{
  "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_brier/lambda": 0.1,
      "aux_brier/loss": -9.102586197069666e-06,
      "aux_brier/mean_r": 0.4665906016643231,
      "aux_brier/n_active_tok": 24.615384615384617,
      "aux_brier/n_step_records": 6.153846153846154,
      "aux_brier/std_r": 0.17818317848902482,
      "calib/answer_extract_rate": 0.08203125,
      "calib/auroc": 0.6944444444444445,
      "calib/avg_num_step_conf": 0.3359375,
      "calib/ece": 0.6230769230769231,
      "calib/final_conf_rate": 0.05078125,
      "calib/format_rate": 0.04296875,
      "calib/frac_conf_gt_0.9": 0.7692307692307693,
      "calib/gap": 0.03861111111111115,
      "calib/mean_conf": 0.9307692307692309,
      "calib/mu_c": 0.9575,
      "calib/mu_w": 0.9188888888888889,
      "calib/nonempty_final_conf_rate": 0.05078125,
      "calib/nonempty_reasoning_rate": 0.09765625,
      "calib/nonempty_step_conf_rate": 0.0703125,
      "calib/pce": 0.6230769230769231,
      "calib/std_conf": 0.07965903671384378,
      "calib/step_conf_rate": 0.0703125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2955.0,
      "completions/max_terminated_length": 2955.0,
      "completions/mean_length": 613.67578125,
      "completions/mean_terminated_length": 674.2532348632812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0010666666666666667,
      "grad_norm": 0.5645009279251099,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.0318,
      "num_tokens": 264685.0,
      "reward": 0.04124843701720238,
      "reward_std": 0.0838509351015091,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.01655624993145466,
      "rewards/format_reward_step": 0.04296875,
      "step": 1
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": 1.2052168732018822e-07,
      "aux_brier/mean_r": 0.43024390700616333,
      "aux_brier/n_active_tok": 28.42105263157895,
      "aux_brier/n_step_records": 7.105263157894737,
      "aux_brier/std_r": 0.16724381615456782,
      "calib/answer_extract_rate": 0.13671875,
      "calib/auroc": 0.5338345864661654,
      "calib/avg_num_step_conf": 0.55078125,
      "calib/ece": 0.6261538461538463,
      "calib/final_conf_rate": 0.1015625,
      "calib/format_rate": 0.08984375,
      "calib/frac_conf_gt_0.9": 0.7692307692307693,
      "calib/gap": 0.002406015037593856,
      "calib/mean_conf": 0.8953846153846153,
      "calib/mu_c": 0.897142857142857,
      "calib/mu_w": 0.8947368421052632,
      "calib/nonempty_final_conf_rate": 0.1015625,
      "calib/nonempty_reasoning_rate": 0.14453125,
      "calib/nonempty_step_conf_rate": 0.109375,
      "calib/pce": 0.6261538461538463,
      "calib/std_conf": 0.18653172073466937,
      "calib/step_conf_rate": 0.109375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3001.0,
      "completions/max_terminated_length": 3001.0,
      "completions/mean_length": 646.4609375,
      "completions/mean_terminated_length": 683.8594970703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0021333333333333334,
      "grad_norm": 0.14317761361598969,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0643,
      "num_tokens": 533467.0,
      "reward": 0.08358447253704071,
      "reward_std": 0.15892045199871063,
      "rewards/accuracy_reward_step": 0.03125,
      "rewards/final_brier_reward_step": 0.02965039201080799,
      "rewards/format_reward_step": 0.08984375,
      "step": 2
    },
    {
      "aux_brier/lambda": 0.09999999999999999,
      "aux_brier/loss": 6.794580712045993e-06,
      "aux_brier/mean_r": 0.45482745360244403,
      "aux_brier/n_active_tok": 24.363636363636363,
      "aux_brier/n_step_records": 6.090909090909091,
      "aux_brier/std_r": 0.16174368424849075,
      "calib/answer_extract_rate": 0.05859375,
      "calib/auroc": 0.5714285714285714,
      "calib/avg_num_step_conf": 0.30078125,
      "calib/ece": 0.4153846153846156,
      "calib/final_conf_rate": 0.05078125,
      "calib/format_rate": 0.046875,
      "calib/frac_conf_gt_0.9": 0.9230769230769231,
      "calib/gap": -0.002142857142857224,
      "calib/mean_conf": 0.953846153846154,
      "calib/mu_c": 0.9528571428571427,
      "calib/mu_w": 0.955,
      "calib/nonempty_final_conf_rate": 0.05078125,
      "calib/nonempty_reasoning_rate": 0.0703125,
      "calib/nonempty_step_conf_rate": 0.06640625,
      "calib/pce": 0.4153846153846156,
      "calib/std_conf": 0.024349525832837542,
      "calib/step_conf_rate": 0.06640625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2999.0,
      "completions/max_terminated_length": 2999.0,
      "completions/mean_length": 624.51953125,
      "completions/mean_terminated_length": 695.1173706054688,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0032,
      "grad_norm": 0.019731556996703148,
      "learning_rate": 7.5e-07,
      "loss": 0.033,
      "num_tokens": 798600.0,
      "reward": 0.058033399283885956,
      "reward_std": 0.11475954204797745,
      "rewards/accuracy_reward_step": 0.02734375,
      "rewards/final_brier_reward_step": 0.029008593410253525,
      "rewards/format_reward_step": 0.046875,
      "step": 3
    },
    {
      "aux_brier/lambda": 0.09999999999999999,
      "aux_brier/loss": 3.005054905003135e-05,
      "aux_brier/mean_r": 0.46613133805138723,
      "aux_brier/n_active_tok": 25.714285714285715,
      "aux_brier/n_step_records": 6.428571428571429,
      "aux_brier/std_r": 0.15916945040225983,
      "calib/answer_extract_rate": 0.05859375,
      "calib/auroc": 0.28125,
      "calib/avg_num_step_conf": 0.17578125,
      "calib/ece": 0.6539999999999999,
      "calib/final_conf_rate": 0.0390625,
      "calib/format_rate": 0.02734375,
      "calib/frac_conf_gt_0.9": 0.6,
      "calib/gap": -0.036250000000000004,
      "calib/mean_conf": 0.8539999999999999,
      "calib/mu_c": 0.825,
      "calib/mu_w": 0.86125,
      "calib/nonempty_final_conf_rate": 0.0390625,
      "calib/nonempty_reasoning_rate": 0.0625,
      "calib/nonempty_step_conf_rate": 0.03515625,
      "calib/pce": 0.6539999999999999,
      "calib/std_conf": 0.1585685971433184,
      "calib/step_conf_rate": 0.03515625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2845.0,
      "completions/max_terminated_length": 2845.0,
      "completions/mean_length": 671.32421875,
      "completions/mean_terminated_length": 716.0792236328125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.004266666666666667,
      "grad_norm": 0.35197219252586365,
      "learning_rate": 1.0000000000000002e-06,
      "loss": -0.0052,
      "num_tokens": 1076627.0,
      "reward": 0.024561328813433647,
      "reward_std": 0.05980735272169113,
      "rewards/accuracy_reward_step": 0.0078125,
      "rewards/final_brier_reward_step": 0.012307812459766865,
      "rewards/format_reward_step": 0.02734375,
      "step": 4
    },
    {
      "aux_brier/lambda": 0.09999999999999999,
      "aux_brier/loss": -1.6236012197134997e-05,
      "aux_brier/mean_r": 0.430947482585907,
      "aux_brier/n_active_tok": 24.0,
      "aux_brier/n_step_records": 6.0,
      "aux_brier/std_r": 0.09453002782538533,
      "calib/answer_extract_rate": 0.04296875,
      "calib/avg_num_step_conf": 0.19140625,
      "calib/ece": 0.7899999999999999,
      "calib/final_conf_rate": 0.02734375,
      "calib/format_rate": 0.01171875,
      "calib/frac_conf_gt_0.9": 0.7142857142857143,
      "calib/mean_conf": 0.79,
      "calib/mu_c": NaN,
      "calib/mu_w": 0.79,
      "calib/nonempty_final_conf_rate": 0.02734375,
      "calib/nonempty_reasoning_rate": 0.05859375,
      "calib/nonempty_step_conf_rate": 0.03125,
      "calib/pce": 0.7899999999999999,
      "calib/std_conf": 0.3361547262794322,
      "calib/step_conf_rate": 0.03125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.1171875,
      "completions/max_length": 3021.0,
      "completions/max_terminated_length": 3021.0,
      "completions/mean_length": 734.9765625,
      "completions/mean_terminated_length": 832.539794921875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.005333333333333333,
      "grad_norm": 0.005658743903040886,
      "learning_rate": 1.25e-06,
      "loss": 0.0077,
      "num_tokens": 1371469.0,
      "reward": 0.00603193324059248,
      "reward_std": 0.013095740228891373,
      "rewards/accuracy_reward_step": 0.0,
      "rewards/final_brier_reward_step": 0.0006902343593537807,
      "rewards/format_reward_step": 0.01171875,
      "step": 5
    },
    {
      "aux_brier/lambda": 0.10000000000000002,
      "aux_brier/loss": 9.501435876657638e-06,
      "aux_brier/mean_r": 0.498914211243391,
      "aux_brier/n_active_tok": 31.0,
      "aux_brier/n_step_records": 7.75,
      "aux_brier/std_r": 0.17871646173298358,
      "calib/answer_extract_rate": 0.1328125,
      "calib/auroc": 0.25,
      "calib/avg_num_step_conf": 0.640625,
      "calib/ece": 0.6983600000000001,
      "calib/final_conf_rate": 0.09765625,
      "calib/format_rate": 0.08984375,
      "calib/frac_conf_gt_0.9": 0.76,
      "calib/gap": -0.044333333333333336,
      "calib/mean_conf": 0.9103600000000001,
      "calib/mu_c": 0.8766666666666666,
      "calib/mu_w": 0.9209999999999999,
      "calib/nonempty_final_conf_rate": 0.09765625,
      "calib/nonempty_reasoning_rate": 0.15625,
      "calib/nonempty_step_conf_rate": 0.1171875,
      "calib/pce": 0.6843600000000001,
      "calib/std_conf": 0.17328609407566434,
      "calib/step_conf_rate": 0.1171875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.09375,
      "completions/max_length": 3035.0,
      "completions/max_terminated_length": 3035.0,
      "completions/mean_length": 602.7265625,
      "completions/mean_terminated_length": 665.0775756835938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 4.0,
      "epoch": 0.0064,
      "grad_norm": 0.1781889647245407,
      "learning_rate": 1.5e-06,
      "loss": 0.0369,
      "num_tokens": 1631719.0,
      "reward": 0.07612704485654831,
      "reward_std": 0.14089490473270416,
      "rewards/accuracy_reward_step": 0.0234375,
      "rewards/final_brier_reward_step": 0.031070686876773834,
      "rewards/format_reward_step": 0.08984375,
      "step": 6
    },
    {
      "aux_brier/lambda": 0.1,
      "aux_brier/loss": 2.105866712052664e-06,
      "aux_brier/mean_r": 0.49490843827907854,
      "aux_brier/n_active_tok": 29.846153846153847,
      "aux_brier/n_step_records": 7.461538461538462,
      "aux_brier/std_r": 0.17764290766074106,
      "calib/answer_extract_rate": 0.09765625,
      "calib/auroc": 0.6384615384615384,
      "calib/avg_num_step_conf": 0.40234375,
      "calib/ece": 0.6325000000000002,
      "calib/final_conf_rate": 0.0703125,
      "calib/format_rate": 0.05859375,
      "calib/frac_conf_gt_0.9": 0.7777777777777778,
      "calib/gap": 0.057769230769230684,
      "calib/mean_conf": 0.9102777777777779,
      "calib/mu_c": 0.952,
      "calib/mu_w": 0.8942307692307693,
      "calib/nonempty_final_conf_rate": 0.0703125,
      "calib/nonempty_reasoning_rate": 0.109375,
      "calib/nonempty_step_conf_rate": 0.07421875,
      "calib/pce": 0.6325000000000002,
      "calib/std_conf": 0.17748608883063213,
      "calib/step_conf_rate": 0.07421875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.109375,
      "completions/max_length": 3053.0,
      "completions/max_terminated_length": 3053.0,
      "completions/mean_length": 693.28515625,
      "completions/mean_terminated_length": 778.4254150390625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.007466666666666667,
      "grad_norm": 0.14674966037273407,
      "learning_rate": 1.75e-06,
      "loss": 0.0137,
      "num_tokens": 1916624.0,
      "reward": 0.054463449865579605,
      "reward_std": 0.13023746013641357,
      "rewards/accuracy_reward_step": 0.01953125,
      "rewards/final_brier_reward_step": 0.022541308775544167,
      "rewards/format_reward_step": 0.05859375,
      "step": 7
    },
    {
      "aux_brier/lambda": 0.1,
      "aux_brier/loss": -1.7897489714194643e-05,
      "aux_brier/mean_r": 0.4835428978715624,
      "aux_brier/n_active_tok": 20.285714285714285,
      "aux_brier/n_step_records": 5.071428571428571,
      "aux_brier/std_r": 0.153486252363239,
      "calib/answer_extract_rate": 0.0703125,
      "calib/auroc": 0.5625,
      "calib/avg_num_step_conf": 0.28125,
      "calib/ece": 0.6809375,
      "calib/final_conf_rate": 0.0625,
      "calib/format_rate": 0.0390625,
      "calib/frac_conf_gt_0.9": 0.875,
      "calib/gap": 0.04874999999999985,
      "calib/mean_conf": 0.9309375,
      "calib/mu_c": 0.9675,
      "calib/mu_w": 0.9187500000000002,
      "calib/nonempty_final_conf_rate": 0.0625,
      "calib/nonempty_reasoning_rate": 0.1015625,
      "calib/nonempty_step_conf_rate": 0.078125,
      "calib/pce": 0.6809375,
      "calib/std_conf": 0.10568554108178659,
      "calib/step_conf_rate": 0.078125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2990.0,
      "completions/max_terminated_length": 2990.0,
      "completions/mean_length": 567.046875,
      "completions/mean_terminated_length": 615.1016845703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.008533333333333334,
      "grad_norm": 0.7247595191001892,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0239,
      "num_tokens": 2168300.0,
      "reward": 0.04034404456615448,
      "reward_std": 0.10075663030147552,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.020751170814037323,
      "rewards/format_reward_step": 0.0390625,
      "step": 8
    },
    {
      "aux_brier/lambda": 0.09999999999999999,
      "aux_brier/loss": -1.0477180734103086e-05,
      "aux_brier/mean_r": 0.5385760571807623,
      "aux_brier/n_active_tok": 20.5,
      "aux_brier/n_step_records": 5.125,
      "aux_brier/std_r": 0.1042425143532455,
      "calib/answer_extract_rate": 0.0546875,
      "calib/auroc": 0.4285714285714286,
      "calib/avg_num_step_conf": 0.16796875,
      "calib/ece": 0.84875,
      "calib/final_conf_rate": 0.03125,
      "calib/format_rate": 0.02734375,
      "calib/frac_conf_gt_0.9": 1.0,
      "calib/gap": -0.004285714285714337,
      "calib/mean_conf": 0.97375,
      "calib/mu_c": 0.97,
      "calib/mu_w": 0.9742857142857143,
      "calib/nonempty_final_conf_rate": 0.03125,
      "calib/nonempty_reasoning_rate": 0.078125,
      "calib/nonempty_step_conf_rate": 0.05078125,
      "calib/pce": 0.84875,
      "calib/std_conf": 0.011110243021644496,
      "calib/step_conf_rate": 0.05078125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.109375,
      "completions/max_length": 3066.0,
      "completions/max_terminated_length": 3066.0,
      "completions/mean_length": 643.45703125,
      "completions/mean_terminated_length": 722.4780883789062,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0096,
      "grad_norm": 0.006357858423143625,
      "learning_rate": 2.25e-06,
      "loss": 0.0304,
      "num_tokens": 2440561.0,
      "reward": 0.01882334053516388,
      "reward_std": 0.041674330830574036,
      "rewards/accuracy_reward_step": 0.00390625,
      "rewards/final_brier_reward_step": 0.004980859346687794,
      "rewards/format_reward_step": 0.02734375,
      "step": 9
    },
    {
      "aux_brier/lambda": 0.10000000000000002,
      "aux_brier/loss": 1.272371829008634e-07,
      "aux_brier/mean_r": 0.47323180735111237,
      "aux_brier/n_active_tok": 37.25,
      "aux_brier/n_step_records": 9.3125,
      "aux_brier/std_r": 0.14950991631485522,
      "calib/answer_extract_rate": 0.0859375,
      "calib/auroc": 0.46875,
      "calib/avg_num_step_conf": 0.59375,
      "calib/ece": 0.8361111111111111,
      "calib/final_conf_rate": 0.0703125,
      "calib/format_rate": 0.05078125,
      "calib/frac_conf_gt_0.9": 0.8333333333333334,
      "calib/gap": -0.17812499999999998,
      "calib/mean_conf": 0.8883333333333333,
      "calib/mu_c": 0.73,
      "calib/mu_w": 0.908125,
      "calib/nonempty_final_conf_rate": 0.0703125,
      "calib/nonempty_reasoning_rate": 0.10546875,
      "calib/nonempty_step_conf_rate": 0.08203125,
      "calib/pce": 0.8066666666666666,
      "calib/std_conf": 0.20097678141185035,
      "calib/step_conf_rate": 0.08203125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0859375,
      "completions/max_length": 3019.0,
      "completions/max_terminated_length": 3019.0,
      "completions/mean_length": 701.59765625,
      "completions/mean_terminated_length": 767.5598754882812,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.010666666666666666,
      "grad_norm": 0.1916794329881668,
      "learning_rate": 2.5e-06,
      "loss": 0.0066,
      "num_tokens": 2726970.0,
      "reward": 0.03645956888794899,
      "reward_std": 0.08734627068042755,
      "rewards/accuracy_reward_step": 0.0078125,
      "rewards/final_brier_reward_step": 0.013025781139731407,
      "rewards/format_reward_step": 0.05078125,
      "step": 10
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": -1.1409345462924586e-05,
      "aux_brier/mean_r": 0.487497550876517,
      "aux_brier/n_active_tok": 32.0,
      "aux_brier/n_step_records": 8.0,
      "aux_brier/std_r": 0.1807484465994333,
      "calib/answer_extract_rate": 0.1484375,
      "calib/auroc": 0.5874999999999999,
      "calib/avg_num_step_conf": 0.61328125,
      "calib/ece": 0.6446249999999999,
      "calib/final_conf_rate": 0.109375,
      "calib/format_rate": 0.09375,
      "calib/frac_conf_gt_0.9": 0.8214285714285714,
      "calib/gap": 0.006524999999999892,
      "calib/mean_conf": 0.9303392857142859,
      "calib/mu_c": 0.9349999999999999,
      "calib/mu_w": 0.928475,
      "calib/nonempty_final_conf_rate": 0.109375,
      "calib/nonempty_reasoning_rate": 0.16796875,
      "calib/nonempty_step_conf_rate": 0.12890625,
      "calib/pce": 0.6446249999999999,
      "calib/std_conf": 0.07888215142714802,
      "calib/step_conf_rate": 0.12890625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2983.0,
      "completions/max_terminated_length": 2983.0,
      "completions/mean_length": 652.19140625,
      "completions/mean_terminated_length": 692.7842407226562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.011733333333333333,
      "grad_norm": 0.3505798280239105,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.0425,
      "num_tokens": 2998411.0,
      "reward": 0.08793935179710388,
      "reward_std": 0.13506345450878143,
      "rewards/accuracy_reward_step": 0.03125,
      "rewards/final_brier_reward_step": 0.03925742208957672,
      "rewards/format_reward_step": 0.09375,
      "step": 11
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": -9.934263243215963e-06,
      "aux_brier/mean_r": 0.5341246542723282,
      "aux_brier/n_active_tok": 22.956521739130434,
      "aux_brier/n_step_records": 5.739130434782608,
      "aux_brier/std_r": 0.17240982081579126,
      "calib/answer_extract_rate": 0.125,
      "calib/auroc": 0.5130718954248366,
      "calib/avg_num_step_conf": 0.515625,
      "calib/ece": 0.5178461538461537,
      "calib/final_conf_rate": 0.1015625,
      "calib/format_rate": 0.0859375,
      "calib/frac_conf_gt_0.9": 0.7307692307692307,
      "calib/gap": 0.1213333333333333,
      "calib/mean_conf": 0.864,
      "calib/mu_c": 0.9433333333333334,
      "calib/mu_w": 0.8220000000000001,
      "calib/nonempty_final_conf_rate": 0.1015625,
      "calib/nonempty_reasoning_rate": 0.15625,
      "calib/nonempty_step_conf_rate": 0.12109375,
      "calib/pce": 0.5178461538461537,
      "calib/std_conf": 0.2585105264930082,
      "calib/step_conf_rate": 0.12109375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3017.0,
      "completions/max_terminated_length": 3017.0,
      "completions/mean_length": 575.91015625,
      "completions/mean_terminated_length": 627.3744506835938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.0128,
      "grad_norm": 0.41277876496315,
      "learning_rate": 3e-06,
      "loss": 0.0412,
      "num_tokens": 3250020.0,
      "reward": 0.0893206000328064,
      "reward_std": 0.16197365522384644,
      "rewards/accuracy_reward_step": 0.03515625,
      "rewards/final_brier_reward_step": 0.04478242248296738,
      "rewards/format_reward_step": 0.0859375,
      "step": 12
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": -4.006175787067563e-08,
      "aux_brier/mean_r": 0.47388083013621246,
      "aux_brier/n_active_tok": 33.09090909090909,
      "aux_brier/n_step_records": 8.272727272727273,
      "aux_brier/std_r": 0.1905684159560637,
      "calib/answer_extract_rate": 0.16015625,
      "calib/auroc": 0.7373737373737373,
      "calib/avg_num_step_conf": 0.734375,
      "calib/ece": 0.8316666666666666,
      "calib/final_conf_rate": 0.140625,
      "calib/format_rate": 0.11328125,
      "calib/frac_conf_gt_0.9": 0.7777777777777778,
      "calib/gap": 0.06000000000000005,
      "calib/mean_conf": 0.9149999999999999,
      "calib/mu_c": 0.9700000000000001,
      "calib/mu_w": 0.91,
      "calib/nonempty_final_conf_rate": 0.140625,
      "calib/nonempty_reasoning_rate": 0.19140625,
      "calib/nonempty_step_conf_rate": 0.15234375,
      "calib/pce": 0.8316666666666666,
      "calib/std_conf": 0.11847409468365272,
      "calib/step_conf_rate": 0.15234375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3028.0,
      "completions/max_terminated_length": 3028.0,
      "completions/mean_length": 668.41015625,
      "completions/mean_terminated_length": 704.168701171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.013866666666666666,
      "grad_norm": 0.12139452993869781,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.0534,
      "num_tokens": 3525725.0,
      "reward": 0.07909873127937317,
      "reward_std": 0.15631961822509766,
      "rewards/accuracy_reward_step": 0.015625,
      "rewards/final_brier_reward_step": 0.027332423254847527,
      "rewards/format_reward_step": 0.11328125,
      "step": 13
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": 2.7218451820911954e-08,
      "aux_brier/mean_r": 0.440531387925148,
      "aux_brier/n_active_tok": 38.08695652173913,
      "aux_brier/n_step_records": 9.521739130434783,
      "aux_brier/std_r": 0.19528335122310597,
      "calib/answer_extract_rate": 0.203125,
      "calib/auroc": 0.46794871794871795,
      "calib/avg_num_step_conf": 0.859375,
      "calib/ece": 0.5763421052631579,
      "calib/final_conf_rate": 0.1484375,
      "calib/format_rate": 0.11328125,
      "calib/frac_conf_gt_0.9": 0.6842105263157895,
      "calib/gap": -0.008628205128205169,
      "calib/mean_conf": 0.8242368421052632,
      "calib/mu_c": 0.8183333333333334,
      "calib/mu_w": 0.8269615384615385,
      "calib/nonempty_final_conf_rate": 0.1484375,
      "calib/nonempty_reasoning_rate": 0.265625,
      "calib/nonempty_step_conf_rate": 0.19140625,
      "calib/pce": 0.5423947368421053,
      "calib/std_conf": 0.27736157611647866,
      "calib/step_conf_rate": 0.19140625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3063.0,
      "completions/max_terminated_length": 3063.0,
      "completions/mean_length": 698.015625,
      "completions/mean_terminated_length": 766.91845703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 7.0,
      "epoch": 0.014933333333333333,
      "grad_norm": 0.4226285517215729,
      "learning_rate": 3.5e-06,
      "loss": 0.0676,
      "num_tokens": 3809817.0,
      "reward": 0.1168142557144165,
      "reward_std": 0.22101931273937225,
      "rewards/accuracy_reward_step": 0.046875,
      "rewards/final_brier_reward_step": 0.05319453030824661,
      "rewards/format_reward_step": 0.11328125,
      "step": 14
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": -3.0660661435681175e-07,
      "aux_brier/mean_r": 0.404715562860171,
      "aux_brier/n_active_tok": 32.833333333333336,
      "aux_brier/n_step_records": 8.208333333333334,
      "aux_brier/std_r": 0.20035543335446468,
      "calib/answer_extract_rate": 0.19140625,
      "calib/auroc": 0.3619047619047619,
      "calib/avg_num_step_conf": 0.78515625,
      "calib/ece": 0.7734146341463415,
