoch": 0.0384,
      "grad_norm": 77.7988510131836,
      "learning_rate": 4.583333333333333e-06,
      "loss": 0.1055,
      "num_tokens": 8152436.0,
      "reward": 0.9992785453796387,
      "reward_std": 0.11146703362464905,
      "rewards/accuracy_reward_step": 0.20703125,
      "rewards/final_brier_reward_step": 0.8032445311546326,
      "rewards/format_reward_step": 0.98828125,
      "step": 36
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8078534454107285,
      "aux_distill/mean_u": 0.2900667277152228,
      "aux_distill/n_active_tok": 104.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 7.70703125,
      "calib/ece": 0.11493769841269842,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.003968253968253968,
      "calib/gap": 0.014841203703703704,
      "calib/mean_conf": 0.20589007936507936,
      "calib/mu_c": 0.21861111111111112,
      "calib/mu_w": 0.2037699074074074,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08898531746031746,
      "calib/std_conf": 0.15356669412485946,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22843478260869562,
      "calib/step_q_c_n": 230.0,
      "calib/step_q_gap": 0.005715333383222304,
      "calib/step_q_w": 0.22271944922547332,
      "calib/step_q_w_n": 1743.0,
      "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": 1109.0,
      "completions/max_terminated_length": 1109.0,
      "completions/mean_length": 262.359375,
      "completions/mean_terminated_length": 265.4703674316406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.039466666666666664,
      "grad_norm": 12.64301872253418,
      "learning_rate": 4.555555555555556e-06,
      "loss": 0.0472,
      "num_tokens": 8326696.0,
      "reward": 0.9826458692550659,
      "reward_std": 0.10451576858758926,
      "rewards/accuracy_reward_step": 0.140625,
      "rewards/final_brier_reward_step": 0.8402917385101318,
      "rewards/format_reward_step": 0.984375,
      "step": 37
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8974905293434858,
      "aux_distill/mean_u": 0.3690486426132548,
      "aux_distill/n_active_tok": 103.625,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 6.98828125,
      "calib/ece": 0.20847222222222223,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.011904761904761904,
      "calib/gap": -0.09633992094861665,
      "calib/mean_conf": 0.2448611111111111,
      "calib/mu_c": 0.15693181818181817,
      "calib/mu_w": 0.2532717391304348,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.18301587301587302,
      "calib/std_conf": 0.20042620894904456,
      "calib/step_conf_rate": 0.98828125,
      "calib/step_q_c": 0.1854905511811024,
      "calib/step_q_c_n": 127.0,
      "calib/step_q_gap": -0.07555920814501069,
      "calib/step_q_w": 0.2610497593261131,
      "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": 0.01171875,
      "completions/max_length": 1125.0,
      "completions/max_terminated_length": 1125.0,
      "completions/mean_length": 256.73046875,
      "completions/mean_terminated_length": 259.77471923828125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 49.0,
      "epoch": 0.04053333333333333,
      "grad_norm": 30.5622615814209,
      "learning_rate": 4.527777777777778e-06,
      "loss": 0.0467,
      "num_tokens": 8499307.0,
      "reward": 0.9338328838348389,
      "reward_std": 0.14586186408996582,
      "rewards/accuracy_reward_step": 0.0859375,
      "rewards/final_brier_reward_step": 0.8129782676696777,
      "rewards/format_reward_step": 0.96875,
      "step": 38
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8440385889261961,
      "aux_distill/mean_u": 0.27543343141819476,
      "aux_distill/n_active_tok": 90.375,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 6.4140625,
      "calib/ece": 0.10633858267716535,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.004351851851851857,
      "calib/mean_conf": 0.1812992125984252,
      "calib/mu_c": 0.185,
      "calib/mu_w": 0.18064814814814814,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.06901574803149607,
      "calib/std_conf": 0.14902664851590786,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23062801932367155,
      "calib/step_q_c_n": 207.0,
      "calib/step_q_gap": 0.04025171270346245,
      "calib/step_q_w": 0.1903763066202091,
      "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": 0.0078125,
      "completions/max_length": 806.0,
      "completions/max_terminated_length": 806.0,
      "completions/mean_length": 242.8828125,
      "completions/mean_terminated_length": 244.79527282714844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 97.0,
      "epoch": 0.0416,
      "grad_norm": 3.65295147895813,
      "learning_rate": 4.5e-06,
      "loss": 0.0743,
      "num_tokens": 8667573.0,
      "reward": 0.9884814023971558,
      "reward_std": 0.09546145796775818,
      "rewards/accuracy_reward_step": 0.1484375,
      "rewards/final_brier_reward_step": 0.8402441740036011,
      "rewards/format_reward_step": 0.98828125,
      "step": 39
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8871491681784391,
      "aux_distill/mean_u": 0.344563438691629,
      "aux_distill/n_active_tok": 92.625,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.74609375,
      "calib/ece": 0.12785156250000004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0375840707964602,
      "calib/mean_conf": 0.2108203125,
      "calib/mu_c": 0.24400000000000005,
      "calib/mu_w": 0.20641592920353985,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11074218750000002,
      "calib/std_conf": 0.1723587253735718,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24912499999999999,
      "calib/step_q_c_n": 160.0,
      "calib/step_q_gap": 0.015387395118230285,
      "calib/step_q_w": 0.2337376048817697,
      "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": 676.0,
      "completions/max_terminated_length": 676.0,
      "completions/mean_length": 232.140625,
      "completions/mean_terminated_length": 233.968505859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.042666666666666665,
      "grad_norm": 2.471982955932617,
      "learning_rate": 4.472222222222223e-06,
      "loss": 0.0831,
      "num_tokens": 8833761.0,
      "reward": 0.9915173649787903,
      "reward_std": 0.07971757650375366,
      "rewards/accuracy_reward_step": 0.1171875,
      "rewards/final_brier_reward_step": 0.8658472299575806,
      "rewards/format_reward_step": 1.0,
      "step": 40
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8687476757913828,
      "aux_distill/mean_u": 0.3529711466997749,
      "aux_distill/n_active_tok": 80.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 5.265625,
      "calib/ece": 0.20875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0024202464202464635,
      "calib/mean_conf": 0.197890625,
      "calib/mu_c": 0.19945054945054944,
      "calib/mu_w": 0.19703030303030297,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.025585937499999992,
      "calib/std_conf": 0.15192335826695438,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2507223194748359,
      "calib/step_q_c_n": 457.0,
      "calib/step_q_gap": 0.02810728019312997,
      "calib/step_q_w": 0.22261503928170595,
      "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": 505.0,
      "completions/max_terminated_length": 505.0,
      "completions/mean_length": 200.63671875,
      "completions/mean_terminated_length": 202.21653747558594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.04373333333333333,
      "grad_norm": 13.761474609375,
      "learning_rate": 4.444444444444444e-06,
      "loss": 0.1044,
      "num_tokens": 8992372.0,
      "reward": 1.035874605178833,
      "reward_std": 0.10525625944137573,
      "rewards/accuracy_reward_step": 0.35546875,
      "rewards/final_brier_reward_step": 0.720186710357666,
      "rewards/format_reward_step": 0.99609375,
      "step": 41
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8690756633877754,
      "aux_distill/mean_u": 0.31748371317377855,
      "aux_distill/n_active_tok": 86.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 5.609375,
      "calib/ece": 0.106,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.00392156862745098,
      "calib/gap": 0.0146590909090909,
      "calib/mean_conf": 0.17941176470588233,
      "calib/mu_c": 0.1909090909090909,
      "calib/mu_w": 0.17625,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0348627450980392,
      "calib/std_conf": 0.140377806255272,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22679531772575254,
      "calib/step_q_c_n": 299.0,
      "calib/step_q_gap": 0.0047460653071069725,
      "calib/step_q_w": 0.22204925241864557,
      "calib/step_q_w_n": 1137.0,
      "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": 482.0,
      "completions/max_terminated_length": 482.0,
      "completions/mean_length": 209.125,
      "completions/mean_terminated_length": 210.7716522216797,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 115.0,
      "epoch": 0.0448,
      "grad_norm": 28.75571060180664,
      "learning_rate": 4.416666666666667e-06,
      "loss": 0.0837,
      "num_tokens": 9150276.0,
      "reward": 1.007413625717163,
      "reward_std": 0.09442334622144699,
      "rewards/accuracy_reward_step": 0.21484375,
      "rewards/final_brier_reward_step": 0.8077961206436157,
      "rewards/format_reward_step": 0.9921875,
      "step": 42
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8431951850652695,
      "aux_distill/mean_u": 0.28573219911255926,
      "aux_distill/n_active_tok": 86.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.27734375,
      "calib/ece": 0.1001953125,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.01317864077669903,
      "calib/mean_conf": 0.1701953125,
      "calib/mu_c": 0.18080000000000002,
      "calib/mu_w": 0.167621359223301,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03753906250000001,
      "calib/std_conf": 0.13265159715596092,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21855513307984792,
      "calib/step_q_c_n": 263.0,
      "calib/step_q_gap": -0.007391558096622675,
      "calib/step_q_w": 0.2259466911764706,
      "calib/step_q_w_n": 1088.0,
      "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": 386.0,
      "completions/max_terminated_length": 386.0,
      "completions/mean_length": 197.3515625,
      "completions/mean_terminated_length": 198.905517578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 123.0,
      "epoch": 0.04586666666666667,
      "grad_norm": 38.969566345214844,
      "learning_rate": 4.388888888888889e-06,
      "loss": 0.099,
      "num_tokens": 9306022.0,
      "reward": 1.012031078338623,
      "reward_std": 0.05379870533943176,
      "rewards/accuracy_reward_step": 0.1953125,
      "rewards/final_brier_reward_step": 0.8287496566772461,
      "rewards/format_reward_step": 1.0,
      "step": 43
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8739052787423134,
      "aux_distill/mean_u": 0.306426789185847,
      "aux_distill/n_active_tok": 73.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 4.73046875,
      "calib/ece": 0.0817578125,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.042027777777777775,
      "calib/mean_conf": 0.2287890625,
      "calib/mu_c": 0.26425,
      "calib/mu_w": 0.2222222222222222,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0771484375,
      "calib/std_conf": 0.1638731003104875,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27608465608465604,
      "calib/step_q_c_n": 189.0,
      "calib/step_q_gap": 0.007816554323403557,
      "calib/step_q_w": 0.2682681017612525,
      "calib/step_q_w_n": 1022.0,
      "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": 386.0,
      "completions/max_terminated_length": 386.0,
      "completions/mean_length": 179.6640625,
      "completions/mean_terminated_length": 181.0787353515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 101.0,
      "epoch": 0.046933333333333334,
      "grad_norm": 6.218605995178223,
      "learning_rate": 4.361111111111112e-06,
      "loss": 0.0856,
      "num_tokens": 9458336.0,
      "reward": 0.9939841628074646,
      "reward_std": 0.10380810499191284,
      "rewards/accuracy_reward_step": 0.15625,
      "rewards/final_brier_reward_step": 0.8395308256149292,
      "rewards/format_reward_step": 0.9921875,
      "step": 44
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8563596252351999,
      "aux_distill/mean_u": 0.30218692557266297,
      "aux_distill/n_active_tok": 75.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 4.83203125,
      "calib/ece": 0.13445312499999998,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": 0.011745819397993296,
      "calib/mean_conf": 0.22406250000000003,
      "calib/mu_c": 0.23461538461538461,
      "calib/mu_w": 0.22286956521739132,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1284765625,
      "calib/std_conf": 0.17754153343302517,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3034090909090909,
      "calib/step_q_c_n": 132.0,
      "calib/step_q_gap": 0.03712646647470175,
      "calib/step_q_w": 0.26628262443438916,
      "calib/step_q_w_n": 1105.0,
      "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": 462.0,
      "completions/max_terminated_length": 462.0,
      "completions/mean_length": 181.8203125,
      "completions/mean_terminated_length": 183.25196838378906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.048,
      "grad_norm": 7.898387908935547,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.0952,
      "num_tokens": 9609930.0,
      "reward": 0.9790830612182617,
      "reward_std": 0.08909302949905396,
      "rewards/accuracy_reward_step": 0.1015625,
      "rewards/final_brier_reward_step": 0.8605098128318787,
      "rewards/format_reward_step": 0.99609375,
      "step": 45
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8764547742903233,
      "aux_distill/mean_u": 0.37825321606828577,
      "aux_distill/n_active_tok": 77.625,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 4.80078125,
      "calib/ece": 0.14183125,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": 0.027243571210424533,
      "calib/mean_conf": 0.22371562500000003,
      "calib/mu_c": 0.24595744680851064,
      "calib/mu_w": 0.2187138755980861,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0909765625,
      "calib/std_conf": 0.1743838162813837,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2587906976744186,
      "calib/step_q_c_n": 215.0,
      "calib/step_q_gap": -0.010274095224989732,
      "calib/step_q_w": 0.2690647928994083,
      "calib/step_q_w_n": 1014.0,
      "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": 521.0,
      "completions/max_terminated_length": 521.0,
      "completions/mean_length": 171.25390625,
      "completions/mean_terminated_length": 172.60235595703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 105.0,
      "epoch": 0.04906666666666667,
      "grad_norm": 9.550283432006836,
      "learning_rate": 4.305555555555556e-06,
      "loss": 0.0684,
      "num_tokens": 9758539.0,
      "reward": 0.9972727298736572,
      "reward_std": 0.09377779811620712,
      "rewards/accuracy_reward_step": 0.18359375,
      "rewards/final_brier_reward_step": 0.8187642693519592,
      "rewards/format_reward_step": 0.9921875,
      "step": 46
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8316536583006382,
      "aux_distill/mean_u": 0.3116091956334562,
      "aux_distill/n_active_tok": 92.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.70703125,
      "calib/ece": 0.14671875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.00349587702331261,
      "calib/mean_conf": 0.182890625,
      "calib/mu_c": 0.18574468085106383,
      "calib/mu_w": 0.18224880382775122,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0730078125,
      "calib/std_conf": 0.15703628334594963,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20154440154440156,
      "calib/step_q_c_n": 259.0,
      "calib/step_q_gap": -0.033763418755099306,
      "calib/step_q_w": 0.23530782029950087,
      "calib/step_q_w_n": 1202.0,
      "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": 456.0,
      "completions/max_terminated_length": 456.0,
      "completions/mean_length": 197.16796875,
      "completions/mean_terminated_length": 198.72047424316406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 117.0,
      "epoch": 0.050133333333333335,
      "grad_norm": 2.7363805770874023,
      "learning_rate": 4.277777777777778e-06,
      "loss": 0.0948,
      "num_tokens": 9914990.0,
      "reward": 1.0050468444824219,
      "reward_std": 0.0753011554479599,
      "rewards/accuracy_reward_step": 0.18359375,
      "rewards/final_brier_reward_step": 0.8264999985694885,
      "rewards/format_reward_step": 1.0,
      "step": 47
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8357707466930151,
      "aux_distill/mean_u": 0.3321379008617859,
      "aux_distill/n_active_tok": 97.625,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.16015625,
      "calib/ece": 0.148203125,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0078125,
      "calib/gap": 0.0021084137013634985,
      "calib/mean_conf": 0.19453125000000002,
      "calib/mu_c": 0.19612903225806452,
      "calib/mu_w": 0.19402061855670102,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.050273437500000004,
      "calib/std_conf": 0.16847365973183315,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23870712401055408,
      "calib/step_q_c_n": 379.0,
      "calib/step_q_gap": -0.0011092365904475932,
      "calib/step_q_w": 0.23981636060100167,
      "calib/step_q_w_n": 1198.0,
      "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": 471.0,
      "completions/max_terminated_length": 471.0,
      "completions/mean_length": 201.66796875,
      "completions/mean_terminated_length": 203.2559051513672,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.0512,
      "grad_norm": 1.6862789392471313,
      "learning_rate": 4.25e-06,
      "loss": 0.1036,
      "num_tokens": 10070305.0,
      "reward": 1.0143871307373047,
      "reward_std": 0.08464100956916809,
      "rewards/accuracy_reward_step": 0.2421875,
      "rewards/final_brier_reward_step": 0.7865867018699646,
      "rewards/format_reward_step": 1.0,
      "step": 48
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.843103788793087,
      "aux_distill/mean_u": 0.36999305802485677,
      "aux_distill/n_active_tok": 95.875,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.1796875,
      "calib/ece": 0.14652343750000002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": -0.016364903801468572,
      "calib/mean_conf": 0.1754296875,
      "calib/mu_c": 0.1624528301886792,
      "calib/mu_w": 0.17881773399014778,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.057460937499999996,
      "calib/std_conf": 0.1460649568467822,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2170418006430868,
      "calib/step_q_c_n": 311.0,
      "calib/step_q_gap": -0.0011414408990060299,
      "calib/step_q_w": 0.21818324154209284,
      "calib/step_q_w_n": 1271.0,
      "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": 449.0,
      "completions/max_terminated_length": 449.0,
      "completions/mean_length": 201.6953125,
      "completions/mean_terminated_length": 203.28346252441406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.05226666666666667,
      "grad_norm": 0.8191512823104858,
      "learning_rate": 4.222222222222223e-06,
      "loss": 0.0786,
      "num_tokens": 10226475.0,
      "reward": 1.0075775384902954,
      "reward_std": 0.07214483618736267,
      "rewards/accuracy_reward_step": 0.20703125,
      "rewards/final_brier_reward_step": 0.8081238269805908,
      "rewards/format_reward_step": 1.0,
      "step": 49
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8194594420492649,
      "aux_distill/mean_u": 0.37592474525324987,
      "aux_distill/n_active_tok": 114.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 6.9921875,
      "calib/ece": 0.1366796875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": -0.012047077492726793,
      "calib/mean_conf": 0.19410156250000002,
      "calib/mu_c": 0.18473684210526314,
      "calib/mu_w": 0.19678391959798994,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.054062500000000006,
      "calib/std_conf": 0.15603977557039295,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20697740112994353,
      "calib/step_q_c_n": 354.0,
      "calib/step_q_gap": -0.058934855137465963,
      "calib/step_q_w": 0.2659122562674095,
      "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": 543.0,
      "completions/max_terminated_length": 543.0,
      "completions/mean_length": 216.328125,
      "completions/mean_terminated_length": 218.031494140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 123.0,
      "epoch": 0.05333333333333334,
      "grad_norm": 1.4355971813201904,
      "learning_rate": 4.194444444444445e-06,
      "loss": 0.0926,
      "num_tokens": 10387215.0,
      "reward": 1.0024211406707764,
      "reward_std": 0.0937405452132225,
      "rewards/accuracy_reward_step": 0.22265625,
      "rewards/final_brier_reward_step": 0.7899988293647766,
      "rewards/format_reward_step": 0.9921875,
      "step": 50
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8286003153771162,
      "aux_distill/mean_u": 0.3526831447679025,
      "aux_distill/n_active_tok": 97.5,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 6.375,
      "calib/ece": 0.16179687499999998,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0013156440022111249,
      "calib/mean_conf": 0.213359375,
      "calib/mu_c": 0.21238805970149258,
      "calib/mu_w": 0.2137037037037037,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.056718749999999984,
      "calib/std_conf": 0.16138452403997533,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.29283333333333333,
      "calib/step_q_c_n": 420.0,
      "calib/step_q_gap": 0.014935313531353134,
      "calib/step_q_w": 0.2778980198019802,
      "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": 472.0,
      "completions/max_terminated_length": 472.0,
      "completions/mean_length": 195.0234375,
      "completions/mean_terminated_length": 196.55905151367188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.0544,
      "grad_norm": 1.832342505455017,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.0536,
      "num_tokens": 10546437.0,
      "reward": 1.0162572860717773,
      "reward_std": 0.09342889487743378,
      "rewards/accuracy_reward_step": 0.26171875,
      "rewards/final_brier_reward_step": 0.7747019529342651,
      "rewards/format_reward_step": 0.99609375,
      "step": 51
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7696451377123594,
      "aux_distill/mean_u": 0.3815194753008485,
      "aux_distill/n_active_tok": 107.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.2890625,
      "calib/ece": 0.12675781249999998,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.007892783054918667,
      "calib/mean_conf": 0.2121484375,
      "calib/mu_c": 0.2055813953488372,
      "calib/mu_w": 0.21347417840375588,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08546875000000001,
      "calib/std_conf": 0.16614562465291885,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28106761565836297,
      "calib/step_q_c_n": 281.0,
      "calib/step_q_gap": -0.003657936392110206,
      "calib/step_q_w": 0.2847255520504732,
      "calib/step_q_w_n": 1585.0,
      "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": 530.0,
      "completions/max_terminated_length": 530.0,
      "completions/mean_length": 205.12109375,
      "completions/mean_terminated_length": 206.73622131347656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.055466666666666664,
      "grad_norm": 0.7373725175857544,
      "learning_rate": 4.138888888888889e-06,
      "loss": 0.0685,
      "num_tokens": 10706900.0,
      "reward": 0.9982255697250366,
      "reward_std": 0.07990739494562149,
      "rewards/accuracy_reward_step": 0.16796875,
      "rewards/final_brier_reward_step": 0.8284823894500732,
      "rewards/format_reward_step": 1.0,
      "step": 52
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7436798084527254,
      "aux_distill/mean_u": 0.3827627011267692,
      "aux_distill/n_active_tok": 108.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.9921875,
      "calib/ece": 0.174921875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.01013415892672867,
      "calib/mean_conf": 0.203828125,
      "calib/mu_c": 0.19705882352941173,
      "calib/mu_w": 0.2071929824561404,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.023359374999999998,
      "calib/std_conf": 0.14288665248715282,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3082001655629139,
      "calib/step_q_c_n": 604.0,
      "calib/step_q_gap": -0.006960036797962976,
      "calib/step_q_w": 0.3151602023608769,
      "calib/step_q_w_n": 1186.0,
      "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": 558.0,
      "completions/max_terminated_length": 558.0,
      "completions/mean_length": 197.81640625,
      "completions/mean_terminated_length": 199.37400817871094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 126.0,
      "epoch": 0.05653333333333333,
      "grad_norm": 0.8482697010040283,
      "learning_rate": 4.111111111111111e-06,
      "loss": 0.0551,
      "num_tokens": 10863365.0,
      "reward": 1.034448504447937,
      "reward_std": 0.09234391897916794,
      "rewards/accuracy_reward_step": 0.33203125,
      "rewards/final_brier_reward_step": 0.7368656396865845,
      "rewards/format_reward_step": 1.0,
      "step": 53
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7325126752257347,
      "aux_distill/mean_u": 0.29912072543487983,
      "aux_distill/n_active_tok": 115.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.32421875,
      "calib/ece": 0.19642734375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.028345783132530084,
      "calib/mean_conf": 0.16661953125,
      "calib/mu_c": 0.185,
      "calib/mu_w": 0.15665421686746991,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.005742187499999999,
      "calib/std_conf": 0.14143963628445977,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.311476148409894,
      "calib/step_q_c_n": 566.0,
      "calib/step_q_gap": 0.023771335575669406,
      "calib/step_q_w": 0.2877048128342246,
      "calib/step_q_w_n": 1309.0,
      "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": 436.0,
      "completions/max_terminated_length": 436.0,
      "completions/mean_length": 201.3359375,
      "completions/mean_terminated_length": 202.9212646484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 129.0,
      "epoch": 0.0576,
      "grad_norm": 1.2834187746047974,
      "learning_rate": 4.083333333333334e-06,
      "loss": 0.0697,
      "num_tokens": 11021139.0,
      "reward": 1.041155457496643,
      "reward_std": 0.06632301211357117,
      "rewards/accuracy_reward_step": 0.3515625,
      "rewards/final_brier_reward_step": 0.7307484149932861,
      "rewards/format_reward_step": 1.0,
      "step": 54
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7187511073425412,
      "aux_distill/mean_u": 0.2917415166701098,
      "aux_distill/n_active_tok": 131.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.2890625,
      "calib/ece": 0.090625,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.01761941747572815,
      "calib/mean_conf": 0.13742187500000003,
      "calib/mu_c": 0.1516,
      "calib/mu_w": 0.13398058252427186,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.016367187499999995,
      "calib/std_conf": 0.12388720987851964,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.306264367816092,
      "calib/step_q_c_n": 348.0,
      "calib/step_q_gap": 0.008286295662766174,
      "calib/step_q_w": 0.29797807215332583,
      "calib/step_q_w_n": 1774.0,
      "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": 606.0,
      "completions/max_terminated_length": 606.0,
      "completions/mean_length": 223.890625,
      "completions/mean_terminated_length": 225.65354919433594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.058666666666666666,
      "grad_norm": 0.8885535001754761,
      "learning_rate": 4.055555555555556e-06,
      "loss": 0.075,
      "num_tokens": 11186279.0,
      "reward": 1.0124928951263428,
      "reward_std": 0.053454600274562836,
      "rewards/accuracy_reward_step": 0.1953125,
      "rewards/final_brier_reward_step": 0.8296734690666199,
      "rewards/format_reward_step": 1.0,
      "step": 55
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7426415104418993,
      "aux_distill/mean_u": 0.34381790715359856,
      "aux_distill/n_active_tok": 122.625,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 8.2421875,
      "calib/ece": 0.11554687500000001,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.021708273553323762,
      "calib/mean_conf": 0.1028125,
      "calib/mu_c": 0.12019607843137255,
      "calib/mu_w": 0.09848780487804878,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.009570312500000002,
      "calib/std_conf": 0.0958998297378572,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3300759493670886,
      "calib/step_q_c_n": 395.0,
      "calib/step_q_gap": 0.04942871904638885,
      "calib/step_q_w": 0.28064723032069977,
      "calib/step_q_w_n": 1715.0,
      "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": 525.0,
      "completions/max_terminated_length": 525.0,
      "completions/mean_length": 229.9921875,
      "completions/mean_terminated_length": 231.8031463623047,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.05973333333333333,
      "grad_norm": 1.1358617544174194,
      "learning_rate": 4.027777777777779e-06,
      "loss": 0.0676,
      "num_tokens": 11351997.0,
      "reward": 1.014061689376831,
      "reward_std": 0.03824378177523613,
      "rewards/accuracy_reward_step": 0.19921875,
      "rewards/final_brier_reward_step": 0.8289046883583069,
      "rewards/format_reward_step": 1.0,
      "step": 56
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7687519304454327,
      "aux_distill/mean_u": 0.3781342782712104,
      "aux_distill/n_active_tok": 113.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.078125,
      "calib/ece": 0.21359375000000003,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.003723165155390215,
      "calib/mean_conf": 0.06953125,
      "calib/mu_c": 0.06681159420289856,
      "calib/mu_w": 0.07053475935828878,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.006796875000000001,
      "calib/std_conf": 0.06692298949865809,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.30905154639175253,
      "calib/step_q_c_n": 485.0,
      "calib/step_q_gap": 0.058286587838625226,
      "calib/step_q_w": 0.2507649585531273,
      "calib/step_q_w_n": 1327.0,
      "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": 542.0,
      "completions/max_terminated_length": 542.0,
      "completions/mean_length": 199.1015625,
      "completions/mean_terminated_length": 200.66929626464844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 131.0,
      "epoch": 0.0608,
      "grad_norm": 0.972671389579773,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0829,
      "num_tokens": 11509759.0,
      "reward": 1.0133512020111084,
      "reward_std": 0.026614950969815254,
      "rewards/accuracy_reward_step": 0.26953125,
      "rewards/final_brier_reward_step": 0.7571711540222168,
      "rewards/format_reward_step": 1.0,
      "step": 57
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7665616478770971,