      "calib/final_conf_rate": 0.16015625,
      "calib/format_rate": 0.12109375,
      "calib/frac_conf_gt_0.9": 0.8536585365853658,
      "calib/gap": -0.0024285714285714466,
      "calib/mean_conf": 0.9070731707317075,
      "calib/mu_c": 0.9049999999999999,
      "calib/mu_w": 0.9074285714285714,
      "calib/nonempty_final_conf_rate": 0.16015625,
      "calib/nonempty_reasoning_rate": 0.2265625,
      "calib/nonempty_step_conf_rate": 0.16796875,
      "calib/pce": 0.7670731707317073,
      "calib/std_conf": 0.1874116885028984,
      "calib/step_conf_rate": 0.16796875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3054.0,
      "completions/max_terminated_length": 3054.0,
      "completions/mean_length": 656.3671875,
      "completions/mean_terminated_length": 685.8367309570312,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.016,
      "grad_norm": 0.18445859849452972,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0866,
      "num_tokens": 4085727.0,
      "reward": 0.09110702574253082,
      "reward_std": 0.17865130305290222,
      "rewards/accuracy_reward_step": 0.0234375,
      "rewards/final_brier_reward_step": 0.028490623459219933,
      "rewards/format_reward_step": 0.12109375,
      "step": 15
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": -6.621526758454621e-06,
      "aux_brier/mean_r": 0.5042844500679237,
      "aux_brier/n_active_tok": 35.23076923076923,
      "aux_brier/n_step_records": 8.807692307692308,
      "aux_brier/std_r": 0.2048667217963017,
      "calib/answer_extract_rate": 0.21875,
      "calib/auroc": 0.3823529411764706,
      "calib/avg_num_step_conf": 0.9140625,
      "calib/ece": 0.7242553191489363,
      "calib/final_conf_rate": 0.18359375,
      "calib/format_rate": 0.16015625,
      "calib/frac_conf_gt_0.9": 0.6808510638297872,
      "calib/gap": -0.1997511312217195,
      "calib/mean_conf": 0.886808510638298,
      "calib/mu_c": 0.7423076923076923,
      "calib/mu_w": 0.9420588235294118,
      "calib/nonempty_final_conf_rate": 0.18359375,
      "calib/nonempty_reasoning_rate": 0.25390625,
      "calib/nonempty_step_conf_rate": 0.203125,
      "calib/pce": 0.6672340425531915,
      "calib/std_conf": 0.210390785161234,
      "calib/step_conf_rate": 0.203125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3022.0,
      "completions/max_terminated_length": 3022.0,
      "completions/mean_length": 643.0234375,
      "completions/mean_terminated_length": 697.5169677734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 2.0,
      "epoch": 0.017066666666666667,
      "grad_norm": 0.18127691745758057,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.1015,
      "num_tokens": 4359189.0,
      "reward": 0.142335444688797,
      "reward_std": 0.24648767709732056,
      "rewards/accuracy_reward_step": 0.05078125,
      "rewards/final_brier_reward_step": 0.04590429738163948,
      "rewards/format_reward_step": 0.16015625,
      "step": 16
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.7248619893495132e-08,
      "aux_brier/mean_r": 0.4430529158562422,
      "aux_brier/n_active_tok": 44.0,
      "aux_brier/n_step_records": 11.0,
      "aux_brier/std_r": 0.2100737655418925,
      "calib/answer_extract_rate": 0.28515625,
      "calib/auroc": 0.44,
      "calib/avg_num_step_conf": 1.37890625,
      "calib/ece": 0.6570769230769231,
      "calib/final_conf_rate": 0.25390625,
      "calib/format_rate": 0.21484375,
      "calib/frac_conf_gt_0.9": 0.7384615384615385,
      "calib/gap": 0.004133333333333433,
      "calib/mean_conf": 0.8721538461538463,
      "calib/mu_c": 0.8753333333333333,
      "calib/mu_w": 0.8711999999999999,
      "calib/nonempty_final_conf_rate": 0.25390625,
      "calib/nonempty_reasoning_rate": 0.32421875,
      "calib/nonempty_step_conf_rate": 0.26953125,
      "calib/pce": 0.6492307692307693,
      "calib/std_conf": 0.2245781844853715,
      "calib/step_conf_rate": 0.26953125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3068.0,
      "completions/max_terminated_length": 3068.0,
      "completions/mean_length": 592.19921875,
      "completions/mean_terminated_length": 613.77734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.018133333333333335,
      "grad_norm": 0.13673274219036102,
      "learning_rate": 4.25e-06,
      "loss": 0.1005,
      "num_tokens": 4614320.0,
      "reward": 0.19617079198360443,
      "reward_std": 0.2713271975517273,
      "rewards/accuracy_reward_step": 0.0703125,
      "rewards/final_brier_reward_step": 0.07374569773674011,
      "rewards/format_reward_step": 0.21484375,
      "step": 17
    },
    {
      "aux_brier/lambda": 0.10000000000000003,
      "aux_brier/loss": 6.56118595898647e-08,
      "aux_brier/mean_r": 0.5370493102073669,
      "aux_brier/n_active_tok": 38.4,
      "aux_brier/n_step_records": 9.6,
      "aux_brier/std_r": 0.20041450768709182,
      "calib/answer_extract_rate": 0.23828125,
      "calib/auroc": 0.5308823529411765,
      "calib/avg_num_step_conf": 0.953125,
      "calib/ece": 0.708409090909091,
      "calib/final_conf_rate": 0.171875,
      "calib/format_rate": 0.15234375,
      "calib/frac_conf_gt_0.9": 0.7045454545454546,
      "calib/gap": -0.03429411764705881,
      "calib/mean_conf": 0.9025000000000002,
      "calib/mu_c": 0.876,
      "calib/mu_w": 0.9102941176470588,
      "calib/nonempty_final_conf_rate": 0.171875,
      "calib/nonempty_reasoning_rate": 0.27734375,
      "calib/nonempty_step_conf_rate": 0.203125,
      "calib/pce": 0.6918181818181819,
      "calib/std_conf": 0.13600008355612406,
      "calib/step_conf_rate": 0.203125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3051.0,
      "completions/max_terminated_length": 3051.0,
      "completions/mean_length": 600.43359375,
      "completions/mean_terminated_length": 643.1422729492188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.0192,
      "grad_norm": 0.48188310861587524,
      "learning_rate": 4.5e-06,
      "loss": 0.0511,
      "num_tokens": 4878751.0,
      "reward": 0.13276375830173492,
      "reward_std": 0.21671216189861298,
      "rewards/accuracy_reward_step": 0.04296875,
      "rewards/final_brier_reward_step": 0.05449257791042328,
      "rewards/format_reward_step": 0.15234375,
      "step": 18
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 8.979131159033812e-09,
      "aux_brier/mean_r": 0.5055194477881154,
      "aux_brier/n_active_tok": 66.45161290322581,
      "aux_brier/n_step_records": 16.612903225806452,
      "aux_brier/std_r": 0.22263317729436582,
      "calib/answer_extract_rate": 0.453125,
      "calib/auroc": 0.30747126436781613,
      "calib/avg_num_step_conf": 2.06640625,
      "calib/ece": 0.7203961904761905,
      "calib/final_conf_rate": 0.41015625,
      "calib/format_rate": 0.34765625,
      "calib/frac_conf_gt_0.9": 0.7238095238095238,
      "calib/gap": -0.028485823754789252,
      "calib/mean_conf": 0.8918247619047619,
      "calib/mu_c": 0.8682222222222222,
      "calib/mu_w": 0.8967080459770115,
      "calib/nonempty_final_conf_rate": 0.41015625,
      "calib/nonempty_reasoning_rate": 0.51953125,
      "calib/nonempty_step_conf_rate": 0.421875,
      "calib/pce": 0.7203961904761905,
      "calib/std_conf": 0.17331358135497102,
      "calib/step_conf_rate": 0.421875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3040.0,
      "completions/max_terminated_length": 3040.0,
      "completions/mean_length": 475.53515625,
      "completions/mean_terminated_length": 492.86236572265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.020266666666666665,
      "grad_norm": 0.3311012387275696,
      "learning_rate": 4.75e-06,
      "loss": 0.1153,
      "num_tokens": 5105248.0,
      "reward": 0.27182435989379883,
      "reward_std": 0.39847370982170105,
      "rewards/accuracy_reward_step": 0.0703125,
      "rewards/final_brier_reward_step": 0.1107349842786789,
      "rewards/format_reward_step": 0.34765625,
      "step": 19
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.4849104437201445e-08,
      "aux_brier/mean_r": 0.5030051426541421,
      "aux_brier/n_active_tok": 77.29032258064517,
      "aux_brier/n_step_records": 19.322580645161292,
      "aux_brier/std_r": 0.26265350657124675,
      "calib/answer_extract_rate": 0.51953125,
      "calib/auroc": 0.5727083333333334,
      "calib/avg_num_step_conf": 2.3984375,
      "calib/ece": 0.6617107438016527,
      "calib/final_conf_rate": 0.47265625,
      "calib/format_rate": 0.38671875,
      "calib/frac_conf_gt_0.9": 0.6446280991735537,
      "calib/gap": 0.07405916666666634,
      "calib/mean_conf": 0.8683223140495869,
      "calib/mu_c": 0.9270799999999998,
      "calib/mu_w": 0.8530208333333335,
      "calib/nonempty_final_conf_rate": 0.47265625,
      "calib/nonempty_reasoning_rate": 0.625,
      "calib/nonempty_step_conf_rate": 0.5234375,
      "calib/pce": 0.6617107438016527,
      "calib/std_conf": 0.2052438265699078,
      "calib/step_conf_rate": 0.5234375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2799.0,
      "completions/max_terminated_length": 2799.0,
      "completions/mean_length": 443.4609375,
      "completions/mean_terminated_length": 455.9277038574219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.021333333333333333,
      "grad_norm": 0.19372797012329102,
      "learning_rate": 5e-06,
      "loss": 0.1248,
      "num_tokens": 5323646.0,
      "reward": 0.3269345164299011,
      "reward_std": 0.4126846492290497,
      "rewards/accuracy_reward_step": 0.09765625,
      "rewards/final_brier_reward_step": 0.14367558062076569,
      "rewards/format_reward_step": 0.38671875,
      "step": 20
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.5966909414365524e-08,
      "aux_brier/mean_r": 0.459222630597651,
      "aux_brier/n_active_tok": 106.5,
      "aux_brier/n_step_records": 26.625,
      "aux_brier/std_r": 0.2647970230318606,
      "calib/answer_extract_rate": 0.66796875,
      "calib/auroc": 0.4911543564794339,
      "calib/avg_num_step_conf": 3.41796875,
      "calib/ece": 0.6863963906581743,
      "calib/final_conf_rate": 0.61328125,
      "calib/format_rate": 0.55078125,
      "calib/frac_conf_gt_0.9": 0.7197452229299363,
      "calib/gap": -0.05793426949727243,
      "calib/mean_conf": 0.8842830148619958,
      "calib/mu_c": 0.840371052631579,
      "calib/mu_w": 0.8983053221288514,
      "calib/nonempty_final_conf_rate": 0.61328125,
      "calib/nonempty_reasoning_rate": 0.75390625,
      "calib/nonempty_step_conf_rate": 0.6796875,
      "calib/pce": 0.6643205944798303,
      "calib/std_conf": 0.19766529344420242,
      "calib/step_conf_rate": 0.6796875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2803.0,
      "completions/max_terminated_length": 2803.0,
      "completions/mean_length": 453.3671875,
      "completions/mean_terminated_length": 467.9919128417969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 24.0,
      "epoch": 0.0224,
      "grad_norm": 0.07202180474996567,
      "learning_rate": 4.9722222222222224e-06,
      "loss": 0.1197,
      "num_tokens": 5542668.0,
      "reward": 0.47843945026397705,
      "reward_std": 0.46075934171676636,
      "rewards/accuracy_reward_step": 0.15625,
      "rewards/final_brier_reward_step": 0.1871953159570694,
      "rewards/format_reward_step": 0.55078125,
      "step": 21
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -4.0624713752712704e-08,
      "aux_brier/mean_r": 0.5024625388905406,
      "aux_brier/n_active_tok": 124.125,
      "aux_brier/n_step_records": 31.03125,
      "aux_brier/std_r": 0.24132200283929706,
      "calib/answer_extract_rate": 0.69921875,
      "calib/auroc": 0.4243589743589744,
      "calib/avg_num_step_conf": 4.12109375,
      "calib/ece": 0.631592380952381,
      "calib/final_conf_rate": 0.68359375,
      "calib/format_rate": 0.59765625,
      "calib/frac_conf_gt_0.9": 0.6685714285714286,
      "calib/gap": -0.010744729344729143,
      "calib/mean_conf": 0.878944761904762,
      "calib/mu_c": 0.870962962962963,
      "calib/mu_w": 0.8817076923076922,
      "calib/nonempty_final_conf_rate": 0.68359375,
      "calib/nonempty_reasoning_rate": 0.81640625,
      "calib/nonempty_step_conf_rate": 0.73828125,
      "calib/pce": 0.626697142857143,
      "calib/std_conf": 0.1736974969330771,
      "calib/step_conf_rate": 0.73828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2448.0,
      "completions/max_terminated_length": 2448.0,
      "completions/mean_length": 379.83203125,
      "completions/mean_terminated_length": 382.8228454589844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 13.0,
      "epoch": 0.023466666666666667,
      "grad_norm": 0.2948322594165802,
      "learning_rate": 4.944444444444445e-06,
      "loss": 0.158,
      "num_tokens": 5741721.0,
      "reward": 0.5376076102256775,
      "reward_std": 0.5166065692901611,
      "rewards/accuracy_reward_step": 0.1796875,
      "rewards/final_brier_reward_step": 0.23636791110038757,
      "rewards/format_reward_step": 0.59765625,
      "step": 22
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -4.414528118057426e-09,
      "aux_brier/mean_r": 0.4795295763760805,
      "aux_brier/n_active_tok": 113.125,
      "aux_brier/n_step_records": 28.28125,
      "aux_brier/std_r": 0.2549027865752578,
      "calib/answer_extract_rate": 0.75,
      "calib/auroc": 0.5606639839034205,
      "calib/avg_num_step_conf": 3.62890625,
      "calib/ece": 0.6868875706214687,
      "calib/final_conf_rate": 0.69140625,
      "calib/format_rate": 0.6015625,
      "calib/frac_conf_gt_0.9": 0.7288135593220338,
      "calib/gap": 0.06581519114688117,
      "calib/mean_conf": 0.8846276836158191,
      "calib/mu_c": 0.9374285714285713,
      "calib/mu_w": 0.8716133802816901,
      "calib/nonempty_final_conf_rate": 0.69140625,
      "calib/nonempty_reasoning_rate": 0.859375,
      "calib/nonempty_step_conf_rate": 0.77734375,
      "calib/pce": 0.6868875706214687,
      "calib/std_conf": 0.19784997561141277,
      "calib/step_conf_rate": 0.77734375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2188.0,
      "completions/max_terminated_length": 2188.0,
      "completions/mean_length": 348.171875,
      "completions/mean_terminated_length": 353.69842529296875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 25.0,
      "epoch": 0.024533333333333334,
      "grad_norm": 0.32816872000694275,
      "learning_rate": 4.9166666666666665e-06,
      "loss": 0.0558,
      "num_tokens": 5934789.0,
      "reward": 0.5142276287078857,
      "reward_std": 0.4823266267776489,
      "rewards/accuracy_reward_step": 0.16015625,
      "rewards/final_brier_reward_step": 0.21316054463386536,
      "rewards/format_reward_step": 0.6015625,
      "step": 23
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.0465101713849094e-08,
      "aux_brier/mean_r": 0.4993862584233284,
      "aux_brier/n_active_tok": 137.125,
      "aux_brier/n_step_records": 34.28125,
      "aux_brier/std_r": 0.27149031683802605,
      "calib/answer_extract_rate": 0.7890625,
      "calib/auroc": 0.5002932551319649,
      "calib/avg_num_step_conf": 4.3359375,
      "calib/ece": 0.7444795918367347,
      "calib/final_conf_rate": 0.765625,
      "calib/format_rate": 0.66796875,
      "calib/frac_conf_gt_0.9": 0.6989795918367347,
      "calib/gap": -0.003604692082111338,
      "calib/mean_conf": 0.893357142857143,
      "calib/mu_c": 0.8903225806451613,
      "calib/mu_w": 0.8939272727272727,
      "calib/nonempty_final_conf_rate": 0.765625,
      "calib/nonempty_reasoning_rate": 0.90234375,
      "calib/nonempty_step_conf_rate": 0.81640625,
      "calib/pce": 0.7398367346938776,
      "calib/std_conf": 0.17031371114529104,
      "calib/step_conf_rate": 0.81640625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3009.0,
      "completions/max_terminated_length": 3009.0,
      "completions/mean_length": 349.859375,
      "completions/mean_terminated_length": 351.23138427734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 8.0,
      "epoch": 0.0256,
      "grad_norm": 0.4393552243709564,
      "learning_rate": 4.888888888888889e-06,
      "loss": 0.0603,
      "num_tokens": 6128865.0,
      "reward": 0.5113178491592407,
      "reward_std": 0.4203681945800781,
      "rewards/accuracy_reward_step": 0.12890625,
      "rewards/final_brier_reward_step": 0.1937090903520584,
      "rewards/format_reward_step": 0.66796875,
      "step": 24
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.080888460848085e-08,
      "aux_brier/mean_r": 0.46861199755221605,
      "aux_brier/n_active_tok": 149.625,
      "aux_brier/n_step_records": 37.40625,
      "aux_brier/std_r": 0.24172480730339885,
      "calib/answer_extract_rate": 0.9296875,
      "calib/auroc": 0.5606768400290667,
      "calib/avg_num_step_conf": 4.78125,
      "calib/ece": 0.6441814159292036,
      "calib/final_conf_rate": 0.8828125,
      "calib/format_rate": 0.8203125,
      "calib/frac_conf_gt_0.9": 0.6858407079646017,
      "calib/gap": 0.034937714107754814,
      "calib/mean_conf": 0.8891371681415929,
      "calib/mu_c": 0.9152631578947369,
      "calib/mu_w": 0.8803254437869821,
      "calib/nonempty_final_conf_rate": 0.8828125,
      "calib/nonempty_reasoning_rate": 0.96875,
      "calib/nonempty_step_conf_rate": 0.9140625,
      "calib/pce": 0.6405530973451328,
      "calib/std_conf": 0.17658599584413187,
      "calib/step_conf_rate": 0.9140625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2853.0,
      "completions/max_terminated_length": 2853.0,
      "completions/mean_length": 294.16796875,
      "completions/mean_terminated_length": 295.32159423828125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 42.0,
      "epoch": 0.02666666666666667,
      "grad_norm": 0.24862326681613922,
      "learning_rate": 4.861111111111111e-06,
      "loss": 0.0723,
      "num_tokens": 6307396.0,
      "reward": 0.7160451412200928,
      "reward_std": 0.5090502500534058,
      "rewards/accuracy_reward_step": 0.2265625,
      "rewards/final_brier_reward_step": 0.31730538606643677,
      "rewards/format_reward_step": 0.8203125,
      "step": 25
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.9125422205955598e-08,
      "aux_brier/mean_r": 0.4840311659500003,
      "aux_brier/n_active_tok": 146.25,
      "aux_brier/n_step_records": 36.5625,
      "aux_brier/std_r": 0.2474837265908718,
      "calib/answer_extract_rate": 0.87890625,
      "calib/auroc": 0.4972375690607735,
      "calib/avg_num_step_conf": 4.62109375,
      "calib/ece": 0.7192286363636364,
      "calib/final_conf_rate": 0.859375,
      "calib/format_rate": 0.796875,
      "calib/frac_conf_gt_0.9": 0.7318181818181818,
      "calib/gap": 0.03853496245927179,
      "calib/mean_conf": 0.8965013636363637,
      "calib/mu_c": 0.9282051282051282,
      "calib/mu_w": 0.8896701657458564,
      "calib/nonempty_final_conf_rate": 0.859375,
      "calib/nonempty_reasoning_rate": 0.953125,
      "calib/nonempty_step_conf_rate": 0.9140625,
      "calib/pce": 0.7192286363636364,
      "calib/std_conf": 0.16700126822587183,
      "calib/step_conf_rate": 0.9140625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2855.0,
      "completions/max_terminated_length": 2855.0,
      "completions/mean_length": 304.125,
      "completions/mean_terminated_length": 306.5196838378906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 20.0,
      "epoch": 0.027733333333333332,
      "grad_norm": 0.8835781812667847,
      "learning_rate": 4.833333333333333e-06,
      "loss": 0.0319,
      "num_tokens": 6490492.0,
      "reward": 0.6220261454582214,
      "reward_std": 0.448641836643219,
      "rewards/accuracy_reward_step": 0.16015625,
      "rewards/final_brier_reward_step": 0.2537296414375305,
      "rewards/format_reward_step": 0.796875,
      "step": 26
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.601770781457402e-08,
      "aux_brier/mean_r": 0.5215900270268321,
      "aux_brier/n_active_tok": 160.375,
      "aux_brier/n_step_records": 40.09375,
      "aux_brier/std_r": 0.24954669224098325,
      "calib/answer_extract_rate": 0.91015625,
      "calib/auroc": 0.5441506051098162,
      "calib/avg_num_step_conf": 5.09765625,
      "calib/ece": 0.6770626388888888,
      "calib/final_conf_rate": 0.9375,
      "calib/format_rate": 0.859375,
      "calib/frac_conf_gt_0.9": 0.6625,
      "calib/gap": 0.03545814283579851,
      "calib/mean_conf": 0.8687293055555556,
      "calib/mu_c": 0.897391304347826,
      "calib/mu_w": 0.8619331615120275,
      "calib/nonempty_final_conf_rate": 0.9375,
      "calib/nonempty_reasoning_rate": 0.96875,
      "calib/nonempty_step_conf_rate": 0.9453125,
      "calib/pce": 0.6770626388888888,
      "calib/std_conf": 0.18192718188034782,
      "calib/step_conf_rate": 0.9453125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2099.0,
      "completions/max_terminated_length": 2099.0,
      "completions/mean_length": 300.140625,
      "completions/mean_terminated_length": 300.140625,
      "completions/min_length": 24.0,
      "completions/min_terminated_length": 24.0,
      "epoch": 0.0288,
      "grad_norm": 0.17994718253612518,
      "learning_rate": 4.805555555555556e-06,
      "loss": 0.0655,
      "num_tokens": 6672544.0,
      "reward": 0.6840643286705017,
      "reward_std": 0.42421311140060425,
      "rewards/accuracy_reward_step": 0.1796875,
      "rewards/final_brier_reward_step": 0.29875725507736206,
      "rewards/format_reward_step": 0.859375,
      "step": 27
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.0590220523318834e-08,
      "aux_brier/mean_r": 0.500780388712883,
      "aux_brier/n_active_tok": 147.0,
      "aux_brier/n_step_records": 36.75,
      "aux_brier/std_r": 0.24309765687212348,
      "calib/answer_extract_rate": 0.9375,
      "calib/auroc": 0.4769244884851821,
      "calib/avg_num_step_conf": 4.61328125,
      "calib/ece": 0.6230042372881356,
      "calib/final_conf_rate": 0.921875,
      "calib/format_rate": 0.8671875,
      "calib/frac_conf_gt_0.9": 0.6652542372881356,
      "calib/gap": -0.005542435085787667,
      "calib/mean_conf": 0.8834279661016948,
      "calib/mu_c": 0.8793650793650793,
      "calib/mu_w": 0.884907514450867,
      "calib/nonempty_final_conf_rate": 0.921875,
      "calib/nonempty_reasoning_rate": 0.96875,
      "calib/nonempty_step_conf_rate": 0.91796875,
      "calib/pce": 0.6197415254237287,
      "calib/std_conf": 0.16116088097575107,
      "calib/step_conf_rate": 0.91796875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2786.0,
      "completions/max_terminated_length": 2786.0,
      "completions/mean_length": 290.82421875,
      "completions/mean_terminated_length": 290.82421875,
      "completions/min_length": 25.0,
      "completions/min_terminated_length": 25.0,
      "epoch": 0.029866666666666666,
      "grad_norm": 1.0136277675628662,
      "learning_rate": 4.777777777777778e-06,
      "loss": 0.064,
      "num_tokens": 6853939.0,
      "reward": 0.771499752998352,
      "reward_std": 0.4849082827568054,
      "rewards/accuracy_reward_step": 0.25,
      "rewards/final_brier_reward_step": 0.35162419080734253,
      "rewards/format_reward_step": 0.8671875,
      "step": 28
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.683060218971022e-09,
      "aux_brier/mean_r": 0.4772613560780883,
      "aux_brier/n_active_tok": 159.0,
      "aux_brier/n_step_records": 39.75,
      "aux_brier/std_r": 0.24786118045449257,
      "calib/answer_extract_rate": 0.9453125,
      "calib/auroc": 0.4210186513629842,
      "calib/avg_num_step_conf": 5.03515625,
      "calib/ece": 0.7571507573221756,
      "calib/final_conf_rate": 0.93359375,
      "calib/format_rate": 0.875,
      "calib/frac_conf_gt_0.9": 0.7364016736401674,
      "calib/gap": -0.0068561053084648105,
      "calib/mean_conf": 0.8994101715481172,
      "calib/mu_c": 0.8935294117647059,
      "calib/mu_w": 0.9003855170731707,
      "calib/nonempty_final_conf_rate": 0.93359375,
      "calib/nonempty_reasoning_rate": 0.98828125,
      "calib/nonempty_step_conf_rate": 0.953125,
      "calib/pce": 0.7571507573221756,
      "calib/std_conf": 0.15954339212531934,
      "calib/step_conf_rate": 0.953125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2037.0,
      "completions/max_terminated_length": 2037.0,
      "completions/mean_length": 298.56640625,
      "completions/mean_terminated_length": 298.56640625,
      "completions/min_length": 28.0,
      "completions/min_terminated_length": 28.0,
      "epoch": 0.030933333333333334,
      "grad_norm": 0.10229931026697159,
      "learning_rate": 4.75e-06,
      "loss": 0.0563,
      "num_tokens": 7037500.0,
      "reward": 0.6297242045402527,
      "reward_std": 0.36027663946151733,
      "rewards/accuracy_reward_step": 0.1328125,
      "rewards/final_brier_reward_step": 0.23764701187610626,
      "rewards/format_reward_step": 0.875,
      "step": 29
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.570245974387042e-08,
      "aux_brier/mean_r": 0.49467434268444777,
      "aux_brier/n_active_tok": 157.5,
      "aux_brier/n_step_records": 39.375,
      "aux_brier/std_r": 0.25863818172365427,
      "calib/answer_extract_rate": 0.953125,
      "calib/auroc": 0.5382694541231127,
      "calib/avg_num_step_conf": 4.98828125,
      "calib/ece": 0.7196153846153844,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.9140625,
      "calib/frac_conf_gt_0.9": 0.7004048582995951,
      "calib/gap": 0.03159233449477361,
      "calib/mean_conf": 0.889493927125506,
      "calib/mu_c": 0.9157142857142857,
      "calib/mu_w": 0.8841219512195121,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 0.984375,
      "calib/nonempty_step_conf_rate": 0.95703125,
      "calib/pce": 0.7195344129554654,
      "calib/std_conf": 0.17370995581297372,
      "calib/step_conf_rate": 0.95703125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2564.0,
      "completions/max_terminated_length": 2564.0,
      "completions/mean_length": 310.22265625,
      "completions/mean_terminated_length": 310.22265625,
      "completions/min_length": 25.0,
      "completions/min_terminated_length": 25.0,
      "epoch": 0.032,
      "grad_norm": 0.9310511350631714,
      "learning_rate": 4.722222222222222e-06,
      "loss": 0.0003,
      "num_tokens": 7223901.0,
      "reward": 0.695044755935669,
      "reward_std": 0.37188243865966797,
      "rewards/accuracy_reward_step": 0.1640625,
      "rewards/final_brier_reward_step": 0.2958042025566101,
      "rewards/format_reward_step": 0.9140625,
      "step": 30
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.1022978224639601e-09,
      "aux_brier/mean_r": 0.4644114878028631,
      "aux_brier/n_active_tok": 163.25,
      "aux_brier/n_step_records": 40.8125,
      "aux_brier/std_r": 0.24233023542910814,
      "calib/answer_extract_rate": 0.93359375,
      "calib/auroc": 0.49856701940035275,
      "calib/avg_num_step_conf": 5.10546875,
      "calib/ece": 0.7009071729957806,
      "calib/final_conf_rate": 0.92578125,
      "calib/format_rate": 0.88671875,
      "calib/frac_conf_gt_0.9": 0.729957805907173,
      "calib/gap": 0.03461309523809519,
      "calib/mean_conf": 0.9034388185654009,
      "calib/mu_c": 0.9310416666666667,
      "calib/mu_w": 0.8964285714285715,
      "calib/nonempty_final_conf_rate": 0.92578125,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.96875,
      "calib/pce": 0.7009071729957806,
      "calib/std_conf": 0.14707015408666976,
      "calib/step_conf_rate": 0.96875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2278.0,
      "completions/max_terminated_length": 2278.0,
      "completions/mean_length": 301.1171875,
      "completions/mean_terminated_length": 301.1171875,
      "completions/min_length": 74.0,
      "completions/min_terminated_length": 74.0,
      "epoch": 0.03306666666666667,
      "grad_norm": 0.325433611869812,
      "learning_rate": 4.694444444444445e-06,
      "loss": 0.038,
      "num_tokens": 7406899.0,
      "reward": 0.7081201076507568,
      "reward_std": 0.3796444237232208,
      "rewards/accuracy_reward_step": 0.19140625,
      "rewards/final_brier_reward_step": 0.29341787099838257,
      "rewards/format_reward_step": 0.88671875,
      "step": 31
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.8724363379241105e-09,
      "aux_brier/mean_r": 0.5441410019993782,
      "aux_brier/n_active_tok": 166.875,
      "aux_brier/n_step_records": 41.71875,
      "aux_brier/std_r": 0.253931765910238,
      "calib/answer_extract_rate": 0.94921875,
      "calib/auroc": 0.44706806282722505,
      "calib/avg_num_step_conf": 5.30859375,
      "calib/ece": 0.6712033195020747,
      "calib/final_conf_rate": 0.94140625,
      "calib/format_rate": 0.9296875,
      "calib/frac_conf_gt_0.9": 0.6224066390041494,
      "calib/gap": 0.008658638743455405,
      "calib/mean_conf": 0.8749377593360996,
      "calib/mu_c": 0.8817999999999999,
      "calib/mu_w": 0.8731413612565445,
      "calib/nonempty_final_conf_rate": 0.94140625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.6693360995850622,
      "calib/std_conf": 0.1636642660348378,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1950.0,
      "completions/max_terminated_length": 1950.0,
      "completions/mean_length": 300.421875,
      "completions/mean_terminated_length": 301.6000061035156,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 75.0,
      "epoch": 0.034133333333333335,
      "grad_norm": 0.19576984643936157,
      "learning_rate": 4.666666666666667e-06,
      "loss": 0.0616,
      "num_tokens": 7590511.0,
      "reward": 0.7490607500076294,
      "reward_std": 0.3445599675178528,
      "rewards/accuracy_reward_step": 0.19921875,
      "rewards/final_brier_reward_step": 0.3399929702281952,
      "rewards/format_reward_step": 0.9296875,
      "step": 32
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.4268494991375178e-08,
      "aux_brier/mean_r": 0.5107473731040955,
      "aux_brier/n_active_tok": 161.625,
      "aux_brier/n_step_records": 40.40625,
      "aux_brier/std_r": 0.2583832503296435,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.5713531894934334,
      "calib/avg_num_step_conf": 5.10546875,
      "calib/ece": 0.6868281124497991,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.9296875,
      "calib/frac_conf_gt_0.9": 0.6104417670682731,
      "calib/gap": 0.0499005393996248,
      "calib/mean_conf": 0.8514867469879518,
      "calib/mu_c": 0.893170731707317,
      "calib/mu_w": 0.8432701923076922,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.97265625,
      "calib/pce": 0.6868281124497991,
      "calib/std_conf": 0.20701952303279791,
      "calib/step_conf_rate": 0.97265625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2055.0,
      "completions/max_terminated_length": 2055.0,
      "completions/mean_length": 290.37109375,
      "completions/mean_terminated_length": 291.50982666015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 53.0,
      "epoch": 0.0352,
      "grad_norm": 0.48981305956840515,
      "learning_rate": 4.638888888888889e-06,
      "loss": 0.0255,
      "num_tokens": 7771718.0,
      "reward": 0.7095664739608765,
      "reward_std": 0.30987101793289185,
      "rewards/accuracy_reward_step": 0.16015625,
      "rewards/final_brier_reward_step": 0.3382660150527954,
      "rewards/format_reward_step": 0.9296875,
      "step": 33
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.904765679379029e-08,
      "aux_brier/mean_r": 0.583881065249443,
      "aux_brier/n_active_tok": 164.375,
      "aux_brier/n_step_records": 41.09375,
      "aux_brier/std_r": 0.25083809066563845,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5058752775721689,
      "calib/avg_num_step_conf": 5.15625,
      "calib/ece": 0.6248995983935743,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.5140562248995983,
      "calib/gap": 0.0061278682457437395,
      "calib/mean_conf": 0.8393574297188755,
      "calib/mu_c": 0.8441071428571428,
      "calib/mu_w": 0.8379792746113991,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.6196787148594378,
      "calib/std_conf": 0.18964665081059243,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1984.0,
      "completions/max_terminated_length": 1984.0,
      "completions/mean_length": 267.8203125,
      "completions/mean_terminated_length": 267.8203125,
      "completions/min_length": 51.0,
      "completions/min_terminated_length": 51.0,
      "epoch": 0.03626666666666667,
      "grad_norm": 0.8077874183654785,
      "learning_rate": 4.611111111111112e-06,
      "loss": 0.0529,
      "num_tokens": 7945392.0,
      "reward": 0.8077145218849182,
      "reward_std": 0.39487701654434204,
      "rewards/accuracy_reward_step": 0.22265625,
      "rewards/final_brier_reward_step": 0.4027332067489624,
      "rewards/format_reward_step": 0.96875,
      "step": 34
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.6753998905171272e-08,
      "aux_brier/mean_r": 0.5733567727729678,
      "aux_brier/n_active_tok": 180.125,
      "aux_brier/n_step_records": 45.03125,
      "aux_brier/std_r": 0.2474320181645453,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5484252900781434,
      "calib/avg_num_step_conf": 5.68359375,
      "calib/ece": 0.6853036437246964,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.94921875,
      "calib/frac_conf_gt_0.9": 0.5587044534412956,
      "calib/gap": 0.045822874733601804,
      "calib/mean_conf": 0.8512955465587045,
      "calib/mu_c": 0.8895121951219513,
      "calib/mu_w": 0.8436893203883495,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.6853036437246964,
      "calib/std_conf": 0.1877760901197293,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2083.0,
      "completions/max_terminated_length": 2083.0,
      "completions/mean_length": 306.234375,
      "completions/mean_terminated_length": 306.234375,
      "completions/min_length": 71.0,
      "completions/min_terminated_length": 71.0,
      "epoch": 0.037333333333333336,
      "grad_norm": 1.0027748346328735,
      "learning_rate": 4.583333333333333e-06,
      "loss": 0.0344,
      "num_tokens": 8133044.0,
      "reward": 0.7308730483055115,
      "reward_std": 0.3583698272705078,
      "rewards/accuracy_reward_step": 0.16796875,
      "rewards/final_brier_reward_step": 0.3531796932220459,
      "rewards/format_reward_step": 0.94921875,
      "step": 35
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 9.42846162255151e-09,
      "aux_brier/mean_r": 0.6210600854828954,
      "aux_brier/n_active_tok": 170.125,
      "aux_brier/n_step_records": 42.53125,
      "aux_brier/std_r": 0.24454850843176246,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.4746660859465738,
      "calib/avg_num_step_conf": 5.34765625,
      "calib/ece": 0.47052000000000005,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.396,
      "calib/gap": -0.006585365853658431,
      "calib/mean_conf": 0.7828399999999999,
      "calib/mu_c": 0.7784146341463415,
      "calib/mu_w": 0.7849999999999999,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.46268000000000004,
      "calib/std_conf": 0.21339244222792897,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1896.0,
      "completions/max_terminated_length": 1896.0,
      "completions/mean_length": 293.53515625,
      "completions/mean_terminated_length": 293.53515625,
      "completions/min_length": 3.0,
      "completions/min_terminated_length": 3.0,
      "epoch": 0.0384,
      "grad_norm": 0.18656954169273376,
      "learning_rate": 4.555555555555556e-06,
      "loss": 0.0579,
      "num_tokens": 8310901.0,
      "reward": 0.9289214611053467,
      "reward_std": 0.4526945948600769,
      "rewards/accuracy_reward_step": 0.3203125,
      "rewards/final_brier_reward_step": 0.5047484636306763,
      "rewards/format_reward_step": 0.96484375,
      "step": 36
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.624792260624842e-08,
      "aux_brier/mean_r": 0.6186469784006476,
      "aux_brier/n_active_tok": 161.125,
      "aux_brier/n_step_records": 40.28125,
      "aux_brier/std_r": 0.2561876643449068,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.5631727107136943,
      "calib/avg_num_step_conf": 5.05078125,
      "calib/ece": 0.5173092369477912,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.953125,
      "calib/frac_conf_gt_0.9": 0.39357429718875503,
      "calib/gap": 0.07312220566318905,
      "calib/mean_conf": 0.7655020080321285,
      "calib/mu_c": 0.819242424242424,
      "calib/mu_w": 0.7461202185792349,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.984375,
      "calib/pce": 0.5088755020080322,
      "calib/std_conf": 0.24703187646118807,
      "calib/step_conf_rate": 0.984375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1983.0,
      "completions/max_terminated_length": 1983.0,
      "completions/mean_length": 280.95703125,
      "completions/mean_terminated_length": 280.95703125,
      "completions/min_length": 37.0,
      "completions/min_terminated_length": 37.0,
      "epoch": 0.039466666666666664,
      "grad_norm": 0.3276523947715759,
      "learning_rate": 4.527777777777778e-06,
      "loss": 0.0233,
      "num_tokens": 8489922.0,
      "reward": 0.8597831726074219,
      "reward_std": 0.3994288146495819,
      "rewards/accuracy_reward_step": 0.2578125,
      "rewards/final_brier_reward_step": 0.501632809638977,
      "rewards/format_reward_step": 0.953125,
      "step": 37
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.8400494062340584e-08,
      "aux_brier/mean_r": 0.6910385163500905,
      "aux_brier/n_active_tok": 173.625,
      "aux_brier/n_step_records": 43.40625,
      "aux_brier/std_r": 0.235653187148273,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5293839758125471,
      "calib/avg_num_step_conf": 5.5,
      "calib/ece": 0.526958,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.284,
      "calib/gap": 0.020413170823885163,
      "calib/mean_conf": 0.7199220000000001,
      "calib/mu_c": 0.735925925925926,
      "calib/mu_w": 0.7155127551020408,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.51544,
      "calib/std_conf": 0.25425096836787076,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1329.0,
      "completions/max_terminated_length": 1329.0,
      "completions/mean_length": 284.35546875,
      "completions/mean_terminated_length": 284.35546875,
      "completions/min_length": 22.0,
      "completions/min_terminated_length": 22.0,
      "epoch": 0.04053333333333333,
      "grad_norm": 0.22487637400627136,
      "learning_rate": 4.5e-06,
      "loss": 0.0591,
      "num_tokens": 8669605.0,
      "reward": 0.8187369108200073,
      "reward_std": 0.3844972848892212,
      "rewards/accuracy_reward_step": 0.2109375,
      "rewards/final_brier_reward_step": 0.5015101432800293,
      "rewards/format_reward_step": 0.96484375,
      "step": 38
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.716412101337994e-08,
      "aux_brier/mean_r": 0.7235443480312824,
      "aux_brier/n_active_tok": 172.875,
      "aux_brier/n_step_records": 43.21875,
      "aux_brier/std_r": 0.2324718649033457,
      "calib/answer_extract_rate": 0.96484375,
      "calib/auroc": 0.5164703874381293,
      "calib/avg_num_step_conf": 5.44921875,
      "calib/ece": 0.45558232931726916,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.9453125,
      "calib/frac_conf_gt_0.9": 0.285140562248996,
      "calib/gap": 0.03605222734254998,
      "calib/mean_conf": 0.7044979919678714,
      "calib/mu_c": 0.7314285714285713,
      "calib/mu_w": 0.6953763440860213,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.4535341365461848,
      "calib/std_conf": 0.25052044020363534,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1514.0,
      "completions/max_terminated_length": 1514.0,
      "completions/mean_length": 287.46484375,
      "completions/mean_terminated_length": 288.5921630859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 84.0,
      "epoch": 0.0416,
      "grad_norm": 0.10107416659593582,
      "learning_rate": 4.472222222222223e-06,
      "loss": -0.001,
      "num_tokens": 8849284.0,
      "reward": 0.8515971899032593,
      "reward_std": 0.3746635913848877,
      "rewards/accuracy_reward_step": 0.24609375,
      "rewards/final_brier_reward_step": 0.5313886404037476,
      "rewards/format_reward_step": 0.9453125,
      "step": 39
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.759384257200459e-08,
      "aux_brier/mean_r": 0.7821152247488499,
      "aux_brier/n_active_tok": 168.25,
      "aux_brier/n_step_records": 42.0625,
      "aux_brier/std_r": 0.19819560437463224,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.47432950191570883,
      "calib/avg_num_step_conf": 5.3046875,
      "calib/ece": 0.4702016129032258,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.1774193548387097,
      "calib/gap": -0.018691844553913484,
      "calib/mean_conf": 0.6364112903225807,
      "calib/mu_c": 0.6211111111111112,
      "calib/mu_w": 0.6398029556650247,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.4625806451612904,
      "calib/std_conf": 0.26290181265699764,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1478.0,
      "completions/max_terminated_length": 1478.0,
      "completions/mean_length": 276.796875,
      "completions/mean_terminated_length": 276.796875,
      "completions/min_length": 62.0,
      "completions/min_terminated_length": 62.0,
      "epoch": 0.042666666666666665,
      "grad_norm": 0.2239619940519333,
      "learning_rate": 4.444444444444444e-06,
      "loss": -0.006,
      "num_tokens": 9026904.0,
      "reward": 0.792715311050415,
      "reward_std": 0.3543362021446228,
      "rewards/accuracy_reward_step": 0.17578125,
      "rewards/final_brier_reward_step": 0.5458613634109497,
      "rewards/format_reward_step": 0.9609375,
      "step": 40
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.5530835445763955e-08,
      "aux_brier/mean_r": 0.796193016692996,
      "aux_brier/n_active_tok": 164.75,
      "aux_brier/n_step_records": 41.1875,
      "aux_brier/std_r": 0.18959426786750555,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.5042929793276534,
      "calib/avg_num_step_conf": 5.16796875,
      "calib/ece": 0.27724000000000004,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.164,
      "calib/gap": 0.0055921009180369685,
      "calib/mean_conf": 0.5824400000000001,
      "calib/mu_c": 0.5857281553398057,
      "calib/mu_w": 0.5801360544217687,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.22384000000000004,
      "calib/std_conf": 0.2798400371640913,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2401.0,
      "completions/max_terminated_length": 2401.0,
      "completions/mean_length": 297.73828125,
      "completions/mean_terminated_length": 297.73828125,
      "completions/min_length": 28.0,
      "completions/min_terminated_length": 28.0,
      "epoch": 0.04373333333333333,
      "grad_norm": 0.06125990301370621,
      "learning_rate": 4.416666666666667e-06,
      "loss": 0.0813,
      "num_tokens": 9210373.0,
      "reward": 1.0402311086654663,
      "reward_std": 0.43915098905563354,
      "rewards/accuracy_reward_step": 0.40234375,
      "rewards/final_brier_reward_step": 0.6296742558479309,
      "rewards/format_reward_step": 0.9609375,
      "step": 41
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -5.5605052118323783e-08,
      "aux_brier/mean_r": 0.8279183488339186,
      "aux_brier/n_active_tok": 162.875,
      "aux_brier/n_step_records": 40.71875,
      "aux_brier/std_r": 0.18691047129686922,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5667389906782384,
      "calib/avg_num_step_conf": 5.1171875,
      "calib/ece": 0.2761075697211155,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.099601593625498,
      "calib/gap": 0.06970475731276105,
      "calib/mean_conf": 0.5225617529880479,
      "calib/mu_c": 0.5733823529411763,
      "calib/mu_w": 0.5036775956284153,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.26387649402390434,
      "calib/std_conf": 0.28826297887753355,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1736.0,
      "completions/max_terminated_length": 1736.0,
      "completions/mean_length": 245.7890625,
      "completions/mean_terminated_length": 245.7890625,
      "completions/min_length": 59.0,
      "completions/min_terminated_length": 59.0,
      "epoch": 0.0448,
      "grad_norm": 0.2336869090795517,
      "learning_rate": 4.388888888888889e-06,
      "loss": -0.0032,
      "num_tokens": 9377663.0,
      "reward": 0.920749306678772,
      "reward_std": 0.3561636507511139,
      "rewards/accuracy_reward_step": 0.26953125,
      "rewards/final_brier_reward_step": 0.6673721075057983,
      "rewards/format_reward_step": 0.96875,
      "step": 42
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.07011273342561e-07,
      "aux_brier/mean_r": 0.8887287676334381,
      "aux_brier/n_active_tok": 169.75,
      "aux_brier/n_step_records": 42.4375,
      "aux_brier/std_r": 0.13937835465185344,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5863016050124453,
      "calib/avg_num_step_conf": 5.3203125,
      "calib/ece": 0.25392857142857145,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0873015873015873,
      "calib/gap": 0.07789631791262563,
      "calib/mean_conf": 0.469484126984127,
      "calib/mu_c": 0.5285245901639345,
      "calib/mu_w": 0.45062827225130886,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.24067460317460315,
      "calib/std_conf": 0.27795049902235747,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2016.0,
      "completions/max_terminated_length": 2016.0,
      "completions/mean_length": 295.72265625,
      "completions/mean_terminated_length": 295.72265625,
      "completions/min_length": 61.0,
      "completions/min_terminated_length": 61.0,
      "epoch": 0.04586666666666667,
      "grad_norm": 0.09244558215141296,
      "learning_rate": 4.361111111111112e-06,
      "loss": 0.0354,
      "num_tokens": 9558592.0,
      "reward": 0.9044816493988037,
      "reward_std": 0.31629693508148193,
      "rewards/accuracy_reward_step": 0.23828125,
      "rewards/final_brier_reward_step": 0.7038640975952148,
      "rewards/format_reward_step": 0.98046875,
      "step": 43
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.058977593977218e-08,
      "aux_brier/mean_r": 0.9013891965150833,
      "aux_brier/n_active_tok": 182.625,
      "aux_brier/n_step_records": 45.65625,
      "aux_brier/std_r": 0.1366663781227544,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.44288488700564965,
      "calib/avg_num_step_conf": 5.71875,
      "calib/ece": 0.2836653386454183,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.07569721115537849,
      "calib/gap": -0.057367584745762756,
      "calib/mean_conf": 0.42557768924302786,
      "calib/mu_c": 0.3816949152542373,
      "calib/mu_w": 0.4390625000000001,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.23709163346613543,
      "calib/std_conf": 0.2705343041121279,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1592.0,
      "completions/max_terminated_length": 1592.0,
      "completions/mean_length": 309.3046875,
      "completions/mean_terminated_length": 309.3046875,
      "completions/min_length": 63.0,
      "completions/min_terminated_length": 63.0,
      "epoch": 0.046933333333333334,
      "grad_norm": 0.026423679664731026,
      "learning_rate": 4.333333333333334e-06,
      "loss": 0.0495,
      "num_tokens": 9744094.0,
      "reward": 0.8851389288902283,
      "reward_std": 0.2901378273963928,
      "rewards/accuracy_reward_step": 0.23046875,
      "rewards/final_brier_reward_step": 0.6733683347702026,
      "rewards/format_reward_step": 0.97265625,
      "step": 44
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.3169754138619982e-07,