      "aux_distill/mean_u": 0.3432783958461975,
      "aux_distill/n_active_tok": 97.875,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.95703125,
      "calib/ece": 0.1603515625,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.01513244725346221,
      "calib/mean_conf": 0.046679687500000004,
      "calib/mu_c": 0.058679245283018856,
      "calib/mu_w": 0.043546798029556646,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.044224115592087804,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27230750853242325,
      "calib/step_q_c_n": 293.0,
      "calib/step_q_gap": 0.03393088515579992,
      "calib/step_q_w": 0.23837662337662333,
      "calib/step_q_w_n": 1232.0,
      "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": 597.0,
      "completions/max_terminated_length": 597.0,
      "completions/mean_length": 176.640625,
      "completions/mean_terminated_length": 178.031494140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 130.0,
      "epoch": 0.06186666666666667,
      "grad_norm": 0.8129911422729492,
      "learning_rate": 3.972222222222223e-06,
      "loss": 0.0853,
      "num_tokens": 11661299.0,
      "reward": 1.0100810527801514,
      "reward_std": 0.014807004481554031,
      "rewards/accuracy_reward_step": 0.20703125,
      "rewards/final_brier_reward_step": 0.8131308555603027,
      "rewards/format_reward_step": 1.0,
      "step": 58
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7977718394249678,
      "aux_distill/mean_u": 0.3172356907549245,
      "aux_distill/n_active_tok": 85.625,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.47265625,
      "calib/ece": 0.2777734375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0054064327485380155,
      "calib/mean_conf": 0.019882812500000003,
      "calib/mu_c": 0.023684210526315794,
      "calib/mu_w": 0.018277777777777778,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.000390625,
      "calib/std_conf": 0.019970337806102422,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26880778588807785,
      "calib/step_q_c_n": 411.0,
      "calib/step_q_gap": 0.004716876797168701,
      "calib/step_q_w": 0.26409090909090915,
      "calib/step_q_w_n": 990.0,
      "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": 321.0,
      "completions/max_terminated_length": 321.0,
      "completions/mean_length": 167.96875,
      "completions/mean_terminated_length": 169.2913360595703,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.06293333333333333,
      "grad_norm": 1.336860179901123,
      "learning_rate": 3.944444444444445e-06,
      "loss": 0.0646,
      "num_tokens": 11810547.0,
      "reward": 1.0066341161727905,
      "reward_std": 0.01024669036269188,
      "rewards/accuracy_reward_step": 0.296875,
      "rewards/final_brier_reward_step": 0.7163933515548706,
      "rewards/format_reward_step": 1.0,
      "step": 59
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8566927462816238,
      "aux_distill/mean_u": 0.34854639763211626,
      "aux_distill/n_active_tok": 82.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.1953125,
      "calib/ece": 0.18334375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0006563106796116502,
      "calib/mean_conf": 0.015328125000000001,
      "calib/mu_c": 0.0148,
      "calib/mu_w": 0.01545631067961165,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0016796875,
      "calib/std_conf": 0.03026371315592941,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2923557692307692,
      "calib/step_q_c_n": 260.0,
      "calib/step_q_gap": 0.004906329978432766,
      "calib/step_q_w": 0.28744943925233646,
      "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": 296.0,
      "completions/max_terminated_length": 296.0,
      "completions/mean_length": 159.59765625,
      "completions/mean_terminated_length": 160.8543243408203,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 129.0,
      "epoch": 0.064,
      "grad_norm": 1.2972129583358765,
      "learning_rate": 3.916666666666667e-06,
      "loss": 0.0789,
      "num_tokens": 11960260.0,
      "reward": 1.0023151636123657,
      "reward_std": 0.0054235332645475864,
      "rewards/accuracy_reward_step": 0.1953125,
      "rewards/final_brier_reward_step": 0.8093178868293762,
      "rewards/format_reward_step": 1.0,
      "step": 60
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8228755127638578,
      "aux_distill/mean_u": 0.3034650743786119,
      "aux_distill/n_active_tok": 81.0,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.015625,
      "calib/ece": 0.315859375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0015278945892907205,
      "calib/mean_conf": 0.004453125,
      "calib/mu_c": 0.003414634146341464,
      "calib/mu_w": 0.004942528735632184,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.008082445653041843,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26985330073349634,
      "calib/step_q_c_n": 409.0,
      "calib/step_q_gap": -0.035374813552218,
      "calib/step_q_w": 0.30522811428571434,
      "calib/step_q_w_n": 875.0,
      "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": 232.0,
      "completions/max_terminated_length": 232.0,
      "completions/mean_length": 160.98046875,
      "completions/mean_terminated_length": 162.24803161621094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 122.0,
      "epoch": 0.06506666666666666,
      "grad_norm": 1.093929648399353,
      "learning_rate": 3.88888888888889e-06,
      "loss": 0.0848,
      "num_tokens": 12105535.0,
      "reward": 1.0010511875152588,
      "reward_std": 0.0020470181480050087,
      "rewards/accuracy_reward_step": 0.3203125,
      "rewards/final_brier_reward_step": 0.6817898154258728,
      "rewards/format_reward_step": 1.0,
      "step": 61
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8450021678581834,
      "aux_distill/mean_u": 0.306607903342974,
      "aux_distill/n_active_tok": 80.875,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.09375,
      "calib/ece": 0.29804375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0013003845316694475,
      "calib/mean_conf": 0.0027375,
      "calib/mu_c": 0.0036467532467532467,
      "calib/mu_w": 0.002346368715083799,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.005092910206846376,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2560384020618557,
      "calib/step_q_c_n": 388.0,
      "calib/step_q_gap": -0.005121095754738181,
      "calib/step_q_w": 0.26115949781659387,
      "calib/step_q_w_n": 916.0,
      "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": 330.0,
      "completions/max_terminated_length": 330.0,
      "completions/mean_length": 175.73828125,
      "completions/mean_terminated_length": 177.12203979492188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.06613333333333334,
      "grad_norm": 0.7330139875411987,
      "learning_rate": 3.861111111111112e-06,
      "loss": 0.0822,
      "num_tokens": 12257604.0,
      "reward": 1.0010801553726196,
      "reward_std": 0.002245472278445959,
      "rewards/accuracy_reward_step": 0.30078125,
      "rewards/final_brier_reward_step": 0.7013790607452393,
      "rewards/format_reward_step": 1.0,
      "step": 62
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8771042171865702,
      "aux_distill/mean_u": 0.3877518793078786,
      "aux_distill/n_active_tok": 80.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.11328125,
      "calib/ece": 0.27490150170769134,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0025972584145657392,
      "calib/mean_conf": 0.006348498292308652,
      "calib/mu_c": 0.008215277777777778,
      "calib/mu_w": 0.005618019363212039,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.00949263115705469,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.263,
      "calib/step_q_c_n": 364.0,
      "calib/step_q_gap": -0.01867301587301584,
      "calib/step_q_w": 0.28167301587301585,
      "calib/step_q_w_n": 945.0,
      "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": 366.0,
      "completions/max_terminated_length": 366.0,
      "completions/mean_length": 173.4453125,
      "completions/mean_terminated_length": 174.81101989746094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 117.0,
      "epoch": 0.0672,
      "grad_norm": 0.8441920280456543,
      "learning_rate": 3.833333333333334e-06,
      "loss": 0.0789,
      "num_tokens": 12410646.0,
      "reward": 1.0022454261779785,
      "reward_std": 0.004115952178835869,
      "rewards/accuracy_reward_step": 0.28125,
      "rewards/final_brier_reward_step": 0.7232406139373779,
      "rewards/format_reward_step": 1.0,
      "step": 63
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7863265555351973,
      "aux_distill/mean_u": 0.2677313118227581,
      "aux_distill/n_active_tok": 82.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.19921875,
      "calib/ece": 0.236480078125,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0026867901731776125,
      "calib/mean_conf": 0.017426171875,
      "calib/mu_c": 0.019430769230769234,
      "calib/mu_w": 0.01674397905759162,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.017271152638734306,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3150280346820809,
      "calib/step_q_c_n": 346.0,
      "calib/step_q_gap": -0.03254881810979726,
      "calib/step_q_w": 0.34757685279187817,
      "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": 349.0,
      "completions/max_terminated_length": 349.0,
      "completions/mean_length": 183.87890625,
      "completions/mean_terminated_length": 185.32676696777344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.06826666666666667,
      "grad_norm": 0.8380021452903748,
      "learning_rate": 3.8055555555555556e-06,
      "loss": 0.0648,
      "num_tokens": 12561495.0,
      "reward": 1.0007264614105225,
      "reward_std": 0.017975719645619392,
      "rewards/accuracy_reward_step": 0.25390625,
      "rewards/final_brier_reward_step": 0.7514531016349792,
      "rewards/format_reward_step": 0.99609375,
      "step": 64
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8005730956792831,
      "aux_distill/mean_u": 0.2784840038174951,
      "aux_distill/n_active_tok": 87.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.53125,
      "calib/ece": 0.23990404668604787,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.953125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.003159976684758207,
      "calib/mean_conf": 0.02392407831395213,
      "calib/mu_c": 0.02625702985074627,
      "calib/mu_w": 0.023097053165988063,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0010546875,
      "calib/std_conf": 0.025948336423693166,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.32175229110512127,
      "calib/step_q_c_n": 371.0,
      "calib/step_q_gap": -0.0038337376030127546,
      "calib/step_q_w": 0.325586028708134,
      "calib/step_q_w_n": 1045.0,
      "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": 461.0,
      "completions/max_terminated_length": 461.0,
      "completions/mean_length": 199.19140625,
      "completions/mean_terminated_length": 200.7598419189453,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 118.0,
      "epoch": 0.06933333333333333,
      "grad_norm": 2.7181830406188965,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.0959,
      "num_tokens": 12717512.0,
      "reward": 0.9650967121124268,
      "reward_std": 0.10888336598873138,
      "rewards/accuracy_reward_step": 0.26171875,
      "rewards/final_brier_reward_step": 0.7153497338294983,
      "rewards/format_reward_step": 0.953125,
      "step": 65
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8654420170933008,
      "aux_distill/mean_u": 0.3045119087832663,
      "aux_distill/n_active_tok": 92.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.90234375,
      "calib/ece": 0.24938525315024132,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0068889948120064645,
      "calib/mean_conf": 0.0288178718497587,
      "calib/mu_c": 0.03382315714285714,
      "calib/mu_w": 0.026934162330850678,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0023828125,
      "calib/std_conf": 0.026407395186875748,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3251741293532339,
      "calib/step_q_c_n": 402.0,
      "calib/step_q_gap": 0.02805050446594809,
      "calib/step_q_w": 0.29712362488728583,
      "calib/step_q_w_n": 1109.0,
      "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": 526.0,
      "completions/max_terminated_length": 526.0,
      "completions/mean_length": 218.2890625,
      "completions/mean_terminated_length": 220.00787353515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 97.0,
      "epoch": 0.0704,
      "grad_norm": 0.7378050684928894,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0816,
      "num_tokens": 12879746.0,
      "reward": 1.0084846019744873,
      "reward_std": 0.01123389508575201,
      "rewards/accuracy_reward_step": 0.2734375,
      "rewards/final_brier_reward_step": 0.7435317039489746,
      "rewards/format_reward_step": 1.0,
      "step": 66
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8232519458979368,
      "aux_distill/mean_u": 0.2802565965867831,
      "aux_distill/n_active_tok": 99.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.30859375,
      "calib/ece": 0.35173385302248666,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.004611640985949166,
      "calib/mean_conf": 0.03156026462457222,
      "calib/mu_c": 0.028684770833333335,
      "calib/mu_w": 0.0332964118192825,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.003411764705882353,
      "calib/std_conf": 0.046358818478338545,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.3011433172302737,
      "calib/step_q_c_n": 621.0,
      "calib/step_q_gap": -0.003653161642965752,
      "calib/step_q_w": 0.30479647887323946,
      "calib/step_q_w_n": 994.0,
      "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": 595.0,
      "completions/max_terminated_length": 595.0,
      "completions/mean_length": 238.3671875,
      "completions/mean_terminated_length": 240.2440948486328,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 34.0,
      "epoch": 0.07146666666666666,
      "grad_norm": 0.6210587024688721,
      "learning_rate": 3.7222222222222225e-06,
      "loss": 0.1009,
      "num_tokens": 13045776.0,
      "reward": 1.0013848543167114,
      "reward_std": 0.038151249289512634,
      "rewards/accuracy_reward_step": 0.375,
      "rewards/final_brier_reward_step": 0.6355822086334229,
      "rewards/format_reward_step": 0.9921875,
      "step": 67
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8016217239201069,
      "aux_distill/mean_u": 0.2958295942596198,
      "aux_distill/n_active_tok": 111.125,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 7.79296875,
      "calib/ece": 0.2818664538399066,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.007091782746271546,
      "calib/mean_conf": 0.032306774506550104,
      "calib/mu_c": 0.037220765700817005,
      "calib/mu_w": 0.03012898295454546,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.003543307086614173,
      "calib/std_conf": 0.03559904487719555,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23322369477911645,
      "calib/step_q_c_n": 498.0,
      "calib/step_q_gap": -0.06897630522088358,
      "calib/step_q_w": 0.3022,
      "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": 0.0078125,
      "completions/max_length": 1821.0,
      "completions/max_terminated_length": 1821.0,
      "completions/mean_length": 261.87890625,
      "completions/mean_terminated_length": 263.9409484863281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 146.0,
      "epoch": 0.07253333333333334,
      "grad_norm": 0.5835959911346436,
      "learning_rate": 3.694444444444445e-06,
      "loss": 0.0625,
      "num_tokens": 13216905.0,
      "reward": 1.0023818016052246,
      "reward_std": 0.03678155690431595,
      "rewards/accuracy_reward_step": 0.3046875,
      "rewards/final_brier_reward_step": 0.7078884840011597,
      "rewards/format_reward_step": 0.9921875,
      "step": 68
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7951194914057851,
      "aux_distill/mean_u": 0.26222809183075535,
      "aux_distill/n_active_tok": 111.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.0390625,
      "calib/ece": 0.2928776784204442,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.006307577900800849,
      "calib/mean_conf": 0.031341071579555756,
      "calib/mu_c": 0.03560361445783132,
      "calib/mu_w": 0.02929603655703047,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.02452726551621772,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25492882562277575,
      "calib/step_q_c_n": 562.0,
      "calib/step_q_gap": 0.04194164820342089,
      "calib/step_q_w": 0.21298717741935486,
      "calib/step_q_w_n": 1240.0,
      "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": 1160.0,
      "completions/max_terminated_length": 1160.0,
      "completions/mean_length": 267.80859375,
      "completions/mean_terminated_length": 269.9173278808594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 142.0,
      "epoch": 0.0736,
      "grad_norm": 0.6954154968261719,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.0831,
      "num_tokens": 13389960.0,
      "reward": 1.0107513666152954,
      "reward_std": 0.015205792151391506,
      "rewards/accuracy_reward_step": 0.32421875,
      "rewards/final_brier_reward_step": 0.6972841024398804,
      "rewards/format_reward_step": 1.0,
      "step": 69
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8393227551132441,
      "aux_distill/mean_u": 0.31709665386517005,
      "aux_distill/n_active_tok": 134.5,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 10.6953125,
      "calib/ece": 0.319916,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0001261223344556639,
      "calib/mean_conf": 0.038484000000000004,
      "calib/mu_c": 0.03840227272727273,
      "calib/mu_w": 0.03852839506172839,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0032,
      "calib/std_conf": 0.056075315638879825,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25926243567753,
      "calib/step_q_c_n": 583.0,
      "calib/step_q_gap": 0.08081201668311658,
      "calib/step_q_w": 0.1784504189944134,
      "calib/step_q_w_n": 2148.0,
      "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": 3059.0,
      "completions/max_terminated_length": 3059.0,
      "completions/mean_length": 338.50390625,
      "completions/mean_terminated_length": 345.24700927734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.07466666666666667,
      "grad_norm": 0.8931414484977722,
      "learning_rate": 3.638888888888889e-06,
      "loss": 0.1097,
      "num_tokens": 13583609.0,
      "reward": 0.9836143255233765,
      "reward_std": 0.08913453668355942,
      "rewards/accuracy_reward_step": 0.34375,
      "rewards/final_brier_reward_step": 0.6508224010467529,
      "rewards/format_reward_step": 0.97265625,
      "step": 70
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8577479831874371,
      "aux_distill/mean_u": 0.36445067461176467,
      "aux_distill/n_active_tok": 126.25,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.2421875,
      "calib/ece": 0.2426550980392157,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0017006375227686665,
      "calib/mean_conf": 0.041501764705882356,
      "calib/mu_c": 0.04272222222222222,
      "calib/mu_w": 0.04102158469945355,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0009019607843137256,
      "calib/std_conf": 0.03134084547240277,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2365654648956357,
      "calib/step_q_c_n": 527.0,
      "calib/step_q_gap": 0.030497240006185283,
      "calib/step_q_w": 0.20606822488945042,
      "calib/step_q_w_n": 1583.0,
      "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": 1184.0,
      "completions/max_terminated_length": 1184.0,
      "completions/mean_length": 323.2265625,
      "completions/mean_terminated_length": 325.77166748046875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.07573333333333333,
      "grad_norm": 0.8447468876838684,
      "learning_rate": 3.6111111111111115e-06,
      "loss": 0.0698,
      "num_tokens": 13770763.0,
      "reward": 1.0067623853683472,
      "reward_std": 0.027596548199653625,
      "rewards/accuracy_reward_step": 0.28125,
      "rewards/final_brier_reward_step": 0.7361809015274048,
      "rewards/format_reward_step": 0.99609375,
      "step": 71
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8418899774551392,
      "aux_distill/mean_u": 0.2976037554519489,
      "aux_distill/n_active_tok": 126.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.8046875,
      "calib/ece": 0.2510117647058823,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0034823055099298186,
      "calib/mean_conf": 0.04420392156862745,
      "calib/mu_c": 0.046675675675675675,
      "calib/mu_w": 0.043193370165745856,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0025098039215686275,
      "calib/std_conf": 0.038148254454884734,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3083068783068783,
      "calib/step_q_c_n": 567.0,
      "calib/step_q_gap": 0.0850304704823377,
      "calib/step_q_w": 0.2232764078245406,
      "calib/step_q_w_n": 1687.0,
      "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": 947.0,
      "completions/max_terminated_length": 947.0,
      "completions/mean_length": 330.25,
      "completions/mean_terminated_length": 332.85040283203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 154.0,
      "epoch": 0.0768,
      "grad_norm": 0.6339616775512695,
      "learning_rate": 3.5833333333333335e-06,
      "loss": 0.0704,
      "num_tokens": 13959715.0,
      "reward": 1.0078879594802856,
      "reward_std": 0.030957698822021484,
      "rewards/accuracy_reward_step": 0.2890625,
      "rewards/final_brier_reward_step": 0.7306196689605713,
      "rewards/format_reward_step": 0.99609375,
      "step": 72
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8132553435862064,
      "aux_distill/mean_u": 0.2690545730196228,
      "aux_distill/n_active_tok": 137.875,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 10.86328125,
      "calib/ece": 0.2577836653386454,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.004715340909090905,
      "calib/mean_conf": 0.049706374501992036,
      "calib/mu_c": 0.046400000000000004,
      "calib/mu_w": 0.05111534090909091,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.004342629482071713,
      "calib/std_conf": 0.047360112733088204,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24456730769230767,
      "calib/step_q_c_n": 624.0,
      "calib/step_q_gap": 0.02669062711743514,
      "calib/step_q_w": 0.21787668057487253,
      "calib/step_q_w_n": 2157.0,
      "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": 0.0234375,
      "completions/max_length": 1120.0,
      "completions/max_terminated_length": 1120.0,
      "completions/mean_length": 355.0390625,
      "completions/mean_terminated_length": 363.5600280761719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 161.0,
      "epoch": 0.07786666666666667,
      "grad_norm": 0.5692130923271179,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0473,
      "num_tokens": 14157637.0,
      "reward": 0.9917516708374023,
      "reward_std": 0.06734663248062134,
      "rewards/accuracy_reward_step": 0.29296875,
      "rewards/final_brier_reward_step": 0.7100658416748047,
      "rewards/format_reward_step": 0.98046875,
      "step": 73
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8305688640102744,
      "aux_distill/mean_u": 0.28209760097768316,
      "aux_distill/n_active_tok": 128.0,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.8125,
      "calib/ece": 0.22303107569721117,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.007932596685082872,
      "calib/mean_conf": 0.0592796812749004,
      "calib/mu_c": 0.065,
      "calib/mu_w": 0.05706740331491713,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0017131474103585656,
      "calib/std_conf": 0.05748070891965741,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.29422413793103447,
      "calib/step_q_c_n": 580.0,
      "calib/step_q_gap": 0.07312682944242166,
      "calib/step_q_w": 0.2210973084886128,
      "calib/step_q_w_n": 1932.0,
      "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": 1695.0,
      "completions/max_terminated_length": 1695.0,
      "completions/mean_length": 344.22265625,
      "completions/mean_terminated_length": 349.6865234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.07893333333333333,
      "grad_norm": 0.7870764136314392,
      "learning_rate": 3.5277777777777784e-06,
      "loss": 0.0325,
      "num_tokens": 14349686.0,
      "reward": 0.9948996901512146,
      "reward_std": 0.08115790784358978,
      "rewards/accuracy_reward_step": 0.2734375,
      "rewards/final_brier_reward_step": 0.735893189907074,
      "rewards/format_reward_step": 0.98046875,
      "step": 74
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8252325486391783,
      "aux_distill/mean_u": 0.2802878247564068,
      "aux_distill/n_active_tok": 136.25,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 11.30078125,
      "calib/ece": 0.43045702811244974,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0024954674586776804,
      "calib/mean_conf": 0.060948594377510046,
      "calib/mu_c": 0.06223140495867769,
      "calib/mu_w": 0.05973593750000001,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.002730923694779116,
      "calib/std_conf": 0.04042357964353128,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23575879396984922,
      "calib/step_q_c_n": 995.0,
      "calib/step_q_gap": -0.031047844597063334,
      "calib/step_q_w": 0.26680663856691256,
      "calib/step_q_w_n": 1898.0,
      "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": 2618.0,
      "completions/max_terminated_length": 2618.0,
      "completions/mean_length": 366.7421875,
      "completions/mean_terminated_length": 377.05218505859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 136.0,
      "epoch": 0.08,
      "grad_norm": 0.5298400521278381,
      "learning_rate": 3.5e-06,
      "loss": 0.0148,
      "num_tokens": 14548324.0,
      "reward": 0.9994690418243408,
      "reward_std": 0.10342110693454742,
      "rewards/accuracy_reward_step": 0.47265625,
      "rewards/final_brier_reward_step": 0.5536255836486816,
      "rewards/format_reward_step": 0.97265625,
      "step": 75
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8440918866544962,
      "aux_distill/mean_u": 0.3047469412064998,
      "aux_distill/n_active_tok": 144.5,
      "calib/answer_extract_rate": 0.93359375,
      "calib/avg_num_step_conf": 14.56640625,
      "calib/ece": 0.321673640167364,
      "calib/final_conf_rate": 0.93359375,
      "calib/format_rate": 0.93359375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.008939708939708968,
      "calib/mean_conf": 0.059079497907949786,
      "calib/mu_c": 0.06461538461538463,
      "calib/mu_w": 0.05567567567567566,
      "calib/nonempty_final_conf_rate": 0.93359375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.03790527113297334,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24815864022662887,
      "calib/step_q_c_n": 706.0,
      "calib/step_q_gap": 0.03367302990575552,
      "calib/step_q_w": 0.21448561032087335,
      "calib/step_q_w_n": 3023.0,
      "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": 975.0,
      "completions/max_terminated_length": 975.0,
      "completions/mean_length": 358.36328125,
      "completions/mean_terminated_length": 383.8535461425781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 159.0,
      "epoch": 0.08106666666666666,
      "grad_norm": 1.081568717956543,
      "learning_rate": 3.4722222222222224e-06,
      "loss": -0.0567,
      "num_tokens": 14743121.0,
      "reward": 0.9542624950408936,
      "reward_std": 0.16976533830165863,
      "rewards/accuracy_reward_step": 0.35546875,
      "rewards/final_brier_reward_step": 0.6194624900817871,
      "rewards/format_reward_step": 0.93359375,
      "step": 76
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7728911470621824,
      "aux_distill/mean_u": 0.25746192158660536,
      "aux_distill/n_active_tok": 178.375,
      "calib/answer_extract_rate": 0.8828125,
      "calib/avg_num_step_conf": 19.1015625,
      "calib/ece": 0.30475770925110135,
      "calib/final_conf_rate": 0.88671875,
      "calib/format_rate": 0.8828125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0009261083743842269,
      "calib/mean_conf": 0.0785022026431718,
      "calib/mu_c": 0.07793103448275862,
      "calib/mu_w": 0.07885714285714285,
      "calib/nonempty_final_conf_rate": 0.88671875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.0533086246916388,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2515760869565218,
      "calib/step_q_c_n": 736.0,
      "calib/step_q_gap": 0.07540889870423481,
      "calib/step_q_w": 0.17616718825228697,
      "calib/step_q_w_n": 4154.0,
      "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": 0.11328125,
      "completions/max_length": 2845.0,
      "completions/max_terminated_length": 2845.0,
      "completions/mean_length": 381.4921875,
      "completions/mean_terminated_length": 430.22906494140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 176.0,
      "epoch": 0.08213333333333334,
      "grad_norm": 0.4085995554924011,
      "learning_rate": 3.444444444444445e-06,
      "loss": -0.135,
      "num_tokens": 14945447.0,
      "reward": 0.9053117036819458,
      "reward_std": 0.27409547567367554,
      "rewards/accuracy_reward_step": 0.33984375,
      "rewards/final_brier_reward_step": 0.5879671573638916,
      "rewards/format_reward_step": 0.8828125,
      "step": 77
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7340011494234204,
      "aux_distill/mean_u": 0.2834807404634447,
      "aux_distill/n_active_tok": 208.375,
      "calib/answer_extract_rate": 0.8203125,
      "calib/avg_num_step_conf": 27.3125,
      "calib/ece": 0.3156872037914692,
      "calib/final_conf_rate": 0.82421875,
      "calib/format_rate": 0.8203125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.02651637801204819,
      "calib/mean_conf": 0.0822274881516588,
      "calib/mu_c": 0.0983132530120482,
      "calib/mu_w": 0.07179687500000001,
      "calib/nonempty_final_conf_rate": 0.82421875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.002274881516587678,
      "calib/std_conf": 0.07320856010629861,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2361008230452675,
      "calib/step_q_c_n": 972.0,
      "calib/step_q_gap": 0.053976304772842265,
      "calib/step_q_w": 0.18212451827242523,
      "calib/step_q_w_n": 6020.0,
      "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": 0.17578125,
      "completions/max_length": 2593.0,
      "completions/max_terminated_length": 2593.0,
      "completions/mean_length": 426.140625,
      "completions/mean_terminated_length": 517.0237426757812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 174.0,
      "epoch": 0.0832,
      "grad_norm": 0.7913491129875183,
      "learning_rate": 3.416666666666667e-06,
      "loss": -0.2688,
      "num_tokens": 15162563.0,
      "reward": 0.8471955060958862,
      "reward_std": 0.3558165431022644,
      "rewards/accuracy_reward_step": 0.32421875,
      "rewards/final_brier_reward_step": 0.5498597621917725,
      "rewards/format_reward_step": 0.8203125,
      "step": 78