      "aux_brier/mean_r": 0.9033907111734152,
      "aux_brier/n_active_tok": 167.875,
      "aux_brier/n_step_records": 41.96875,
      "aux_brier/std_r": 0.13565030810423195,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5077584148961566,
      "calib/avg_num_step_conf": 5.24609375,
      "calib/ece": 0.20056451612903223,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.953125,
      "calib/frac_conf_gt_0.9": 0.016129032258064516,
      "calib/gap": -0.0029776398504018364,
      "calib/mean_conf": 0.3462096774193549,
      "calib/mu_c": 0.3440845070422536,
      "calib/mu_w": 0.3470621468926554,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.13024193548387097,
      "calib/std_conf": 0.25285337055914786,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2129.0,
      "completions/max_terminated_length": 2129.0,
      "completions/mean_length": 311.640625,
      "completions/mean_terminated_length": 311.640625,
      "completions/min_length": 39.0,
      "completions/min_terminated_length": 39.0,
      "epoch": 0.048,
      "grad_norm": 0.056766387075185776,
      "learning_rate": 4.305555555555556e-06,
      "loss": 0.0236,
      "num_tokens": 9928922.0,
      "reward": 0.9266473054885864,
      "reward_std": 0.39333993196487427,
      "rewards/accuracy_reward_step": 0.27734375,
      "rewards/final_brier_reward_step": 0.6909644603729248,
      "rewards/format_reward_step": 0.953125,
      "step": 45
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.73603463794847e-07,
      "aux_brier/mean_r": 0.9203233141452074,
      "aux_brier/n_active_tok": 184.625,
      "aux_brier/n_step_records": 46.15625,
      "aux_brier/std_r": 0.1193378433235921,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.43169398907103823,
      "calib/avg_num_step_conf": 5.7890625,
      "calib/ece": 0.24682142857142858,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.047619047619047616,
      "calib/gap": -0.06515633167023044,
      "calib/mean_conf": 0.33992460317460316,
      "calib/mu_c": 0.29260869565217396,
      "calib/mu_w": 0.3577650273224044,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.15646825396825398,
      "calib/std_conf": 0.25365768371132563,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1465.0,
      "completions/max_terminated_length": 1465.0,
      "completions/mean_length": 312.875,
      "completions/mean_terminated_length": 312.875,
      "completions/min_length": 91.0,
      "completions/min_terminated_length": 91.0,
      "epoch": 0.04906666666666667,
      "grad_norm": 0.23408041894435883,
      "learning_rate": 4.277777777777778e-06,
      "loss": 0.0456,
      "num_tokens": 10113786.0,
      "reward": 0.9338980913162231,
      "reward_std": 0.30101126432418823,
      "rewards/accuracy_reward_step": 0.2734375,
      "rewards/final_brier_reward_step": 0.6887174844741821,
      "rewards/format_reward_step": 0.9765625,
      "step": 46
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.192369668561737e-07,
      "aux_brier/mean_r": 0.9377381801605225,
      "aux_brier/n_active_tok": 190.5,
      "aux_brier/n_step_records": 47.625,
      "aux_brier/std_r": 0.0988645100442227,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5192012908430819,
      "calib/avg_num_step_conf": 5.984375,
      "calib/ece": 0.1789285714285714,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.03968253968253968,
      "calib/gap": 0.014729326341266669,
      "calib/mean_conf": 0.27575396825396825,
      "calib/mu_c": 0.2865671641791045,
      "calib/mu_w": 0.2718378378378378,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.09440476190476191,
      "calib/std_conf": 0.23050975899555634,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1042.0,
      "completions/max_terminated_length": 1042.0,
      "completions/mean_length": 320.44140625,
      "completions/mean_terminated_length": 321.69805908203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 103.0,
      "epoch": 0.050133333333333335,
      "grad_norm": 0.15534737706184387,
      "learning_rate": 4.25e-06,
      "loss": -0.028,
      "num_tokens": 10301795.0,
      "reward": 0.9412600994110107,
      "reward_std": 0.3168777823448181,
      "rewards/accuracy_reward_step": 0.265625,
      "rewards/final_brier_reward_step": 0.7416031360626221,
      "rewards/format_reward_step": 0.98046875,
      "step": 47
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -4.0683897054027973e-07,
      "aux_brier/mean_r": 0.9527103621512651,
      "aux_brier/n_active_tok": 174.75,
      "aux_brier/n_step_records": 43.6875,
      "aux_brier/std_r": 0.07518987158255186,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.4866093366093366,
      "calib/avg_num_step_conf": 5.51171875,
      "calib/ece": 0.15785099601593625,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.00398406374501992,
      "calib/gap": 0.0015961998361998797,
      "calib/mean_conf": 0.22230836653386454,
      "calib/mu_c": 0.2234848484848485,
      "calib/mu_w": 0.22188864864864863,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.05860557768924303,
      "calib/std_conf": 0.19427346410229035,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2620.0,
      "completions/max_terminated_length": 2620.0,
      "completions/mean_length": 295.82421875,
      "completions/mean_terminated_length": 295.82421875,
      "completions/min_length": 63.0,
      "completions/min_terminated_length": 63.0,
      "epoch": 0.0512,
      "grad_norm": 0.02304792031645775,
      "learning_rate": 4.222222222222223e-06,
      "loss": 0.0558,
      "num_tokens": 10481214.0,
      "reward": 0.937140703201294,
      "reward_std": 0.2706897556781769,
      "rewards/accuracy_reward_step": 0.26171875,
      "rewards/final_brier_reward_step": 0.7485626935958862,
      "rewards/format_reward_step": 0.9765625,
      "step": 48
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.166361513262661e-07,
      "aux_brier/mean_r": 0.946187186986208,
      "aux_brier/n_active_tok": 196.25,
      "aux_brier/n_step_records": 49.0625,
      "aux_brier/std_r": 0.08926301234168932,
      "calib/answer_extract_rate": 0.96484375,
      "calib/auroc": 0.5330502476999293,
      "calib/avg_num_step_conf": 6.29296875,
      "calib/ece": 0.21672064777327932,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.008097165991902834,
      "calib/gap": 0.030697806086341095,
      "calib/mean_conf": 0.21137651821862352,
      "calib/mu_c": 0.2308888888888889,
      "calib/mu_w": 0.2001910828025478,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.03186234817813765,
      "calib/std_conf": 0.19012601812598928,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2560.0,
      "completions/max_terminated_length": 2560.0,
      "completions/mean_length": 336.3984375,
      "completions/mean_terminated_length": 339.0472412109375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 82.0,
      "epoch": 0.05226666666666667,
      "grad_norm": 0.11247003078460693,
      "learning_rate": 4.194444444444445e-06,
      "loss": 0.0271,
      "num_tokens": 10671868.0,
      "reward": 1.0026288032531738,
      "reward_std": 0.32875940203666687,
      "rewards/accuracy_reward_step": 0.3515625,
      "rewards/final_brier_reward_step": 0.6902027130126953,
      "rewards/format_reward_step": 0.95703125,
      "step": 49
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 9.116142907217295e-08,
      "aux_brier/mean_r": 0.9486544150859118,
      "aux_brier/n_active_tok": 230.875,
      "aux_brier/n_step_records": 57.71875,
      "aux_brier/std_r": 0.09179092763224617,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.4349324145242512,
      "calib/avg_num_step_conf": 7.234375,
      "calib/ece": 0.30401190476190476,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.011904761904761904,
      "calib/gap": -0.06128107606679034,
      "calib/mean_conf": 0.2044801587301587,
      "calib/mu_c": 0.16703061224489796,
      "calib/mu_w": 0.2283116883116883,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05980158730158729,
      "calib/std_conf": 0.2061868749074533,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2431.0,
      "completions/max_terminated_length": 2431.0,
      "completions/mean_length": 430.69921875,
      "completions/mean_terminated_length": 430.69921875,
      "completions/min_length": 109.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.05333333333333334,
      "grad_norm": 0.02688698098063469,
      "learning_rate": 4.166666666666667e-06,
      "loss": 0.1138,
      "num_tokens": 10887487.0,
      "reward": 1.0308661460876465,
      "reward_std": 0.3271676301956177,
      "rewards/accuracy_reward_step": 0.3828125,
      "rewards/final_brier_reward_step": 0.6390897035598755,
      "rewards/format_reward_step": 0.9765625,
      "step": 50
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.7028624504922263e-07,
      "aux_brier/mean_r": 0.961110221222043,
      "aux_brier/n_active_tok": 204.625,
      "aux_brier/n_step_records": 51.15625,
      "aux_brier/std_r": 0.07517423389072064,
      "calib/answer_extract_rate": 0.9609375,
      "calib/auroc": 0.5988922841864017,
      "calib/avg_num_step_conf": 6.45703125,
      "calib/ece": 0.1960263157894737,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.016194331983805668,
      "calib/gap": 0.02218292589763174,
      "calib/mean_conf": 0.17239473684210524,
      "calib/mu_c": 0.18766233766233764,
      "calib/mu_w": 0.1654794117647059,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.028340080971659923,
      "calib/std_conf": 0.17546301689012933,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2552.0,
      "completions/max_terminated_length": 2552.0,
      "completions/mean_length": 376.6953125,
      "completions/mean_terminated_length": 379.6614074707031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 5.0,
      "epoch": 0.0544,
      "grad_norm": 0.17208071053028107,
      "learning_rate": 4.138888888888889e-06,
      "loss": 0.0557,
      "num_tokens": 11093217.0,
      "reward": 0.9575573801994324,
      "reward_std": 0.2885648310184479,
      "rewards/accuracy_reward_step": 0.30078125,
      "rewards/final_brier_reward_step": 0.7130419015884399,
      "rewards/format_reward_step": 0.95703125,
      "step": 51
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -5.3009054812314105e-08,
      "aux_brier/mean_r": 0.9536644276231527,
      "aux_brier/n_active_tok": 182.625,
      "aux_brier/n_step_records": 45.65625,
      "aux_brier/std_r": 0.09595581494795624,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5271350306155334,
      "calib/avg_num_step_conf": 5.73046875,
      "calib/ece": 0.31847222222222205,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.005066387367064162,
      "calib/mean_conf": 0.13049603174603178,
      "calib/mu_c": 0.13341121495327105,
      "calib/mu_w": 0.12834482758620688,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.012182539682539682,
      "calib/std_conf": 0.1386136430794062,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2326.0,
      "completions/max_terminated_length": 2326.0,
      "completions/mean_length": 358.49609375,
      "completions/mean_terminated_length": 358.49609375,
      "completions/min_length": 87.0,
      "completions/min_terminated_length": 87.0,
      "epoch": 0.055466666666666664,
      "grad_norm": 0.20132383704185486,
      "learning_rate": 4.111111111111111e-06,
      "loss": 0.0894,
      "num_tokens": 11292944.0,
      "reward": 1.0649170875549316,
      "reward_std": 0.3719336986541748,
      "rewards/accuracy_reward_step": 0.41796875,
      "rewards/final_brier_reward_step": 0.6346686482429504,
      "rewards/format_reward_step": 0.9765625,
      "step": 52
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 9.958885474059542e-07,
      "aux_brier/mean_r": 0.9550592508167028,
      "aux_brier/n_active_tok": 199.75,
      "aux_brier/n_step_records": 49.9375,
      "aux_brier/std_r": 0.09451603806519415,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.505686485120364,
      "calib/avg_num_step_conf": 6.30078125,
      "calib/ece": 0.3805555555555554,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.007936507936507936,
      "calib/gap": -0.020843874391862,
      "calib/mean_conf": 0.14412698412698416,
      "calib/mu_c": 0.1331260504201681,
      "calib/mu_w": 0.1539699248120301,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.02623015873015873,
      "calib/std_conf": 0.15765589332602928,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3004.0,
      "completions/max_terminated_length": 3004.0,
      "completions/mean_length": 398.421875,
      "completions/mean_terminated_length": 398.421875,
      "completions/min_length": 97.0,
      "completions/min_terminated_length": 97.0,
      "epoch": 0.05653333333333333,
      "grad_norm": 0.1364300400018692,
      "learning_rate": 4.083333333333334e-06,
      "loss": 0.031,
      "num_tokens": 11500764.0,
      "reward": 1.101509928703308,
      "reward_std": 0.29608970880508423,
      "rewards/accuracy_reward_step": 0.46484375,
      "rewards/final_brier_reward_step": 0.5935396552085876,
      "rewards/format_reward_step": 0.9765625,
      "step": 53
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.238592622967218e-07,
      "aux_brier/mean_r": 0.960156099870801,
      "aux_brier/n_active_tok": 184.75,
      "aux_brier/n_step_records": 46.1875,
      "aux_brier/std_r": 0.08469618789968081,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.4676079734219269,
      "calib/avg_num_step_conf": 5.7734375,
      "calib/ece": 0.3886274509803922,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.00392156862745098,
      "calib/gap": 0.0037246216315983682,
      "calib/mean_conf": 0.1116078431372549,
      "calib/mu_c": 0.11349206349206349,
      "calib/mu_w": 0.10976744186046512,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.003058823529411765,
      "calib/std_conf": 0.133687384600157,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2505.0,
      "completions/max_terminated_length": 2505.0,
      "completions/mean_length": 332.48046875,
      "completions/mean_terminated_length": 332.48046875,
      "completions/min_length": 87.0,
      "completions/min_terminated_length": 87.0,
      "epoch": 0.0576,
      "grad_norm": 0.204210564494133,
      "learning_rate": 4.055555555555556e-06,
      "loss": 0.019,
      "num_tokens": 11692111.0,
      "reward": 1.1365880966186523,
      "reward_std": 0.3041144013404846,
      "rewards/accuracy_reward_step": 0.4921875,
      "rewards/final_brier_reward_step": 0.5854148864746094,
      "rewards/format_reward_step": 0.99609375,
      "step": 54
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.125977435096793e-06,
      "aux_brier/mean_r": 0.9559526536613703,
      "aux_brier/n_active_tok": 189.625,
      "aux_brier/n_step_records": 47.40625,
      "aux_brier/std_r": 0.08679731689335313,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5260857595154519,
      "calib/avg_num_step_conf": 5.97265625,
      "calib/ece": 0.2925236220472441,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.003937007874015748,
      "calib/gap": -0.013067175992841865,
      "calib/mean_conf": 0.10905118110236221,
      "calib/mu_c": 0.10045977011494255,
      "calib/mu_w": 0.11352694610778441,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.029527559055118113,
      "calib/std_conf": 0.1356200739828324,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1762.0,
      "completions/max_terminated_length": 1762.0,
      "completions/mean_length": 356.78515625,
      "completions/mean_terminated_length": 358.184326171875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 107.0,
      "epoch": 0.058666666666666666,
      "grad_norm": 0.06138739734888077,
      "learning_rate": 4.027777777777779e-06,
      "loss": 0.0092,
      "num_tokens": 11891272.0,
      "reward": 1.006003499031067,
      "reward_std": 0.28058964014053345,
      "rewards/accuracy_reward_step": 0.33984375,
      "rewards/final_brier_reward_step": 0.6880766153335571,
      "rewards/format_reward_step": 0.98828125,
      "step": 55
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -7.937277767405826e-07,
      "aux_brier/mean_r": 0.9601368252187967,
      "aux_brier/n_active_tok": 198.625,
      "aux_brier/n_step_records": 49.65625,
      "aux_brier/std_r": 0.09625902345578652,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.5453077995450877,
      "calib/avg_num_step_conf": 6.23046875,
      "calib/ece": 0.2386956692913386,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.007874015748031496,
      "calib/gap": 0.007739371927507524,
      "calib/mean_conf": 0.09791850393700786,
      "calib/mu_c": 0.10331168831168831,
      "calib/mu_w": 0.09557231638418079,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.016732283464566927,
      "calib/std_conf": 0.1293169244185342,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2104.0,
      "completions/max_terminated_length": 2104.0,
      "completions/mean_length": 389.4453125,
      "completions/mean_terminated_length": 390.9725646972656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 68.0,
      "epoch": 0.05973333333333333,
      "grad_norm": 0.5584831237792969,
      "learning_rate": 4.000000000000001e-06,
      "loss": 0.0456,
      "num_tokens": 12097810.0,
      "reward": 0.9787373542785645,
      "reward_std": 0.24299530684947968,
      "rewards/accuracy_reward_step": 0.30078125,
      "rewards/final_brier_reward_step": 0.727449357509613,
      "rewards/format_reward_step": 0.9921875,
      "step": 56
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.7471648068867474e-06,
      "aux_brier/mean_r": 0.9713607523590326,
      "aux_brier/n_active_tok": 194.25,
      "aux_brier/n_step_records": 48.5625,
      "aux_brier/std_r": 0.0657299466402037,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.4891007287870901,
      "calib/avg_num_step_conf": 6.07421875,
      "calib/ece": 0.3910441767068273,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.011974882873503373,
      "calib/mean_conf": 0.07008032128514056,
      "calib/mu_c": 0.06353982300884957,
      "calib/mu_w": 0.07551470588235294,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.003654618473895582,
      "calib/std_conf": 0.09132224583657442,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2615.0,
      "completions/max_terminated_length": 2615.0,
      "completions/mean_length": 392.71484375,
      "completions/mean_terminated_length": 394.2549133300781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 109.0,
      "epoch": 0.0608,
      "grad_norm": 0.08221420645713806,
      "learning_rate": 3.972222222222223e-06,
      "loss": 0.0511,
      "num_tokens": 12305137.0,
      "reward": 1.0669505596160889,
      "reward_std": 0.3339952826499939,
      "rewards/accuracy_reward_step": 0.44140625,
      "rewards/final_brier_reward_step": 0.572489857673645,
      "rewards/format_reward_step": 0.96484375,
      "step": 57
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.4384147526413926e-06,
      "aux_brier/mean_r": 0.9715827684849501,
      "aux_brier/n_active_tok": 227.5,
      "aux_brier/n_step_records": 56.875,
      "aux_brier/std_r": 0.0746084921520378,
      "calib/answer_extract_rate": 0.95703125,
      "calib/auroc": 0.5481687014428412,
      "calib/avg_num_step_conf": 7.44921875,
      "calib/ece": 0.29920855327868856,
      "calib/final_conf_rate": 0.953125,
      "calib/format_rate": 0.953125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.00722200480947096,
      "calib/mean_conf": 0.06152915163934426,
      "calib/mu_c": 0.06623529411764706,
      "calib/mu_w": 0.0590132893081761,
      "calib/nonempty_final_conf_rate": 0.953125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.006188524590163935,
      "calib/std_conf": 0.08642366508624892,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2412.0,
      "completions/max_terminated_length": 2412.0,
      "completions/mean_length": 473.35546875,
      "completions/mean_terminated_length": 480.86907958984375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 97.0,
      "epoch": 0.06186666666666667,
      "grad_norm": 0.06282463669776917,
      "learning_rate": 3.944444444444445e-06,
      "loss": 0.0398,
      "num_tokens": 12532636.0,
      "reward": 0.9721814393997192,
      "reward_std": 0.30911701917648315,
      "rewards/accuracy_reward_step": 0.33203125,
      "rewards/final_brier_reward_step": 0.654350757598877,
      "rewards/format_reward_step": 0.953125,
      "step": 58
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.87842881334577e-06,
      "aux_brier/mean_r": 0.9620572626590729,
      "aux_brier/n_active_tok": 194.75,
      "aux_brier/n_step_records": 48.6875,
      "aux_brier/std_r": 0.07372519691853086,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.4679160125588697,
      "calib/avg_num_step_conf": 6.1484375,
      "calib/ece": 0.36913147410358566,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.01195219123505976,
      "calib/gap": 0.0004322344322344185,
      "calib/mean_conf": 0.06513147410358565,
      "calib/mu_c": 0.06538461538461539,
      "calib/mu_w": 0.06495238095238097,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.0099601593625498,
      "calib/std_conf": 0.1323152799648236,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2243.0,
      "completions/max_terminated_length": 2243.0,
      "completions/mean_length": 406.8671875,
      "completions/mean_terminated_length": 408.4627685546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 98.0,
      "epoch": 0.06293333333333333,
      "grad_norm": 0.13648149371147156,
      "learning_rate": 3.916666666666667e-06,
      "loss": 0.037,
      "num_tokens": 12743042.0,
      "reward": 1.0425814390182495,
      "reward_std": 0.25638192892074585,
      "rewards/accuracy_reward_step": 0.40625,
      "rewards/final_brier_reward_step": 0.600013256072998,
      "rewards/format_reward_step": 0.97265625,
      "step": 59
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.848851273333345e-07,
      "aux_brier/mean_r": 0.9681909456849098,
      "aux_brier/n_active_tok": 187.5,
      "aux_brier/n_step_records": 46.875,
      "aux_brier/std_r": 0.07947756828070851,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.47326698695950586,
      "calib/avg_num_step_conf": 5.9140625,
      "calib/ece": 0.34575502008032133,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.004016064257028112,
      "calib/gap": -0.017937062457103622,
      "calib/mean_conf": 0.05111244979919679,
      "calib/mu_c": 0.039946808510638296,
      "calib/mu_w": 0.05788387096774192,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.00967871485943775,
      "calib/std_conf": 0.09965484311687559,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2884.0,
      "completions/max_terminated_length": 2884.0,
      "completions/mean_length": 419.39453125,
      "completions/mean_terminated_length": 421.03924560546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 133.0,
      "epoch": 0.064,
      "grad_norm": 0.05861761420965195,
      "learning_rate": 3.88888888888889e-06,
      "loss": 0.0734,
      "num_tokens": 12959263.0,
      "reward": 1.0062377452850342,
      "reward_std": 0.31737709045410156,
      "rewards/accuracy_reward_step": 0.3671875,
      "rewards/final_brier_reward_step": 0.6187013387680054,
      "rewards/format_reward_step": 0.96875,
      "step": 60
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 8.383347991774981e-07,
      "aux_brier/mean_r": 0.9603624045848846,
      "aux_brier/n_active_tok": 206.625,
      "aux_brier/n_step_records": 51.65625,
      "aux_brier/std_r": 0.08457234721845452,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.46875,
      "calib/avg_num_step_conf": 6.515625,
      "calib/ece": 0.4553147410358566,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.005514989837398378,
      "calib/mean_conf": 0.048828685258964145,
      "calib/mu_c": 0.04601626016260163,
      "calib/mu_w": 0.05153125000000001,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.007051792828685259,
      "calib/std_conf": 0.08277776212398787,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2914.0,
      "completions/max_terminated_length": 2914.0,
      "completions/mean_length": 384.80078125,
      "completions/mean_terminated_length": 384.80078125,
      "completions/min_length": 130.0,
      "completions/min_terminated_length": 130.0,
      "epoch": 0.06506666666666666,
      "grad_norm": 0.012749969959259033,
      "learning_rate": 3.861111111111112e-06,
      "loss": 0.0474,
      "num_tokens": 13161836.0,
      "reward": 1.1015645265579224,
      "reward_std": 0.23647527396678925,
      "rewards/accuracy_reward_step": 0.48046875,
      "rewards/final_brier_reward_step": 0.5312580466270447,
      "rewards/format_reward_step": 0.9765625,
      "step": 61
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -4.5184355332850146e-06,
      "aux_brier/mean_r": 0.9696124047040939,
      "aux_brier/n_active_tok": 204.5,
      "aux_brier/n_step_records": 51.125,
      "aux_brier/std_r": 0.07563763821781322,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.5045647093193106,
      "calib/avg_num_step_conf": 6.578125,
      "calib/ece": 0.30546153846153845,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.008953403447268471,
      "calib/mean_conf": 0.03721052631578947,
      "calib/mu_c": 0.04311904761904761,
      "calib/mu_w": 0.03416564417177914,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0012955465587044534,
      "calib/std_conf": 0.06996016948448516,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2187.0,
      "completions/max_terminated_length": 2187.0,
      "completions/mean_length": 413.3359375,
      "completions/mean_terminated_length": 419.8968505859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 106.0,
      "epoch": 0.06613333333333334,
      "grad_norm": 0.593379557132721,
      "learning_rate": 3.833333333333334e-06,
      "loss": 0.0366,
      "num_tokens": 13374730.0,
      "reward": 0.9733339548110962,
      "reward_std": 0.308281809091568,
      "rewards/accuracy_reward_step": 0.33203125,
      "rewards/final_brier_reward_step": 0.6511483192443848,
      "rewards/format_reward_step": 0.95703125,
      "step": 62
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.1979301490772087e-06,
      "aux_brier/mean_r": 0.9618304893374443,
      "aux_brier/n_active_tok": 196.5,
      "aux_brier/n_step_records": 49.125,
      "aux_brier/std_r": 0.08810602781704802,
      "calib/answer_extract_rate": 0.96484375,
      "calib/auroc": 0.532903395630429,
      "calib/avg_num_step_conf": 6.25,
      "calib/ece": 0.4359109311740891,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0025671229270860796,
      "calib/mean_conf": 0.0337246963562753,
      "calib/mu_c": 0.035086206896551723,
      "calib/mu_w": 0.032519083969465644,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.052133559379460834,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2898.0,
      "completions/max_terminated_length": 2898.0,
      "completions/mean_length": 455.8828125,
      "completions/mean_terminated_length": 459.4724426269531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 152.0,
      "epoch": 0.0672,
      "grad_norm": 0.4723755717277527,
      "learning_rate": 3.8055555555555556e-06,
      "loss": 0.1104,
      "num_tokens": 13600076.0,
      "reward": 1.068447470664978,
      "reward_std": 0.33388417959213257,
      "rewards/accuracy_reward_step": 0.453125,
      "rewards/final_brier_reward_step": 0.5394150018692017,
      "rewards/format_reward_step": 0.9609375,
      "step": 63
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 8.365504952811964e-06,
      "aux_brier/mean_r": 0.9629658255726099,
      "aux_brier/n_active_tok": 196.0,
      "aux_brier/n_step_records": 49.0,
      "aux_brier/std_r": 0.06713463319829316,
      "calib/answer_extract_rate": 0.9609375,
      "calib/auroc": 0.54032474503489,
      "calib/avg_num_step_conf": 6.4765625,
      "calib/ece": 0.4112642276422764,
      "calib/final_conf_rate": 0.9609375,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.0040650406504065045,
      "calib/gap": -0.004680354267310796,
      "calib/mean_conf": 0.03832926829268292,
      "calib/mu_c": 0.0357037037037037,
      "calib/mu_w": 0.0403840579710145,
      "calib/nonempty_final_conf_rate": 0.9609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.005284552845528456,
      "calib/std_conf": 0.07393686411442102,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2802.0,
      "completions/max_terminated_length": 2802.0,
      "completions/mean_length": 435.5234375,
      "completions/mean_terminated_length": 442.4365234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 155.0,
      "epoch": 0.06826666666666667,
      "grad_norm": 0.084785595536232,
      "learning_rate": 3.777777777777778e-06,
      "loss": 0.0216,
      "num_tokens": 13815346.0,
      "reward": 1.0429743528366089,
      "reward_std": 0.33013078570365906,
      "rewards/accuracy_reward_step": 0.421875,
      "rewards/final_brier_reward_step": 0.5625226497650146,
      "rewards/format_reward_step": 0.9609375,
      "step": 64
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.2754387730179495e-06,
      "aux_brier/mean_r": 0.9693706203252077,
      "aux_brier/n_active_tok": 177.625,
      "aux_brier/n_step_records": 44.40625,
      "aux_brier/std_r": 0.06981034247201023,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5354100445009535,
      "calib/avg_num_step_conf": 5.56640625,
      "calib/ece": 0.41743873517786556,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.005405594405594405,
      "calib/mean_conf": 0.030782608695652174,
      "calib/mu_c": 0.027727272727272725,
      "calib/mu_w": 0.03313286713286713,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.006719367588932806,
      "calib/std_conf": 0.0697802605099112,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2557.0,
      "completions/max_terminated_length": 2557.0,
      "completions/mean_length": 369.671875,
      "completions/mean_terminated_length": 369.671875,
      "completions/min_length": 87.0,
      "completions/min_terminated_length": 87.0,
      "epoch": 0.06933333333333333,
      "grad_norm": 0.028543710708618164,
      "learning_rate": 3.7500000000000005e-06,
      "loss": 0.0658,
      "num_tokens": 14015006.0,
      "reward": 1.0593637228012085,
      "reward_std": 0.24863088130950928,
      "rewards/accuracy_reward_step": 0.4296875,
      "rewards/final_brier_reward_step": 0.5655798316001892,
      "rewards/format_reward_step": 0.9765625,
      "step": 65
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.1284482016326045e-06,
      "aux_brier/mean_r": 0.9776174630969763,
      "aux_brier/n_active_tok": 205.875,
      "aux_brier/n_step_records": 51.46875,
      "aux_brier/std_r": 0.06463263387195184,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.4761224489795918,
      "calib/avg_num_step_conf": 6.453125,
      "calib/ece": 0.36782258064516127,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.004032258064516129,
      "calib/gap": -0.010451700680272118,
      "calib/mean_conf": 0.03540322580645162,
      "calib/mu_c": 0.02908163265306122,
      "calib/mu_w": 0.03953333333333334,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.004032258064516129,
      "calib/std_conf": 0.07831799522900817,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2626.0,
      "completions/max_terminated_length": 2626.0,
      "completions/mean_length": 463.76171875,
      "completions/mean_terminated_length": 465.5804138183594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 105.0,
      "epoch": 0.0704,
      "grad_norm": 0.3943735957145691,
      "learning_rate": 3.7222222222222225e-06,
      "loss": 0.0598,
      "num_tokens": 14240081.0,
      "reward": 1.0145626068115234,
      "reward_std": 0.2606889605522156,
      "rewards/accuracy_reward_step": 0.3828125,
      "rewards/final_brier_reward_step": 0.5973129272460938,
      "rewards/format_reward_step": 0.96484375,
      "step": 66
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.8116742162298536e-06,
      "aux_brier/mean_r": 0.96447335742414,
      "aux_brier/n_active_tok": 174.75,
      "aux_brier/n_step_records": 43.6875,
      "aux_brier/std_r": 0.08723837620618724,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.4643442622950819,
      "calib/avg_num_step_conf": 5.5546875,
      "calib/ece": 0.4696746031746032,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.003968253968253968,
      "calib/gap": -0.015695081967213104,
      "calib/mean_conf": 0.030801587301587302,
      "calib/mu_c": 0.022704918032786885,
      "calib/mu_w": 0.03839999999999999,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.008174603174603175,
      "calib/std_conf": 0.07869218914480082,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2463.0,
      "completions/max_terminated_length": 2463.0,
      "completions/mean_length": 422.4375,
      "completions/mean_terminated_length": 424.0941467285156,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.07146666666666666,
      "grad_norm": 0.1314437985420227,
      "learning_rate": 3.694444444444445e-06,
      "loss": 0.0244,
      "num_tokens": 14453233.0,
      "reward": 1.0993558168411255,
      "reward_std": 0.247754767537117,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.522423505783081,
      "rewards/format_reward_step": 0.984375,
      "step": 67
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.061379815318734e-06,
      "aux_brier/mean_r": 0.980652455240488,
      "aux_brier/n_active_tok": 191.0,
      "aux_brier/n_step_records": 47.75,
      "aux_brier/std_r": 0.04698441309847112,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.44688311688311694,
      "calib/avg_num_step_conf": 6.03125,
      "calib/ece": 0.42776,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.004,
      "calib/gap": -0.012435064935064936,
      "calib/mean_conf": 0.0216,
      "calib/mu_c": 0.014636363636363637,
      "calib/mu_w": 0.027071428571428573,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.00468,
      "calib/std_conf": 0.06551824173464975,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2527.0,
      "completions/max_terminated_length": 2527.0,
      "completions/mean_length": 428.94921875,
      "completions/mean_terminated_length": 428.94921875,
      "completions/min_length": 110.0,
      "completions/min_terminated_length": 110.0,
      "epoch": 0.07253333333333334,
      "grad_norm": 0.13681000471115112,
      "learning_rate": 3.6666666666666666e-06,
      "loss": 0.1456,
      "num_tokens": 14667132.0,
      "reward": 1.0546975135803223,
      "reward_std": 0.2470514327287674,
      "rewards/accuracy_reward_step": 0.4296875,
      "rewards/final_brier_reward_step": 0.5547277331352234,
      "rewards/format_reward_step": 0.97265625,
      "step": 68
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.1646586658985445e-06,
      "aux_brier/mean_r": 0.9609247855842113,
      "aux_brier/n_active_tok": 178.125,
      "aux_brier/n_step_records": 44.53125,
      "aux_brier/std_r": 0.09590224422208848,
      "calib/answer_extract_rate": 0.9296875,
      "calib/auroc": 0.49823047998230474,
      "calib/avg_num_step_conf": 5.7421875,
      "calib/ece": 0.40915254237288134,
      "calib/final_conf_rate": 0.921875,
      "calib/format_rate": 0.91796875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.008397847083978466,
      "calib/mean_conf": 0.02144067796610169,
      "calib/mu_c": 0.016565656565656568,
      "calib/mu_w": 0.024963503649635035,
      "calib/nonempty_final_conf_rate": 0.921875,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.005550847457627118,
      "calib/std_conf": 0.05749392633697977,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2986.0,
      "completions/max_terminated_length": 2986.0,
      "completions/mean_length": 490.15625,
      "completions/mean_terminated_length": 503.93572998046875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 133.0,
      "epoch": 0.0736,
      "grad_norm": 0.20296375453472137,
      "learning_rate": 3.638888888888889e-06,
      "loss": 0.0887,
      "num_tokens": 14897108.0,
      "reward": 0.9813294410705566,
      "reward_std": 0.3530961871147156,
      "rewards/accuracy_reward_step": 0.38671875,
      "rewards/final_brier_reward_step": 0.5425054430961609,
      "rewards/format_reward_step": 0.91796875,
      "step": 69
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.3152611430755634e-05,
      "aux_brier/mean_r": 0.9708932712674141,
      "aux_brier/n_active_tok": 195.25,
      "aux_brier/n_step_records": 48.8125,
      "aux_brier/std_r": 0.0706544549825594,
      "calib/answer_extract_rate": 0.953125,
      "calib/auroc": 0.5127907769970667,
      "calib/avg_num_step_conf": 6.12890625,
      "calib/ece": 0.42109180327868856,
      "calib/final_conf_rate": 0.953125,
      "calib/format_rate": 0.94921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0014895422607271967,
      "calib/mean_conf": 0.019154098360655736,
      "calib/mu_c": 0.018317757009345795,
      "calib/mu_w": 0.019807299270072992,
      "calib/nonempty_final_conf_rate": 0.953125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0008606557377049181,
      "calib/std_conf": 0.02696463727794811,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3036.0,
      "completions/max_terminated_length": 3036.0,
      "completions/mean_length": 474.51171875,
      "completions/mean_terminated_length": 478.2480163574219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 81.0,
      "epoch": 0.07466666666666667,
      "grad_norm": 0.015685321763157845,
      "learning_rate": 3.6111111111111115e-06,
      "loss": 0.0665,
      "num_tokens": 15125575.0,
      "reward": 1.0289678573608398,
      "reward_std": 0.2912682294845581,
      "rewards/accuracy_reward_step": 0.41796875,
      "rewards/final_brier_reward_step": 0.5455588698387146,
      "rewards/format_reward_step": 0.94921875,
      "step": 70
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.660947077627208e-06,
      "aux_brier/mean_r": 0.9693093933165073,
      "aux_brier/n_active_tok": 217.5,
      "aux_brier/n_step_records": 54.375,
      "aux_brier/std_r": 0.06705048960066051,
      "calib/answer_extract_rate": 0.9609375,
      "calib/auroc": 0.5066202566202567,
      "calib/avg_num_step_conf": 7.0625,
      "calib/ece": 0.38995951417004054,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.004048582995951417,
      "calib/gap": -0.010331695331695333,
      "calib/mean_conf": 0.024372469635627527,
      "calib/mu_c": 0.018181818181818184,
      "calib/mu_w": 0.028513513513513517,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.006761133603238867,
      "calib/std_conf": 0.07421049318498833,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2672.0,
      "completions/max_terminated_length": 2672.0,
      "completions/mean_length": 486.51953125,
      "completions/mean_terminated_length": 490.35040283203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 126.0,
      "epoch": 0.07573333333333333,
      "grad_norm": 0.031791944056749344,
      "learning_rate": 3.5833333333333335e-06,
      "loss": 0.0882,
      "num_tokens": 15354532.0,
      "reward": 1.0098564624786377,
      "reward_std": 0.3248251974582672,
      "rewards/accuracy_reward_step": 0.38671875,
      "rewards/final_brier_reward_step": 0.5784882307052612,
      "rewards/format_reward_step": 0.95703125,
      "step": 71
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.385538210944674e-07,
      "aux_brier/mean_r": 0.9742661193013191,
      "aux_brier/n_active_tok": 200.625,
      "aux_brier/n_step_records": 50.15625,
      "aux_brier/std_r": 0.0674651956433081,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.4851190476190476,
      "calib/avg_num_step_conf": 6.31640625,
      "calib/ece": 0.4344763779527559,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0055828320802005,
      "calib/mean_conf": 0.018515748031496068,
      "calib/mu_c": 0.015438596491228073,
      "calib/mu_w": 0.021021428571428573,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.0020866141732283467,
      "calib/std_conf": 0.040990177285409095,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1937.0,
      "completions/max_terminated_length": 1937.0,
      "completions/mean_length": 417.8984375,
      "completions/mean_terminated_length": 417.8984375,
      "completions/min_length": 125.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.0768,
      "grad_norm": 0.055013444274663925,
      "learning_rate": 3.555555555555556e-06,
      "loss": 0.0454,
      "num_tokens": 15565922.0,
      "reward": 1.0758342742919922,
      "reward_std": 0.25904011726379395,
      "rewards/accuracy_reward_step": 0.4453125,
      "rewards/final_brier_reward_step": 0.5533369183540344,
      "rewards/format_reward_step": 0.984375,
      "step": 72
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.074364908360217e-06,
      "aux_brier/mean_r": 0.9756907764822245,
      "aux_brier/n_active_tok": 203.25,
      "aux_brier/n_step_records": 50.8125,
      "aux_brier/std_r": 0.06860575730752316,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.5050731707317073,
      "calib/avg_num_step_conf": 6.37890625,
      "calib/ece": 0.485866935483871,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0004139837398373959,
      "calib/mean_conf": 0.018165322580645165,
      "calib/mu_c": 0.01796,
      "calib/mu_w": 0.018373983739837396,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.02829782718307253,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2450.0,
      "completions/max_terminated_length": 2450.0,
      "completions/mean_length": 467.484375,
      "completions/mean_terminated_length": 467.484375,
      "completions/min_length": 106.0,
      "completions/min_terminated_length": 106.0,
      "epoch": 0.07786666666666667,
      "grad_norm": 0.01597970351576805,
      "learning_rate": 3.5277777777777784e-06,
      "loss": 0.0943,
      "num_tokens": 15792630.0,
      "reward": 1.0968843698501587,
      "reward_std": 0.2916446328163147,
      "rewards/accuracy_reward_step": 0.48828125,
      "rewards/final_brier_reward_step": 0.49691241979599,
      "rewards/format_reward_step": 0.96875,
      "step": 73
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.9740423112257766e-06,
      "aux_brier/mean_r": 0.9756334722042084,
      "aux_brier/n_active_tok": 204.125,
      "aux_brier/n_step_records": 51.03125,
      "aux_brier/std_r": 0.0775755551783277,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.48560000000000003,
      "calib/avg_num_step_conf": 6.3984375,
      "calib/ece": 0.38628,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.004733333333333329,
      "calib/mean_conf": 0.021240000000000002,
      "calib/mu_c": 0.018400000000000003,
      "calib/mu_w": 0.023133333333333332,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.00376,
      "calib/std_conf": 0.053736974235622896,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2635.0,
      "completions/max_terminated_length": 2635.0,
      "completions/mean_length": 428.98046875,
      "completions/mean_terminated_length": 434.06719970703125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 152.0,
      "epoch": 0.07893333333333333,
      "grad_norm": 0.038290224969387054,
      "learning_rate": 3.5e-06,
      "loss": 0.0614,
      "num_tokens": 16006377.0,
      "reward": 1.0281691551208496,
      "reward_std": 0.27312523126602173,
      "rewards/accuracy_reward_step": 0.390625,
      "rewards/final_brier_reward_step": 0.5970519185066223,
      "rewards/format_reward_step": 0.9765625,
      "step": 74
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.118100940658011e-06,
      "aux_brier/mean_r": 0.9475213903933764,
      "aux_brier/n_active_tok": 199.75,
      "aux_brier/n_step_records": 49.9375,
      "aux_brier/std_r": 0.11736969361845695,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.4985180590019299,
      "calib/avg_num_step_conf": 6.26171875,
      "calib/ece": 0.6072289156626507,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0026985111662531022,
      "calib/mean_conf": 0.01927710843373494,
      "calib/mu_c": 0.01826923076923077,
      "calib/mu_w": 0.020967741935483872,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.04188672233661531,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2719.0,
      "completions/max_terminated_length": 2719.0,
      "completions/mean_length": 417.76171875,
      "completions/mean_terminated_length": 417.76171875,
      "completions/min_length": 93.0,
      "completions/min_terminated_length": 93.0,
      "epoch": 0.08,
      "grad_norm": 0.015792354941368103,
      "learning_rate": 3.4722222222222224e-06,
      "loss": 0.048,
      "num_tokens": 16218076.0,
      "reward": 1.191572904586792,
      "reward_std": 0.27867376804351807,
      "rewards/accuracy_reward_step": 0.609375,
      "rewards/final_brier_reward_step": 0.38347890973091125,
      "rewards/format_reward_step": 0.97265625,
      "step": 75
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -4.665835495520909e-06,
      "aux_brier/mean_r": 0.9634039793163538,
      "aux_brier/n_active_tok": 205.875,
      "aux_brier/n_step_records": 51.46875,
      "aux_brier/std_r": 0.08791857168216666,
      "calib/answer_extract_rate": 0.95703125,
      "calib/auroc": 0.4433262711864407,
      "calib/avg_num_step_conf": 6.43359375,
      "calib/ece": 0.4641056910569105,
      "calib/final_conf_rate": 0.9609375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.008158103813559319,
      "calib/mean_conf": 0.018821138211382118,
      "calib/mu_c": 0.014576271186440682,
      "calib/mu_w": 0.022734375,
      "calib/nonempty_final_conf_rate": 0.9609375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0016260162601626018,
      "calib/std_conf": 0.03622188453172582,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2531.0,
      "completions/max_terminated_length": 2531.0,
      "completions/mean_length": 470.32421875,
      "completions/mean_terminated_length": 474.0275573730469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 71.0,
      "epoch": 0.08106666666666666,
      "grad_norm": 0.13059166073799133,
      "learning_rate": 3.444444444444445e-06,
      "loss": 0.13,
      "num_tokens": 16441535.0,
      "reward": 1.0665919780731201,
      "reward_std": 0.2876865863800049,
      "rewards/accuracy_reward_step": 0.4609375,
      "rewards/final_brier_reward_step": 0.5085550546646118,
      "rewards/format_reward_step": 0.95703125,
      "step": 76
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 7.768571664484858e-06,
      "aux_brier/mean_r": 0.9557140190154314,
      "aux_brier/n_active_tok": 210.0,
      "aux_brier/n_step_records": 52.5,
      "aux_brier/std_r": 0.09832711402032146,
      "calib/answer_extract_rate": 0.96484375,
      "calib/auroc": 0.47872410175714664,
      "calib/avg_num_step_conf": 6.73046875,
      "calib/ece": 0.48611336032388663,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.005589430894308949,
      "calib/mean_conf": 0.0202834008097166,
      "calib/mu_c": 0.017499999999999995,
      "calib/mu_w": 0.023089430894308944,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0021862348178137655,
      "calib/std_conf": 0.039309843723114794,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3008.0,
      "completions/max_terminated_length": 3008.0,
      "completions/mean_length": 438.40625,