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8018683372065425,
      "aux_distill/mean_u": 0.3224408325214972,
      "aux_distill/n_active_tok": 233.5,
      "calib/answer_extract_rate": 0.79296875,
      "calib/avg_num_step_conf": 29.0078125,
      "calib/ece": 0.27322205882352935,
      "calib/final_conf_rate": 0.796875,
      "calib/format_rate": 0.7890625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.003921666840949509,
      "calib/mean_conf": 0.09638578431372549,
      "calib/mu_c": 0.09890410958904111,
      "calib/mu_w": 0.0949824427480916,
      "calib/nonempty_final_conf_rate": 0.796875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0058823529411764705,
      "calib/std_conf": 0.08280033060860721,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2973452380952381,
      "calib/step_q_c_n": 840.0,
      "calib/step_q_gap": 0.11331534134455484,
      "calib/step_q_w": 0.18402989675068326,
      "calib/step_q_w_n": 6586.0,
      "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": 0.203125,
      "completions/max_length": 2227.0,
      "completions/max_terminated_length": 2227.0,
      "completions/mean_length": 437.3046875,
      "completions/mean_terminated_length": 548.7745361328125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 34.0,
      "epoch": 0.08426666666666667,
      "grad_norm": 0.5817393064498901,
      "learning_rate": 3.3888888888888893e-06,
      "loss": -0.3343,
      "num_tokens": 15380889.0,
      "reward": 0.811467170715332,
      "reward_std": 0.3905557692050934,
      "rewards/accuracy_reward_step": 0.28515625,
      "rewards/final_brier_reward_step": 0.5487155914306641,
      "rewards/format_reward_step": 0.7890625,
      "step": 79
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.735305467620492,
      "aux_distill/mean_u": 0.2073347987640853,
      "aux_distill/n_active_tok": 170.125,
      "calib/answer_extract_rate": 0.86328125,
      "calib/avg_num_step_conf": 22.1328125,
      "calib/ece": 0.31369369369369365,
      "calib/final_conf_rate": 0.8671875,
      "calib/format_rate": 0.86328125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.010581327069876675,
      "calib/mean_conf": 0.1300900900900901,
      "calib/mu_c": 0.12384615384615384,
      "calib/mu_w": 0.13442748091603052,
      "calib/nonempty_final_conf_rate": 0.8671875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.016936936936936934,
      "calib/std_conf": 0.11121763013634592,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2721401673640168,
      "calib/step_q_c_n": 956.0,
      "calib/step_q_gap": 0.06341783190754122,
      "calib/step_q_w": 0.2087223354564756,
      "calib/step_q_w_n": 4710.0,
      "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": 2560.0,
      "completions/max_terminated_length": 2560.0,
      "completions/mean_length": 389.6640625,
      "completions/mean_terminated_length": 449.34234619140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 178.0,
      "epoch": 0.08533333333333333,
      "grad_norm": 2.6019041538238525,
      "learning_rate": 3.3611111111111117e-06,
      "loss": -0.2047,
      "num_tokens": 15582803.0,
      "reward": 0.8946083784103394,
      "reward_std": 0.3351396322250366,
      "rewards/accuracy_reward_step": 0.35546875,
      "rewards/final_brier_reward_step": 0.5704667568206787,
      "rewards/format_reward_step": 0.86328125,
      "step": 80
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7553578931838274,
      "aux_distill/mean_u": 0.20833043231980844,
      "aux_distill/n_active_tok": 177.5,
      "calib/answer_extract_rate": 0.91015625,
      "calib/avg_num_step_conf": 17.73046875,
      "calib/ece": 0.28060085836909876,
      "calib/final_conf_rate": 0.91015625,
      "calib/format_rate": 0.91015625,
      "calib/frac_conf_gt_0.9": 0.008583690987124463,
      "calib/gap": 0.012084093673965901,
      "calib/mean_conf": 0.15424892703862664,
      "calib/mu_c": 0.16135416666666666,
      "calib/mu_w": 0.14927007299270076,
      "calib/nonempty_final_conf_rate": 0.91015625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.011416309012875537,
      "calib/std_conf": 0.14436621236072153,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27140337922403,
      "calib/step_q_c_n": 799.0,
      "calib/step_q_gap": 0.019378860507452467,
      "calib/step_q_w": 0.25202451871657755,
      "calib/step_q_w_n": 3740.0,
      "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": 1838.0,
      "completions/max_terminated_length": 1838.0,
      "completions/mean_length": 385.62890625,
      "completions/mean_terminated_length": 423.6952819824219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 178.0,
      "epoch": 0.0864,
      "grad_norm": 0.4589557945728302,
      "learning_rate": 3.3333333333333333e-06,
      "loss": -0.0633,
      "num_tokens": 15787772.0,
      "reward": 0.9503519535064697,
      "reward_std": 0.24027007818222046,
      "rewards/accuracy_reward_step": 0.375,
      "rewards/final_brier_reward_step": 0.6155476570129395,
      "rewards/format_reward_step": 0.91015625,
      "step": 81
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7371373372152448,
      "aux_distill/mean_u": 0.21507642836182117,
      "aux_distill/n_active_tok": 143.75,
      "calib/answer_extract_rate": 0.9453125,
      "calib/avg_num_step_conf": 13.14453125,
      "calib/ece": 0.30004115226337447,
      "calib/final_conf_rate": 0.94921875,
      "calib/format_rate": 0.9453125,
      "calib/frac_conf_gt_0.9": 0.012345679012345678,
      "calib/gap": 0.048083049693669166,
      "calib/mean_conf": 0.18020576131687246,
      "calib/mu_c": 0.205929203539823,
      "calib/mu_w": 0.15784615384615383,
      "calib/nonempty_final_conf_rate": 0.94921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.007613168724279836,
      "calib/std_conf": 0.16698105794269957,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3053854625550661,
      "calib/step_q_c_n": 908.0,
      "calib/step_q_gap": -0.011894920025316502,
      "calib/step_q_w": 0.3172803825803826,
      "calib/step_q_w_n": 2457.0,
      "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": 1641.0,
      "completions/max_terminated_length": 1641.0,
      "completions/mean_length": 369.72265625,
      "completions/mean_terminated_length": 389.5020446777344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 172.0,
      "epoch": 0.08746666666666666,
      "grad_norm": 2.2132375240325928,
      "learning_rate": 3.3055555555555558e-06,
      "loss": -0.0156,
      "num_tokens": 15987973.0,
      "reward": 1.007659673690796,
      "reward_std": 0.20080611109733582,
      "rewards/accuracy_reward_step": 0.44140625,
      "rewards/final_brier_reward_step": 0.6286004185676575,
      "rewards/format_reward_step": 0.9453125,
      "step": 82
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.794050301425159,
      "aux_distill/mean_u": 0.25047651542267396,
      "aux_distill/n_active_tok": 146.5,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 11.8515625,
      "calib/ece": 0.20485714285714285,
      "calib/final_conf_rate": 0.95703125,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.00816326530612245,
      "calib/gap": 0.07423508771929829,
      "calib/mean_conf": 0.1979183673469388,
      "calib/mu_c": 0.24336842105263162,
      "calib/mu_w": 0.16913333333333333,
      "calib/nonempty_final_conf_rate": 0.95703125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.007510204081632653,
      "calib/std_conf": 0.18803862355472306,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.306963216957606,
      "calib/step_q_c_n": 802.0,
      "calib/step_q_gap": 0.035586156025706306,
      "calib/step_q_w": 0.27137706093189967,
      "calib/step_q_w_n": 2232.0,
      "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": 0.04296875,
      "completions/max_length": 1898.0,
      "completions/max_terminated_length": 1898.0,
      "completions/mean_length": 400.6875,
      "completions/mean_terminated_length": 418.6775207519531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 160.0,
      "epoch": 0.08853333333333334,
      "grad_norm": 0.5713662505149841,
      "learning_rate": 3.277777777777778e-06,
      "loss": -0.0261,
      "num_tokens": 16197813.0,
      "reward": 1.0116798877716064,
      "reward_std": 0.18111923336982727,
      "rewards/accuracy_reward_step": 0.37109375,
      "rewards/final_brier_reward_step": 0.6952347755432129,
      "rewards/format_reward_step": 0.95703125,
      "step": 83
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8229641187936068,
      "aux_distill/mean_u": 0.2732668140313402,
      "aux_distill/n_active_tok": 147.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.39453125,
      "calib/ece": 0.2481746031746032,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.011904761904761904,
      "calib/gap": 0.02588289292505408,
      "calib/mean_conf": 0.17785714285714288,
      "calib/mu_c": 0.19336633663366334,
      "calib/mu_w": 0.16748344370860926,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.012619047619047615,
      "calib/std_conf": 0.1744565582113434,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25671852825229957,
      "calib/step_q_c_n": 761.0,
      "calib/step_q_gap": 0.017268345770547727,
      "calib/step_q_w": 0.23945018248175184,
      "calib/step_q_w_n": 1644.0,
      "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": 900.0,
      "completions/max_terminated_length": 900.0,
      "completions/mean_length": 352.953125,
      "completions/mean_terminated_length": 359.98406982421875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 154.0,
      "epoch": 0.0896,
      "grad_norm": 1.077708125114441,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.0584,
      "num_tokens": 16394089.0,
      "reward": 1.0301148891448975,
      "reward_std": 0.13591772317886353,
      "rewards/accuracy_reward_step": 0.39453125,
      "rewards/final_brier_reward_step": 0.6813234090805054,
      "rewards/format_reward_step": 0.984375,
      "step": 84
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8132430296391249,
      "aux_distill/mean_u": 0.2754530526020663,
      "aux_distill/n_active_tok": 135.875,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 9.79296875,
      "calib/ece": 0.3206719367588933,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.015810276679841896,
      "calib/gap": -0.010492700729926974,
      "calib/mean_conf": 0.18818181818181817,
      "calib/mu_c": 0.18250000000000002,
      "calib/mu_w": 0.192992700729927,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.025177865612648224,
      "calib/std_conf": 0.18922312191444773,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.33029307359307364,
      "calib/step_q_c_n": 924.0,
      "calib/step_q_gap": 0.01789042040292832,
      "calib/step_q_w": 0.3124026531901453,
      "calib/step_q_w_n": 1583.0,
      "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": 1078.0,
      "completions/max_terminated_length": 1078.0,
      "completions/mean_length": 375.21484375,
      "completions/mean_terminated_length": 379.6640319824219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.09066666666666667,
      "grad_norm": 0.6015828847885132,
      "learning_rate": 3.2222222222222227e-06,
      "loss": 0.0769,
      "num_tokens": 16597968.0,
      "reward": 1.0357849597930908,
      "reward_std": 0.12696963548660278,
      "rewards/accuracy_reward_step": 0.453125,
      "rewards/final_brier_reward_step": 0.6301636695861816,
      "rewards/format_reward_step": 0.98828125,
      "step": 85
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7791156647726893,
      "aux_distill/mean_u": 0.22830810682716612,
      "aux_distill/n_active_tok": 148.125,
      "calib/answer_extract_rate": 0.96484375,
      "calib/avg_num_step_conf": 11.9296875,
      "calib/ece": 0.2980566801619433,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.008097165991902834,
      "calib/gap": -0.009987755102040835,
      "calib/mean_conf": 0.16534412955465586,
      "calib/mu_c": 0.1594,
      "calib/mu_w": 0.16938775510204082,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.029271255060728744,
      "calib/std_conf": 0.17124391001302797,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2538078590785908,
      "calib/step_q_c_n": 738.0,
      "calib/step_q_gap": 0.021425907437830904,
      "calib/step_q_w": 0.23238195164075992,
      "calib/step_q_w_n": 2316.0,
      "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": 1861.0,
      "completions/max_terminated_length": 1861.0,
      "completions/mean_length": 383.796875,
      "completions/mean_terminated_length": 397.7813720703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 123.0,
      "epoch": 0.09173333333333333,
      "grad_norm": 1.1985909938812256,
      "learning_rate": 3.1944444444444443e-06,
      "loss": 0.0005,
      "num_tokens": 16801732.0,
      "reward": 0.9997738003730774,
      "reward_std": 0.16255082190036774,
      "rewards/accuracy_reward_step": 0.390625,
      "rewards/final_brier_reward_step": 0.6440788507461548,
      "rewards/format_reward_step": 0.96484375,
      "step": 86
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7769611058756709,
      "aux_distill/mean_u": 0.2535630387433568,
      "aux_distill/n_active_tok": 129.75,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 8.62109375,
      "calib/ece": 0.4245238095238096,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.007936507936507936,
      "calib/gap": -0.016798926586160584,
      "calib/mean_conf": 0.17222222222222225,
      "calib/mu_c": 0.16482269503546101,
      "calib/mu_w": 0.1816216216216216,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.018611111111111113,
      "calib/std_conf": 0.18810833120326595,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24014472876151483,
      "calib/step_q_c_n": 977.0,
      "calib/step_q_gap": 0.04641269624118963,
      "calib/step_q_w": 0.1937320325203252,
      "calib/step_q_w_n": 1230.0,
      "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": 1153.0,
      "completions/max_terminated_length": 1153.0,
      "completions/mean_length": 331.5390625,
      "completions/mean_terminated_length": 335.4703674316406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 135.0,
      "epoch": 0.0928,
      "grad_norm": 0.9272819757461548,
      "learning_rate": 3.1666666666666667e-06,
      "loss": 0.0383,
      "num_tokens": 16992102.0,
      "reward": 1.0431418418884277,
      "reward_std": 0.1451120376586914,
      "rewards/accuracy_reward_step": 0.55078125,
      "rewards/final_brier_reward_step": 0.5511273145675659,
      "rewards/format_reward_step": 0.984375,
      "step": 87
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8187052253633738,
      "aux_distill/mean_u": 0.23038298876880164,
      "aux_distill/n_active_tok": 117.375,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 8.796875,
      "calib/ece": 0.2947628458498024,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.003952569169960474,
      "calib/gap": 0.041013986013985954,
      "calib/mean_conf": 0.1631818181818182,
      "calib/mu_c": 0.18636363636363634,
      "calib/mu_w": 0.1453496503496504,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.011581027667984188,
      "calib/std_conf": 0.17293974865677617,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2335702380952381,
      "calib/step_q_c_n": 840.0,
      "calib/step_q_gap": 0.024612731013085143,
      "calib/step_q_w": 0.20895750708215297,
      "calib/step_q_w_n": 1412.0,
      "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": 941.0,
      "completions/max_terminated_length": 941.0,
      "completions/mean_length": 370.21875,
      "completions/mean_terminated_length": 376.0952453613281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 179.0,
      "epoch": 0.09386666666666667,
      "grad_norm": 0.47781553864479065,
      "learning_rate": 3.138888888888889e-06,
      "loss": 0.0596,
      "num_tokens": 17196726.0,
      "reward": 1.0404224395751953,
      "reward_std": 0.12067631632089615,
      "rewards/accuracy_reward_step": 0.4296875,
      "rewards/final_brier_reward_step": 0.6628760695457458,
      "rewards/format_reward_step": 0.98828125,
      "step": 88
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8449487295001745,
      "aux_distill/mean_u": 0.30785709829699454,
      "aux_distill/n_active_tok": 117.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.51953125,
      "calib/ece": 0.23152343750000004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0078125,
      "calib/gap": 0.04241266088783821,
      "calib/mean_conf": 0.19082031249999998,
      "calib/mu_c": 0.21765957446808512,
      "calib/mu_w": 0.1752469135802469,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.027578125000000002,
      "calib/std_conf": 0.18813982223974365,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2421522693997072,
      "calib/step_q_c_n": 683.0,
      "calib/step_q_gap": 0.019251142185536474,
      "calib/step_q_w": 0.2229011272141707,
      "calib/step_q_w_n": 1242.0,
      "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": 1279.0,
      "completions/max_terminated_length": 1279.0,
      "completions/mean_length": 357.10546875,
      "completions/mean_terminated_length": 359.9173278808594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.09493333333333333,
      "grad_norm": 0.6366252899169922,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0585,
      "num_tokens": 17397033.0,
      "reward": 1.0440173149108887,
      "reward_std": 0.10052931308746338,
      "rewards/accuracy_reward_step": 0.3671875,
      "rewards/final_brier_reward_step": 0.7208472490310669,
      "rewards/format_reward_step": 1.0,
      "step": 89
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8226105701178312,
      "aux_distill/mean_u": 0.2654090760485537,
      "aux_distill/n_active_tok": 115.5,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 7.99609375,
      "calib/ece": 0.32437007874015744,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.011811023622047244,
      "calib/gap": -0.01987114845938376,
      "calib/mean_conf": 0.2060236220472441,
      "calib/mu_c": 0.1954621848739496,
      "calib/mu_w": 0.21533333333333335,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.030944881889763784,
      "calib/std_conf": 0.1961006074975203,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2793063829787234,
      "calib/step_q_c_n": 799.0,
      "calib/step_q_gap": 0.004115036824877294,
      "calib/step_q_w": 0.2751913461538461,
      "calib/step_q_w_n": 1248.0,
      "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": 1888.0,
      "completions/max_terminated_length": 1888.0,
      "completions/mean_length": 357.85546875,
      "completions/mean_terminated_length": 360.6732177734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.096,
      "grad_norm": 0.4627319872379303,
      "learning_rate": 3.0833333333333336e-06,
      "loss": 0.0536,
      "num_tokens": 17591964.0,
      "reward": 1.042912244796753,
      "reward_std": 0.13357305526733398,
      "rewards/accuracy_reward_step": 0.46484375,
      "rewards/final_brier_reward_step": 0.6287933588027954,
      "rewards/format_reward_step": 0.9921875,
      "step": 90
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8454980216920376,
      "aux_distill/mean_u": 0.241689415357186,
      "aux_distill/n_active_tok": 106.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.84375,
      "calib/ece": 0.33191406250000005,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": -0.007789215686274442,
      "calib/mean_conf": 0.2439453125,
      "calib/mu_c": 0.24029411764705888,
      "calib/mu_w": 0.24808333333333332,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.022304687500000007,
      "calib/std_conf": 0.2144680486326048,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3091839393939394,
      "calib/step_q_c_n": 990.0,
      "calib/step_q_gap": 0.03560769267477931,
      "calib/step_q_w": 0.2735762467191601,
      "calib/step_q_w_n": 762.0,
      "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": 1337.0,
      "completions/max_terminated_length": 1337.0,
      "completions/mean_length": 339.36328125,
      "completions/mean_terminated_length": 342.0354309082031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 120.0,
      "epoch": 0.09706666666666666,
      "grad_norm": 0.7517727613449097,
      "learning_rate": 3.055555555555556e-06,
      "loss": 0.1172,
      "num_tokens": 17786553.0,
      "reward": 1.0749033689498901,
      "reward_std": 0.13493086397647858,
      "rewards/accuracy_reward_step": 0.53125,
      "rewards/final_brier_reward_step": 0.6185566186904907,
      "rewards/format_reward_step": 1.0,
      "step": 91
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8148021893575788,
      "aux_distill/mean_u": 0.21645658987542504,
      "aux_distill/n_active_tok": 86.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.10546875,
      "calib/ece": 0.3235976562499999,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": -0.006712854349951114,
      "calib/mean_conf": 0.23257421875,
      "calib/mu_c": 0.22911290322580646,
      "calib/mu_w": 0.23582575757575758,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0358984375,
      "calib/std_conf": 0.2086469112018607,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28883021390374336,
      "calib/step_q_c_n": 748.0,
      "calib/step_q_gap": -0.03785101309012162,
      "calib/step_q_w": 0.326681226993865,
      "calib/step_q_w_n": 815.0,
      "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": 1504.0,
      "completions/max_terminated_length": 1504.0,
      "completions/mean_length": 300.97265625,
      "completions/mean_terminated_length": 303.342529296875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 118.0,
      "epoch": 0.09813333333333334,
      "grad_norm": 4.081338405609131,
      "learning_rate": 3.0277777777777776e-06,
      "loss": 0.0894,
      "num_tokens": 17970322.0,
      "reward": 1.0621644258499146,
      "reward_std": 0.13498783111572266,
      "rewards/accuracy_reward_step": 0.484375,
      "rewards/final_brier_reward_step": 0.6399537920951843,
      "rewards/format_reward_step": 1.0,
      "step": 92
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8725457116961479,
      "aux_distill/mean_u": 0.2621401066955912,
      "aux_distill/n_active_tok": 100.5,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.234375,
      "calib/ece": 0.26855468749999994,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0390625,
      "calib/gap": 0.04151270278543004,
      "calib/mean_conf": 0.2576953125,
      "calib/mu_c": 0.2795867768595041,
      "calib/mu_w": 0.23807407407407408,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.026796875,
      "calib/std_conf": 0.21198043315487244,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3198336434108527,
      "calib/step_q_c_n": 645.0,
      "calib/step_q_gap": -0.02491346489619256,
      "calib/step_q_w": 0.34474710830704525,
      "calib/step_q_w_n": 951.0,
      "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": 1035.0,
      "completions/max_terminated_length": 1035.0,
      "completions/mean_length": 302.5078125,
      "completions/mean_terminated_length": 304.8897705078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 88.0,
      "epoch": 0.0992,
      "grad_norm": 0.8950026035308838,
      "learning_rate": 3e-06,
      "loss": 0.101,
      "num_tokens": 18153540.0,
      "reward": 1.0764771699905396,
      "reward_std": 0.13514947891235352,
      "rewards/accuracy_reward_step": 0.47265625,
      "rewards/final_brier_reward_step": 0.6802980899810791,
      "rewards/format_reward_step": 1.0,
      "step": 93
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8754433235153556,
      "aux_distill/mean_u": 0.2528739918905104,
      "aux_distill/n_active_tok": 77.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 4.98046875,
      "calib/ece": 0.32996093749999994,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": 0.02433778107239931,
      "calib/mean_conf": 0.25042968749999994,
      "calib/mu_c": 0.2612676056338028,
      "calib/mu_w": 0.2369298245614035,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0128515625,
      "calib/std_conf": 0.17700252397537256,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27266932952924394,
      "calib/step_q_c_n": 701.0,
      "calib/step_q_gap": -0.04155540914671424,
      "calib/step_q_w": 0.3142247386759582,
      "calib/step_q_w_n": 574.0,
      "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": 628.0,
      "completions/max_terminated_length": 628.0,
      "completions/mean_length": 257.109375,
      "completions/mean_terminated_length": 259.13385009765625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 98.0,
      "epoch": 0.10026666666666667,
      "grad_norm": 2.6950621604919434,
      "learning_rate": 2.9722222222222225e-06,
      "loss": 0.0702,
      "num_tokens": 18328040.0,
      "reward": 1.0978994369506836,
      "reward_std": 0.11955973505973816,
      "rewards/accuracy_reward_step": 0.5546875,
      "rewards/final_brier_reward_step": 0.6411113739013672,
      "rewards/format_reward_step": 1.0,
      "step": 94
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9074766356498003,
      "aux_distill/mean_u": 0.2802716825590698,
      "aux_distill/n_active_tok": 85.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.55078125,
      "calib/ece": 0.2603125,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": 0.0869743589743589,
      "calib/mean_conf": 0.2475,
      "calib/mu_c": 0.29030769230769227,
      "calib/mu_w": 0.20333333333333337,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.17327047137351478,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2969854466858789,
      "calib/step_q_c_n": 694.0,
      "calib/step_q_gap": 0.055014607621229644,
      "calib/step_q_w": 0.24197083906464925,
      "calib/step_q_w_n": 727.0,
      "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": 676.0,
      "completions/max_terminated_length": 676.0,
      "completions/mean_length": 285.51953125,
      "completions/mean_terminated_length": 287.7677307128906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 93.0,
      "epoch": 0.10133333333333333,
      "grad_norm": 1.4201782941818237,
      "learning_rate": 2.944444444444445e-06,
      "loss": 0.0709,
      "num_tokens": 18507261.0,
      "reward": 1.1017823219299316,
      "reward_std": 0.12060728669166565,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.6957523226737976,
      "rewards/format_reward_step": 1.0,
      "step": 95
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8319747727364302,
      "aux_distill/mean_u": 0.25121979181888454,
      "aux_distill/n_active_tok": 77.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.3125,
      "calib/ece": 0.31843338671875004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.01171875,
      "calib/gap": 0.03609005367372353,
      "calib/mean_conf": 0.26641036328125,
      "calib/mu_c": 0.2819178082191781,
      "calib/mu_w": 0.24582775454545458,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.007265625000000001,
      "calib/std_conf": 0.18174814973820932,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3027449602122016,
      "calib/step_q_c_n": 754.0,
      "calib/step_q_gap": 0.02137933644982537,
      "calib/step_q_w": 0.2813656237623762,
      "calib/step_q_w_n": 606.0,
      "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": 615.0,
      "completions/max_terminated_length": 615.0,
      "completions/mean_length": 270.23046875,
      "completions/mean_terminated_length": 272.3582763671875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.1024,
      "grad_norm": 6.897700309753418,
      "learning_rate": 2.916666666666667e-06,
      "loss": 0.0658,
      "num_tokens": 18682256.0,
      "reward": 1.1087777614593506,
      "reward_std": 0.1243782639503479,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6472431421279907,
      "rewards/format_reward_step": 1.0,
      "step": 96
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9163010641932487,
      "aux_distill/mean_u": 0.2991586962353635,
      "aux_distill/n_active_tok": 80.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 5.1640625,
      "calib/ece": 0.25729411764705884,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00784313725490196,
      "calib/gap": 0.02928615384615388,
      "calib/mean_conf": 0.25650980392156864,
      "calib/mu_c": 0.27144,
      "calib/mu_w": 0.24215384615384614,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.011803921568627454,
      "calib/std_conf": 0.16145167097539087,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2120297674418605,
      "calib/step_q_c_n": 645.0,
      "calib/step_q_gap": -0.03514866682697257,
      "calib/step_q_w": 0.24717843426883307,
      "calib/step_q_w_n": 677.0,
      "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": 530.0,
      "completions/max_terminated_length": 530.0,
      "completions/mean_length": 265.76953125,
      "completions/mean_terminated_length": 267.8622131347656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 120.0,
      "epoch": 0.10346666666666667,
      "grad_norm": 4.537506103515625,
      "learning_rate": 2.888888888888889e-06,
      "loss": 0.0987,
      "num_tokens": 18855365.0,
      "reward": 1.0828802585601807,
      "reward_std": 0.12252596020698547,
      "rewards/accuracy_reward_step": 0.48828125,
      "rewards/final_brier_reward_step": 0.6813855171203613,
      "rewards/format_reward_step": 0.99609375,
      "step": 97
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8761290339753032,
      "aux_distill/mean_u": 0.27758280371669913,
      "aux_distill/n_active_tok": 86.875,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.71484375,
      "calib/ece": 0.23363281249999998,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0078125,
      "calib/gap": 0.03753549190535499,
      "calib/mean_conf": 0.3366796875,
      "calib/mu_c": 0.35280821917808225,
      "calib/mu_w": 0.31527272727272726,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.1586894290742214,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25460158730158733,
      "calib/step_q_c_n": 819.0,
      "calib/step_q_gap": 0.05324690814891295,
      "calib/step_q_w": 0.20135467915267438,
      "calib/step_q_w_n": 644.0,
      "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": 657.0,
      "completions/max_terminated_length": 657.0,
      "completions/mean_length": 284.09765625,
      "completions/mean_terminated_length": 286.33465576171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 98.0,
      "epoch": 0.10453333333333334,
      "grad_norm": 3.4571356773376465,
      "learning_rate": 2.861111111111111e-06,
      "loss": 0.0864,
      "num_tokens": 19034278.0,
      "reward": 1.1319432258605957,
      "reward_std": 0.12014371156692505,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6935738325119019,
      "rewards/format_reward_step": 1.0,
      "step": 98
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8914099540561438,
      "aux_distill/mean_u": 0.31211058932284613,
      "aux_distill/n_active_tok": 95.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 6.05859375,
      "calib/ece": 0.0665333333333333,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.08531102003642999,
      "calib/mean_conf": 0.32974901960784314,
      "calib/mu_c": 0.3909722222222223,
      "calib/mu_w": 0.3056612021857923,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05696470588235291,