      "completions/mean_terminated_length": 443.6047668457031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.08213333333333334,
      "grad_norm": 0.21531686186790466,
      "learning_rate": 3.416666666666667e-06,
      "loss": 0.0249,
      "num_tokens": 16658431.0,
      "reward": 1.0906803607940674,
      "reward_std": 0.3182448744773865,
      "rewards/accuracy_reward_step": 0.484375,
      "rewards/final_brier_reward_step": 0.49553394317626953,
      "rewards/format_reward_step": 0.96484375,
      "step": 77
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.6879490613862522e-06,
      "aux_brier/mean_r": 0.9649478234350681,
      "aux_brier/n_active_tok": 202.875,
      "aux_brier/n_step_records": 50.71875,
      "aux_brier/std_r": 0.09124442109714437,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.4816045807555614,
      "calib/avg_num_step_conf": 6.33984375,
      "calib/ece": 0.412289156626506,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.001884296432802423,
      "calib/mean_conf": 0.017429718875502008,
      "calib/mu_c": 0.01635514018691589,
      "calib/mu_w": 0.018239436619718313,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.02097570023438015,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3041.0,
      "completions/max_terminated_length": 3041.0,
      "completions/mean_length": 474.3203125,
      "completions/mean_terminated_length": 479.9446716308594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 180.0,
      "epoch": 0.0832,
      "grad_norm": 0.04863213747739792,
      "learning_rate": 3.3888888888888893e-06,
      "loss": 0.0253,
      "num_tokens": 16887881.0,
      "reward": 1.0462058782577515,
      "reward_std": 0.2661869525909424,
      "rewards/accuracy_reward_step": 0.41796875,
      "rewards/final_brier_reward_step": 0.5676359534263611,
      "rewards/format_reward_step": 0.97265625,
      "step": 78
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.1245986737734626e-06,
      "aux_brier/mean_r": 0.9615942277014256,
      "aux_brier/n_active_tok": 215.0,
      "aux_brier/n_step_records": 53.75,
      "aux_brier/std_r": 0.07969562334346847,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5287981859410431,
      "calib/avg_num_step_conf": 6.734375,
      "calib/ece": 0.568452380952381,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.003968253968253968,
      "calib/gap": -0.007768707482993201,
      "calib/mean_conf": 0.02670634920634921,
      "calib/mu_c": 0.02346938775510204,
      "calib/mu_w": 0.031238095238095242,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.005912698412698413,
      "calib/std_conf": 0.07395747034617287,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2119.0,
      "completions/max_terminated_length": 2119.0,
      "completions/mean_length": 454.57421875,
      "completions/mean_terminated_length": 458.1535339355469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.08426666666666667,
      "grad_norm": 0.0486554317176342,
      "learning_rate": 3.3611111111111117e-06,
      "loss": 0.0315,
      "num_tokens": 17110628.0,
      "reward": 1.1722087860107422,
      "reward_std": 0.2507716715335846,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.4310230612754822,
      "rewards/format_reward_step": 0.98046875,
      "step": 79
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.122955576458474e-06,
      "aux_brier/mean_r": 0.9465570580214262,
      "aux_brier/n_active_tok": 220.625,
      "aux_brier/n_step_records": 55.15625,
      "aux_brier/std_r": 0.11152001099799236,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5113501776549546,
      "calib/avg_num_step_conf": 6.9296875,
      "calib/ece": 0.5724541832669323,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0008170811948940704,
      "calib/mean_conf": 0.021171314741035858,
      "calib/mu_c": 0.02150335570469799,
      "calib/mu_w": 0.02068627450980392,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0,
      "calib/std_conf": 0.023020139585986626,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2556.0,
      "completions/max_terminated_length": 2556.0,
      "completions/mean_length": 429.8125,
      "completions/mean_terminated_length": 433.19683837890625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 136.0,
      "epoch": 0.08533333333333333,
      "grad_norm": 0.1881030946969986,
      "learning_rate": 3.3333333333333333e-06,
      "loss": 0.0251,
      "num_tokens": 17322820.0,
      "reward": 1.173011064529419,
      "reward_std": 0.2777586281299591,
      "rewards/accuracy_reward_step": 0.58203125,
      "rewards/final_brier_reward_step": 0.4186069667339325,
      "rewards/format_reward_step": 0.97265625,
      "step": 80
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.5029714336443085e-06,
      "aux_brier/mean_r": 0.9561555609107018,
      "aux_brier/n_active_tok": 224.25,
      "aux_brier/n_step_records": 56.0625,
      "aux_brier/std_r": 0.1031700312596513,
      "calib/answer_extract_rate": 0.953125,
      "calib/auroc": 0.4685198946159562,
      "calib/avg_num_step_conf": 7.140625,
      "calib/ece": 0.5130737704918032,
      "calib/final_conf_rate": 0.953125,
      "calib/format_rate": 0.953125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0039444707153955295,
      "calib/mean_conf": 0.023811475409836066,
      "calib/mu_c": 0.02198473282442748,
      "calib/mu_w": 0.02592920353982301,
      "calib/nonempty_final_conf_rate": 0.953125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.022557333861472195,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2848.0,
      "completions/max_terminated_length": 2848.0,
      "completions/mean_length": 472.19140625,
      "completions/mean_terminated_length": 479.6865234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 135.0,
      "epoch": 0.0864,
      "grad_norm": 0.4952904284000397,
      "learning_rate": 3.3055555555555558e-06,
      "loss": 0.0761,
      "num_tokens": 17549949.0,
      "reward": 1.104001522064209,
      "reward_std": 0.2605801224708557,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.46288084983825684,
      "rewards/format_reward_step": 0.953125,
      "step": 81
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.9720755916718957e-05,
      "aux_brier/mean_r": 0.9714564047753811,
      "aux_brier/n_active_tok": 191.125,
      "aux_brier/n_step_records": 47.78125,
      "aux_brier/std_r": 0.06360979704913916,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.5121765096217651,
      "calib/avg_num_step_conf": 6.1015625,
      "calib/ece": 0.5317813765182187,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0006751824817518327,
      "calib/mean_conf": 0.022874493927125508,
      "calib/mu_c": 0.023175182481751825,
      "calib/mu_w": 0.022499999999999992,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0,
      "calib/std_conf": 0.021027176295511332,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3020.0,
      "completions/max_terminated_length": 3020.0,
      "completions/mean_length": 428.13671875,
      "completions/mean_terminated_length": 431.50787353515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 146.0,
      "epoch": 0.08746666666666666,
      "grad_norm": 0.15230073034763336,
      "learning_rate": 3.277777777777778e-06,
      "loss": 0.0657,
      "num_tokens": 17765104.0,
      "reward": 1.1348745822906494,
      "reward_std": 0.283991277217865,
      "rewards/accuracy_reward_step": 0.5390625,
      "rewards/final_brier_reward_step": 0.4535607695579529,
      "rewards/format_reward_step": 0.96484375,
      "step": 82
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.247837684585252e-06,
      "aux_brier/mean_r": 0.9476204346865416,
      "aux_brier/n_active_tok": 225.875,
      "aux_brier/n_step_records": 56.46875,
      "aux_brier/std_r": 0.10184100706737809,
      "calib/answer_extract_rate": 0.96484375,
      "calib/auroc": 0.47741402116402115,
      "calib/avg_num_step_conf": 7.34765625,
      "calib/ece": 0.4845528455284553,
      "calib/final_conf_rate": 0.9609375,
      "calib/format_rate": 0.95703125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.003908730158730167,
      "calib/mean_conf": 0.030081300813008128,
      "calib/mu_c": 0.028174603174603165,
      "calib/mu_w": 0.03208333333333333,
      "calib/nonempty_final_conf_rate": 0.9609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0012195121951219512,
      "calib/std_conf": 0.03064339701178227,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2680.0,
      "completions/max_terminated_length": 2680.0,
      "completions/mean_length": 466.0859375,
      "completions/mean_terminated_length": 475.37054443359375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 108.0,
      "epoch": 0.08853333333333334,
      "grad_norm": 0.031010059639811516,
      "learning_rate": 3.2500000000000002e-06,
      "loss": 0.0327,
      "num_tokens": 17991686.0,
      "reward": 1.093408226966858,
      "reward_std": 0.2682487368583679,
      "rewards/accuracy_reward_step": 0.4921875,
      "rewards/final_brier_reward_step": 0.4908202886581421,
      "rewards/format_reward_step": 0.95703125,
      "step": 83
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -9.938589001223797e-07,
      "aux_brier/mean_r": 0.9530369993299246,
      "aux_brier/n_active_tok": 200.25,
      "aux_brier/n_step_records": 50.0625,
      "aux_brier/std_r": 0.09934906920716458,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.45075270200720535,
      "calib/avg_num_step_conf": 6.2890625,
      "calib/ece": 0.501596,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.004,
      "calib/gap": 0.004099073597529604,
      "calib/mean_conf": 0.034404000000000004,
      "calib/mu_c": 0.03630597014925374,
      "calib/mu_w": 0.03220689655172414,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.06647568566024724,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2984.0,
      "completions/max_terminated_length": 2984.0,
      "completions/mean_length": 435.32421875,
      "completions/mean_terminated_length": 437.0314025878906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 97.0,
      "epoch": 0.0896,
      "grad_norm": 0.20255745947360992,
      "learning_rate": 3.2222222222222227e-06,
      "loss": 0.0305,
      "num_tokens": 18209049.0,
      "reward": 1.133134126663208,
      "reward_std": 0.23849549889564514,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.48566144704818726,
      "rewards/format_reward_step": 0.9765625,
      "step": 84
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.3112116958591269e-06,
      "aux_brier/mean_r": 0.9549486991018057,
      "aux_brier/n_active_tok": 202.375,
      "aux_brier/n_step_records": 50.59375,
      "aux_brier/std_r": 0.0979152055697341,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.5352040489461208,
      "calib/avg_num_step_conf": 6.359375,
      "calib/ece": 0.4508999999999999,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.003602088538663596,
      "calib/mean_conf": 0.033100000000000004,
      "calib/mu_c": 0.034958677685950415,
      "calib/mu_w": 0.03135658914728682,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.025816467612746714,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2836.0,
      "completions/max_terminated_length": 2836.0,
      "completions/mean_length": 479.80859375,
      "completions/mean_terminated_length": 481.6902160644531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 91.0,
      "epoch": 0.09066666666666667,
      "grad_norm": 0.20410680770874023,
      "learning_rate": 3.1944444444444443e-06,
      "loss": 0.07,
      "num_tokens": 18439704.0,
      "reward": 1.0888876914978027,
      "reward_std": 0.2575167715549469,
      "rewards/accuracy_reward_step": 0.47265625,
      "rewards/final_brier_reward_step": 0.5274260640144348,
      "rewards/format_reward_step": 0.96875,
      "step": 85
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -5.420407020173545e-07,
      "aux_brier/mean_r": 0.9533153977245092,
      "aux_brier/n_active_tok": 201.75,
      "aux_brier/n_step_records": 50.4375,
      "aux_brier/std_r": 0.09625982833676971,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5537828426717315,
      "calib/avg_num_step_conf": 6.37109375,
      "calib/ece": 0.4328730158730159,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0016501424501424514,
      "calib/mean_conf": 0.03784126984126985,
      "calib/mu_c": 0.03695726495726495,
      "calib/mu_w": 0.0386074074074074,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0032142857142857142,
      "calib/std_conf": 0.05499847222963427,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2693.0,
      "completions/max_terminated_length": 2693.0,
      "completions/mean_length": 435.80078125,
      "completions/mean_terminated_length": 440.9683837890625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 92.0,
      "epoch": 0.09173333333333333,
      "grad_norm": 0.19507144391536713,
      "learning_rate": 3.1666666666666667e-06,
      "loss": 0.0046,
      "num_tokens": 18656781.0,
      "reward": 1.0854783058166504,
      "reward_std": 0.24163693189620972,
      "rewards/accuracy_reward_step": 0.45703125,
      "rewards/final_brier_reward_step": 0.5528507232666016,
      "rewards/format_reward_step": 0.98046875,
      "step": 86
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -8.30398129769705e-06,
      "aux_brier/mean_r": 0.9406522922217846,
      "aux_brier/n_active_tok": 178.5,
      "aux_brier/n_step_records": 44.625,
      "aux_brier/std_r": 0.11897473732096842,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5404861111111111,
      "calib/avg_num_step_conf": 5.62890625,
      "calib/ece": 0.6040599999999999,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0038125000000000034,
      "calib/mean_conf": 0.03594000000000001,
      "calib/mu_c": 0.0373125,
      "calib/mu_w": 0.033499999999999995,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0,
      "calib/std_conf": 0.024446193977795398,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2861.0,
      "completions/max_terminated_length": 2861.0,
      "completions/mean_length": 408.0859375,
      "completions/mean_terminated_length": 411.2992248535156,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.0928,
      "grad_norm": 0.35734379291534424,
      "learning_rate": 3.138888888888889e-06,
      "loss": 0.0698,
      "num_tokens": 18866747.0,
      "reward": 1.216277003288269,
      "reward_std": 0.2729073166847229,
      "rewards/accuracy_reward_step": 0.62890625,
      "rewards/final_brier_reward_step": 0.3963581323623657,
      "rewards/format_reward_step": 0.9765625,
      "step": 87
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.586885609306977e-06,
      "aux_brier/mean_r": 0.958948154002428,
      "aux_brier/n_active_tok": 186.5,
      "aux_brier/n_step_records": 46.625,
      "aux_brier/std_r": 0.09168809103903186,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5831846565566458,
      "calib/avg_num_step_conf": 5.9375,
      "calib/ece": 0.426390438247012,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.009595514209251955,
      "calib/mean_conf": 0.04372908366533864,
      "calib/mu_c": 0.048813559322033906,
      "calib/mu_w": 0.03921804511278195,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0,
      "calib/std_conf": 0.027815955873425644,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2427.0,
      "completions/max_terminated_length": 2427.0,
      "completions/mean_length": 440.56640625,
      "completions/mean_terminated_length": 444.0354309082031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 142.0,
      "epoch": 0.09386666666666667,
      "grad_norm": 0.03470201417803764,
      "learning_rate": 3.1111111111111116e-06,
      "loss": 0.0571,
      "num_tokens": 19089380.0,
      "reward": 1.0857967138290405,
      "reward_std": 0.2128307819366455,
      "rewards/accuracy_reward_step": 0.4609375,
      "rewards/final_brier_reward_step": 0.5541242957115173,
      "rewards/format_reward_step": 0.97265625,
      "step": 88
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.0660022709512744e-07,
      "aux_brier/mean_r": 0.9549915064126253,
      "aux_brier/n_active_tok": 195.25,
      "aux_brier/n_step_records": 48.8125,
      "aux_brier/std_r": 0.10538984932281892,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.48639942734430924,
      "calib/avg_num_step_conf": 6.4921875,
      "calib/ece": 0.4424931451612903,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.00365990108674432,
      "calib/mean_conf": 0.04702298387096775,
      "calib/mu_c": 0.04514876033057852,
      "calib/mu_w": 0.04880866141732284,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.0008064516129032258,
      "calib/std_conf": 0.02516101046357493,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2154.0,
      "completions/max_terminated_length": 2154.0,
      "completions/mean_length": 447.8359375,
      "completions/mean_terminated_length": 451.3622131347656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 105.0,
      "epoch": 0.09493333333333333,
      "grad_norm": 0.03613610938191414,
      "learning_rate": 3.0833333333333336e-06,
      "loss": 0.0916,
      "num_tokens": 19312914.0,
      "reward": 1.0920158624649048,
      "reward_std": 0.27613985538482666,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.5321258902549744,
      "rewards/format_reward_step": 0.96484375,
      "step": 89
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.1211321207030278e-06,
      "aux_brier/mean_r": 0.9305798541754484,
      "aux_brier/n_active_tok": 225.5,
      "aux_brier/n_step_records": 56.375,
      "aux_brier/std_r": 0.12866863430645026,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5555231805231805,
      "calib/avg_num_step_conf": 7.27734375,
      "calib/ece": 0.48307630522088363,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.001032634032634025,
      "calib/mean_conf": 0.051060240963855426,
      "calib/mu_c": 0.05154545454545454,
      "calib/mu_w": 0.050512820512820515,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.002008032128514056,
      "calib/std_conf": 0.03211526031847574,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2640.0,
      "completions/max_terminated_length": 2640.0,
      "completions/mean_length": 446.48828125,
      "completions/mean_terminated_length": 450.0039367675781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 129.0,
      "epoch": 0.096,
      "grad_norm": 0.04400340095162392,
      "learning_rate": 3.055555555555556e-06,
      "loss": 0.0506,
      "num_tokens": 19530535.0,
      "reward": 1.1325215101242065,
      "reward_std": 0.23195365071296692,
      "rewards/accuracy_reward_step": 0.51953125,
      "rewards/final_brier_reward_step": 0.5066484808921814,
      "rewards/format_reward_step": 0.97265625,
      "step": 90
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.524275157789191e-07,
      "aux_brier/mean_r": 0.9710229635238647,
      "aux_brier/n_active_tok": 211.875,
      "aux_brier/n_step_records": 52.96875,
      "aux_brier/std_r": 0.07653339493845124,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5369927629878521,
      "calib/avg_num_step_conf": 6.75,
      "calib/ece": 0.5215674603174604,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0021583096407340255,
      "calib/mean_conf": 0.057797619047619056,
      "calib/mu_c": 0.05870547945205478,
      "calib/mu_w": 0.056547169811320755,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.0,
      "calib/std_conf": 0.02801889091843786,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2377.0,
      "completions/max_terminated_length": 2377.0,
      "completions/mean_length": 455.1796875,
      "completions/mean_terminated_length": 456.9647216796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 177.0,
      "epoch": 0.09706666666666666,
      "grad_norm": 0.10957203805446625,
      "learning_rate": 3.0277777777777776e-06,
      "loss": 0.0056,
      "num_tokens": 19754773.0,
      "reward": 1.1797873973846436,
      "reward_std": 0.24657493829727173,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.4769623279571533,
      "rewards/format_reward_step": 0.98046875,
      "step": 91
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.0264232747081437e-06,
      "aux_brier/mean_r": 0.9800701979547739,
      "aux_brier/n_active_tok": 195.625,
      "aux_brier/n_step_records": 48.90625,
      "aux_brier/std_r": 0.05496317755387281,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.4971723011185119,
      "calib/avg_num_step_conf": 6.15625,
      "calib/ece": 0.5177254901960784,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 2.796280005027546e-05,
      "calib/mean_conf": 0.05639215686274511,
      "calib/mu_c": 0.056404109589041095,
      "calib/mu_w": 0.05637614678899082,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0007843137254901962,
      "calib/std_conf": 0.031105814620277675,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1535.0,
      "completions/max_terminated_length": 1535.0,
      "completions/mean_length": 403.2578125,
      "completions/mean_terminated_length": 403.2578125,
      "completions/min_length": 139.0,
      "completions/min_terminated_length": 139.0,
      "epoch": 0.09813333333333334,
      "grad_norm": 0.0068860137835145,
      "learning_rate": 3e-06,
      "loss": 0.0433,
      "num_tokens": 19964727.0,
      "reward": 1.1898558139801025,
      "reward_std": 0.22652819752693176,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.48598572611808777,
      "rewards/format_reward_step": 0.99609375,
      "step": 92
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.3416964775878633e-06,
      "aux_brier/mean_r": 0.9667877722531557,
      "aux_brier/n_active_tok": 223.5,
      "aux_brier/n_step_records": 55.875,
      "aux_brier/std_r": 0.08430354389565764,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.43957713293650796,
      "calib/avg_num_step_conf": 7.0703125,
      "calib/ece": 0.42933070866141726,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.011117311507936503,
      "calib/mean_conf": 0.0690944881889764,
      "calib/mu_c": 0.0634920634920635,
      "calib/mu_w": 0.074609375,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0011811023622047244,
      "calib/std_conf": 0.04065854222957474,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2462.0,
      "completions/max_terminated_length": 2462.0,
      "completions/mean_length": 446.9296875,
      "completions/mean_terminated_length": 448.682373046875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 122.0,
      "epoch": 0.0992,
      "grad_norm": 0.024433419108390808,
      "learning_rate": 2.9722222222222225e-06,
      "loss": -0.0094,
      "num_tokens": 20184917.0,
      "reward": 1.1273119449615479,
      "reward_std": 0.2513663172721863,
      "rewards/accuracy_reward_step": 0.4921875,
      "rewards/final_brier_reward_step": 0.5561230182647705,
      "rewards/format_reward_step": 0.9921875,
      "step": 93
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.637672211343924e-06,
      "aux_brier/mean_r": 0.9576106257736683,
      "aux_brier/n_active_tok": 188.5,
      "aux_brier/n_step_records": 47.125,
      "aux_brier/std_r": 0.08591984666600183,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5248251748251748,
      "calib/avg_num_step_conf": 5.89453125,
      "calib/ece": 0.4592231075697212,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0014259376986649475,
      "calib/mean_conf": 0.06069721115537849,
      "calib/mu_c": 0.06138461538461538,
      "calib/mu_w": 0.05995867768595043,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.00099601593625498,
      "calib/std_conf": 0.02886311891768817,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2821.0,
      "completions/max_terminated_length": 2821.0,
      "completions/mean_length": 418.40625,
      "completions/mean_terminated_length": 421.7007751464844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 86.0,
      "epoch": 0.10026666666666667,
      "grad_norm": 0.027087682858109474,
      "learning_rate": 2.944444444444445e-06,
      "loss": 0.0689,
      "num_tokens": 20400709.0,
      "reward": 1.1306896209716797,
      "reward_std": 0.1795433759689331,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.5305709838867188,
      "rewards/format_reward_step": 0.98046875,
      "step": 94
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.1341438536104373e-06,
      "aux_brier/mean_r": 0.9487221036106348,
      "aux_brier/n_active_tok": 201.75,
      "aux_brier/n_step_records": 50.4375,
      "aux_brier/std_r": 0.12240388198188157,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5189329863370202,
      "calib/avg_num_step_conf": 6.36328125,
      "calib/ece": 0.5137051792828684,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0009847104749511892,
      "calib/mean_conf": 0.06844621513944223,
      "calib/mu_c": 0.06886206896551723,
      "calib/mu_w": 0.06787735849056604,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0022310756972111555,
      "calib/std_conf": 0.03630416129191106,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2231.0,
      "completions/max_terminated_length": 2231.0,
      "completions/mean_length": 434.8671875,
      "completions/mean_terminated_length": 436.57257080078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 70.0,
      "epoch": 0.10133333333333333,
      "grad_norm": 0.07396937161684036,
      "learning_rate": 2.916666666666667e-06,
      "loss": 0.0509,
      "num_tokens": 20618163.0,
      "reward": 1.1752595901489258,
      "reward_std": 0.28123950958251953,
      "rewards/accuracy_reward_step": 0.56640625,
      "rewards/final_brier_reward_step": 0.4822881817817688,
      "rewards/format_reward_step": 0.9765625,
      "step": 95
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.7296195456716354e-08,
      "aux_brier/mean_r": 0.9515257924795151,
      "aux_brier/n_active_tok": 198.5,
      "aux_brier/n_step_records": 49.625,
      "aux_brier/std_r": 0.09161369614230352,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.5040392785012884,
      "calib/avg_num_step_conf": 6.30078125,
      "calib/ece": 0.6062304687500001,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0011181140747962975,
      "calib/mean_conf": 0.06955078125,
      "calib/mu_c": 0.06991329479768786,
      "calib/mu_w": 0.06879518072289156,
      "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": 0.035771264427814795,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.00390625,
      "completions/max_length": 1379.0,
      "completions/max_terminated_length": 1379.0,
      "completions/mean_length": 394.8515625,
      "completions/mean_terminated_length": 396.4000244140625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.1024,
      "grad_norm": 0.1715133637189865,
      "learning_rate": 2.888888888888889e-06,
      "loss": -0.022,
      "num_tokens": 20825061.0,
      "reward": 1.2789297103881836,
      "reward_std": 0.2132999300956726,
      "rewards/accuracy_reward_step": 0.67578125,
      "rewards/final_brier_reward_step": 0.4125940203666687,
      "rewards/format_reward_step": 1.0,
      "step": 96
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.5436490251329325e-06,
      "aux_brier/mean_r": 0.9722522143274546,
      "aux_brier/n_active_tok": 203.75,
      "aux_brier/n_step_records": 50.9375,
      "aux_brier/std_r": 0.07436182858145912,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.4908001711596063,
      "calib/avg_num_step_conf": 6.48046875,
      "calib/ece": 0.4502734375,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.00048352588789044026,
      "calib/mean_conf": 0.07082031250000001,
      "calib/mu_c": 0.07105263157894737,
      "calib/mu_w": 0.07056910569105693,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.00078125,
      "calib/std_conf": 0.03331477093426194,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 993.0,
      "completions/max_terminated_length": 993.0,
      "completions/mean_length": 395.16015625,
      "completions/mean_terminated_length": 396.7098388671875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 99.0,
      "epoch": 0.10346666666666667,
      "grad_norm": 0.07863761484622955,
      "learning_rate": 2.861111111111111e-06,
      "loss": 0.049,
      "num_tokens": 21031294.0,
      "reward": 1.1536492109298706,
      "reward_std": 0.2430051863193512,
      "rewards/accuracy_reward_step": 0.51953125,
      "rewards/final_brier_reward_step": 0.5442843437194824,
      "rewards/format_reward_step": 0.99609375,
      "step": 97
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.8480245942152926e-07,
      "aux_brier/mean_r": 0.941277053207159,
      "aux_brier/n_active_tok": 204.25,
      "aux_brier/n_step_records": 51.0625,
      "aux_brier/std_r": 0.11857475936994888,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5042275905912269,
      "calib/avg_num_step_conf": 6.55078125,
      "calib/ece": 0.45173625498007963,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0036596312778131163,
      "calib/mean_conf": 0.07161035856573705,
      "calib/mu_c": 0.06984615384615384,
      "calib/mu_w": 0.07350578512396695,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0027091633466135457,
      "calib/std_conf": 0.03385360638999163,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2572.0,
      "completions/max_terminated_length": 2572.0,
      "completions/mean_length": 447.6796875,
      "completions/mean_terminated_length": 449.4353332519531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 65.0,
      "epoch": 0.10453333333333334,
      "grad_norm": 0.026394343003630638,
      "learning_rate": 2.8333333333333335e-06,
      "loss": 0.0615,
      "num_tokens": 21252084.0,
      "reward": 1.1324074268341064,
      "reward_std": 0.2503637373447418,
      "rewards/accuracy_reward_step": 0.5078125,
      "rewards/final_brier_reward_step": 0.5374422073364258,
      "rewards/format_reward_step": 0.98046875,
      "step": 98
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.126785942054623e-06,
      "aux_brier/mean_r": 0.9733042698353529,
      "aux_brier/n_active_tok": 218.125,
      "aux_brier/n_step_records": 54.53125,
      "aux_brier/std_r": 0.07072518536369898,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.46119114910043424,
      "calib/avg_num_step_conf": 6.90625,
      "calib/ece": 0.27119047619047615,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.005099882815192663,
      "calib/mean_conf": 0.08198412698412698,
      "calib/mu_c": 0.07868539325842697,
      "calib/mu_w": 0.08378527607361963,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.03978144341412035,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1775.0,
      "completions/max_terminated_length": 1775.0,
      "completions/mean_length": 475.3203125,
      "completions/mean_terminated_length": 480.95654296875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 108.0,
      "epoch": 0.1056,
      "grad_norm": 0.08703938871622086,
      "learning_rate": 2.805555555555556e-06,
      "loss": -0.0333,
      "num_tokens": 21479566.0,
      "reward": 1.010657548904419,
      "reward_std": 0.24851438403129578,
      "rewards/accuracy_reward_step": 0.34765625,
      "rewards/final_brier_reward_step": 0.6832554936408997,
      "rewards/format_reward_step": 0.984375,
      "step": 99
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -9.474406214016184e-07,
      "aux_brier/mean_r": 0.9467994384467602,
      "aux_brier/n_active_tok": 200.25,
      "aux_brier/n_step_records": 50.0625,
      "aux_brier/std_r": 0.09086217200092506,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5883014623172104,
      "calib/avg_num_step_conf": 6.31640625,
      "calib/ece": 0.42170355731225295,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.012717597800274955,
      "calib/mean_conf": 0.0822490118577075,
      "calib/mu_c": 0.08858267716535434,
      "calib/mu_w": 0.07586507936507939,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0009881422924901185,
      "calib/std_conf": 0.046348964720305544,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2220.0,
      "completions/max_terminated_length": 2220.0,
      "completions/mean_length": 427.08984375,
      "completions/mean_terminated_length": 428.7647399902344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 110.0,
      "epoch": 0.10666666666666667,
      "grad_norm": 0.23760278522968292,
      "learning_rate": 2.7777777777777783e-06,
      "loss": 0.0292,
      "num_tokens": 21696309.0,
      "reward": 1.1313462257385254,
      "reward_std": 0.3002142310142517,
      "rewards/accuracy_reward_step": 0.5,
      "rewards/final_brier_reward_step": 0.5644475817680359,
      "rewards/format_reward_step": 0.98046875,
      "step": 100
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.2005507928902759e-06,
      "aux_brier/mean_r": 0.9751377236098051,
      "aux_brier/n_active_tok": 224.625,
      "aux_brier/n_step_records": 56.15625,
      "aux_brier/std_r": 0.06352466254975297,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.6312695674397801,
      "calib/avg_num_step_conf": 7.25,
      "calib/ece": 0.3567539682539684,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.015325666091623544,
      "calib/mean_conf": 0.08372222222222221,
      "calib/mu_c": 0.09229729729729728,
      "calib/mu_w": 0.07697163120567374,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0,
      "calib/std_conf": 0.04128407950358775,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1664.0,
      "completions/max_terminated_length": 1664.0,
      "completions/mean_length": 455.83203125,
      "completions/mean_terminated_length": 459.4212646484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 117.0,
      "epoch": 0.10773333333333333,
      "grad_norm": 0.1534683108329773,
      "learning_rate": 2.7500000000000004e-06,
      "loss": 0.0202,
      "num_tokens": 21919994.0,
      "reward": 1.0813418626785278,
      "reward_std": 0.2274094820022583,
      "rewards/accuracy_reward_step": 0.43359375,
      "rewards/final_brier_reward_step": 0.6222426891326904,
      "rewards/format_reward_step": 0.984375,
      "step": 101
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.4301347423261177e-06,
      "aux_brier/mean_r": 0.960718834772706,
      "aux_brier/n_active_tok": 195.75,
      "aux_brier/n_step_records": 48.9375,
      "aux_brier/std_r": 0.0864135812225868,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.47873786407766994,
      "calib/avg_num_step_conf": 6.30078125,
      "calib/ece": 0.511999209486166,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.003679961165048512,
      "calib/mean_conf": 0.08088616600790514,
      "calib/mu_c": 0.079388,
      "calib/mu_w": 0.08306796116504851,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.03451517175602543,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2431.0,
      "completions/max_terminated_length": 2431.0,
      "completions/mean_length": 391.41015625,
      "completions/mean_terminated_length": 394.49212646484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 67.0,
      "epoch": 0.1088,
      "grad_norm": 0.2126937061548233,
      "learning_rate": 2.7222222222222224e-06,
      "loss": -0.0243,
      "num_tokens": 22126891.0,
      "reward": 1.1990914344787598,
      "reward_std": 0.19276341795921326,
      "rewards/accuracy_reward_step": 0.5859375,
      "rewards/final_brier_reward_step": 0.4838661551475525,
      "rewards/format_reward_step": 0.984375,
      "step": 102
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.9599245464907433e-06,
      "aux_brier/mean_r": 0.9712262339890003,
      "aux_brier/n_active_tok": 204.75,
      "aux_brier/n_step_records": 51.1875,
      "aux_brier/std_r": 0.07889097868610406,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.41545345686014995,
      "calib/avg_num_step_conf": 6.67578125,
      "calib/ece": 0.42543650793650795,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.009837398373983758,
      "calib/mean_conf": 0.08646825396825399,
      "calib/mu_c": 0.08166666666666667,
      "calib/mu_w": 0.09150406504065042,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0,
      "calib/std_conf": 0.03863785595532429,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2845.0,
      "completions/max_terminated_length": 2845.0,
      "completions/mean_length": 471.1796875,
      "completions/mean_terminated_length": 476.7668151855469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 159.0,
      "epoch": 0.10986666666666667,
      "grad_norm": 0.20270833373069763,
      "learning_rate": 2.6944444444444444e-06,
      "loss": 0.0215,
      "num_tokens": 22352065.0,
      "reward": 1.1316890716552734,
      "reward_std": 0.17910169064998627,
      "rewards/accuracy_reward_step": 0.50390625,
      "rewards/final_brier_reward_step": 0.5501939654350281,
      "rewards/format_reward_step": 0.98046875,
      "step": 103
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 7.972208633244549e-07,
      "aux_brier/mean_r": 0.9698248207569122,
      "aux_brier/n_active_tok": 205.625,
      "aux_brier/n_step_records": 51.40625,
      "aux_brier/std_r": 0.06784799284650944,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.5757894066717596,
      "calib/avg_num_step_conf": 6.49609375,
      "calib/ece": 0.31007421875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.006632161955691385,
      "calib/mean_conf": 0.08836328125000001,
      "calib/mu_c": 0.09235294117647061,
      "calib/mu_w": 0.08572077922077923,
      "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": 0.033913678036795104,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1112.0,
      "completions/max_terminated_length": 1112.0,
      "completions/mean_length": 401.39453125,
      "completions/mean_terminated_length": 402.9686584472656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 64.0,
      "epoch": 0.11093333333333333,
      "grad_norm": 0.2142917662858963,
      "learning_rate": 2.666666666666667e-06,
      "loss": 0.043,
      "num_tokens": 22561502.0,
      "reward": 1.0649869441986084,
      "reward_std": 0.1767987608909607,
      "rewards/accuracy_reward_step": 0.3984375,
      "rewards/final_brier_reward_step": 0.6661980152130127,
      "rewards/format_reward_step": 1.0,
      "step": 104
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.2800965493164895e-06,
      "aux_brier/mean_r": 0.9729265086352825,
      "aux_brier/n_active_tok": 216.25,
      "aux_brier/n_step_records": 54.0625,
      "aux_brier/std_r": 0.06955023356567835,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.44589785214785216,
      "calib/avg_num_step_conf": 6.78515625,
      "calib/ece": 0.3487843137254902,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.008751873126873116,
      "calib/mean_conf": 0.09356862745098041,
      "calib/mu_c": 0.0886607142857143,
      "calib/mu_w": 0.09741258741258742,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0015686274509803923,
      "calib/std_conf": 0.04703993084810543,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2305.0,
      "completions/max_terminated_length": 2305.0,
      "completions/mean_length": 453.078125,
      "completions/mean_terminated_length": 453.078125,
      "completions/min_length": 113.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.112,
      "grad_norm": 0.29024821519851685,
      "learning_rate": 2.6388888888888893e-06,
      "loss": 0.0712,
      "num_tokens": 22783250.0,
      "reward": 1.0889368057250977,
      "reward_std": 0.2602972984313965,
      "rewards/accuracy_reward_step": 0.4375,
      "rewards/final_brier_reward_step": 0.6213722825050354,
      "rewards/format_reward_step": 0.9921875,
      "step": 105
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -5.263094973451121e-07,
      "aux_brier/mean_r": 0.9607481062412262,
      "aux_brier/n_active_tok": 196.0,
      "aux_brier/n_step_records": 49.0,
      "aux_brier/std_r": 0.0893817414471414,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.638063909774436,
      "calib/avg_num_step_conf": 6.140625,
      "calib/ece": 0.3584448818897637,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.01055513784461154,
      "calib/mean_conf": 0.09659448818897638,
      "calib/mu_c": 0.1024122807017544,
      "calib/mu_w": 0.09185714285714286,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.003110236220472441,
      "calib/std_conf": 0.03486358298337732,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2765.0,
      "completions/max_terminated_length": 2765.0,
      "completions/mean_length": 419.21875,
      "completions/mean_terminated_length": 419.21875,
      "completions/min_length": 170.0,
      "completions/min_terminated_length": 170.0,
      "epoch": 0.11306666666666666,
      "grad_norm": 0.4787517786026001,
      "learning_rate": 2.6111111111111113e-06,
      "loss": 0.0234,
      "num_tokens": 22995154.0,
      "reward": 1.0962450504302979,
      "reward_std": 0.23281504213809967,
      "rewards/accuracy_reward_step": 0.4453125,
      "rewards/final_brier_reward_step": 0.6271676421165466,
      "rewards/format_reward_step": 0.98828125,
      "step": 106
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.7233121986925948e-06,
      "aux_brier/mean_r": 0.9619654640555382,
      "aux_brier/n_active_tok": 203.125,
      "aux_brier/n_step_records": 50.78125,
      "aux_brier/std_r": 0.07960818279389059,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.5041942604856512,
      "calib/avg_num_step_conf": 6.390625,
      "calib/ece": 0.49173828125000013,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0005017344686218811,
      "calib/mean_conf": 0.09810546875000001,
      "calib/mu_c": 0.09831125827814569,
      "calib/mu_w": 0.0978095238095238,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.0,
      "calib/std_conf": 0.038725100485250824,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1079.0,
      "completions/max_terminated_length": 1079.0,
      "completions/mean_length": 392.1796875,
      "completions/mean_terminated_length": 393.7176818847656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 106.0,
      "epoch": 0.11413333333333334,
      "grad_norm": 0.9041869640350342,
      "learning_rate": 2.5833333333333337e-06,
      "loss": 0.0233,
      "num_tokens": 23200168.0,
      "reward": 1.216104507446289,
      "reward_std": 0.22070017457008362,
      "rewards/accuracy_reward_step": 0.58984375,
      "rewards/final_brier_reward_step": 0.5128557682037354,
      "rewards/format_reward_step": 0.99609375,
      "step": 107
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.099712111485296e-06,
      "aux_brier/mean_r": 0.9625500831753016,
      "aux_brier/n_active_tok": 206.5,
      "aux_brier/n_step_records": 51.625,
      "aux_brier/std_r": 0.08628429169039009,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.519775641025641,
      "calib/avg_num_step_conf": 6.59765625,
      "calib/ece": 0.4774133858267717,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.001210256410256419,
      "calib/mean_conf": 0.11313779527559056,
      "calib/mu_c": 0.11363333333333335,
      "calib/mu_w": 0.11242307692307693,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.044309584518234434,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1412.0,
      "completions/max_terminated_length": 1412.0,
      "completions/mean_length": 422.6171875,
      "completions/mean_terminated_length": 424.2745361328125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 99.0,
      "epoch": 0.1152,
      "grad_norm": 0.5872453451156616,
      "learning_rate": 2.5555555555555557e-06,
      "loss": -0.0107,
      "num_tokens": 23411590.0,
      "reward": 1.2171289920806885,
      "reward_std": 0.24834440648555756,
      "rewards/accuracy_reward_step": 0.58984375,
      "rewards/final_brier_reward_step": 0.5247659087181091,
      "rewards/format_reward_step": 0.9921875,
      "step": 108
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.118077545687491e-06,
      "aux_brier/mean_r": 0.9741557370871305,
      "aux_brier/n_active_tok": 238.75,
      "aux_brier/n_step_records": 59.6875,
      "aux_brier/std_r": 0.059286125950166024,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.5429352068696331,
      "calib/avg_num_step_conf": 7.73046875,
      "calib/ece": 0.3932725806451613,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.002741451990632332,
      "calib/mean_conf": 0.11543709677419357,
      "calib/mu_c": 0.11404426229508198,
      "calib/mu_w": 0.11678571428571431,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.00838709677419355,
      "calib/std_conf": 0.046235695203231446,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2604.0,
      "completions/max_terminated_length": 2604.0,
      "completions/mean_length": 470.63671875,
      "completions/mean_terminated_length": 478.107177734375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 163.0,
      "epoch": 0.11626666666666667,
      "grad_norm": 0.24365048110485077,
      "learning_rate": 2.5277777777777778e-06,
      "loss": 0.0144,
      "num_tokens": 23636673.0,
      "reward": 1.1074140071868896,
      "reward_std": 0.15337789058685303,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.5859057307243347,
      "rewards/format_reward_step": 0.96875,
      "step": 109
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.6531132100983115e-06,
      "aux_brier/mean_r": 0.9645422715693712,
      "aux_brier/n_active_tok": 186.5,
      "aux_brier/n_step_records": 46.625,
      "aux_brier/std_r": 0.07064221261680359,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5730846144231971,
      "calib/avg_num_step_conf": 5.875,
      "calib/ece": 0.38774110671936757,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.008528058992625903,
      "calib/mean_conf": 0.11028260869565219,
      "calib/mu_c": 0.11456349206349208,
      "calib/mu_w": 0.10603543307086617,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0,
      "calib/std_conf": 0.04433911020295698,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2529.0,
      "completions/max_terminated_length": 2529.0,
      "completions/mean_length": 410.0625,
      "completions/mean_terminated_length": 411.6706237792969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 100.0,
      "epoch": 0.11733333333333333,
      "grad_norm": 0.929890513420105,
      "learning_rate": 2.5e-06,
      "loss": 0.0519,
      "num_tokens": 23846569.0,
      "reward": 1.135054349899292,
      "reward_std": 0.29901665449142456,
      "rewards/accuracy_reward_step": 0.4921875,
      "rewards/final_brier_reward_step": 0.5949045419692993,
      "rewards/format_reward_step": 0.98828125,
      "step": 110
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 9.971097242739146e-07,
      "aux_brier/mean_r": 0.9534461908042431,
      "aux_brier/n_active_tok": 213.875,
      "aux_brier/n_step_records": 53.46875,
      "aux_brier/std_r": 0.09113074054766912,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.4310224089635854,
      "calib/avg_num_step_conf": 6.9375,
      "calib/ece": 0.40815537848605576,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.0287392411510059,
      "calib/mean_conf": 0.13877689243027888,
      "calib/mu_c": 0.12515151515151515,
      "calib/mu_w": 0.15389075630252105,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.01051792828685259,
      "calib/std_conf": 0.07677393487116559,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2807.0,
      "completions/max_terminated_length": 2807.0,
      "completions/mean_length": 459.984375,