      "calib/std_conf": 0.1499662698972071,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.18594666666666668,
      "calib/step_q_c_n": 375.0,
      "calib/step_q_gap": -0.021685221088435347,
      "calib/step_q_w": 0.20763188775510202,
      "calib/step_q_w_n": 1176.0,
      "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": 1261.0,
      "completions/max_terminated_length": 1261.0,
      "completions/mean_length": 307.6875,
      "completions/mean_terminated_length": 310.1102294921875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 85.0,
      "epoch": 0.1056,
      "grad_norm": 1.3000514507293701,
      "learning_rate": 2.8333333333333335e-06,
      "loss": 0.0563,
      "num_tokens": 19218846.0,
      "reward": 1.040698766708374,
      "reward_std": 0.11782108247280121,
      "rewards/accuracy_reward_step": 0.28125,
      "rewards/final_brier_reward_step": 0.8040539026260376,
      "rewards/format_reward_step": 0.99609375,
      "step": 99
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.9092141855508089,
      "aux_distill/mean_u": 0.3021948987905085,
      "aux_distill/n_active_tok": 87.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.5390625,
      "calib/ece": 0.19819607843137255,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.08460488595288546,
      "calib/mean_conf": 0.36525490196078425,
      "calib/mu_c": 0.40274647887323944,
      "calib/mu_w": 0.318141592920354,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0032941176470588232,
      "calib/std_conf": 0.1601170947918998,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20816040609137054,
      "calib/step_q_c_n": 788.0,
      "calib/step_q_gap": -0.0857559431149787,
      "calib/step_q_w": 0.29391634920634924,
      "calib/step_q_w_n": 630.0,
      "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": 842.0,
      "completions/max_terminated_length": 842.0,
      "completions/mean_length": 288.59375,
      "completions/mean_terminated_length": 290.86614990234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 101.0,
      "epoch": 0.10666666666666667,
      "grad_norm": 0.8430328369140625,
      "learning_rate": 2.805555555555556e-06,
      "loss": 0.0924,
      "num_tokens": 19400134.0,
      "reward": 1.1422317028045654,
      "reward_std": 0.13820312917232513,
      "rewards/accuracy_reward_step": 0.55859375,
      "rewards/final_brier_reward_step": 0.7297757863998413,
      "rewards/format_reward_step": 0.99609375,
      "step": 100
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8671903610229492,
      "aux_distill/mean_u": 0.28633546362227774,
      "aux_distill/n_active_tok": 101.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.55859375,
      "calib/ece": 0.18968749999999995,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.00390625,
      "calib/gap": 0.019789987789987862,
      "calib/mean_conf": 0.406875,
      "calib/mu_c": 0.4166153846153846,
      "calib/mu_w": 0.39682539682539675,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04437499999999998,
      "calib/std_conf": 0.1950210325195721,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25435709606986895,
      "calib/step_q_c_n": 916.0,
      "calib/step_q_gap": -0.06151865753432506,
      "calib/step_q_w": 0.315875753604194,
      "calib/step_q_w_n": 763.0,
      "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": 908.0,
      "completions/max_terminated_length": 908.0,
      "completions/mean_length": 335.12109375,
      "completions/mean_terminated_length": 337.75982666015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.10773333333333333,
      "grad_norm": 1.9569110870361328,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 0.0976,
      "num_tokens": 19592917.0,
      "reward": 1.1097722053527832,
      "reward_std": 0.13801291584968567,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.711732029914856,
      "rewards/format_reward_step": 1.0,
      "step": 101
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8509639389812946,
      "aux_distill/mean_u": 0.25738222654461373,
      "aux_distill/n_active_tok": 87.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.234375,
      "calib/ece": 0.09972103409260036,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0390625,
      "calib/gap": 0.17604035958991537,
      "calib/mean_conf": 0.48027896590739966,
      "calib/mu_c": 0.5586719385372838,
      "calib/mu_w": 0.38263157894736843,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.01265625,
      "calib/std_conf": 0.23970905810145982,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26079555555555556,
      "calib/step_q_c_n": 720.0,
      "calib/step_q_gap": -0.07308089605734769,
      "calib/step_q_w": 0.33387645161290325,
      "calib/step_q_w_n": 620.0,
      "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": 810.0,
      "completions/max_terminated_length": 810.0,
      "completions/mean_length": 266.01953125,
      "completions/mean_terminated_length": 268.1141662597656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 66.0,
      "epoch": 0.1088,
      "grad_norm": 1.224706768989563,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.0801,
      "num_tokens": 19767714.0,
      "reward": 1.1658241748809814,
      "reward_std": 0.1503104567527771,
      "rewards/accuracy_reward_step": 0.5546875,
      "rewards/final_brier_reward_step": 0.7769608497619629,
      "rewards/format_reward_step": 1.0,
      "step": 102
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8516613645479083,
      "aux_distill/mean_u": 0.23035033735915972,
      "aux_distill/n_active_tok": 108.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.08203125,
      "calib/ece": 0.09985117187500002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.08203125,
      "calib/gap": 0.1841112862862863,
      "calib/mean_conf": 0.5346800781250001,
      "calib/mu_c": 0.6123520270270271,
      "calib/mu_w": 0.42824074074074076,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.028203124999999996,
      "calib/std_conf": 0.2673069865020486,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.28984356796116506,
      "calib/step_q_c_n": 824.0,
      "calib/step_q_gap": -0.02840131607703411,
      "calib/step_q_w": 0.31824488403819917,
      "calib/step_q_w_n": 733.0,
      "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": 1320.0,
      "completions/max_terminated_length": 1320.0,
      "completions/mean_length": 334.1796875,
      "completions/mean_terminated_length": 336.81103515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 100.0,
      "epoch": 0.10986666666666667,
      "grad_norm": 1.4121969938278198,
      "learning_rate": 2.7222222222222224e-06,
      "loss": 0.0908,
      "num_tokens": 19957816.0,
      "reward": 1.1753480434417725,
      "reward_std": 0.15455801784992218,
      "rewards/accuracy_reward_step": 0.578125,
      "rewards/final_brier_reward_step": 0.7725712060928345,
      "rewards/format_reward_step": 1.0,
      "step": 103
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8432601001113653,
      "aux_distill/mean_u": 0.3180295561600359,
      "aux_distill/n_active_tok": 93.875,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 5.63671875,
      "calib/ece": 0.30630980392156865,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.24705882352941178,
      "calib/gap": 0.15246913496683379,
      "calib/mean_conf": 0.6391725490196078,
      "calib/mu_c": 0.738426966292135,
      "calib/mu_w": 0.5859578313253012,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.29823137254901966,
      "calib/std_conf": 0.3083625724843869,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.30692622950819665,
      "calib/step_q_c_n": 488.0,
      "calib/step_q_gap": 0.002957014848510797,
      "calib/step_q_w": 0.30396921465968585,
      "calib/step_q_w_n": 955.0,
      "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": 890.0,
      "completions/max_terminated_length": 890.0,
      "completions/mean_length": 296.39453125,
      "completions/mean_terminated_length": 298.72833251953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 96.0,
      "epoch": 0.11093333333333333,
      "grad_norm": 3.2531869411468506,
      "learning_rate": 2.6944444444444444e-06,
      "loss": 0.0498,
      "num_tokens": 20140373.0,
      "reward": 1.0019816160202026,
      "reward_std": 0.24969851970672607,
      "rewards/accuracy_reward_step": 0.34765625,
      "rewards/final_brier_reward_step": 0.6602132320404053,
      "rewards/format_reward_step": 0.99609375,
      "step": 104
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8673386499285698,
      "aux_distill/mean_u": 0.26861608940092646,
      "aux_distill/n_active_tok": 100.25,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 6.59765625,
      "calib/ece": 0.23828627450980394,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.3764705882352941,
      "calib/gap": 0.17266130820399095,
      "calib/mean_conf": 0.7040117647058823,
      "calib/mu_c": 0.7872954545454546,
      "calib/mu_w": 0.6146341463414636,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.21232549019607844,
      "calib/std_conf": 0.3037210082328581,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2936667892156863,
      "calib/step_q_c_n": 816.0,
      "calib/step_q_gap": -0.002520381460144161,
      "calib/step_q_w": 0.29618717067583045,
      "calib/step_q_w_n": 873.0,
      "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": 1256.0,
      "completions/max_terminated_length": 1256.0,
      "completions/mean_length": 345.3359375,
      "completions/mean_terminated_length": 348.05511474609375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 137.0,
      "epoch": 0.112,
      "grad_norm": 1.3178426027297974,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.0841,
      "num_tokens": 20334539.0,
      "reward": 1.102736473083496,
      "reward_std": 0.305696964263916,
      "rewards/accuracy_reward_step": 0.515625,
      "rewards/final_brier_reward_step": 0.7015665769577026,
      "rewards/format_reward_step": 0.98828125,
      "step": 105
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8582525728270411,
      "aux_distill/mean_u": 0.2746546685465814,
      "aux_distill/n_active_tok": 93.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 5.69921875,
      "calib/ece": 0.2409288884941433,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.453125,
      "calib/gap": 0.1835228672186019,
      "calib/mean_conf": 0.7755554865058568,
      "calib/mu_c": 0.8587142857142857,
      "calib/mu_w": 0.6751914184956838,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.23480468750000005,
      "calib/std_conf": 0.261164994761232,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.3307972053462941,
      "calib/step_q_c_n": 823.0,
      "calib/step_q_gap": 0.03137896635258336,
      "calib/step_q_w": 0.29941823899371073,
      "calib/step_q_w_n": 636.0,
      "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": 792.0,
      "completions/max_terminated_length": 792.0,
      "completions/mean_length": 309.47265625,
      "completions/mean_terminated_length": 311.9094543457031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 120.0,
      "epoch": 0.11306666666666666,
      "grad_norm": 0.6310628652572632,
      "learning_rate": 2.6388888888888893e-06,
      "loss": 0.0944,
      "num_tokens": 20518348.0,
      "reward": 1.1347626447677612,
      "reward_std": 0.25099045038223267,
      "rewards/accuracy_reward_step": 0.546875,
      "rewards/final_brier_reward_step": 0.7226502895355225,
      "rewards/format_reward_step": 1.0,
      "step": 106
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8792069256305695,
      "aux_distill/mean_u": 0.34678596406839074,
      "aux_distill/n_active_tok": 98.75,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.50390625,
      "calib/ece": 0.23453125000000002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.46875,
      "calib/gap": 0.16238051187172375,
      "calib/mean_conf": 0.7342187499999999,
      "calib/mu_c": 0.8071631205673758,
      "calib/mu_w": 0.6447826086956521,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.208984375,
      "calib/std_conf": 0.3006714085150723,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.32097870762711866,
      "calib/step_q_c_n": 944.0,
      "calib/step_q_gap": 0.033084255477326685,
      "calib/step_q_w": 0.287894452149792,
      "calib/step_q_w_n": 721.0,
      "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": 1056.0,
      "completions/max_terminated_length": 1056.0,
      "completions/mean_length": 321.4921875,
      "completions/mean_terminated_length": 324.02362060546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.11413333333333334,
      "grad_norm": 0.6019020080566406,
      "learning_rate": 2.6111111111111113e-06,
      "loss": 0.0717,
      "num_tokens": 20705266.0,
      "reward": 1.1298301219940186,
      "reward_std": 0.23322361707687378,
      "rewards/accuracy_reward_step": 0.55078125,
      "rewards/final_brier_reward_step": 0.7088788747787476,
      "rewards/format_reward_step": 1.0,
      "step": 107
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7998718656599522,
      "aux_distill/mean_u": 0.25381538008380866,
      "aux_distill/n_active_tok": 105.5,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 6.96484375,
      "calib/ece": 0.16759842519685036,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.5866141732283464,
      "calib/gap": 0.2511341772151898,
      "calib/mean_conf": 0.7742913385826772,
      "calib/mu_c": 0.8524,
      "calib/mu_w": 0.6012658227848102,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1264566929133858,
      "calib/std_conf": 0.30715498178991935,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.26813386243386245,
      "calib/step_q_c_n": 1134.0,
      "calib/step_q_gap": -0.08027908055535171,
      "calib/step_q_w": 0.34841294298921416,
      "calib/step_q_w_n": 649.0,
      "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": 1208.0,
      "completions/max_terminated_length": 1208.0,
      "completions/mean_length": 353.328125,
      "completions/mean_terminated_length": 356.1102294921875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 85.0,
      "epoch": 0.1152,
      "grad_norm": 0.7394620776176453,
      "learning_rate": 2.5833333333333337e-06,
      "loss": 0.0762,
      "num_tokens": 20898950.0,
      "reward": 1.2267515659332275,
      "reward_std": 0.24268360435962677,
      "rewards/accuracy_reward_step": 0.68359375,
      "rewards/final_brier_reward_step": 0.7816281318664551,
      "rewards/format_reward_step": 0.98828125,
      "step": 108
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8652790896594524,
      "aux_distill/mean_u": 0.2592223691426132,
      "aux_distill/n_active_tok": 120.0,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 7.63671875,
      "calib/ece": 0.2507929133858267,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.5433070866141733,
      "calib/gap": 0.3318445442635224,
      "calib/mean_conf": 0.6992181102362205,
      "calib/mu_c": 0.8782051282051282,
      "calib/mu_w": 0.5463605839416058,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2446905511811023,
      "calib/std_conf": 0.36504522142847917,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24525593008739074,
      "calib/step_q_c_n": 801.0,
      "calib/step_q_gap": -0.012741115574985612,
      "calib/step_q_w": 0.25799704566237636,
      "calib/step_q_w_n": 1154.0,
      "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": 2066.0,
      "completions/max_terminated_length": 2066.0,
      "completions/mean_length": 368.7734375,
      "completions/mean_terminated_length": 373.1462707519531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.11626666666666667,
      "grad_norm": 0.8723954558372498,
      "learning_rate": 2.5555555555555557e-06,
      "loss": 0.0752,
      "num_tokens": 21097956.0,
      "reward": 1.0849031209945679,
      "reward_std": 0.17993712425231934,
      "rewards/accuracy_reward_step": 0.45703125,
      "rewards/final_brier_reward_step": 0.7205872535705566,
      "rewards/format_reward_step": 0.9921875,
      "step": 109
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8240192476660013,
      "aux_distill/mean_u": 0.19362646090396934,
      "aux_distill/n_active_tok": 96.125,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 6.3671875,
      "calib/ece": 0.3861378906250001,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.7890625,
      "calib/gap": 0.10433952366412202,
      "calib/mean_conf": 0.856762890625,
      "calib/mu_c": 0.9077099236641221,
      "calib/mu_w": 0.8033704,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.36559101562500007,
      "calib/std_conf": 0.27331232918943993,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.275311004784689,
      "calib/step_q_c_n": 836.0,
      "calib/step_q_gap": -0.021312073734772063,
      "calib/step_q_w": 0.29662307851946107,
      "calib/step_q_w_n": 794.0,
      "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": 349.12109375,
      "completions/mean_terminated_length": 351.8700866699219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.11733333333333333,
      "grad_norm": 0.7028835415840149,
      "learning_rate": 2.5277777777777778e-06,
      "loss": 0.0932,
      "num_tokens": 21292251.0,
      "reward": 1.060120940208435,
      "reward_std": 0.25176236033439636,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.6085233688354492,
      "rewards/format_reward_step": 1.0,
      "step": 110
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8519136030226946,
      "aux_distill/mean_u": 0.2509307208367431,
      "aux_distill/n_active_tok": 110.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.21484375,
      "calib/ece": 0.26287401574803154,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.7834645669291339,
      "calib/gap": 0.26822805682394724,
      "calib/mean_conf": 0.8201968503937009,
      "calib/mu_c": 0.9342465753424658,
      "calib/mu_w": 0.6660185185185186,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2541338582677166,
      "calib/std_conf": 0.32242370859491043,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21969259259259255,
      "calib/step_q_c_n": 945.0,
      "calib/step_q_gap": -0.05836239632093293,
      "calib/step_q_w": 0.2780549889135255,
      "calib/step_q_w_n": 902.0,
      "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": 1508.0,
      "completions/max_terminated_length": 1508.0,
      "completions/mean_length": 374.2421875,
      "completions/mean_terminated_length": 377.18896484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 102.0,
      "epoch": 0.1184,
      "grad_norm": 1.3935075998306274,
      "learning_rate": 2.5e-06,
      "loss": 0.0767,
      "num_tokens": 21495465.0,
      "reward": 1.1396938562393188,
      "reward_std": 0.2531803250312805,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.716887891292572,
      "rewards/format_reward_step": 0.9921875,
      "step": 111
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7935940055176616,
      "aux_distill/mean_u": 0.27512142615491214,
      "aux_distill/n_active_tok": 119.0,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 7.546875,
      "calib/ece": 0.3260714285714286,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.6944444444444444,
      "calib/gap": 0.23053679435483887,
      "calib/mean_conf": 0.7325793650793652,
      "calib/mu_c": 0.8496774193548389,
      "calib/mu_w": 0.619140625,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2832936507936508,
      "calib/std_conf": 0.38686993534751096,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21454656862745097,
      "calib/step_q_c_n": 816.0,
      "calib/step_q_gap": -0.016405492304448682,
      "calib/step_q_w": 0.23095206093189966,
      "calib/step_q_w_n": 1116.0,
      "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": 2146.0,
      "completions/max_terminated_length": 2146.0,
      "completions/mean_length": 401.26171875,
      "completions/mean_terminated_length": 406.019775390625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 147.0,
      "epoch": 0.11946666666666667,
      "grad_norm": 0.6045711636543274,
      "learning_rate": 2.4722222222222226e-06,
      "loss": 0.0266,
      "num_tokens": 21706108.0,
      "reward": 1.058129072189331,
      "reward_std": 0.25581157207489014,
      "rewards/accuracy_reward_step": 0.484375,
      "rewards/final_brier_reward_step": 0.6475081443786621,
      "rewards/format_reward_step": 0.984375,
      "step": 112
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7709256280213594,
      "aux_distill/mean_u": 0.19761347736346177,
      "aux_distill/n_active_tok": 100.375,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.0859375,
      "calib/ece": 0.34094117647058825,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.6549019607843137,
      "calib/gap": 0.16162422360248452,
      "calib/mean_conf": 0.6900392156862746,
      "calib/mu_c": 0.7629285714285714,
      "calib/mu_w": 0.6013043478260869,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24098039215686273,
      "calib/std_conf": 0.4073862146340922,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.1886348054679285,
      "calib/step_q_c_n": 951.0,
      "calib/step_q_gap": 0.002134226093652686,
      "calib/step_q_w": 0.1865005793742758,
      "calib/step_q_w_n": 863.0,
      "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": 1383.0,
      "completions/max_terminated_length": 1383.0,
      "completions/mean_length": 344.7734375,
      "completions/mean_terminated_length": 348.8616638183594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.12053333333333334,
      "grad_norm": 0.6675346493721008,
      "learning_rate": 2.4444444444444447e-06,
      "loss": 0.0652,
      "num_tokens": 21899570.0,
      "reward": 1.0935156345367432,
      "reward_std": 0.25981682538986206,
      "rewards/accuracy_reward_step": 0.546875,
      "rewards/final_brier_reward_step": 0.6440625190734863,
      "rewards/format_reward_step": 0.99609375,
      "step": 113
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7940104631707072,
      "aux_distill/mean_u": 0.22985165603566762,
      "aux_distill/n_active_tok": 125.0,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 7.71875,
      "calib/ece": 0.18881023622047247,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.7244094488188977,
      "calib/gap": 0.37727772357723577,
      "calib/mean_conf": 0.7528433070866142,
      "calib/mu_c": 0.8865243902439024,
      "calib/mu_w": 0.5092466666666666,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.147992125984252,
      "calib/std_conf": 0.38016471463101426,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21010340699815838,
      "calib/step_q_c_n": 1086.0,
      "calib/step_q_gap": 0.012003069919506698,
      "calib/step_q_w": 0.19810033707865168,
      "calib/step_q_w_n": 890.0,
      "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": 2818.0,
      "completions/max_terminated_length": 2818.0,
      "completions/mean_length": 377.83203125,
      "completions/mean_terminated_length": 382.312255859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.1216,
      "grad_norm": 0.6961753368377686,
      "learning_rate": 2.4166666666666667e-06,
      "loss": 0.0901,
      "num_tokens": 22101319.0,
      "reward": 1.2033731937408447,
      "reward_std": 0.18372264504432678,
      "rewards/accuracy_reward_step": 0.640625,
      "rewards/final_brier_reward_step": 0.7778403759002686,
      "rewards/format_reward_step": 0.98828125,
      "step": 114
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7925838343799114,
      "aux_distill/mean_u": 0.21411440185498998,
      "aux_distill/n_active_tok": 129.5,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 7.734375,
      "calib/ece": 0.3548208661417324,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.7007874015748031,
      "calib/gap": 0.12842446727549484,
      "calib/mean_conf": 0.7268326771653543,
      "calib/mu_c": 0.7814383561643836,
      "calib/mu_w": 0.6530138888888888,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.25342519685039383,
      "calib/std_conf": 0.3961359234375426,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23115589285714283,
      "calib/step_q_c_n": 1120.0,
      "calib/step_q_gap": -0.027523642026578088,
      "calib/step_q_w": 0.2586795348837209,
      "calib/step_q_w_n": 860.0,
      "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": 1121.0,
      "completions/max_terminated_length": 1121.0,
      "completions/mean_length": 379.71484375,
      "completions/mean_terminated_length": 382.7047119140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 143.0,
      "epoch": 0.12266666666666666,
      "grad_norm": 0.566737711429596,
      "learning_rate": 2.388888888888889e-06,
      "loss": 0.0575,
      "num_tokens": 22303790.0,
      "reward": 1.0979235172271729,
      "reward_std": 0.22625698149204254,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6333469152450562,
      "rewards/format_reward_step": 0.9921875,
      "step": 115
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8143215561285615,
      "aux_distill/mean_u": 0.22442365785776003,
      "aux_distill/n_active_tok": 110.375,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 7.9296875,
      "calib/ece": 0.3188671874999999,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.6796875,
      "calib/gap": 0.2612024427480917,
      "calib/mean_conf": 0.7157421875000001,
      "calib/mu_c": 0.8432824427480916,
      "calib/mu_w": 0.5820799999999999,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.26144531249999997,
      "calib/std_conf": 0.4012434724175769,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.19985699693564862,
      "calib/step_q_c_n": 979.0,
      "calib/step_q_gap": -0.012093526375483638,
      "calib/step_q_w": 0.21195052331113226,
      "calib/step_q_w_n": 1051.0,
      "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": 1051.0,
      "completions/max_terminated_length": 1051.0,
      "completions/mean_length": 416.99609375,
      "completions/mean_terminated_length": 420.279541015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 86.0,
      "epoch": 0.12373333333333333,
      "grad_norm": 0.4892609119415283,
      "learning_rate": 2.361111111111111e-06,
      "loss": 0.0871,
      "num_tokens": 22515061.0,
      "reward": 1.0948817729949951,
      "reward_std": 0.2809078097343445,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.6780449151992798,
      "rewards/format_reward_step": 1.0,
      "step": 116
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.814437004737556,
      "aux_distill/mean_u": 0.22517764619918906,
      "aux_distill/n_active_tok": 124.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 8.37109375,
      "calib/ece": 0.3279015450952209,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.7193675889328063,
      "calib/gap": 0.25905888442679836,
      "calib/mean_conf": 0.7500431189363996,
      "calib/mu_c": 0.8841803278688525,
      "calib/mu_w": 0.6251214434420541,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.29786561264822126,
      "calib/std_conf": 0.38683644816285084,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21326670366259712,
      "calib/step_q_c_n": 901.0,
      "calib/step_q_gap": -0.03654110632129984,
      "calib/step_q_w": 0.24980780998389696,
      "calib/step_q_w_n": 1242.0,
      "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": 1547.0,
      "completions/max_terminated_length": 1547.0,
      "completions/mean_length": 394.45703125,
      "completions/mean_terminated_length": 400.7182922363281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 151.0,
      "epoch": 0.1248,
      "grad_norm": 0.7132987380027771,
      "learning_rate": 2.3333333333333336e-06,
      "loss": 0.0144,
      "num_tokens": 22722642.0,
      "reward": 1.0538398027420044,
      "reward_std": 0.23345988988876343,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.6467421650886536,
      "rewards/format_reward_step": 0.984375,
      "step": 117
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8741880189627409,
      "aux_distill/mean_u": 0.30376733348721374,
      "aux_distill/n_active_tok": 120.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.91796875,
      "calib/ece": 0.2701960784313725,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.6862745098039216,
      "calib/gap": 0.2668129032258063,
      "calib/mean_conf": 0.7229803921568628,
      "calib/mu_c": 0.8276129032258063,
      "calib/mu_w": 0.5608,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1926666666666666,
      "calib/std_conf": 0.3975500210330464,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25283794037940377,
      "calib/step_q_c_n": 1107.0,
      "calib/step_q_gap": -0.0044661835336397004,
      "calib/step_q_w": 0.25730412391304347,
      "calib/step_q_w_n": 920.0,
      "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": 1858.0,
      "completions/max_terminated_length": 1858.0,
      "completions/mean_length": 412.984375,
      "completions/mean_terminated_length": 416.2362060546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.12586666666666665,
      "grad_norm": 0.6454364061355591,
      "learning_rate": 2.305555555555556e-06,
      "loss": 0.0796,
      "num_tokens": 22932374.0,
      "reward": 1.1581437587738037,
      "reward_std": 0.17849767208099365,
      "rewards/accuracy_reward_step": 0.60546875,
      "rewards/final_brier_reward_step": 0.7147250175476074,
      "rewards/format_reward_step": 0.99609375,
      "step": 118
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7642288217321038,
      "aux_distill/mean_u": 0.1817720144921363,
      "aux_distill/n_active_tok": 118.5,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 8.58984375,
      "calib/ece": 0.23912350597609555,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.5338645418326693,
      "calib/gap": 0.3801702544031311,
      "calib/mean_conf": 0.5819920318725099,
      "calib/mu_c": 0.741027397260274,
      "calib/mu_w": 0.3608571428571429,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11972111553784856,
      "calib/std_conf": 0.43989315971992726,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2825702912621359,
      "calib/step_q_c_n": 1030.0,
      "calib/step_q_gap": 0.032339324623983645,
      "calib/step_q_w": 0.25023096663815225,
      "calib/step_q_w_n": 1169.0,
      "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": 1918.0,
      "completions/max_terminated_length": 1918.0,
      "completions/mean_length": 441.87890625,
      "completions/mean_terminated_length": 450.6812744140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.12693333333333334,
      "grad_norm": 0.6065315008163452,
      "learning_rate": 2.277777777777778e-06,
      "loss": 0.018,
      "num_tokens": 23150559.0,
      "reward": 1.142172932624817,
      "reward_std": 0.2332339584827423,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.7335648536682129,
      "rewards/format_reward_step": 0.98046875,
      "step": 119
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8176003079861403,
      "aux_distill/mean_u": 0.22354107810224982,
      "aux_distill/n_active_tok": 123.875,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 8.140625,
      "calib/ece": 0.19992094861660073,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.6047430830039525,
      "calib/gap": 0.423583066425289,
      "calib/mean_conf": 0.6472727272727272,
      "calib/mu_c": 0.7912574850299401,
      "calib/mu_w": 0.3676744186046511,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.09355731225296439,
      "calib/std_conf": 0.4286647076080423,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2459582322357019,
      "calib/step_q_c_n": 1154.0,
      "calib/step_q_gap": -0.01889520862451316,
      "calib/step_q_w": 0.26485344086021506,
      "calib/step_q_w_n": 930.0,
      "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": 1255.0,
      "completions/max_terminated_length": 1255.0,
      "completions/mean_length": 411.6640625,
      "completions/mean_terminated_length": 418.19842529296875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 161.0,
      "epoch": 0.128,
      "grad_norm": 0.7400944828987122,
      "learning_rate": 2.25e-06,
      "loss": 0.0619,
      "num_tokens": 23362633.0,
      "reward": 1.2066268920898438,
      "reward_std": 0.21622836589813232,
      "rewards/accuracy_reward_step": 0.65234375,
      "rewards/final_brier_reward_step": 0.772628903388977,
      "rewards/format_reward_step": 0.98828125,
      "step": 120
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7588197402656078,
      "aux_distill/mean_u": 0.23710005069366824,
      "aux_distill/n_active_tok": 141.125,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 9.17578125,
      "calib/ece": 0.2766963855421687,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.5542168674698795,
      "calib/gap": 0.2997629651390521,
      "calib/mean_conf": 0.6042859437751005,
      "calib/mu_c": 0.7379152173913044,
      "calib/mu_w": 0.4381522522522523,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1633827309236948,
      "calib/std_conf": 0.43625114971277235,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25048216007714563,
      "calib/step_q_c_n": 1037.0,
      "calib/step_q_gap": 0.0437382576381212,
      "calib/step_q_w": 0.20674390243902444,
      "calib/step_q_w_n": 1312.0,
      "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": 1584.0,
      "completions/max_terminated_length": 1584.0,
      "completions/mean_length": 462.2578125,
      "completions/mean_terminated_length": 475.25299072265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 158.0,
      "epoch": 0.12906666666666666,
      "grad_norm": 0.8531492352485657,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0536,
      "num_tokens": 23586027.0,
      "reward": 1.100294828414917,
      "reward_std": 0.25339704751968384,
      "rewards/accuracy_reward_step": 0.5390625,
      "rewards/final_brier_reward_step": 0.6888708472251892,
      "rewards/format_reward_step": 0.97265625,
      "step": 121
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8276457553729415,
      "aux_distill/mean_u": 0.2641006238337189,
      "aux_distill/n_active_tok": 126.375,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 9.01171875,
      "calib/ece": 0.21577510040160652,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.7469879518072289,
      "calib/gap": 0.35342051649928263,
      "calib/mean_conf": 0.765269076305221,
      "calib/mu_c": 0.8859146341463416,
      "calib/mu_w": 0.532494117647059,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.16120481927710853,
      "calib/std_conf": 0.3897101743185824,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23407108239095317,
      "calib/step_q_c_n": 1238.0,
      "calib/step_q_gap": 0.05060990371929741,
      "calib/step_q_w": 0.18346117867165576,
      "calib/step_q_w_n": 1069.0,
      "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": 2604.0,
      "completions/max_terminated_length": 2604.0,
      "completions/mean_length": 418.12890625,
      "completions/mean_terminated_length": 429.8835144042969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 182.0,
      "epoch": 0.13013333333333332,
      "grad_norm": 0.5496814250946045,
      "learning_rate": 2.1944444444444445e-06,
      "loss": 0.0207,
      "num_tokens": 23800412.0,
      "reward": 1.181523084640503,
      "reward_std": 0.2424289733171463,
      "rewards/accuracy_reward_step": 0.640625,
      "rewards/final_brier_reward_step": 0.7497649192810059,
      "rewards/format_reward_step": 0.97265625,
      "step": 122
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7614726750180125,
      "aux_distill/mean_u": 0.19446524486235212,
      "aux_distill/n_active_tok": 157.375,
      "calib/answer_extract_rate": 0.96484375,
      "calib/avg_num_step_conf": 9.91015625,
      "calib/ece": 0.21352225086617072,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.5991902834008097,
      "calib/gap": 0.41288853684169874,
      "calib/mean_conf": 0.6422671901374257,
      "calib/mu_c": 0.7977272465191181,
      "calib/mu_w": 0.38483870967741934,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.11615382981353915,
      "calib/std_conf": 0.4308034461532648,
      "calib/step_conf_rate": 0.9921875,
      "calib/step_q_c": 0.2534844621513944,
      "calib/step_q_c_n": 1255.0,
      "calib/step_q_gap": -0.07045063925266176,
      "calib/step_q_w": 0.3239351014040562,
      "calib/step_q_w_n": 1282.0,
      "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": 2566.0,
      "completions/max_terminated_length": 2566.0,
      "completions/mean_length": 494.14453125,
      "completions/mean_terminated_length": 510.08465576171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 171.0,
      "epoch": 0.1312,
      "grad_norm": 1.6037333011627197,
      "learning_rate": 2.166666666666667e-06,
      "loss": 0.0292,
      "num_tokens": 24032201.0,
      "reward": 1.1502926349639893,
      "reward_std": 0.24385644495487213,
      "rewards/accuracy_reward_step": 0.6015625,
      "rewards/final_brier_reward_step": 0.741991400718689,
      "rewards/format_reward_step": 0.95703125,
      "step": 123
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.824720211327076,
      "aux_distill/mean_u": 0.24733909527073586,
      "aux_distill/n_active_tok": 115.125,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 7.8828125,
      "calib/ece": 0.24675849802371536,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.7193675889328063,
      "calib/gap": 0.29872926829268287,
      "calib/mean_conf": 0.7500786561264823,
      "calib/mu_c": 0.8469000000000001,
      "calib/mu_w": 0.5481707317073172,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1604739130434782,
      "calib/std_conf": 0.3967623338775554,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24739656301145663,
      "calib/step_q_c_n": 1222.0,
      "calib/step_q_gap": -0.024907205832764512,
      "calib/step_q_w": 0.27230376884422114,
      "calib/step_q_w_n": 796.0,
      "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": 1346.0,
      "completions/max_terminated_length": 1346.0,
      "completions/mean_length": 439.67578125,
      "completions/mean_terminated_length": 444.88934326171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 168.0,
      "epoch": 0.13226666666666667,
      "grad_norm": 0.6115875840187073,
      "learning_rate": 2.138888888888889e-06,
      "loss": 0.0567,
      "num_tokens": 24251574.0,
      "reward": 1.1981837749481201,
      "reward_std": 0.2457866370677948,
      "rewards/accuracy_reward_step": 0.66796875,
      "rewards/final_brier_reward_step": 0.7401175498962402,
      "rewards/format_reward_step": 0.98828125,
      "step": 124
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7017071144655347,
      "aux_distill/mean_u": 0.15763443178974784,
      "aux_distill/n_active_tok": 116.5,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 8.8984375,
      "calib/ece": 0.3583236274653614,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.7991967871485943,
      "calib/gap": 0.12134742182954572,
      "calib/mean_conf": 0.8181421013609436,
      "calib/mu_c": 0.8663886666666666,
      "calib/mu_w": 0.7450412448371209,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.28702804513604413,
      "calib/std_conf": 0.3617559594279362,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23328895814781833,
      "calib/step_q_c_n": 1123.0,
      "calib/step_q_gap": 0.03644211832097849,
      "calib/step_q_w": 0.19684683982683984,
      "calib/step_q_w_n": 1155.0,
      "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": 2637.0,
      "completions/max_terminated_length": 2637.0,
      "completions/mean_length": 444.05859375,
      "completions/mean_terminated_length": 456.5421447753906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 165.0,
      "epoch": 0.13333333333333333,
      "grad_norm": 0.5504864454269409,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 0.0249,
      "num_tokens": 24470061.0,
      "reward": 1.0911345481872559,
      "reward_std": 0.22019320726394653,
      "rewards/accuracy_reward_step": 0.5859375,
      "rewards/final_brier_reward_step": 0.6236752271652222,
      "rewards/format_reward_step": 0.97265625,
      "step": 125
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7042123721912503,
      "aux_distill/mean_u": 0.19295040940696292,
      "aux_distill/n_active_tok": 128.5,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 8.70703125,
      "calib/ece": 0.3480152000000001,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.788,
      "calib/gap": 0.18341957186544344,
      "calib/mean_conf": 0.8060624000000001,
      "calib/mu_c": 0.8860333333333333,
      "calib/mu_w": 0.7026137614678899,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2950388000000001,
      "calib/std_conf": 0.37368177331285507,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2567842105263158,
      "calib/step_q_c_n": 1045.0,
      "calib/step_q_gap": 0.005554987553342827,
      "calib/step_q_w": 0.25122922297297295,
      "calib/step_q_w_n": 1184.0,
      "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": 0.0234375,
      "completions/max_length": 1498.0,
      "completions/max_terminated_length": 1498.0,
      "completions/mean_length": 438.39453125,
      "completions/mean_terminated_length": 448.916015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 145.0,
      "epoch": 0.1344,
      "grad_norm": 0.5972703695297241,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.0183,
      "num_tokens": 24687754.0,
      "reward": 1.0752307176589966,
      "reward_std": 0.20789799094200134,
      "rewards/accuracy_reward_step": 0.55078125,
      "rewards/final_brier_reward_step": 0.6270239353179932,
      "rewards/format_reward_step": 0.97265625,
      "step": 126
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.6871523316949606,
      "aux_distill/mean_u": 0.17274493110363137,
      "aux_distill/n_active_tok": 130.25,
      "calib/answer_extract_rate": 0.9375,
      "calib/avg_num_step_conf": 10.15625,
      "calib/ece": 0.3320694891576859,
      "calib/final_conf_rate": 0.9375,
      "calib/format_rate": 0.9375,
      "calib/frac_conf_gt_0.9": 0.8166666666666667,
      "calib/gap": 0.21511819003880417,
      "calib/mean_conf": 0.8359305108423143,
      "calib/mu_c": 0.9309410447761194,
      "calib/mu_w": 0.7158228547373152,
      "calib/nonempty_final_conf_rate": 0.9375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.30483333333333346,
      "calib/std_conf": 0.3469884047970489,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25272319277108435,
      "calib/step_q_c_n": 996.0,
      "calib/step_q_gap": 0.0026969132996860767,
      "calib/step_q_w": 0.25002627947139827,
      "calib/step_q_w_n": 1604.0,
      "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": 0.0625,
      "completions/max_length": 1793.0,
      "completions/max_terminated_length": 1793.0,
      "completions/mean_length": 419.29296875,
      "completions/mean_terminated_length": 447.245849609375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 142.0,
      "epoch": 0.13546666666666668,
      "grad_norm": 0.6025396585464478,
      "learning_rate": 2.0555555555555555e-06,
      "loss": -0.0473,
      "num_tokens": 24898765.0,
      "reward": 1.0407984256744385,
      "reward_std": 0.25408264994621277,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.6206594109535217,
      "rewards/format_reward_step": 0.9375,
      "step": 127
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8674097079783678,
      "aux_distill/mean_u": 0.26414057126039797,
      "aux_distill/n_active_tok": 113.75,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 8.3203125,
      "calib/ece": 0.26362549800796825,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.8326693227091634,
      "calib/gap": 0.3066482431898936,
      "calib/mean_conf": 0.8496812749003984,
      "calib/mu_c": 0.974295302013423,
      "calib/mu_w": 0.6676470588235294,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2598406374501993,
      "calib/std_conf": 0.3359945367014103,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24466067864271454,
      "calib/step_q_c_n": 1002.0,
      "calib/step_q_gap": -0.0230578497260798,
      "calib/step_q_w": 0.26771852836879434,
      "calib/step_q_w_n": 1128.0,
      "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": 0.0234375,
      "completions/max_length": 1386.0,
      "completions/max_terminated_length": 1386.0,
      "completions/mean_length": 423.8828125,
      "completions/mean_terminated_length": 434.0560302734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.13653333333333334,
      "grad_norm": 0.6722913980484009,
      "learning_rate": 2.027777777777778e-06,
      "loss": 0.0224,
      "num_tokens": 25113943.0,
      "reward": 1.1382665634155273,
      "reward_std": 0.22327424585819244,
      "rewards/accuracy_reward_step": 0.58203125,
      "rewards/final_brier_reward_step": 0.7140331268310547,
      "rewards/format_reward_step": 0.98046875,
      "step": 128
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8107666317373514,
      "aux_distill/mean_u": 0.30531200243128254,
      "aux_distill/n_active_tok": 140.625,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.48828125,
      "calib/ece": 0.2943433070866142,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8818897637795275,
      "calib/gap": 0.16955083199141174,
      "calib/mean_conf": 0.897703937007874,
      "calib/mu_c": 0.9591160493827161,
      "calib/mu_w": 0.7895652173913044,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.27712598425196855,
      "calib/std_conf": 0.2804184666550919,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2683291239147592,
      "calib/step_q_c_n": 1267.0,
      "calib/step_q_gap": 0.026084090802176413,
      "calib/step_q_w": 0.2422450331125828,
      "calib/step_q_w_n": 906.0,
      "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": 1864.0,
      "completions/max_terminated_length": 1864.0,
      "completions/mean_length": 430.71875,
      "completions/mean_terminated_length": 435.82611083984375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 146.0,
      "epoch": 0.1376,
      "grad_norm": 0.5686946511268616,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0528,
      "num_tokens": 25326591.0,
      "reward": 1.160329818725586,
      "reward_std": 0.20995208621025085,
      "rewards/accuracy_reward_step": 0.6328125,
      "rewards/final_brier_reward_step": 0.6956596374511719,
      "rewards/format_reward_step": 0.9921875,
      "step": 129
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8531761262565851,
      "aux_distill/mean_u": 0.21606510653578878,
      "aux_distill/n_active_tok": 113.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.3671875,
      "calib/ece": 0.25872795275590554,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8937007874015748,
      "calib/gap": 0.20760023250898318,
      "calib/mean_conf": 0.907074409448819,
      "calib/mu_c": 0.9749122807017543,
      "calib/mu_w": 0.7673120481927711,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24628700787401575,
      "calib/std_conf": 0.26971461735983954,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22322997746055598,
      "calib/step_q_c_n": 1331.0,
      "calib/step_q_gap": 0.031155748114070153,
      "calib/step_q_w": 0.19207422934648583,
      "calib/step_q_w_n": 811.0,
      "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": 1929.0,
      "completions/max_terminated_length": 1929.0,
      "completions/mean_length": 432.1796875,
      "completions/mean_terminated_length": 437.3043518066406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.13866666666666666,
      "grad_norm": 0.35849037766456604,
      "learning_rate": 1.9722222222222224e-06,
      "loss": 0.0914,
      "num_tokens": 25542517.0,
      "reward": 1.1991316080093384,
      "reward_std": 0.1761476993560791,
      "rewards/accuracy_reward_step": 0.66796875,
      "rewards/final_brier_reward_step": 0.7381070256233215,
      "rewards/format_reward_step": 0.9921875,
      "step": 130
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7969897780567408,
      "aux_distill/mean_u": 0.21326160858040874,
      "aux_distill/n_active_tok": 132.375,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 8.53125,
      "calib/ece": 0.3938095238095238,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.8571428571428571,
      "calib/gap": 0.1807711693548386,
      "calib/mean_conf": 0.8728571428571428,
      "calib/mu_c": 0.9646774193548386,
      "calib/mu_w": 0.78390625,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.38730158730158726,
      "calib/std_conf": 0.30949358826821977,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23247561761546723,
      "calib/step_q_c_n": 931.0,
      "calib/step_q_gap": 0.012266838684900583,
      "calib/step_q_w": 0.22020877893056665,
      "calib/step_q_w_n": 1253.0,
      "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": 2988.0,
      "completions/max_terminated_length": 2988.0,
      "completions/mean_length": 432.69140625,
      "completions/mean_terminated_length": 437.8221435546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 174.0,
      "epoch": 0.13973333333333332,
      "grad_norm": 0.6910375952720642,
      "learning_rate": 1.944444444444445e-06,
      "loss": 0.0493,
      "num_tokens": 25759494.0,
      "reward": 1.029508113861084,
      "reward_std": 0.18506643176078796,
      "rewards/accuracy_reward_step": 0.484375,
      "rewards/final_brier_reward_step": 0.5902664065361023,
      "rewards/format_reward_step": 0.984375,
      "step": 131
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8498719027265906,
      "aux_distill/mean_u": 0.290067270262824,
      "aux_distill/n_active_tok": 120.125,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 8.43359375,
      "calib/ece": 0.2372379211918686,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.8849206349206349,
      "calib/gap": 0.2252981316928686,
      "calib/mean_conf": 0.8945874756335284,
      "calib/mu_c": 0.9634285714285715,
      "calib/mu_w": 0.7381304397357029,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2186904761904762,
      "calib/std_conf": 0.2894467810542942,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2468620178041543,
      "calib/step_q_c_n": 1348.0,
      "calib/step_q_gap": 0.04331084641081273,
      "calib/step_q_w": 0.20355117139334156,
      "calib/step_q_w_n": 811.0,
      "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": 1766.0,
      "completions/max_terminated_length": 1766.0,
      "completions/mean_length": 452.33984375,
      "completions/mean_terminated_length": 461.3506164550781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 147.0,
      "epoch": 0.1408,
      "grad_norm": 0.5889887809753418,
      "learning_rate": 1.916666666666667e-06,
      "loss": 0.0537,
      "num_tokens": 25980885.0,
      "reward": 1.207842469215393,
      "reward_std": 0.29401665925979614,
      "rewards/accuracy_reward_step": 0.68359375,
      "rewards/final_brier_reward_step": 0.7477160692214966,
      "rewards/format_reward_step": 0.984375,
      "step": 132
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.799846863374114,
      "aux_distill/mean_u": 0.26558087362637905,
      "aux_distill/n_active_tok": 160.625,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.68359375,
      "calib/ece": 0.36370354330708676,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.8937007874015748,
      "calib/gap": 0.14455491085491112,
      "calib/mean_conf": 0.9048003937007874,
      "calib/mu_c": 0.9679720279720281,
      "calib/mu_w": 0.823417117117117,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.3527559055118112,
      "calib/std_conf": 0.2743696667909248,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.24034103299856527,
      "calib/step_q_c_n": 1394.0,
      "calib/step_q_gap": -0.003924036125858704,
      "calib/step_q_w": 0.24426506912442397,
      "calib/step_q_w_n": 1085.0,
      "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": 1330.0,
      "completions/max_terminated_length": 1330.0,
      "completions/mean_length": 534.34375,
      "completions/mean_terminated_length": 540.6798706054688,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 191.0,
      "epoch": 0.14186666666666667,
      "grad_norm": 0.40780898928642273,
      "learning_rate": 1.888888888888889e-06,
      "loss": 0.0611,
      "num_tokens": 26224021.0,
      "reward": 1.0855050086975098,
      "reward_std": 0.33582109212875366,
      "rewards/accuracy_reward_step": 0.55859375,
      "rewards/final_brier_reward_step": 0.6241352558135986,
      "rewards/format_reward_step": 0.98828125,
      "step": 133
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7864964846521616,
      "aux_distill/mean_u": 0.21162502770453695,
      "aux_distill/n_active_tok": 146.625,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 10.14453125,
      "calib/ece": 0.34138379446640316,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.7707509881422925,
      "calib/gap": 0.25076517499999995,
      "calib/mean_conf": 0.7975885375494071,
      "calib/mu_c": 0.9214843749999999,
      "calib/mu_w": 0.6707192,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.31652173913043474,
      "calib/std_conf": 0.37247587807807786,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24591915444348578,
      "calib/step_q_c_n": 1159.0,
      "calib/step_q_gap": 0.02796393886629528,
      "calib/step_q_w": 0.2179552155771905,
      "calib/step_q_w_n": 1438.0,
      "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": 2285.0,
      "completions/max_terminated_length": 2285.0,
      "completions/mean_length": 553.34765625,
      "completions/mean_terminated_length": 559.9091186523438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.14293333333333333,
      "grad_norm": 0.34763002395629883,
      "learning_rate": 1.8611111111111113e-06,
      "loss": 0.0455,
      "num_tokens": 26474630.0,
      "reward": 1.0661208629608154,
      "reward_std": 0.26195263862609863,
      "rewards/accuracy_reward_step": 0.5,
      "rewards/final_brier_reward_step": 0.6439605355262756,
      "rewards/format_reward_step": 0.98828125,
      "step": 134
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7998809479176998,
      "aux_distill/mean_u": 0.23606169165538424,
      "aux_distill/n_active_tok": 139.875,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 10.41015625,
      "calib/ece": 0.3348012096774194,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.8709677419354839,
      "calib/gap": 0.1498497541590893,
      "calib/mean_conf": 0.8940697580645163,
      "calib/mu_c": 0.9550972789115645,
      "calib/mu_w": 0.8052475247524752,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.3180645161290323,
      "calib/std_conf": 0.2821037821153988,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24970825892857146,
      "calib/step_q_c_n": 1344.0,
      "calib/step_q_gap": -0.002882049343265969,
      "calib/step_q_w": 0.2525903082718374,
      "calib/step_q_w_n": 1321.0,
      "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": 1928.0,
      "completions/max_terminated_length": 1928.0,
      "completions/mean_length": 491.953125,
      "completions/mean_terminated_length": 507.82257080078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 173.0,
      "epoch": 0.144,
      "grad_norm": 0.7895566821098328,
      "learning_rate": 1.8333333333333333e-06,
      "loss": 0.0048,
      "num_tokens": 26706450.0,
      "reward": 1.0914466381072998,
      "reward_std": 0.31719958782196045,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.6399244666099548,
      "rewards/format_reward_step": 0.96875,
      "step": 135
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8619387056678534,
      "aux_distill/mean_u": 0.23585160573000902,
      "aux_distill/n_active_tok": 147.375,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 11.0859375,
      "calib/ece": 0.22354448979591832,
      "calib/final_conf_rate": 0.95703125,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.636734693877551,
      "calib/gap": 0.40863818132757695,
      "calib/mean_conf": 0.6740065306122449,
      "calib/mu_c": 0.8558088235294118,
      "calib/mu_w": 0.44717064220183483,
      "calib/nonempty_final_conf_rate": 0.95703125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1712244897959183,
      "calib/std_conf": 0.43286631545939436,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2619659426585577,
      "calib/step_q_c_n": 1151.0,
      "calib/step_q_gap": 0.0020058952370994865,
      "calib/step_q_w": 0.25996004742145823,
      "calib/step_q_w_n": 1687.0,
      "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": 0.04296875,
      "completions/max_length": 1701.0,
      "completions/max_terminated_length": 1701.0,
      "completions/mean_length": 472.9140625,
      "completions/mean_terminated_length": 494.14691162109375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.14506666666666668,
      "grad_norm": 0.5085462927818298,
      "learning_rate": 1.8055555555555557e-06,
      "loss": 0.036,
      "num_tokens": 26936004.0,
      "reward": 1.1046361923217773,
      "reward_std": 0.20290738344192505,
      "rewards/accuracy_reward_step": 0.53125,
      "rewards/final_brier_reward_step": 0.720991313457489,
      "rewards/format_reward_step": 0.95703125,
      "step": 136
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7779526645317674,
      "aux_distill/mean_u": 0.24285273704166116,
      "aux_distill/n_active_tok": 143.25,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.35546875,
      "calib/ece": 0.23234280236066573,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.5866141732283464,
      "calib/gap": 0.3843206332307751,
      "calib/mean_conf": 0.6412005047259485,
      "calib/mu_c": 0.8257954545454547,
      "calib/mu_w": 0.4414748213146796,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.17692913385826775,
      "calib/std_conf": 0.4305953745895403,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25945538847117794,
      "calib/step_q_c_n": 1197.0,
      "calib/step_q_gap": 0.03860481105481495,
      "calib/step_q_w": 0.220850577416363,
      "calib/step_q_w_n": 1198.0,
      "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": 2020.0,
      "completions/max_terminated_length": 2020.0,
      "completions/mean_length": 485.73046875,
      "completions/mean_terminated_length": 489.55511474609375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 139.0,
      "epoch": 0.14613333333333334,
      "grad_norm": 0.46715131402015686,
      "learning_rate": 1.777777777777778e-06,
      "loss": 0.0587,
      "num_tokens": 27167335.0,
      "reward": 1.1220433712005615,
      "reward_std": 0.19661134481430054,
      "rewards/accuracy_reward_step": 0.515625,
      "rewards/final_brier_reward_step": 0.7362741827964783,
      "rewards/format_reward_step": 0.9921875,
      "step": 137
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8091667117550969,
      "aux_distill/mean_u": 0.2716275383094954,
      "aux_distill/n_active_tok": 127.875,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 10.17578125,
      "calib/ece": 0.2742096385542168,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.5863453815261044,
      "calib/gap": 0.30169601889338743,
      "calib/mean_conf": 0.6436208835341366,
      "calib/mu_c": 0.7381280701754387,
      "calib/mu_w": 0.43643205128205126,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11554176706827307,
      "calib/std_conf": 0.42848028822246104,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2989956616052061,
      "calib/step_q_c_n": 1383.0,
      "calib/step_q_gap": 0.02238592347099988,
      "calib/step_q_w": 0.2766097381342062,
      "calib/step_q_w_n": 1222.0,
      "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": 2664.0,
      "completions/max_terminated_length": 2664.0,
      "completions/mean_length": 497.33203125,
      "completions/mean_terminated_length": 511.313232421875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 123.0,
      "epoch": 0.1472,
      "grad_norm": 0.4575296640396118,
      "learning_rate": 1.75e-06,
      "loss": 0.0189,
      "num_tokens": 27398988.0,
      "reward": 1.174954891204834,
      "reward_std": 0.2306070327758789,
      "rewards/accuracy_reward_step": 0.66796875,
      "rewards/final_brier_reward_step": 0.7092845439910889,
      "rewards/format_reward_step": 0.97265625,
      "step": 138
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8192444164305925,
      "aux_distill/mean_u": 0.2686307010511263,
      "aux_distill/n_active_tok": 122.625,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 8.55078125,
      "calib/ece": 0.19777051792828676,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.5179282868525896,
      "calib/gap": 0.4742080321285141,
      "calib/mean_conf": 0.5787235059760957,
      "calib/mu_c": 0.7355333333333334,
      "calib/mu_w": 0.2613253012048193,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05358565737051783,
      "calib/std_conf": 0.43468715302071437,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2363246855345912,
      "calib/step_q_c_n": 1272.0,
      "calib/step_q_gap": -0.03599483463989081,
      "calib/step_q_w": 0.272319520174482,
      "calib/step_q_w_n": 917.0,
      "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": 1611.0,
      "completions/max_terminated_length": 1611.0,
      "completions/mean_length": 423.83203125,
      "completions/mean_terminated_length": 432.27490234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 160.0,
      "epoch": 0.14826666666666666,
      "grad_norm": 0.47275084257125854,
      "learning_rate": 1.7222222222222224e-06,
      "loss": 0.0267,
      "num_tokens": 27610585.0,
      "reward": 1.2063415050506592,
      "reward_std": 0.16486656665802002,
      "rewards/accuracy_reward_step": 0.65625,
      "rewards/final_brier_reward_step": 0.7759643197059631,
      "rewards/format_reward_step": 0.98046875,
      "step": 139
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7892370466142893,
      "aux_distill/mean_u": 0.22641714346360392,
      "aux_distill/n_active_tok": 144.625,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 9.5234375,
      "calib/ece": 0.1941599999999999,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.764,
      "calib/gap": 0.333031746031746,
      "calib/mean_conf": 0.79464,
      "calib/mu_c": 0.8878888888888888,
      "calib/mu_w": 0.5548571428571428,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1343999999999999,
      "calib/std_conf": 0.36501406876995857,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2524066025641026,
      "calib/step_q_c_n": 1560.0,
      "calib/step_q_gap": 0.0037251674843759297,
      "calib/step_q_w": 0.24868143507972665,
      "calib/step_q_w_n": 878.0,
      "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": 0.0234375,
      "completions/max_length": 1140.0,
      "completions/max_terminated_length": 1140.0,
      "completions/mean_length": 462.64453125,
      "completions/mean_terminated_length": 473.7480163574219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 147.0,
      "epoch": 0.14933333333333335,
      "grad_norm": 0.3123050034046173,
      "learning_rate": 1.6944444444444446e-06,
      "loss": 0.0137,
      "num_tokens": 27834038.0,
      "reward": 1.2274765968322754,
      "reward_std": 0.21313458681106567,
      "rewards/accuracy_reward_step": 0.703125,
      "rewards/final_brier_reward_step": 0.7752655744552612,
      "rewards/format_reward_step": 0.9765625,
      "step": 140
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7888972610235214,
      "aux_distill/mean_u": 0.2564832817743102,
      "aux_distill/n_active_tok": 138.5,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 9.1015625,
      "calib/ece": 0.12324479999999999,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.74,
      "calib/gap": 0.5349787951345657,
      "calib/mean_conf": 0.7746048000000001,
      "calib/mu_c": 0.8880203045685279,
      "calib/mu_w": 0.35304150943396223,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05492480000000001,
      "calib/std_conf": 0.3798887015652875,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2544185230769231,
      "calib/step_q_c_n": 1625.0,
      "calib/step_q_gap": 0.0246826365521004,
      "calib/step_q_w": 0.22973588652482269,
      "calib/step_q_w_n": 705.0,
      "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": 0.0234375,
      "completions/max_length": 1647.0,
      "completions/max_terminated_length": 1647.0,
      "completions/mean_length": 497.48828125,
      "completions/mean_terminated_length": 509.4280090332031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 188.0,
      "epoch": 0.1504,
      "grad_norm": 0.26303279399871826,
      "learning_rate": 1.6666666666666667e-06,
      "loss": 0.0068,
      "num_tokens": 28068491.0,
      "reward": 1.296480417251587,
      "reward_std": 0.19722458720207214,
      "rewards/accuracy_reward_step": 0.76953125,
      "rewards/final_brier_reward_step": 0.8468671441078186,
      "rewards/format_reward_step": 0.9765625,
      "step": 141
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7698502484709024,
      "aux_distill/mean_u": 0.20123263576425643,
      "aux_distill/n_active_tok": 141.75,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 9.453125,
      "calib/ece": 0.18946000000000002,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.664,
      "calib/gap": 0.45957737148913624,
      "calib/mean_conf": 0.70006,
      "calib/mu_c": 0.8875675675675676,
      "calib/mu_w": 0.42799019607843136,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.14876000000000003,
      "calib/std_conf": 0.41952937489525094,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2402120717781403,
      "calib/step_q_c_n": 1226.0,
      "calib/step_q_gap": -0.004463807618844656,
      "calib/step_q_w": 0.24467587939698496,
      "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": 0.01953125,
      "completions/max_length": 1467.0,
      "completions/max_terminated_length": 1467.0,
      "completions/mean_length": 506.0,
      "completions/mean_terminated_length": 516.0797119140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 177.0,
      "epoch": 0.15146666666666667,
      "grad_norm": 0.32701781392097473,
      "learning_rate": 1.638888888888889e-06,
      "loss": 0.0514,
      "num_tokens": 28303187.0,
      "reward": 1.1644487380981445,
      "reward_std": 0.23320737481117249,
      "rewards/accuracy_reward_step": 0.578125,
      "rewards/final_brier_reward_step": 0.7742100954055786,
      "rewards/format_reward_step": 0.9765625,
      "step": 142
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8147965837270021,
      "aux_distill/mean_u": 0.22807440234532067,
      "aux_distill/n_active_tok": 126.375,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 9.73828125,
      "calib/ece": 0.2558064516129033,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.7782258064516129,
      "calib/gap": 0.344020680055677,
      "calib/mean_conf": 0.807741935483871,
      "calib/mu_c": 0.9561702127659574,
      "calib/mu_w": 0.6121495327102804,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.24750000000000003,
      "calib/std_conf": 0.3621215236889883,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.24809659090909092,
      "calib/step_q_c_n": 1056.0,
      "calib/step_q_gap": -0.049370354115265414,
      "calib/step_q_w": 0.29746694502435633,
      "calib/step_q_w_n": 1437.0,
      "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": 2738.0,
      "completions/max_terminated_length": 2738.0,
      "completions/mean_length": 480.73828125,
      "completions/mean_terminated_length": 494.25299072265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 179.0,
      "epoch": 0.15253333333333333,
      "grad_norm": 0.4245584011077881,
      "learning_rate": 1.6111111111111113e-06,
      "loss": 0.0173,
      "num_tokens": 28533592.0,
      "reward": 1.112665057182312,
      "reward_std": 0.22011610865592957,
      "rewards/accuracy_reward_step": 0.55078125,
      "rewards/final_brier_reward_step": 0.709705114364624,
      "rewards/format_reward_step": 0.96484375,
      "step": 143
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8023251444101334,
      "aux_distill/mean_u": 0.24155563343629866,
      "aux_distill/n_active_tok": 126.375,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 9.10546875,
      "calib/ece": 0.20767667984189725,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.7351778656126482,
      "calib/gap": 0.3498372549019608,
      "calib/mean_conf": 0.774299604743083,
      "calib/mu_c": 0.8655614973262032,
      "calib/mu_w": 0.5157242424242424,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12142292490118577,
      "calib/std_conf": 0.38275148920672053,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20582251407129454,
      "calib/step_q_c_n": 1599.0,
      "calib/step_q_gap": -0.04698650232214807,
      "calib/step_q_w": 0.2528090163934426,
      "calib/step_q_w_n": 732.0,
      "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": 2042.0,
      "completions/max_terminated_length": 2042.0,
      "completions/mean_length": 497.22265625,
      "completions/mean_terminated_length": 505.1151123046875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.1536,
      "grad_norm": 0.3189343810081482,
      "learning_rate": 1.5833333333333333e-06,
      "loss": 0.0464,
      "num_tokens": 28765009.0,
      "reward": 1.251899003982544,
      "reward_std": 0.1643814742565155,
      "rewards/accuracy_reward_step": 0.73046875,
      "rewards/final_brier_reward_step": 0.7850477695465088,
      "rewards/format_reward_step": 0.98828125,
      "step": 144
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7487006606534123,
      "aux_distill/mean_u": 0.22828780529426004,
      "aux_distill/n_active_tok": 154.25,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 9.53515625,
      "calib/ece": 0.24680318725099604,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.8406374501992032,
      "calib/gap": 0.24577970233876667,
      "calib/mean_conf": 0.8661051792828686,
      "calib/mu_c": 0.9493373493975902,
      "calib/mu_w": 0.7035576470588235,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.22577689243027887,
      "calib/std_conf": 0.31133066281148913,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2027402123424021,
      "calib/step_q_c_n": 1507.0,
      "calib/step_q_gap": -0.055428952539824855,
      "calib/step_q_w": 0.25816916488222696,
      "calib/step_q_w_n": 934.0,
      "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": 1423.0,
      "completions/max_terminated_length": 1423.0,
      "completions/mean_length": 486.19921875,
      "completions/mean_terminated_length": 495.8844909667969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.15466666666666667,
      "grad_norm": 0.3531167805194855,
      "learning_rate": 1.5555555555555558e-06,
      "loss": 0.0375,
      "num_tokens": 28992180.0,
      "reward": 1.1807942390441895,
      "reward_std": 0.2349383682012558,
      "rewards/accuracy_reward_step": 0.6484375,
      "rewards/final_brier_reward_step": 0.7326824069023132,
      "rewards/format_reward_step": 0.98046875,
      "step": 145
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8367437515407801,
      "aux_distill/mean_u": 0.26077006896482124,
      "aux_distill/n_active_tok": 185.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.51171875,
      "calib/ece": 0.31686507936507935,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.753968253968254,
      "calib/gap": 0.34171163202122956,
      "calib/mean_conf": 0.7851984126984127,
      "calib/mu_c": 0.965546218487395,
      "calib/mu_w": 0.6238345864661654,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.3149206349206349,
      "calib/std_conf": 0.3751057527041757,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.20598406593406593,
      "calib/step_q_c_n": 1092.0,
      "calib/step_q_gap": -0.025591919056553197,
      "calib/step_q_w": 0.23157598499061913,
      "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": 0.01953125,
      "completions/max_length": 1575.0,
      "completions/max_terminated_length": 1575.0,
      "completions/mean_length": 532.8828125,
      "completions/mean_terminated_length": 543.498046875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 184.0,
      "epoch": 0.15573333333333333,
      "grad_norm": 0.3752824366092682,
      "learning_rate": 1.527777777777778e-06,
      "loss": 0.04,
      "num_tokens": 29235814.0,
      "reward": 1.0565924644470215,
      "reward_std": 0.2701501250267029,
      "rewards/accuracy_reward_step": 0.46484375,
      "rewards/final_brier_reward_step": 0.6678722500801086,
      "rewards/format_reward_step": 0.98046875,
      "step": 146
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8423489462584257,
      "aux_distill/mean_u": 0.25418225832492053,
      "aux_distill/n_active_tok": 157.625,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 9.234375,
      "calib/ece": 0.33116553015758,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.8152610441767069,
      "calib/gap": 0.16782206675464206,
      "calib/mean_conf": 0.8400649919307734,
      "calib/mu_c": 0.9108333333333333,
      "calib/mu_w": 0.7430112665786912,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2964586345381526,
      "calib/std_conf": 0.33575434298393025,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2360386038687973,
      "calib/step_q_c_n": 1189.0,
      "calib/step_q_gap": 0.021529858653178607,
      "calib/step_q_w": 0.2145087452156187,
      "calib/step_q_w_n": 1175.0,
      "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": 0.0234375,
      "completions/max_length": 2809.0,
      "completions/max_terminated_length": 2809.0,
      "completions/mean_length": 502.1875,
      "completions/mean_terminated_length": 514.2400512695312,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 157.0,
      "epoch": 0.1568,
      "grad_norm": 0.33426088094711304,
      "learning_rate": 1.5e-06,
      "loss": 0.0549,
      "num_tokens": 29468054.0,
      "reward": 1.0869696140289307,
      "reward_std": 0.1892041116952896,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.638782799243927,
      "rewards/format_reward_step": 0.97265625,
      "step": 147
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7472190484404564,
      "aux_distill/mean_u": 0.20466814080151272,
      "aux_distill/n_active_tok": 133.5,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 9.3359375,
      "calib/ece": 0.19012244897959188,
      "calib/final_conf_rate": 0.95703125,
      "calib/format_rate": 0.953125,
      "calib/frac_conf_gt_0.9": 0.8816326530612245,
      "calib/gap": 0.2648595848595847,
      "calib/mean_conf": 0.894530612244898,
      "calib/mu_c": 0.9626373626373625,
      "calib/mu_w": 0.6977777777777778,
      "calib/nonempty_final_conf_rate": 0.95703125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1708979591836735,
      "calib/std_conf": 0.28752619061118634,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2247715395480226,
      "calib/step_q_c_n": 1416.0,
      "calib/step_q_gap": -0.08958164320351741,
      "calib/step_q_w": 0.31435318275154,
      "calib/step_q_w_n": 974.0,
      "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": 1790.0,
      "completions/max_terminated_length": 1790.0,
      "completions/mean_length": 455.68359375,
      "completions/mean_terminated_length": 474.2073059082031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 90.0,
      "epoch": 0.15786666666666666,
      "grad_norm": 0.41133859753608704,
      "learning_rate": 1.4722222222222225e-06,
      "loss": -0.002,
      "num_tokens": 29689821.0,
      "reward": 1.2150394916534424,
      "reward_std": 0.21160367131233215,
      "rewards/accuracy_reward_step": 0.7109375,
      "rewards/final_brier_reward_step": 0.76601642370224,
      "rewards/format_reward_step": 0.953125,
      "step": 148
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7386438930407166,
      "aux_distill/mean_u": 0.16422576102609746,
      "aux_distill/n_active_tok": 129.75,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 9.6015625,
      "calib/ece": 0.25140161752988055,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.8326693227091634,
      "calib/gap": 0.26694276221587043,
      "calib/mean_conf": 0.8634979840637451,
      "calib/mu_c": 0.9624050632911393,
      "calib/mu_w": 0.6954623010752689,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.2427087649402391,
      "calib/std_conf": 0.3119481918486996,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.22707812499999996,
      "calib/step_q_c_n": 1280.0,
      "calib/step_q_gap": 0.023896622453310673,
      "calib/step_q_w": 0.2031815025466893,
      "calib/step_q_w_n": 1178.0,
      "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": 0.0234375,
      "completions/max_length": 2860.0,
      "completions/max_terminated_length": 2860.0,
      "completions/mean_length": 525.71484375,
      "completions/mean_terminated_length": 538.33203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.15893333333333334,
      "grad_norm": 0.2715623676776886,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 0.0575,
      "num_tokens": 29928860.0,
      "reward": 1.157318115234375,
      "reward_std": 0.22797748446464539,
      "rewards/accuracy_reward_step": 0.6171875,
      "rewards/final_brier_reward_step": 0.7208861112594604,
      "rewards/format_reward_step": 0.9765625,
      "step": 149
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8586932197213173,
      "aux_distill/mean_u": 0.2912806011103142,
      "aux_distill/n_active_tok": 152.125,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 9.04296875,
      "calib/ece": 0.2258000000000001,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.904,
      "calib/gap": 0.23218693284936498,
      "calib/mean_conf": 0.91476,
      "calib/mu_c": 0.985344827586207,
      "calib/mu_w": 0.753157894736842,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.22228000000000012,
      "calib/std_conf": 0.2608864549952718,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24752561071710008,
      "calib/step_q_c_n": 1269.0,
      "calib/step_q_gap": -0.04520861490431485,
      "calib/step_q_w": 0.29273422562141493,
      "calib/step_q_w_n": 1046.0,
      "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": 2462.0,
      "completions/max_terminated_length": 2462.0,
      "completions/mean_length": 428.546875,
      "completions/mean_terminated_length": 440.5943603515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 154.0,
      "epoch": 0.16,
      "grad_norm": 0.25994738936424255,
      "learning_rate": 1.4166666666666667e-06,
      "loss": 0.0239,
      "num_tokens": 30143528.0,
      "reward": 1.2044689655303955,
      "reward_std": 0.1882563829421997,
      "rewards/accuracy_reward_step": 0.6796875,
      "rewards/final_brier_reward_step": 0.752687931060791,
      "rewards/format_reward_step": 0.9765625,
      "step": 150
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7531673237681389,
      "aux_distill/mean_u": 0.2139044884626973,
      "aux_distill/n_active_tok": 176.0,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 10.06640625,
      "calib/ece": 0.30708063241106714,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.8102766798418972,
      "calib/gap": 0.18073650793650786,
      "calib/mean_conf": 0.8372142292490119,
      "calib/mu_c": 0.902222222222222,
      "calib/mu_w": 0.7214857142857142,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.2519893280632411,
      "calib/std_conf": 0.34190325867934074,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2547523510971787,
      "calib/step_q_c_n": 1595.0,
      "calib/step_q_gap": 0.004969968205121689,
      "calib/step_q_w": 0.249782382892057,
      "calib/step_q_w_n": 982.0,
      "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": 2939.0,
      "completions/max_terminated_length": 2939.0,
      "completions/mean_length": 562.5859375,
      "completions/mean_terminated_length": 569.2569580078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 179.0,
      "epoch": 0.16106666666666666,
      "grad_norm": 0.2744375765323639,
      "learning_rate": 1.3888888888888892e-06,
      "loss": 0.0603,
      "num_tokens": 30394574.0,
      "reward": 1.151229977607727,
      "reward_std": 0.18375049531459808,
      "rewards/accuracy_reward_step": 0.6328125,
      "rewards/final_brier_reward_step": 0.6852724552154541,
      "rewards/format_reward_step": 0.984375,
      "step": 151
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8130042403936386,
      "aux_distill/mean_u": 0.24612967474911507,
      "aux_distill/n_active_tok": 140.0,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.87109375,
      "calib/ece": 0.3172618954450678,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.8214285714285714,
      "calib/gap": 0.18971410155844615,
      "calib/mean_conf": 0.8573412605244328,
      "calib/mu_c": 0.9333774678950798,
      "calib/mu_w": 0.7436633663366337,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.2876984033815757,
      "calib/std_conf": 0.32317201413168617,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.23970624673771612,
      "calib/step_q_c_n": 1354.0,
      "calib/step_q_gap": -0.08222896212843905,
      "calib/step_q_w": 0.3219352088661552,
      "calib/step_q_w_n": 1173.0,
      "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": 2213.0,
      "completions/max_terminated_length": 2213.0,
      "completions/mean_length": 514.3984375,
      "completions/mean_terminated_length": 522.5635375976562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 168.0,
      "epoch": 0.16213333333333332,
      "grad_norm": 0.5380045175552368,
      "learning_rate": 1.3611111111111112e-06,
      "loss": 0.061,
      "num_tokens": 30631652.0,
      "reward": 1.1183058023452759,
      "reward_std": 0.2768405079841614,
      "rewards/accuracy_reward_step": 0.58984375,
      "rewards/final_brier_reward_step": 0.6662992238998413,
      "rewards/format_reward_step": 0.98046875,
      "step": 152
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8390794601291418,
      "aux_distill/mean_u": 0.26110398384518824,
      "aux_distill/n_active_tok": 144.125,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 10.3125,
      "calib/ece": 0.28084677419354837,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.8588709677419355,
      "calib/gap": 0.17451219512195115,
      "calib/mean_conf": 0.8722983870967742,
      "calib/mu_c": 0.9299999999999999,
      "calib/mu_w": 0.7554878048780488,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24189516129032254,
      "calib/std_conf": 0.31580535756754713,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2394113524877365,
      "calib/step_q_c_n": 1427.0,
      "calib/step_q_gap": 7.087433439767477e-05,
      "calib/step_q_w": 0.23934047815333884,
      "calib/step_q_w_n": 1213.0,
      "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": 0.0234375,
      "completions/max_length": 2672.0,
      "completions/max_terminated_length": 2672.0,
      "completions/mean_length": 527.36328125,
      "completions/mean_terminated_length": 540.02001953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 150.0,
      "epoch": 0.1632,
      "grad_norm": 0.38069459795951843,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0659,
      "num_tokens": 30873977.0,
      "reward": 1.1549255847930908,
      "reward_std": 0.22423149645328522,
      "rewards/accuracy_reward_step": 0.6484375,
      "rewards/final_brier_reward_step": 0.6926636695861816,
      "rewards/format_reward_step": 0.96875,
      "step": 153
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7914398023858666,
      "aux_distill/mean_u": 0.26568597258707727,
      "aux_distill/n_active_tok": 134.875,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 8.640625,
      "calib/ece": 0.3906719367588934,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.8616600790513834,
      "calib/gap": 0.17957687500000008,
      "calib/mean_conf": 0.8755731225296443,
      "calib/mu_c": 0.964296875,
      "calib/mu_w": 0.78472,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.3801581027667985,
      "calib/std_conf": 0.31197725857903,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22771812080536916,
      "calib/step_q_c_n": 1043.0,
      "calib/step_q_gap": -0.026976233343476014,
      "calib/step_q_w": 0.25469435414884517,
      "calib/step_q_w_n": 1169.0,
      "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": 1539.0,
      "completions/max_terminated_length": 1539.0,
      "completions/mean_length": 485.9296875,
      "completions/mean_terminated_length": 491.69171142578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 194.0,
      "epoch": 0.16426666666666667,
      "grad_norm": 0.36173632740974426,
      "learning_rate": 1.3055555555555556e-06,
      "loss": 0.0564,
      "num_tokens": 31102815.0,
      "reward": 1.0435129404067993,
      "reward_std": 0.19690637290477753,
      "rewards/accuracy_reward_step": 0.5,
      "rewards/final_brier_reward_step": 0.5987445116043091,
      "rewards/format_reward_step": 0.98828125,
      "step": 154
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7353885294869542,
      "aux_distill/mean_u": 0.17460190229666994,
      "aux_distill/n_active_tok": 140.0,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 9.6171875,
      "calib/ece": 0.37886639676113354,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.8259109311740891,
      "calib/gap": 0.17066701680672258,
      "calib/mean_conf": 0.8434008097165993,
      "calib/mu_c": 0.9256249999999999,
      "calib/mu_w": 0.7549579831932773,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.35202429149797565,
      "calib/std_conf": 0.3417278276311164,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25213857998289135,
      "calib/step_q_c_n": 1169.0,
      "calib/step_q_gap": 0.007281658095807025,
      "calib/step_q_w": 0.24485692188708433,
      "calib/step_q_w_n": 1293.0,
      "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": 1362.0,
      "completions/max_terminated_length": 1362.0,
      "completions/mean_length": 454.2578125,
      "completions/mean_terminated_length": 468.9112854003906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 160.0,
      "epoch": 0.16533333333333333,
      "grad_norm": 0.3328990936279297,
      "learning_rate": 1.2777777777777779e-06,
      "loss": 0.0022,
      "num_tokens": 31326321.0,
      "reward": 1.0281612873077393,
      "reward_std": 0.30114853382110596,
      "rewards/accuracy_reward_step": 0.5,
      "rewards/final_brier_reward_step": 0.5914789438247681,
      "rewards/format_reward_step": 0.96484375,
      "step": 155
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7826054934412241,
      "aux_distill/mean_u": 0.23677276982612502,
      "aux_distill/n_active_tok": 145.375,
      "calib/answer_extract_rate": 0.9609375,
      "calib/avg_num_step_conf": 9.84765625,
      "calib/ece": 0.32630081300813013,
      "calib/final_conf_rate": 0.9609375,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.8252032520325203,
      "calib/gap": 0.042313883299798793,
      "calib/mean_conf": 0.8420731707317072,
      "calib/mu_c": 0.8542857142857143,
      "calib/mu_w": 0.8119718309859155,
      "calib/nonempty_final_conf_rate": 0.9609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2284959349593496,
      "calib/std_conf": 0.34261810427610406,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22662320262747246,
      "calib/step_q_c_n": 1454.0,
      "calib/step_q_gap": -0.03780041499202144,
      "calib/step_q_w": 0.2644236176194939,
      "calib/step_q_w_n": 1067.0,
      "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": 2093.0,
      "completions/max_terminated_length": 2093.0,
      "completions/mean_length": 478.33984375,
      "completions/mean_terminated_length": 493.7701416015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 151.0,
      "epoch": 0.1664,
      "grad_norm": 0.48554399609565735,
      "learning_rate": 1.25e-06,
      "loss": 0.0245,
      "num_tokens": 31553536.0,
      "reward": 1.1478266716003418,
      "reward_std": 0.24589796364307404,
      "rewards/accuracy_reward_step": 0.68359375,
      "rewards/final_brier_reward_step": 0.6511222720146179,
      "rewards/format_reward_step": 0.9609375,
      "step": 156
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7190311551094055,
      "aux_distill/mean_u": 0.16654506377717385,
      "aux_distill/n_active_tok": 156.0,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 10.51171875,
      "calib/ece": 0.2189558232931727,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.7831325301204819,
      "calib/gap": 0.27367736486486494,
      "calib/mean_conf": 0.8105220883534137,
      "calib/mu_c": 0.8808648648648649,
      "calib/mu_w": 0.6071875,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1432530120481928,
      "calib/std_conf": 0.36241067105153557,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2257866342648846,
      "calib/step_q_c_n": 1646.0,
      "calib/step_q_gap": -0.09673547099827334,
      "calib/step_q_w": 0.32252210526315794,
      "calib/step_q_w_n": 1045.0,
      "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": 1836.0,
      "completions/max_terminated_length": 1836.0,
      "completions/mean_length": 491.25,
      "completions/mean_terminated_length": 505.0602111816406,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 131.0,
      "epoch": 0.16746666666666668,
      "grad_norm": 0.3559320271015167,
      "learning_rate": 1.2222222222222223e-06,
      "loss": 0.0533,
      "num_tokens": 31783024.0,
      "reward": 1.2219460010528564,
      "reward_std": 0.23536671698093414,
      "rewards/accuracy_reward_step": 0.72265625,
      "rewards/final_brier_reward_step": 0.7524859309196472,
      "rewards/format_reward_step": 0.96875,
      "step": 157
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.6704925671219826,
      "aux_distill/mean_u": 0.20121584100111614,
      "aux_distill/n_active_tok": 163.375,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.32421875,
      "calib/ece": 0.2774206349206349,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.9166666666666666,
      "calib/gap": 0.1412540610786226,
      "calib/mean_conf": 0.9242460317460317,
      "calib/mu_c": 0.9696491228070175,
      "calib/mu_w": 0.8283950617283949,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.26154761904761903,
      "calib/std_conf": 0.25085691825547474,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22334996299037752,
      "calib/step_q_c_n": 1351.0,
      "calib/step_q_gap": 0.04315691279732731,
      "calib/step_q_w": 0.1801930501930502,
      "calib/step_q_w_n": 1036.0,
      "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": 2685.0,
      "completions/max_terminated_length": 2685.0,
      "completions/mean_length": 494.26171875,
      "completions/mean_terminated_length": 502.107177734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 129.0,
      "epoch": 0.16853333333333334,
      "grad_norm": 0.5057961940765381,
      "learning_rate": 1.1944444444444446e-06,
      "loss": 0.0602,
      "num_tokens": 32014795.0,
      "reward": 1.1806557178497314,
      "reward_std": 0.22892944514751434,
      "rewards/accuracy_reward_step": 0.66796875,
      "rewards/final_brier_reward_step": 0.7089675664901733,
      "rewards/format_reward_step": 0.984375,
      "step": 158
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8052110997959971,
      "aux_distill/mean_u": 0.23945831289057012,
      "aux_distill/n_active_tok": 123.875,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 9.0078125,
      "calib/ece": 0.29416000000000003,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.836,
      "calib/gap": 0.25487091675447815,
      "calib/mean_conf": 0.85096,
      "calib/mu_c": 0.9569863013698628,
      "calib/mu_w": 0.7021153846153847,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.28056000000000003,
      "calib/std_conf": 0.3366788950914506,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23610054844606945,
      "calib/step_q_c_n": 1094.0,
      "calib/step_q_gap": 0.00771771016224107,
      "calib/step_q_w": 0.22838283828382838,
      "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": 0.0234375,
      "completions/max_length": 1796.0,
      "completions/max_terminated_length": 1796.0,
      "completions/mean_length": 451.89453125,
      "completions/mean_terminated_length": 462.7400207519531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 155.0,
      "epoch": 0.1696,
      "grad_norm": 0.3434606194496155,
      "learning_rate": 1.1666666666666668e-06,
      "loss": 0.0371,
      "num_tokens": 32235264.0,
      "reward": 1.113415241241455,
      "reward_std": 0.22357870638370514,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6799554824829102,
      "rewards/format_reward_step": 0.9765625,
      "step": 159
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.825240034610033,
      "aux_distill/mean_u": 0.2686492410873999,
      "aux_distill/n_active_tok": 147.75,
      "calib/answer_extract_rate": 0.96484375,
      "calib/avg_num_step_conf": 9.66015625,
      "calib/ece": 0.3138866396761135,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.8421052631578947,
      "calib/gap": 0.15572612921669515,
      "calib/mean_conf": 0.8561538461538462,
      "calib/mu_c": 0.9116352201257861,
      "calib/mu_w": 0.755909090909091,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.26315789473684226,
      "calib/std_conf": 0.3342975362280166,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.1681236203090508,
      "calib/step_q_c_n": 1359.0,
      "calib/step_q_gap": -0.03851013193511438,
      "calib/step_q_w": 0.20663375224416516,
      "calib/step_q_w_n": 1114.0,
      "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": 2083.0,
      "completions/max_terminated_length": 2083.0,
      "completions/mean_length": 469.82421875,
      "completions/mean_terminated_length": 486.9433288574219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 140.0,
      "epoch": 0.17066666666666666,
      "grad_norm": 0.406261146068573,
      "learning_rate": 1.138888888888889e-06,
      "loss": 0.0152,
      "num_tokens": 32460379.0,
      "reward": 1.1235268115997314,
      "reward_std": 0.24834099411964417,
      "rewards/accuracy_reward_step": 0.62109375,
      "rewards/final_brier_reward_step": 0.6611160039901733,
      "rewards/format_reward_step": 0.96484375,
      "step": 160
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7200377890840173,
      "aux_distill/mean_u": 0.20223440044865573,
      "aux_distill/n_active_tok": 121.75,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 7.734375,
      "calib/ece": 0.1636862745098039,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.9098039215686274,
      "calib/gap": 0.2220439024390245,
      "calib/mean_conf": 0.9207058823529413,
      "calib/mu_c": 0.9642439024390245,
      "calib/mu_w": 0.7422,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.140235294117647,
      "calib/std_conf": 0.25349689978516904,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.17341063274624918,
      "calib/step_q_c_n": 1533.0,
      "calib/step_q_gap": -0.04077281244390743,
      "calib/step_q_w": 0.2141834451901566,
      "calib/step_q_w_n": 447.0,
      "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": 1486.0,
      "completions/max_terminated_length": 1486.0,
      "completions/mean_length": 425.61328125,
      "completions/mean_terminated_length": 428.9645690917969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 166.0,
      "epoch": 0.17173333333333332,
      "grad_norm": 0.3874889314174652,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.0667,
      "num_tokens": 32673256.0,
      "reward": 1.314043402671814,
      "reward_std": 0.12282635271549225,
      "rewards/accuracy_reward_step": 0.80078125,
      "rewards/final_brier_reward_step": 0.8312117457389832,
      "rewards/format_reward_step": 0.99609375,
      "step": 161
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7323440834879875,
      "aux_distill/mean_u": 0.17847615366741984,
      "aux_distill/n_active_tok": 126.0,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.1796875,
      "calib/ece": 0.1663529411764706,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.8980392156862745,
      "calib/gap": 0.3037251984126985,
      "calib/mean_conf": 0.9102745098039215,
      "calib/mu_c": 0.9853125,
      "calib/mu_w": 0.6815873015873015,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.16184313725490196,
      "calib/std_conf": 0.26846317003264225,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.1819727705922396,
      "calib/step_q_c_n": 1469.0,
      "calib/step_q_gap": -0.0333872294077604,
      "calib/step_q_w": 0.21536,
      "calib/step_q_w_n": 625.0,
      "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": 1650.0,
      "completions/max_terminated_length": 1650.0,
      "completions/mean_length": 420.1328125,
      "completions/mean_terminated_length": 423.4409484863281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.1728,
      "grad_norm": 0.27647775411605835,
      "learning_rate": 1.0833333333333335e-06,
      "loss": 0.0436,
      "num_tokens": 32884954.0,
      "reward": 1.286501169204712,
      "reward_std": 0.16905483603477478,
      "rewards/accuracy_reward_step": 0.75,
      "rewards/final_brier_reward_step": 0.8269085884094238,
      "rewards/format_reward_step": 0.99609375,
      "step": 162
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8068936876952648,
      "aux_distill/mean_u": 0.2444127955766234,
      "aux_distill/n_active_tok": 155.125,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.10546875,
      "calib/ece": 0.20146808055212456,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.7142857142857143,
      "calib/gap": 0.4506941576834671,
      "calib/mean_conf": 0.7395633186473626,
      "calib/mu_c": 0.9201986754966889,
      "calib/mu_w": 0.4695045178132218,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.170912524996569,
      "calib/std_conf": 0.41283393997796536,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20412260536398466,
      "calib/step_q_c_n": 1305.0,
      "calib/step_q_gap": -0.02532357248312922,
      "calib/step_q_w": 0.22944617784711388,
      "calib/step_q_w_n": 1282.0,
      "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": 2390.0,
      "completions/max_terminated_length": 2390.0,
      "completions/mean_length": 533.078125,
      "completions/mean_terminated_length": 541.5397338867188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 158.0,
      "epoch": 0.17386666666666667,
      "grad_norm": 0.5397083759307861,
      "learning_rate": 1.0555555555555557e-06,
      "loss": 0.0504,
      "num_tokens": 33126254.0,
      "reward": 1.1740601062774658,
      "reward_std": 0.18381524085998535,
      "rewards/accuracy_reward_step": 0.58984375,
      "rewards/final_brier_reward_step": 0.7739015221595764,
      "rewards/format_reward_step": 0.984375,
      "step": 163
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8425343632698059,
      "aux_distill/mean_u": 0.253130894448872,
      "aux_distill/n_active_tok": 147.5,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 11.15625,
      "calib/ece": 0.16938775510204088,
      "calib/final_conf_rate": 0.95703125,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.6489795918367347,
      "calib/gap": 0.5163950770077008,
      "calib/mean_conf": 0.6816326530612244,
      "calib/mu_c": 0.8945138888888889,
      "calib/mu_w": 0.3781188118811881,
      "calib/nonempty_final_conf_rate": 0.95703125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.13163265306122457,
      "calib/std_conf": 0.43294127866318344,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20181670721816705,
      "calib/step_q_c_n": 1233.0,
      "calib/step_q_gap": -0.04759795698392785,
      "calib/step_q_w": 0.2494146642020949,
      "calib/step_q_w_n": 1623.0,
      "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": 0.04296875,
      "completions/max_length": 2876.0,
      "completions/max_terminated_length": 2876.0,
      "completions/mean_length": 541.52734375,
      "completions/mean_terminated_length": 565.8408203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 199.0,
      "epoch": 0.17493333333333333,
      "grad_norm": 0.738795280456543,
      "learning_rate": 1.0277777777777777e-06,
      "loss": -0.0139,
      "num_tokens": 33371021.0,
      "reward": 1.1481738090515137,
      "reward_std": 0.20172443985939026,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.7768163681030273,
      "rewards/format_reward_step": 0.95703125,
      "step": 164
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.782092509791255,
      "aux_distill/mean_u": 0.2121803635043166,
      "aux_distill/n_active_tok": 142.375,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 9.75,
      "calib/ece": 0.2654612840032003,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.678714859437751,
      "calib/gap": 0.3788403057693124,
      "calib/mean_conf": 0.7032926093044052,
      "calib/mu_c": 0.88282335661677,
      "calib/mu_w": 0.5039830508474576,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.22132473781846138,
      "calib/std_conf": 0.43055869360944665,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.15882145077672744,
      "calib/step_q_c_n": 1092.0,
      "calib/step_q_gap": -0.046880757200480516,
      "calib/step_q_w": 0.20570220797720795,
      "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": 0.02734375,
      "completions/max_length": 3063.0,
      "completions/max_terminated_length": 3063.0,
      "completions/mean_length": 509.64453125,
      "completions/mean_terminated_length": 523.9718627929688,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 155.0,
      "epoch": 0.176,
      "grad_norm": 0.41305992007255554,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0166,
      "num_tokens": 33607066.0,
      "reward": 1.0937097072601318,
      "reward_std": 0.20582950115203857,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.7030445337295532,
      "rewards/format_reward_step": 0.97265625,
      "step": 165
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7836181567981839,
      "aux_distill/mean_u": 0.2615975476655479,
      "aux_distill/n_active_tok": 181.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 11.03125,
      "calib/ece": 0.15141732283464576,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.7007874015748031,
      "calib/gap": 0.5479756401790299,
      "calib/mean_conf": 0.728740157480315,
      "calib/mu_c": 0.8948587570621468,
      "calib/mu_w": 0.3468831168831169,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0916535433070867,
      "calib/std_conf": 0.4158020276317233,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20698069046225864,
      "calib/step_q_c_n": 1709.0,
      "calib/step_q_gap": -0.004050699672270508,
      "calib/step_q_w": 0.21103139013452915,
      "calib/step_q_w_n": 1115.0,
      "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": 1957.0,
      "completions/max_terminated_length": 1957.0,
      "completions/mean_length": 558.41796875,
      "completions/mean_terminated_length": 565.03955078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 175.0,
      "epoch": 0.17706666666666668,
      "grad_norm": 0.3993739187717438,
      "learning_rate": 9.722222222222224e-07,
      "loss": 0.0384,
      "num_tokens": 33856205.0,
      "reward": 1.261671543121338,
      "reward_std": 0.1854332834482193,
      "rewards/accuracy_reward_step": 0.69140625,
      "rewards/final_brier_reward_step": 0.8397492170333862,
      "rewards/format_reward_step": 0.9921875,
      "step": 166
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7592298779636621,
      "aux_distill/mean_u": 0.22087851047994195,
      "aux_distill/n_active_tok": 140.0,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 9.703125,
      "calib/ece": 0.20209486166007906,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.8221343873517787,
      "calib/gap": 0.3174265767934522,
      "calib/mean_conf": 0.8354545454545454,
      "calib/mu_c": 0.9195161290322581,
      "calib/mu_w": 0.6020895522388059,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.15118577075098819,
      "calib/std_conf": 0.34964732500972096,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2433490011750881,
      "calib/step_q_c_n": 1702.0,
      "calib/step_q_gap": -0.06839012925969448,
      "calib/step_q_w": 0.3117391304347826,
      "calib/step_q_w_n": 782.0,
      "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": 1761.0,
      "completions/max_terminated_length": 1761.0,
      "completions/mean_length": 495.65234375,
      "completions/mean_terminated_length": 501.5296630859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 146.0,
      "epoch": 0.17813333333333334,
      "grad_norm": 0.36381569504737854,
      "learning_rate": 9.444444444444445e-07,
      "loss": 0.0563,
      "num_tokens": 34088700.0,
      "reward": 1.2510545253753662,
      "reward_std": 0.18535427749156952,
      "rewards/accuracy_reward_step": 0.7265625,
      "rewards/final_brier_reward_step": 0.7872651815414429,
      "rewards/format_reward_step": 0.98828125,
      "step": 167
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8132620574906468,
      "aux_distill/mean_u": 0.31038632137509864,
      "aux_distill/n_active_tok": 166.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.69140625,
      "calib/ece": 0.1765869047619048,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.6825396825396826,
      "calib/gap": 0.4661973958333333,
      "calib/mean_conf": 0.7226980158730157,
      "calib/mu_c": 0.8336973958333332,
      "calib/mu_w": 0.36749999999999994,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.06869007936507943,
      "calib/std_conf": 0.4104957785922956,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22837146783947224,
      "calib/step_q_c_n": 1819.0,
      "calib/step_q_gap": -0.004222214077739123,
      "calib/step_q_w": 0.23259368191721136,
      "calib/step_q_w_n": 918.0,
      "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": 1883.0,
      "completions/max_terminated_length": 1883.0,
      "completions/mean_length": 547.50390625,
      "completions/mean_terminated_length": 556.1944580078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 173.0,
      "epoch": 0.1792,
      "grad_norm": 0.35889482498168945,
      "learning_rate": 9.166666666666666e-07,
      "loss": 0.0273,
      "num_tokens": 34333533.0,
      "reward": 1.2696452140808105,
      "reward_std": 0.17705899477005005,
      "rewards/accuracy_reward_step": 0.75,
      "rewards/final_brier_reward_step": 0.8049156069755554,
      "rewards/format_reward_step": 0.984375,
      "step": 168
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8193460293114185,
      "aux_distill/mean_u": 0.2598738432131758,
      "aux_distill/n_active_tok": 144.0,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.30859375,
      "calib/ece": 0.2248809523809523,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.7222222222222222,
      "calib/gap": 0.3845061728395063,
      "calib/mean_conf": 0.7449603174603175,
      "calib/mu_c": 0.882283950617284,
      "calib/mu_w": 0.49777777777777776,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1634920634920634,
      "calib/std_conf": 0.40791718154145534,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25499312500000004,
      "calib/step_q_c_n": 1600.0,
      "calib/step_q_gap": 0.03606502105389803,
      "calib/step_q_w": 0.21892810394610202,
      "calib/step_q_w_n": 1039.0,
      "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": 2248.0,
      "completions/max_terminated_length": 2248.0,
      "completions/mean_length": 554.4375,
      "completions/mean_terminated_length": 558.8031616210938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 172.0,
      "epoch": 0.18026666666666666,
      "grad_norm": 0.267215371131897,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0669,
      "num_tokens": 34579653.0,
      "reward": 1.1876498460769653,
      "reward_std": 0.14423486590385437,
      "rewards/accuracy_reward_step": 0.6328125,
      "rewards/final_brier_reward_step": 0.7581120729446411,
      "rewards/format_reward_step": 0.984375,
      "step": 169
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7837082333862782,
      "aux_distill/mean_u": 0.2228750338044867,
      "aux_distill/n_active_tok": 160.75,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 10.40234375,
      "calib/ece": 0.1924701195219125,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.6812749003984063,
      "calib/gap": 0.45554193093727974,
      "calib/mean_conf": 0.7129482071713147,
      "calib/mu_c": 0.8690303030303029,
      "calib/mu_w": 0.4134883720930232,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.12402390438247027,
      "calib/std_conf": 0.41963472223631293,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.23673456397199238,
      "calib/step_q_c_n": 1571.0,
      "calib/step_q_gap": -0.018082285844857438,
      "calib/step_q_w": 0.2548168498168498,
      "calib/step_q_w_n": 1092.0,
      "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": 1662.0,
      "completions/max_terminated_length": 1662.0,
      "completions/mean_length": 513.48828125,
      "completions/mean_terminated_length": 523.7171630859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 135.0,
      "epoch": 0.18133333333333335,
      "grad_norm": 0.4967786371707916,
      "learning_rate": 8.611111111111112e-07,
      "loss": 0.0372,
      "num_tokens": 34815258.0,
      "reward": 1.2031220197677612,
      "reward_std": 0.16210168600082397,
      "rewards/accuracy_reward_step": 0.64453125,
      "rewards/final_brier_reward_step": 0.785150408744812,
      "rewards/format_reward_step": 0.9765625,
      "step": 170
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7938511082902551,
      "aux_distill/mean_u": 0.29352478921058617,
      "aux_distill/n_active_tok": 132.375,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.83203125,
      "calib/ece": 0.20181102362204736,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.6141732283464567,
      "calib/gap": 0.4881854636591479,
      "calib/mean_conf": 0.6434645669291339,
      "calib/mu_c": 0.8625714285714287,
      "calib/mu_w": 0.37438596491228077,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1470472440944883,
      "calib/std_conf": 0.4485635480269359,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2466503667481663,
      "calib/step_q_c_n": 1227.0,
      "calib/step_q_gap": 0.004727886128011255,
      "calib/step_q_w": 0.24192248062015503,
      "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": 0.01171875,
      "completions/max_length": 1692.0,
      "completions/max_terminated_length": 1692.0,
      "completions/mean_length": 515.5390625,
      "completions/mean_terminated_length": 521.6522216796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 162.0,
      "epoch": 0.1824,
      "grad_norm": 0.3575134873390198,
      "learning_rate": 8.333333333333333e-07,
      "loss": 0.0805,
      "num_tokens": 35054132.0,
      "reward": 1.1586816310882568,
      "reward_std": 0.12559093534946442,
      "rewards/accuracy_reward_step": 0.546875,
      "rewards/final_brier_reward_step": 0.7783007621765137,
      "rewards/format_reward_step": 0.9921875,
      "step": 171
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7738157333806157,
      "aux_distill/mean_u": 0.1982485849683548,
      "aux_distill/n_active_tok": 145.75,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 9.375,
      "calib/ece": 0.16667999999999994,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.752,
      "calib/gap": 0.4780542529399034,
      "calib/mean_conf": 0.77428,
      "calib/mu_c": 0.9329940119760479,
      "calib/mu_w": 0.45493975903614453,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.13647999999999996,
      "calib/std_conf": 0.39090009158351446,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24634110787172012,
      "calib/step_q_c_n": 1372.0,
      "calib/step_q_gap": -0.009428541933727358,
      "calib/step_q_w": 0.2557696498054475,
      "calib/step_q_w_n": 1028.0,
      "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": 0.0234375,
      "completions/max_length": 1691.0,
      "completions/max_terminated_length": 1691.0,
      "completions/mean_length": 472.53515625,
      "completions/mean_terminated_length": 483.87603759765625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 201.0,
      "epoch": 0.18346666666666667,
      "grad_norm": 0.2834770977497101,
      "learning_rate": 8.055555555555557e-07,
      "loss": 0.0369,
      "num_tokens": 35278453.0,
      "reward": 1.2178552150726318,
      "reward_std": 0.18590107560157776,
      "rewards/accuracy_reward_step": 0.65234375,
      "rewards/final_brier_reward_step": 0.8068042993545532,
      "rewards/format_reward_step": 0.9765625,
      "step": 172
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.6961722830310464,
      "aux_distill/mean_u": 0.18221597900992778,
      "aux_distill/n_active_tok": 160.875,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 9.8515625,
      "calib/ece": 0.19749003984063757,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.7609561752988048,
      "calib/gap": 0.40618126574773983,
      "calib/mean_conf": 0.7747011952191236,
      "calib/mu_c": 0.9009248554913296,
      "calib/mu_w": 0.4947435897435898,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.14147410358565748,
      "calib/std_conf": 0.3952963112197103,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25296927822331894,
      "calib/step_q_c_n": 1621.0,
      "calib/step_q_gap": 0.04764186423885722,
      "calib/step_q_w": 0.20532741398446172,
      "calib/step_q_w_n": 901.0,
      "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": 1634.0,
      "completions/max_terminated_length": 1634.0,
      "completions/mean_length": 514.015625,
      "completions/mean_terminated_length": 524.2550048828125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 118.0,
      "epoch": 0.18453333333333333,
      "grad_norm": 0.4571469724178314,
      "learning_rate": 7.777777777777779e-07,
      "loss": -0.0008,
      "num_tokens": 35513201.0,
      "reward": 1.218473196029663,
      "reward_std": 0.18532350659370422,
      "rewards/accuracy_reward_step": 0.67578125,
      "rewards/final_brier_reward_step": 0.7806965112686157,
      "rewards/format_reward_step": 0.98046875,
      "step": 173
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8278934434056282,
      "aux_distill/mean_u": 0.24226153047754964,
      "aux_distill/n_active_tok": 142.0,
      "calib/answer_extract_rate": 0.94921875,
      "calib/avg_num_step_conf": 11.37890625,
      "calib/ece": 0.280493827160494,
      "calib/final_conf_rate": 0.94921875,
      "calib/format_rate": 0.9453125,
      "calib/frac_conf_gt_0.9": 0.49382716049382713,
      "calib/gap": 0.37458674609084147,
      "calib/mean_conf": 0.5306172839506172,
      "calib/mu_c": 0.6616455696202532,
      "calib/mu_w": 0.28705882352941176,
      "calib/nonempty_final_conf_rate": 0.94921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.08045267489711945,
      "calib/std_conf": 0.4577571345912221,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.27785419532324623,
      "calib/step_q_c_n": 1454.0,
      "calib/step_q_gap": -0.038314413312805884,
      "calib/step_q_w": 0.3161686086360521,
      "calib/step_q_w_n": 1459.0,
      "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": 2799.0,
      "completions/max_terminated_length": 2799.0,
      "completions/mean_length": 527.046875,
      "completions/mean_terminated_length": 552.9671630859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 171.0,
      "epoch": 0.1856,
      "grad_norm": 0.4958760142326355,
      "learning_rate": 7.5e-07,
      "loss": 0.0298,
      "num_tokens": 35752357.0,
      "reward": 1.1205930709838867,
      "reward_std": 0.2511502206325531,
      "rewards/accuracy_reward_step": 0.6171875,
      "rewards/final_brier_reward_step": 0.6786863207817078,
      "rewards/format_reward_step": 0.9453125,
      "step": 174
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7686019539833069,
      "aux_distill/mean_u": 0.2133538263066181,
      "aux_distill/n_active_tok": 129.125,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 11.33203125,
      "calib/ece": 0.3505306122448979,
      "calib/final_conf_rate": 0.95703125,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.5061224489795918,
      "calib/gap": 0.2137369349803177,
      "calib/mean_conf": 0.536734693877551,
      "calib/mu_c": 0.6292086330935253,
      "calib/mu_w": 0.41547169811320755,
      "calib/nonempty_final_conf_rate": 0.95703125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1599591836734693,
      "calib/std_conf": 0.4636016400241371,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2988099808061421,
      "calib/step_q_c_n": 1042.0,
      "calib/step_q_gap": -0.01760206868283054,
      "calib/step_q_w": 0.3164120494889726,
      "calib/step_q_w_n": 1859.0,
      "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": 0.04296875,
      "completions/max_length": 2092.0,
      "completions/max_terminated_length": 2092.0,
      "completions/mean_length": 487.95703125,
      "completions/mean_terminated_length": 509.86529541015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 185.0,
      "epoch": 0.18666666666666668,
      "grad_norm": 0.434774786233902,
      "learning_rate": 7.222222222222222e-07,
      "loss": 0.0155,
      "num_tokens": 35983098.0,
      "reward": 1.0579733848571777,
      "reward_std": 0.18678444623947144,
      "rewards/accuracy_reward_step": 0.54296875,
      "rewards/final_brier_reward_step": 0.615946888923645,
      "rewards/format_reward_step": 0.95703125,
      "step": 175
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8209396656602621,
      "aux_distill/mean_u": 0.3008576483306273,
      "aux_distill/n_active_tok": 128.875,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 10.97265625,
      "calib/ece": 0.23026572580645144,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.5403225806451613,
      "calib/gap": 0.45330585785485106,
      "calib/mean_conf": 0.5659439516129033,
      "calib/mu_c": 0.7432450331125829,
      "calib/mu_w": 0.2899391752577319,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.09366935483870951,
      "calib/std_conf": 0.4650180486912076,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2576894223555889,
      "calib/step_q_c_n": 1333.0,
      "calib/step_q_gap": -0.02182954783411306,
      "calib/step_q_w": 0.27951897018970195,
      "calib/step_q_w_n": 1476.0,
      "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": 2911.0,
      "completions/max_terminated_length": 2911.0,
      "completions/mean_length": 507.81640625,
      "completions/mean_terminated_length": 524.1975708007812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 167.0,
      "epoch": 0.18773333333333334,
      "grad_norm": 0.24123993515968323,
      "learning_rate": 6.944444444444446e-07,
      "loss": 0.0473,
      "num_tokens": 36217163.0,
      "reward": 1.1472645998001099,
      "reward_std": 0.14933854341506958,
      "rewards/accuracy_reward_step": 0.58984375,
      "rewards/final_brier_reward_step": 0.7359354496002197,
      "rewards/format_reward_step": 0.96875,
      "step": 176
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8030837085098028,
      "aux_distill/mean_u": 0.2057992942512779,
      "aux_distill/n_active_tok": 130.625,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 9.81640625,
      "calib/ece": 0.11093669354838712,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.6008064516129032,
      "calib/gap": 0.6747169407894738,
      "calib/mean_conf": 0.6255955645161291,
      "calib/mu_c": 0.8867763157894738,
      "calib/mu_w": 0.21205937500000002,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.061814516129032264,
      "calib/std_conf": 0.4567104782993171,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.26664710813076276,
      "calib/step_q_c_n": 1193.0,
      "calib/step_q_gap": 0.03225316873682332,
      "calib/step_q_w": 0.23439393939393943,
      "calib/step_q_w_n": 1320.0,
      "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": 1856.0,
      "completions/max_terminated_length": 1856.0,
      "completions/mean_length": 506.3359375,
      "completions/mean_terminated_length": 522.6693115234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 120.0,
      "epoch": 0.1888,
      "grad_norm": 0.4691639840602875,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.0359,
      "num_tokens": 36450617.0,
      "reward": 1.2046705484390259,
      "reward_std": 0.1261509507894516,
      "rewards/accuracy_reward_step": 0.59375,
      "rewards/final_brier_reward_step": 0.8468411564826965,
      "rewards/format_reward_step": 0.96875,
      "step": 177
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8158742636442184,
      "aux_distill/mean_u": 0.2410105040041707,
      "aux_distill/n_active_tok": 172.5,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 10.5390625,
      "calib/ece": 0.21916996047430828,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.5928853754940712,
      "calib/gap": 0.48188432267884324,
      "calib/mean_conf": 0.6205138339920949,
      "calib/mu_c": 0.7595555555555555,
      "calib/mu_w": 0.2776712328767123,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.06411067193675889,
      "calib/std_conf": 0.4542858409665569,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.23063775510204082,
      "calib/step_q_c_n": 1568.0,
      "calib/step_q_gap": -0.01617020949972911,
      "calib/step_q_w": 0.24680796460176993,
      "calib/step_q_w_n": 1130.0,
      "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": 2971.0,
      "completions/max_terminated_length": 2971.0,
      "completions/mean_length": 541.33984375,
      "completions/mean_terminated_length": 549.9325561523438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 143.0,
      "epoch": 0.18986666666666666,
      "grad_norm": 0.27010011672973633,
      "learning_rate": 6.388888888888889e-07,
      "loss": 0.0427,
      "num_tokens": 36695272.0,
      "reward": 1.2301025390625,
      "reward_std": 0.16309009492397308,
      "rewards/accuracy_reward_step": 0.703125,
      "rewards/final_brier_reward_step": 0.768798828125,
      "rewards/format_reward_step": 0.98828125,
      "step": 178
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7807596288621426,
      "aux_distill/mean_u": 0.25980170657934115,
      "aux_distill/n_active_tok": 185.625,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 10.578125,
      "calib/ece": 0.16274193548387095,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.6895161290322581,
      "calib/gap": 0.5231907514450866,
      "calib/mean_conf": 0.7109677419354838,
      "calib/mu_c": 0.8691907514450867,
      "calib/mu_w": 0.34600000000000003,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08806451612903225,
      "calib/std_conf": 0.4258197395287359,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.24101198402130491,
      "calib/step_q_c_n": 1502.0,
      "calib/step_q_gap": 0.015945649029596776,
      "calib/step_q_w": 0.22506633499170814,
      "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": 0.03125,
      "completions/max_length": 2596.0,
      "completions/max_terminated_length": 2596.0,
      "completions/mean_length": 501.41796875,
      "completions/mean_terminated_length": 517.5927124023438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 225.0,
      "epoch": 0.19093333333333334,
      "grad_norm": 0.5760951638221741,
      "learning_rate": 6.111111111111112e-07,
      "loss": 0.0235,
      "num_tokens": 36929899.0,
      "reward": 1.2234652042388916,
      "reward_std": 0.22339525818824768,
      "rewards/accuracy_reward_step": 0.67578125,
      "rewards/final_brier_reward_step": 0.802399218082428,
      "rewards/format_reward_step": 0.96875,
      "step": 179
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7465346790850163,
      "aux_distill/mean_u": 0.195027078707378,
      "aux_distill/n_active_tok": 164.125,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 11.4609375,
      "calib/ece": 0.20440000000000014,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.748,
      "calib/gap": 0.3948253968253968,
      "calib/mean_conf": 0.76456,
      "calib/mu_c": 0.8751111111111111,
      "calib/mu_w": 0.48028571428571426,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12448000000000012,
      "calib/std_conf": 0.4017452008425241,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2172933549432739,
      "calib/step_q_c_n": 1851.0,
      "calib/step_q_gap": 0.001965561776145569,
      "calib/step_q_w": 0.21532779316712833,
      "calib/step_q_w_n": 1083.0,
      "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": 0.0234375,
      "completions/max_length": 2143.0,
      "completions/max_terminated_length": 2143.0,
      "completions/mean_length": 615.8203125,
      "completions/mean_terminated_length": 630.6000366210938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 196.0,
      "epoch": 0.192,
      "grad_norm": 0.5065380334854126,
      "learning_rate": 5.833333333333334e-07,
      "loss": 0.0231,
      "num_tokens": 37191405.0,
      "reward": 1.2276408672332764,
      "reward_std": 0.21545855700969696,
      "rewards/accuracy_reward_step": 0.703125,
      "rewards/final_brier_reward_step": 0.7755945324897766,
      "rewards/format_reward_step": 0.9765625,
      "step": 180
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7801644625142217,
      "aux_distill/mean_u": 0.21185583020148446,
      "aux_distill/n_active_tok": 139.75,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.74609375,
      "calib/ece": 0.2775062992125985,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.7598425196850394,
      "calib/gap": 0.25334444140197143,
      "calib/mean_conf": 0.7777692913385827,
      "calib/mu_c": 0.8655421686746988,
      "calib/mu_w": 0.6121977272727274,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.20086614173228357,
      "calib/std_conf": 0.39219924853706833,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25771151178918167,
      "calib/step_q_c_n": 1442.0,
      "calib/step_q_gap": 0.04990724579169106,
      "calib/step_q_w": 0.2078042659974906,
      "calib/step_q_w_n": 797.0,
      "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": 1499.0,
      "completions/max_terminated_length": 1499.0,
      "completions/mean_length": 473.95703125,
      "completions/mean_terminated_length": 479.57708740234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 183.0,
      "epoch": 0.19306666666666666,
      "grad_norm": 0.44195133447647095,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.0639,
      "num_tokens": 37419002.0,
      "reward": 1.1770286560058594,
      "reward_std": 0.19044727087020874,
      "rewards/accuracy_reward_step": 0.6484375,
      "rewards/final_brier_reward_step": 0.7134323716163635,
      "rewards/format_reward_step": 0.9921875,
      "step": 181
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8013272704556584,
      "aux_distill/mean_u": 0.23888848999273496,
      "aux_distill/n_active_tok": 161.75,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 9.640625,
      "calib/ece": 0.2185039370078741,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8110236220472441,
      "calib/gap": 0.3231657925407927,
      "calib/mean_conf": 0.8245669291338582,
      "calib/mu_c": 0.9238068181818183,
      "calib/mu_w": 0.6006410256410256,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1750787401574804,
      "calib/std_conf": 0.3598145977209505,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22807674463866073,
      "calib/step_q_c_n": 1515.0,
      "calib/step_q_gap": -0.10879628789019552,
      "calib/step_q_w": 0.33687303252885625,
      "calib/step_q_w_n": 953.0,
      "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": 2919.0,
      "completions/max_terminated_length": 2919.0,
      "completions/mean_length": 528.34375,
      "completions/mean_terminated_length": 532.50390625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 161.0,
      "epoch": 0.19413333333333332,
      "grad_norm": 0.9773353338241577,
      "learning_rate": 5.277777777777779e-07,
      "loss": 0.0592,
      "num_tokens": 37660418.0,
      "reward": 1.2257777452468872,
      "reward_std": 0.17774598300457,
      "rewards/accuracy_reward_step": 0.6875,
      "rewards/final_brier_reward_step": 0.7718679904937744,
      "rewards/format_reward_step": 0.9921875,
      "step": 182
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7841676194220781,
      "aux_distill/mean_u": 0.2744013052323904,
      "aux_distill/n_active_tok": 180.5,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 11.27734375,
      "calib/ece": 0.22580645161290328,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.7862903225806451,
      "calib/gap": 0.2790196078431372,
      "calib/mean_conf": 0.8076612903225807,
      "calib/mu_c": 0.8841666666666667,
      "calib/mu_w": 0.6051470588235295,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.15383064516129039,
      "calib/std_conf": 0.36774839941970433,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.27389355742296917,
      "calib/step_q_c_n": 1785.0,
      "calib/step_q_gap": 0.039664882286853026,
      "calib/step_q_w": 0.23422867513611614,
      "calib/step_q_w_n": 1102.0,
      "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": 2245.0,
      "completions/max_terminated_length": 2245.0,
      "completions/mean_length": 559.9375,
      "completions/mean_terminated_length": 578.0,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 232.0,
      "epoch": 0.1952,
      "grad_norm": 0.32870110869407654,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0065,
      "num_tokens": 37910442.0,
      "reward": 1.2089574337005615,
      "reward_std": 0.265907347202301,
      "rewards/accuracy_reward_step": 0.703125,
      "rewards/final_brier_reward_step": 0.746039867401123,
      "rewards/format_reward_step": 0.96875,
      "step": 183
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8048588903620839,
      "aux_distill/mean_u": 0.26784059578838393,
      "aux_distill/n_active_tok": 139.375,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.76953125,
      "calib/ece": 0.2058730158730159,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.8650793650793651,
      "calib/gap": 0.26646551028640586,
      "calib/mean_conf": 0.8741269841269841,
      "calib/mu_c": 0.9449729729729729,
      "calib/mu_w": 0.678507462686567,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.17293650793650797,
      "calib/std_conf": 0.31427485150379103,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.2232556750298686,
      "calib/step_q_c_n": 1674.0,
      "calib/step_q_gap": -0.07265726330144945,
      "calib/step_q_w": 0.29591293833131804,
      "calib/step_q_w_n": 827.0,
      "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": 2177.0,
      "completions/max_terminated_length": 2177.0,
      "completions/mean_length": 515.34375,
      "completions/mean_terminated_length": 521.45458984375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 184.0,
      "epoch": 0.19626666666666667,
      "grad_norm": 0.32190728187561035,
      "learning_rate": 4.7222222222222226e-07,
      "loss": 0.1,
      "num_tokens": 38147650.0,
      "reward": 1.2386672496795654,
      "reward_std": 0.24186360836029053,
      "rewards/accuracy_reward_step": 0.72265625,
      "rewards/final_brier_reward_step": 0.7742093801498413,
      "rewards/format_reward_step": 0.98046875,
      "step": 184
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.740824181586504,
      "aux_distill/mean_u": 0.2537910775115197,
      "aux_distill/n_active_tok": 157.875,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 10.3359375,
      "calib/ece": 0.20019841269841274,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.75,
      "calib/gap": 0.4772727272727273,
      "calib/mean_conf": 0.7669444444444443,
      "calib/mu_c": 0.9544444444444444,
      "calib/mu_w": 0.47717171717171714,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.18000000000000005,
      "calib/std_conf": 0.4024061914848307,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22706200432588317,
      "calib/step_q_c_n": 1387.0,
      "calib/step_q_gap": -0.0007854936884139074,
      "calib/step_q_w": 0.22784749801429707,
      "calib/step_q_w_n": 1259.0,
      "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": 1760.0,
      "completions/max_terminated_length": 1760.0,
      "completions/mean_length": 515.328125,
      "completions/mean_terminated_length": 525.5936279296875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 172.0,
      "epoch": 0.19733333333333333,
      "grad_norm": 0.3035050928592682,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0102,
      "num_tokens": 38386494.0,
      "reward": 1.1855978965759277,
      "reward_std": 0.2117016315460205,
      "rewards/accuracy_reward_step": 0.59765625,
      "rewards/final_brier_reward_step": 0.7891644239425659,
      "rewards/format_reward_step": 0.984375,
      "step": 185
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7983589041978121,
      "aux_distill/mean_u": 0.25068229766670735,
      "aux_distill/n_active_tok": 128.375,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 9.6875,
      "calib/ece": 0.18947791164658628,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.8192771084337349,
      "calib/gap": 0.3847062820900029,
      "calib/mean_conf": 0.8313253012048193,
      "calib/mu_c": 0.9502906976744185,
      "calib/mu_w": 0.5655844155844156,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1650200803212851,
      "calib/std_conf": 0.3565624738998539,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2070189701897019,
      "calib/step_q_c_n": 1476.0,
      "calib/step_q_gap": -0.035988997937788175,
      "calib/step_q_w": 0.24300796812749006,
      "calib/step_q_w_n": 1004.0,
      "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": 1498.0,
      "completions/max_terminated_length": 1498.0,
      "completions/mean_length": 513.640625,
      "completions/mean_terminated_length": 528.080322265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.1984,
      "grad_norm": 0.29591473937034607,
      "learning_rate": 4.1666666666666667e-07,
      "loss": 0.0232,
      "num_tokens": 38623026.0,
      "reward": 1.213200330734253,
      "reward_std": 0.22806373238563538,
      "rewards/accuracy_reward_step": 0.671875,
      "rewards/final_brier_reward_step": 0.7818695306777954,
      "rewards/format_reward_step": 0.97265625,
      "step": 186
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7855865238234401,
      "aux_distill/mean_u": 0.27453464204647304,
      "aux_distill/n_active_tok": 147.875,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 10.73046875,
      "calib/ece": 0.29519718875502016,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.8594377510040161,
      "calib/gap": 0.2540338741853969,
      "calib/mean_conf": 0.871904016064257,
      "calib/mu_c": 0.976986301369863,
      "calib/mu_w": 0.7229524271844661,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2903779116465864,
      "calib/std_conf": 0.3149587983252822,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.1995216400911162,
      "calib/step_q_c_n": 1317.0,
      "calib/step_q_gap": 0.00855919253866863,
      "calib/step_q_w": 0.19096244755244757,
      "calib/step_q_w_n": 1430.0,
      "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": 2264.0,
      "completions/max_terminated_length": 2264.0,
      "completions/mean_length": 530.7421875,
      "completions/mean_terminated_length": 545.6626586914062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 151.0,
      "epoch": 0.19946666666666665,
      "grad_norm": 0.5334581732749939,
      "learning_rate": 3.8888888888888895e-07,
      "loss": 0.0159,
      "num_tokens": 38860440.0,
      "reward": 1.1118857860565186,
      "reward_std": 0.29268577694892883,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6808027029037476,
      "rewards/format_reward_step": 0.97265625,
      "step": 187
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7995928060263395,
      "aux_distill/mean_u": 0.2580252115995233,
      "aux_distill/n_active_tok": 169.25,
      "calib/answer_extract_rate": 1.0,
      "calib/avg_num_step_conf": 9.96875,
      "calib/ece": 0.20089843750000014,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.921875,
      "calib/gap": 0.1752726970402394,
      "calib/mean_conf": 0.9362109375000001,
      "calib/mu_c": 0.978659793814433,
      "calib/mu_w": 0.8033870967741936,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.18964843750000013,
      "calib/std_conf": 0.22592020688812034,
      "calib/step_conf_rate": 0.99609375,
      "calib/step_q_c": 0.22686349115801482,
      "calib/step_q_c_n": 1753.0,
      "calib/step_q_gap": -0.017692203460258016,
      "calib/step_q_w": 0.24455569461827284,
      "calib/step_q_w_n": 799.0,
      "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": 2698.0,
      "completions/max_terminated_length": 2698.0,
      "completions/mean_length": 581.9609375,
      "completions/mean_terminated_length": 586.5433349609375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 174.0,
      "epoch": 0.20053333333333334,
      "grad_norm": 0.24379847943782806,
      "learning_rate": 3.611111111111111e-07,
      "loss": 0.0696,
      "num_tokens": 39113494.0,
      "reward": 1.2739689350128174,
      "reward_std": 0.14255055785179138,
      "rewards/accuracy_reward_step": 0.7578125,
      "rewards/final_brier_reward_step": 0.7940316200256348,
      "rewards/format_reward_step": 0.99609375,
      "step": 188
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7341479770839214,
      "aux_distill/mean_u": 0.23114255566897401,
      "aux_distill/n_active_tok": 134.125,
      "calib/answer_extract_rate": 0.9921875,
      "calib/avg_num_step_conf": 8.6171875,
      "calib/ece": 0.24019685039370073,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.8346456692913385,
      "calib/gap": 0.24245139664804471,
      "calib/mean_conf": 0.8536614173228346,
      "calib/mu_c": 0.9252513966480447,
      "calib/mu_w": 0.6828,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1945669291338582,
      "calib/std_conf": 0.33396859016985175,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.21389307745030842,
      "calib/step_q_c_n": 1459.0,
      "calib/step_q_gap": -0.06229166150551488,
      "calib/step_q_w": 0.2761847389558233,
      "calib/step_q_w_n": 747.0,
      "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": 1380.0,
      "completions/max_terminated_length": 1380.0,
      "completions/mean_length": 481.73046875,
      "completions/mean_terminated_length": 487.4427185058594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 180.0,
      "epoch": 0.2016,
      "grad_norm": 0.43488699197769165,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.0585,
      "num_tokens": 39344585.0,
      "reward": 1.222286581993103,
      "reward_std": 0.1747114658355713,
      "rewards/accuracy_reward_step": 0.69921875,
      "rewards/final_brier_reward_step": 0.7531667947769165,
      "rewards/format_reward_step": 0.9921875,
      "step": 189
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7943196892738342,
      "aux_distill/mean_u": 0.28323400437481766,
      "aux_distill/n_active_tok": 155.0,
      "calib/answer_extract_rate": 0.98046875,
      "calib/avg_num_step_conf": 10.48828125,
      "calib/ece": 0.24609561752988054,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.8645418326693227,
      "calib/gap": 0.3083033891384239,
      "calib/mean_conf": 0.8734262948207171,
      "calib/mu_c": 0.9876582278481013,
      "calib/mu_w": 0.6793548387096774,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24501992031872516,
      "calib/std_conf": 0.3173474059160562,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25361627260083447,
      "calib/step_q_c_n": 1438.0,
      "calib/step_q_gap": -0.031147159636535193,
      "calib/step_q_w": 0.28476343223736966,
      "calib/step_q_w_n": 1247.0,
      "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": 1389.0,
      "completions/max_terminated_length": 1389.0,
      "completions/mean_length": 550.19921875,
      "completions/mean_terminated_length": 561.1593627929688,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 187.0,
      "epoch": 0.20266666666666666,
      "grad_norm": 0.3425717055797577,
      "learning_rate": 3.055555555555556e-07,
      "loss": 0.0328,
      "num_tokens": 39591044.0,
      "reward": 1.1666810512542725,
      "reward_std": 0.2564879357814789,
      "rewards/accuracy_reward_step": 0.6171875,
      "rewards/final_brier_reward_step": 0.7357058525085449,
      "rewards/format_reward_step": 0.98046875,
      "step": 190
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7623467231169343,
      "aux_distill/mean_u": 0.20448913400170793,
      "aux_distill/n_active_tok": 195.125,
      "calib/answer_extract_rate": 0.96875,
      "calib/avg_num_step_conf": 11.90625,
      "calib/ece": 0.298991935483871,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.8951612903225806,
      "calib/gap": 0.20007046145344032,
      "calib/mean_conf": 0.8995564516129033,
      "calib/mu_c": 0.9753896103896105,
      "calib/mu_w": 0.7753191489361702,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.2887903225806452,
      "calib/std_conf": 0.2888328236870281,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22404383077160353,
      "calib/step_q_c_n": 1430.0,
      "calib/step_q_gap": -0.04238225081059674,
      "calib/step_q_w": 0.26642608158220027,
      "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": 0.03125,
      "completions/max_length": 2723.0,
      "completions/max_terminated_length": 2723.0,
      "completions/mean_length": 548.07421875,
      "completions/mean_terminated_length": 565.7540283203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 147.0,
      "epoch": 0.20373333333333332,
      "grad_norm": 0.23078249394893646,
      "learning_rate": 2.7777777777777776e-07,
      "loss": 0.011,
      "num_tokens": 39835519.0,
      "reward": 1.1231420040130615,
      "reward_std": 0.21966737508773804,
      "rewards/accuracy_reward_step": 0.6015625,
      "rewards/final_brier_reward_step": 0.675971508026123,
      "rewards/format_reward_step": 0.96875,
      "step": 191
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7831371380016208,
      "aux_distill/mean_u": 0.20444626811200128,
      "aux_distill/n_active_tok": 146.5,
      "calib/answer_extract_rate": 0.984375,
      "calib/avg_num_step_conf": 9.640625,
      "calib/ece": 0.20777777777777776,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.8690476190476191,
      "calib/gap": 0.3182753246753247,
      "calib/mean_conf": 0.8820634920634921,
      "calib/mu_c": 0.9793142857142857,
      "calib/mu_w": 0.661038961038961,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.19769841269841268,
      "calib/std_conf": 0.30560457244438705,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22681944672962867,
      "calib/step_q_c_n": 1373.0,
      "calib/step_q_gap": -0.038139457379960384,
      "calib/step_q_w": 0.26495890410958906,
      "calib/step_q_w_n": 1095.0,
      "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": 2369.0,
      "completions/max_terminated_length": 2369.0,
      "completions/mean_length": 551.92578125,
      "completions/mean_terminated_length": 560.6865234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 132.0,
      "epoch": 0.2048,
      "grad_norm": 0.24515196681022644,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.0516,
      "num_tokens": 40081788.0,
      "reward": 1.2249211072921753,
      "reward_std": 0.18614095449447632,
      "rewards/accuracy_reward_step": 0.68359375,
      "rewards/final_brier_reward_step": 0.7818734049797058,
      "rewards/format_reward_step": 0.984375,
      "step": 192
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7868331456556916,
      "aux_distill/mean_u": 0.223379436485949,
      "aux_distill/n_active_tok": 125.25,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 10.15234375,
      "calib/ece": 0.33192771084337347,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.9397590361445783,
      "calib/gap": 0.07807935393258403,
      "calib/mean_conf": 0.9436546184738956,
      "calib/mu_c": 0.9715624999999999,
      "calib/mu_w": 0.8934831460674159,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.3165060240963855,
      "calib/std_conf": 0.21971480507931,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2257960056061668,
      "calib/step_q_c_n": 1427.0,
      "calib/step_q_gap": 0.017220920281934726,
      "calib/step_q_w": 0.20857508532423208,
      "calib/step_q_w_n": 1172.0,
      "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": 1447.0,
      "completions/max_terminated_length": 1447.0,
      "completions/mean_length": 492.33984375,
      "completions/mean_terminated_length": 506.1806945800781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 204.0,
      "epoch": 0.20586666666666667,
      "grad_norm": 0.3053726553916931,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.0685,
      "num_tokens": 40313539.0,
      "reward": 1.123337984085083,
      "reward_std": 0.25543320178985596,
      "rewards/accuracy_reward_step": 0.625,
      "rewards/final_brier_reward_step": 0.6490199565887451,
      "rewards/format_reward_step": 0.97265625,
      "step": 193
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8021860085427761,
      "aux_distill/mean_u": 0.2652982515224455,
      "aux_distill/n_active_tok": 133.625,
      "calib/answer_extract_rate": 0.98828125,
      "calib/avg_num_step_conf": 9.234375,
      "calib/ece": 0.23719367588932805,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.8814229249011858,
      "calib/gap": 0.2746909992912827,
      "calib/mean_conf": 0.8877075098814229,
      "calib/mu_c": 0.9778235294117648,
      "calib/mu_w": 0.703132530120482,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.22648221343873517,
      "calib/std_conf": 0.30340668096512574,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20807055435565155,
      "calib/step_q_c_n": 1389.0,
      "calib/step_q_gap": -0.01781662513152793,
      "calib/step_q_w": 0.22588717948717948,
      "calib/step_q_w_n": 975.0,
      "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": 1787.0,
      "completions/max_terminated_length": 1787.0,
      "completions/mean_length": 479.71484375,
      "completions/mean_terminated_length": 485.4031677246094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 210.0,
      "epoch": 0.20693333333333333,
      "grad_norm": 0.29109108448028564,
      "learning_rate": 1.9444444444444447e-07,
      "loss": 0.0248,
      "num_tokens": 40542290.0,
      "reward": 1.2027337551116943,
      "reward_std": 0.17572781443595886,
      "rewards/accuracy_reward_step": 0.6640625,
      "rewards/final_brier_reward_step": 0.7531238198280334,
      "rewards/format_reward_step": 0.98828125,
      "step": 194
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7851871335878968,
      "aux_distill/mean_u": 0.24789486043876469,
      "aux_distill/n_active_tok": 156.625,
      "calib/answer_extract_rate": 0.95703125,
      "calib/avg_num_step_conf": 11.4375,
      "calib/ece": 0.33289795918367343,
      "calib/final_conf_rate": 0.95703125,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.8979591836734694,
      "calib/gap": 0.16312925170068038,
      "calib/mean_conf": 0.9019591836734694,
      "calib/mu_c": 0.9672108843537416,
      "calib/mu_w": 0.8040816326530612,
      "calib/nonempty_final_conf_rate": 0.95703125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.3174285714285714,
      "calib/std_conf": 0.2854159842200734,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.25532035842293904,
      "calib/step_q_c_n": 1395.0,
      "calib/step_q_gap": 0.05504638582019933,
      "calib/step_q_w": 0.2002739726027397,
      "calib/step_q_w_n": 1533.0,
      "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": 0.04296875,
      "completions/max_length": 2072.0,
      "completions/max_terminated_length": 2072.0,
      "completions/mean_length": 528.6640625,
      "completions/mean_terminated_length": 552.3999633789062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 158.0,
      "epoch": 0.208,
      "grad_norm": 0.37227320671081543,
      "learning_rate": 1.6666666666666668e-07,
      "loss": -0.0018,
      "num_tokens": 40783612.0,
      "reward": 1.0841538906097412,
      "reward_std": 0.26760923862457275,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.6370578408241272,
      "rewards/format_reward_step": 0.95703125,
      "step": 195
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.8097772235050797,
      "aux_distill/mean_u": 0.250261366145977,
      "aux_distill/n_active_tok": 138.125,
      "calib/answer_extract_rate": 0.99609375,
      "calib/avg_num_step_conf": 8.828125,
      "calib/ece": 0.273461921647051,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.9490196078431372,
      "calib/gap": 0.1354934395147117,
      "calib/mean_conf": 0.9515011373333255,
      "calib/mu_c": 0.9945402298850575,
      "calib/mu_w": 0.8590467903703458,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.271305058901953,
      "calib/std_conf": 0.20554215236891232,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.20119723183391006,
      "calib/step_q_c_n": 1445.0,
      "calib/step_q_gap": -0.04124816693909611,
      "calib/step_q_w": 0.24244539877300617,
      "calib/step_q_w_n": 815.0,
      "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": 1477.0,
      "completions/max_terminated_length": 1477.0,
      "completions/mean_length": 439.47265625,
      "completions/mean_terminated_length": 442.9330749511719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 165.0,
      "epoch": 0.20906666666666668,
      "grad_norm": 0.32618144154548645,
      "learning_rate": 1.3888888888888888e-07,
      "loss": 0.0864,
      "num_tokens": 40998661.0,
      "reward": 1.2001200914382935,
      "reward_std": 0.16445036232471466,
      "rewards/accuracy_reward_step": 0.6796875,
      "rewards/final_brier_reward_step": 0.7244589328765869,
      "rewards/format_reward_step": 0.99609375,
      "step": 196
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.806681577116251,
      "aux_distill/mean_u": 0.28128467985306665,
      "aux_distill/n_active_tok": 136.0,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 10.25,
      "calib/ece": 0.35570281124497993,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.9156626506024096,
      "calib/gap": 0.10875353535353527,
      "calib/mean_conf": 0.9234939759036144,
      "calib/mu_c": 0.9667333333333332,
      "calib/mu_w": 0.857979797979798,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.33839357429718875,
      "calib/std_conf": 0.25452420562463374,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2311550632911392,
      "calib/step_q_c_n": 1264.0,
      "calib/step_q_gap": -0.009418466120625552,
      "calib/step_q_w": 0.24057352941176474,
      "calib/step_q_w_n": 1360.0,
      "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": 1759.0,
      "completions/max_terminated_length": 1759.0,
      "completions/mean_length": 501.43359375,
      "completions/mean_terminated_length": 515.5300903320312,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 196.0,
      "epoch": 0.21013333333333334,
      "grad_norm": 0.31283727288246155,
      "learning_rate": 1.1111111111111112e-07,
      "loss": 0.004,
      "num_tokens": 41232084.0,
      "reward": 1.0928354263305664,
      "reward_std": 0.29931554198265076,
      "rewards/accuracy_reward_step": 0.5859375,
      "rewards/final_brier_reward_step": 0.6270769238471985,
      "rewards/format_reward_step": 0.97265625,
      "step": 197
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7634249338880181,
      "aux_distill/mean_u": 0.20665586842475142,
      "aux_distill/n_active_tok": 138.625,
      "calib/answer_extract_rate": 0.9765625,
      "calib/avg_num_step_conf": 9.84765625,
      "calib/ece": 0.26354581673306765,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.900398406374502,
      "calib/gap": 0.1527189393939392,
      "calib/mean_conf": 0.9052191235059761,
      "calib/mu_c": 0.9508522727272727,
      "calib/mu_w": 0.7981333333333335,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.23378486055776887,
      "calib/std_conf": 0.2810514049492244,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.2543802749600839,
      "calib/step_q_c_n": 1548.0,
      "calib/step_q_gap": -0.016051379716175096,
      "calib/step_q_w": 0.270431654676259,
      "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": 0.0234375,
      "completions/max_length": 1370.0,
      "completions/max_terminated_length": 1370.0,
      "completions/mean_length": 476.20703125,
      "completions/mean_terminated_length": 487.6360168457031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 143.0,
      "epoch": 0.2112,
      "grad_norm": 0.27833330631256104,
      "learning_rate": 8.333333333333334e-08,
      "loss": 0.0462,
      "num_tokens": 41459377.0,
      "reward": 1.1917555332183838,
      "reward_std": 0.2348858118057251,
      "rewards/accuracy_reward_step": 0.6875,
      "rewards/final_brier_reward_step": 0.719448447227478,
      "rewards/format_reward_step": 0.9765625,
      "step": 198
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7831273702904582,
      "aux_distill/mean_u": 0.23099978917405076,
      "aux_distill/n_active_tok": 133.875,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 9.484375,
      "calib/ece": 0.19971999999999998,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.888,
      "calib/gap": 0.2780963154492566,
      "calib/mean_conf": 0.8971600000000001,
      "calib/mu_c": 0.9728021978021979,
      "calib/mu_w": 0.6947058823529413,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.18443999999999997,
      "calib/std_conf": 0.2912763883324565,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22950873846153844,
      "calib/step_q_c_n": 1625.0,
      "calib/step_q_gap": -0.006829991301848831,
      "calib/step_q_w": 0.23633872976338727,
      "calib/step_q_w_n": 803.0,
      "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": 0.0234375,
      "completions/max_length": 1574.0,
      "completions/max_terminated_length": 1574.0,
      "completions/mean_length": 529.06640625,
      "completions/mean_terminated_length": 541.7640380859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 143.0,
      "epoch": 0.21226666666666666,
      "grad_norm": 0.24881617724895477,
      "learning_rate": 5.555555555555556e-08,
      "loss": 0.0162,
      "num_tokens": 41699018.0,
      "reward": 1.2316912412643433,
      "reward_std": 0.23456567525863647,
      "rewards/accuracy_reward_step": 0.7109375,
      "rewards/final_brier_reward_step": 0.7797886729240417,
      "rewards/format_reward_step": 0.97265625,
      "step": 199
    },
    {
      "aux_distill/lambda": 0.10000000000000005,
      "aux_distill/loss": 0.7529831556603312,
      "aux_distill/mean_u": 0.220185384755039,
      "aux_distill/n_active_tok": 161.875,
      "calib/answer_extract_rate": 0.97265625,
      "calib/avg_num_step_conf": 10.34375,
      "calib/ece": 0.23032128514056222,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.8875502008032129,
      "calib/gap": 0.22098116790631428,
      "calib/mean_conf": 0.8986746987951807,
      "calib/mu_c": 0.9616853932584271,
      "calib/mu_w": 0.7407042253521128,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.20706827309236944,
      "calib/std_conf": 0.28799865190030877,
      "calib/step_conf_rate": 1.0,
      "calib/step_q_c": 0.22154745529573588,
      "calib/step_q_c_n": 1454.0,
      "calib/step_q_gap": -0.06527540902587214,
      "calib/step_q_w": 0.28682286432160803,
      "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": 0.02734375,
      "completions/max_length": 2556.0,
      "completions/max_terminated_length": 2556.0,
      "completions/mean_length": 553.09375,
      "completions/mean_terminated_length": 568.642578125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 130.0,
      "epoch": 0.21333333333333335,
      "grad_norm": 0.2786003649234772,
      "learning_rate": 2.777777777777778e-08,
      "loss": 0.0135,
      "num_tokens": 41948658.0,
      "reward": 1.2082239389419556,
      "reward_std": 0.16592733561992645,
      "rewards/accuracy_reward_step": 0.6953125,
      "rewards/final_brier_reward_step": 0.7484792470932007,
      "rewards/format_reward_step": 0.97265625,
      "step": 200
    },
    {
      "epoch": 0.21333333333333335,
      "step": 200,
      "total_flos": 0.0,
      "train_loss": 0.04889526396640576,
      "train_runtime": 15994.4075,
      "train_samples_per_second": 3.201,
      "train_steps_per_second": 0.013
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 41948658,
  "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
}