      "completions/mean_terminated_length": 463.6062927246094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 122.0,
      "epoch": 0.1184,
      "grad_norm": 0.3018098473548889,
      "learning_rate": 2.4722222222222226e-06,
      "loss": 0.0499,
      "num_tokens": 24071733.0,
      "reward": 1.1431818008422852,
      "reward_std": 0.2569875717163086,
      "rewards/accuracy_reward_step": 0.515625,
      "rewards/final_brier_reward_step": 0.5649149417877197,
      "rewards/format_reward_step": 0.97265625,
      "step": 111
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.5166348797955873e-07,
      "aux_brier/mean_r": 0.9389750920236111,
      "aux_brier/n_active_tok": 206.875,
      "aux_brier/n_step_records": 51.71875,
      "aux_brier/std_r": 0.11770817166689085,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.6010658914728682,
      "calib/avg_num_step_conf": 6.86328125,
      "calib/ece": 0.3773734939759036,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.015600387596899218,
      "calib/mean_conf": 0.1406987951807229,
      "calib/mu_c": 0.1482170542635659,
      "calib/mu_w": 0.1326166666666667,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.0,
      "calib/std_conf": 0.06453119197337234,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2913.0,
      "completions/max_terminated_length": 2913.0,
      "completions/mean_length": 457.41015625,
      "completions/mean_terminated_length": 462.8340148925781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 70.0,
      "epoch": 0.11946666666666667,
      "grad_norm": 1.4498605728149414,
      "learning_rate": 2.4444444444444447e-06,
      "loss": -0.0034,
      "num_tokens": 24296750.0,
      "reward": 1.1367487907409668,
      "reward_std": 0.27346163988113403,
      "rewards/accuracy_reward_step": 0.50390625,
      "rewards/final_brier_reward_step": 0.5938701033592224,
      "rewards/format_reward_step": 0.96875,
      "step": 112
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.1186343113210881e-07,
      "aux_brier/mean_r": 0.9530810378491879,
      "aux_brier/n_active_tok": 213.75,
      "aux_brier/n_step_records": 53.4375,
      "aux_brier/std_r": 0.09911039125290699,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5730155774840191,
      "calib/avg_num_step_conf": 6.72265625,
      "calib/ece": 0.3657086614173228,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.01889344007943894,
      "calib/mean_conf": 0.15161417322834647,
      "calib/mu_c": 0.16076335877862594,
      "calib/mu_w": 0.141869918699187,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0007874015748031494,
      "calib/std_conf": 0.06573593646116775,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2937.0,
      "completions/max_terminated_length": 2937.0,
      "completions/mean_length": 407.609375,
      "completions/mean_terminated_length": 409.2078552246094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 105.0,
      "epoch": 0.12053333333333334,
      "grad_norm": 0.08281253278255463,
      "learning_rate": 2.4166666666666667e-06,
      "loss": 0.0149,
      "num_tokens": 24506298.0,
      "reward": 1.1622889041900635,
      "reward_std": 0.24908068776130676,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.6179052591323853,
      "rewards/format_reward_step": 0.9921875,
      "step": 113
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 8.571375050836494e-07,
      "aux_brier/mean_r": 0.9518765844404697,
      "aux_brier/n_active_tok": 209.0,
      "aux_brier/n_step_records": 52.25,
      "aux_brier/std_r": 0.08919659565435722,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.48012173012173015,
      "calib/avg_num_step_conf": 6.62109375,
      "calib/ece": 0.47306078431372556,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.014660353535353554,
      "calib/mean_conf": 0.15023333333333333,
      "calib/mu_c": 0.14454166666666668,
      "calib/mu_w": 0.15920202020202023,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.005764705882352941,
      "calib/std_conf": 0.08733099907073553,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2707.0,
      "completions/max_terminated_length": 2707.0,
      "completions/mean_length": 412.65625,
      "completions/mean_terminated_length": 412.65625,
      "completions/min_length": 133.0,
      "completions/min_terminated_length": 133.0,
      "epoch": 0.1216,
      "grad_norm": 0.1044851541519165,
      "learning_rate": 2.388888888888889e-06,
      "loss": 0.0548,
      "num_tokens": 24716962.0,
      "reward": 1.2406219244003296,
      "reward_std": 0.22638444602489471,
      "rewards/accuracy_reward_step": 0.609375,
      "rewards/final_brier_reward_step": 0.5328000783920288,
      "rewards/format_reward_step": 0.99609375,
      "step": 114
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.011089377835674e-07,
      "aux_brier/mean_r": 0.9587736297398806,
      "aux_brier/n_active_tok": 196.0,
      "aux_brier/n_step_records": 49.0,
      "aux_brier/std_r": 0.0770155581558356,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5316535433070866,
      "calib/avg_num_step_conf": 6.14453125,
      "calib/ece": 0.3045238095238096,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.003968253968253968,
      "calib/gap": 0.016079370078740163,
      "calib/mean_conf": 0.19293650793650796,
      "calib/mu_c": 0.20104,
      "calib/mu_w": 0.18496062992125983,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.98046875,
      "calib/pce": 0.0007142857142857148,
      "calib/std_conf": 0.11020142064245368,
      "calib/step_conf_rate": 0.98046875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3058.0,
      "completions/max_terminated_length": 3058.0,
      "completions/mean_length": 407.12890625,
      "completions/mean_terminated_length": 408.72552490234375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 88.0,
      "epoch": 0.12266666666666666,
      "grad_norm": 0.09419503808021545,
      "learning_rate": 2.361111111111111e-06,
      "loss": 0.0633,
      "num_tokens": 24926451.0,
      "reward": 1.1358131170272827,
      "reward_std": 0.2533056437969208,
      "rewards/accuracy_reward_step": 0.48828125,
      "rewards/final_brier_reward_step": 0.6370025277137756,
      "rewards/format_reward_step": 0.9765625,
      "step": 115
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.5253489710497234e-07,
      "aux_brier/mean_r": 0.9548651613295078,
      "aux_brier/n_active_tok": 222.5,
      "aux_brier/n_step_records": 55.625,
      "aux_brier/std_r": 0.07094483741093427,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5422413793103449,
      "calib/avg_num_step_conf": 7.2578125,
      "calib/ece": 0.3584189723320158,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": 0.0227292464878672,
      "calib/mean_conf": 0.2239525691699605,
      "calib/mu_c": 0.23365517241379313,
      "calib/mu_w": 0.21092592592592593,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.004624505928853755,
      "calib/std_conf": 0.12984100083314182,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3068.0,
      "completions/max_terminated_length": 3068.0,
      "completions/mean_length": 459.1484375,
      "completions/mean_terminated_length": 462.7637634277344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 88.0,
      "epoch": 0.12373333333333333,
      "grad_norm": 0.10471275448799133,
      "learning_rate": 2.3333333333333336e-06,
      "loss": -0.0039,
      "num_tokens": 25148513.0,
      "reward": 1.2133257389068604,
      "reward_std": 0.23516668379306793,
      "rewards/accuracy_reward_step": 0.56640625,
      "rewards/final_brier_reward_step": 0.618928074836731,
      "rewards/format_reward_step": 0.984375,
      "step": 116
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.8026882203935344e-07,
      "aux_brier/mean_r": 0.9457707889378071,
      "aux_brier/n_active_tok": 212.0,
      "aux_brier/n_step_records": 53.0,
      "aux_brier/std_r": 0.08542424446204677,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5064516129032257,
      "calib/avg_num_step_conf": 7.203125,
      "calib/ece": 0.1695486055776892,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.00398406374501992,
      "calib/gap": 0.0043745430107527405,
      "calib/mean_conf": 0.23969442231075697,
      "calib/mu_c": 0.24239583333333337,
      "calib/mu_w": 0.23802129032258063,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.013386454183266934,
      "calib/std_conf": 0.12561702456800855,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2788.0,
      "completions/max_terminated_length": 2788.0,
      "completions/mean_length": 419.0546875,
      "completions/mean_terminated_length": 425.70635986328125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 162.0,
      "epoch": 0.1248,
      "grad_norm": 0.07769742608070374,
      "learning_rate": 2.305555555555556e-06,
      "loss": -0.025,
      "num_tokens": 25362391.0,
      "reward": 1.041290044784546,
      "reward_std": 0.3035169839859009,
      "rewards/accuracy_reward_step": 0.375,
      "rewards/final_brier_reward_step": 0.7120351791381836,
      "rewards/format_reward_step": 0.9765625,
      "step": 117
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.1297898944162554e-07,
      "aux_brier/mean_r": 0.9387212097644806,
      "aux_brier/n_active_tok": 224.625,
      "aux_brier/n_step_records": 56.15625,
      "aux_brier/std_r": 0.09899964329088107,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.48184617358393744,
      "calib/avg_num_step_conf": 7.3828125,
      "calib/ece": 0.3282188,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.004,
      "calib/gap": -0.019473622426069648,
      "calib/mean_conf": 0.25194120000000003,
      "calib/mu_c": 0.24267175572519084,
      "calib/mu_w": 0.2621453781512605,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.028080000000000008,
      "calib/std_conf": 0.16203874876880528,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1644.0,
      "completions/max_terminated_length": 1644.0,
      "completions/mean_length": 424.3125,
      "completions/mean_terminated_length": 431.0476379394531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 101.0,
      "epoch": 0.12586666666666665,
      "grad_norm": 0.1443454772233963,
      "learning_rate": 2.277777777777778e-06,
      "loss": -0.0635,
      "num_tokens": 25575023.0,
      "reward": 1.1563937664031982,
      "reward_std": 0.27992963790893555,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.6255753040313721,
      "rewards/format_reward_step": 0.9765625,
      "step": 118
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 9.348430286826304e-08,
      "aux_brier/mean_r": 0.928545979782939,
      "aux_brier/n_active_tok": 202.25,
      "aux_brier/n_step_records": 50.5625,
      "aux_brier/std_r": 0.10552759381243959,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5303793328973185,
      "calib/avg_num_step_conf": 6.421875,
      "calib/ece": 0.2990763052208835,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.004016064257028112,
      "calib/gap": 0.015157619359058239,
      "calib/mean_conf": 0.2671887550200803,
      "calib/mu_c": 0.273884892086331,
      "calib/mu_w": 0.25872727272727275,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.004016064257028112,
      "calib/std_conf": 0.143208109983746,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2854.0,
      "completions/max_terminated_length": 2854.0,
      "completions/mean_length": 479.15234375,
      "completions/mean_terminated_length": 484.8340148925781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 123.0,
      "epoch": 0.12693333333333334,
      "grad_norm": 0.1274876594543457,
      "learning_rate": 2.25e-06,
      "loss": 0.0035,
      "num_tokens": 25802750.0,
      "reward": 1.1830272674560547,
      "reward_std": 0.2851863503456116,
      "rewards/accuracy_reward_step": 0.54296875,
      "rewards/final_brier_reward_step": 0.6305460929870605,
      "rewards/format_reward_step": 0.96484375,
      "step": 119
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.848828588117527e-08,
      "aux_brier/mean_r": 0.9395423028618097,
      "aux_brier/n_active_tok": 196.875,
      "aux_brier/n_step_records": 49.21875,
      "aux_brier/std_r": 0.09219324978766963,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.4797689646651841,
      "calib/avg_num_step_conf": 6.1796875,
      "calib/ece": 0.29363281249999995,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0,
      "calib/gap": -0.008479120336051416,
      "calib/mean_conf": 0.2822265625,
      "calib/mu_c": 0.2784507042253522,
      "calib/mu_w": 0.2869298245614036,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.010585937499999996,
      "calib/std_conf": 0.13133376581988956,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1104.0,
      "completions/max_terminated_length": 1104.0,
      "completions/mean_length": 399.3984375,
      "completions/mean_terminated_length": 400.9647216796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 165.0,
      "epoch": 0.128,
      "grad_norm": 0.22246916592121124,
      "learning_rate": 2.222222222222222e-06,
      "loss": 0.0094,
      "num_tokens": 26011684.0,
      "reward": 1.2160890102386475,
      "reward_std": 0.2542319595813751,
      "rewards/accuracy_reward_step": 0.5546875,
      "rewards/final_brier_reward_step": 0.6534183025360107,
      "rewards/format_reward_step": 0.99609375,
      "step": 120
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.904218475765276e-09,
      "aux_brier/mean_r": 0.919608561322093,
      "aux_brier/n_active_tok": 219.625,
      "aux_brier/n_step_records": 54.90625,
      "aux_brier/std_r": 0.11453270737547427,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5790712468193384,
      "calib/avg_num_step_conf": 7.08984375,
      "calib/ece": 0.19111553784860558,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.02390438247011952,
      "calib/gap": 0.045514529262086545,
      "calib/mean_conf": 0.3282820717131474,
      "calib/mu_c": 0.3520366666666667,
      "calib/mu_w": 0.3065221374045802,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.02065498007968128,
      "calib/std_conf": 0.18741276801443255,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2559.0,
      "completions/max_terminated_length": 2559.0,
      "completions/mean_length": 460.58203125,
      "completions/mean_terminated_length": 466.0434875488281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 94.0,
      "epoch": 0.12906666666666666,
      "grad_norm": 0.20854102075099945,
      "learning_rate": 2.1944444444444445e-06,
      "loss": -0.0333,
      "num_tokens": 26234649.0,
      "reward": 1.131608486175537,
      "reward_std": 0.3270259201526642,
      "rewards/accuracy_reward_step": 0.46875,
      "rewards/final_brier_reward_step": 0.6983091831207275,
      "rewards/format_reward_step": 0.9765625,
      "step": 121
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.3580222608944315e-07,
      "aux_brier/mean_r": 0.9207999929785728,
      "aux_brier/n_active_tok": 213.25,
      "aux_brier/n_step_records": 53.3125,
      "aux_brier/std_r": 0.10475145361851901,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.6422525200310157,
      "calib/avg_num_step_conf": 6.984375,
      "calib/ece": 0.27222380952380953,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.003968253968253968,
      "calib/gap": 0.06454241406048072,
      "calib/mean_conf": 0.32182380952380957,
      "calib/mu_c": 0.348972602739726,
      "calib/mu_w": 0.2844301886792453,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.007341269841269841,
      "calib/std_conf": 0.15098661689277507,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2416.0,
      "completions/max_terminated_length": 2416.0,
      "completions/mean_length": 416.7421875,
      "completions/mean_terminated_length": 418.3764953613281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 143.0,
      "epoch": 0.13013333333333332,
      "grad_norm": 0.0847122073173523,
      "learning_rate": 2.166666666666667e-06,
      "loss": 0.0377,
      "num_tokens": 26448679.0,
      "reward": 1.2359998226165771,
      "reward_std": 0.24089466035366058,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.686186671257019,
      "rewards/format_reward_step": 0.98046875,
      "step": 122
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 8.972941983875771e-08,
      "aux_brier/mean_r": 0.8995146565139294,
      "aux_brier/n_active_tok": 221.25,
      "aux_brier/n_step_records": 55.3125,
      "aux_brier/std_r": 0.12771078263176605,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.47579493835171965,
      "calib/avg_num_step_conf": 7.09375,
      "calib/ece": 0.2025301204819277,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.020080321285140562,
      "calib/gap": -0.016073329007138204,
      "calib/mean_conf": 0.35160642570281125,
      "calib/mu_c": 0.34295652173913044,
      "calib/mu_w": 0.35902985074626864,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.046144578313253026,
      "calib/std_conf": 0.19232834029544033,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2124.0,
      "completions/max_terminated_length": 2124.0,
      "completions/mean_length": 490.41015625,
      "completions/mean_terminated_length": 494.2716369628906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 150.0,
      "epoch": 0.1312,
      "grad_norm": 0.05073205381631851,
      "learning_rate": 2.138888888888889e-06,
      "loss": 0.0109,
      "num_tokens": 26679512.0,
      "reward": 1.0989124774932861,
      "reward_std": 0.31424033641815186,
      "rewards/accuracy_reward_step": 0.44921875,
      "rewards/final_brier_reward_step": 0.6690871119499207,
      "rewards/format_reward_step": 0.96484375,
      "step": 123
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -5.925066599854478e-08,
      "aux_brier/mean_r": 0.9220362659543753,
      "aux_brier/n_active_tok": 207.375,
      "aux_brier/n_step_records": 51.84375,
      "aux_brier/std_r": 0.10577379714231938,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.4525448345541804,
      "calib/avg_num_step_conf": 6.546875,
      "calib/ece": 0.3172352941176471,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.01568627450980392,
      "calib/gap": -0.03454723415003791,
      "calib/mean_conf": 0.3455098039215686,
      "calib/mu_c": 0.3310135135135135,
      "calib/mu_w": 0.3655607476635514,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04117647058823531,
      "calib/std_conf": 0.16270246783972891,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1435.0,
      "completions/max_terminated_length": 1435.0,
      "completions/mean_length": 442.12890625,
      "completions/mean_terminated_length": 443.8627624511719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 135.0,
      "epoch": 0.13226666666666667,
      "grad_norm": 0.27319470047950745,
      "learning_rate": 2.1111111111111114e-06,
      "loss": 0.0268,
      "num_tokens": 26899513.0,
      "reward": 1.243934154510498,
      "reward_std": 0.23813574016094208,
      "rewards/accuracy_reward_step": 0.58203125,
      "rewards/final_brier_reward_step": 0.6554237604141235,
      "rewards/format_reward_step": 0.99609375,
      "step": 124
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.6478172820976198e-08,
      "aux_brier/mean_r": 0.880499541759491,
      "aux_brier/n_active_tok": 203.875,
      "aux_brier/n_step_records": 50.96875,
      "aux_brier/std_r": 0.14444965438451618,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.4911504424778761,
      "calib/avg_num_step_conf": 6.625,
      "calib/ece": 0.17086152610441765,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.028112449799196786,
      "calib/gap": -0.0063922917751171315,
      "calib/mean_conf": 0.3800621686746988,
      "calib/mu_c": 0.376570796460177,
      "calib/mu_w": 0.3829630882352941,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.048554216867469885,
      "calib/std_conf": 0.19724339864989018,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2910.0,
      "completions/max_terminated_length": 2910.0,
      "completions/mean_length": 478.03125,
      "completions/mean_terminated_length": 481.7952880859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 60.0,
      "epoch": 0.13333333333333333,
      "grad_norm": 0.2627008855342865,
      "learning_rate": 2.0833333333333334e-06,
      "loss": 0.0082,
      "num_tokens": 27126697.0,
      "reward": 1.091411828994751,
      "reward_std": 0.29050981998443604,
      "rewards/accuracy_reward_step": 0.44140625,
      "rewards/final_brier_reward_step": 0.6781472563743591,
      "rewards/format_reward_step": 0.9609375,
      "step": 125
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.8760992945288102e-07,
      "aux_brier/mean_r": 0.907262934371829,
      "aux_brier/n_active_tok": 220.625,
      "aux_brier/n_step_records": 55.15625,
      "aux_brier/std_r": 0.11587508826050907,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.6095294706544863,
      "calib/avg_num_step_conf": 7.06640625,
      "calib/ece": 0.16498023715415017,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.015810276679841896,
      "calib/gap": 0.05830121386559883,
      "calib/mean_conf": 0.3815810276679843,
      "calib/mu_c": 0.40969465648854964,
      "calib/mu_w": 0.3513934426229508,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.014387351778656125,
      "calib/std_conf": 0.16757474397389432,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2576.0,
      "completions/max_terminated_length": 2576.0,
      "completions/mean_length": 451.58984375,
      "completions/mean_terminated_length": 453.3608093261719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 106.0,
      "epoch": 0.1344,
      "grad_norm": 0.28045162558555603,
      "learning_rate": 2.0555555555555555e-06,
      "loss": -0.0043,
      "num_tokens": 27347768.0,
      "reward": 1.1840758323669434,
      "reward_std": 0.27042731642723083,
      "rewards/accuracy_reward_step": 0.51171875,
      "rewards/final_brier_reward_step": 0.7206785082817078,
      "rewards/format_reward_step": 0.984375,
      "step": 126
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.482128341751988e-08,
      "aux_brier/mean_r": 0.9151476826518774,
      "aux_brier/n_active_tok": 222.75,
      "aux_brier/n_step_records": 55.6875,
      "aux_brier/std_r": 0.10804212570656091,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5121724646037542,
      "calib/avg_num_step_conf": 7.16015625,
      "calib/ece": 0.199314,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.012,
      "calib/gap": -0.004110545198283022,
      "calib/mean_conf": 0.35652600000000007,
      "calib/mu_c": 0.35440495867768596,
      "calib/mu_w": 0.358515503875969,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.03592000000000002,
      "calib/std_conf": 0.1663230240345575,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2832.0,
      "completions/max_terminated_length": 2832.0,
      "completions/mean_length": 438.4609375,
      "completions/mean_terminated_length": 443.66009521484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 129.0,
      "epoch": 0.13546666666666668,
      "grad_norm": 0.2871302366256714,
      "learning_rate": 2.027777777777778e-06,
      "loss": 0.0481,
      "num_tokens": 27563686.0,
      "reward": 1.1367894411087036,
      "reward_std": 0.2733447253704071,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.6877830624580383,
      "rewards/format_reward_step": 0.9765625,
      "step": 127
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.683733678152336e-08,
      "aux_brier/mean_r": 0.9009036216884851,
      "aux_brier/n_active_tok": 204.0,
      "aux_brier/n_step_records": 51.0,
      "aux_brier/std_r": 0.12476722081191838,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.515438733632988,
      "calib/avg_num_step_conf": 6.5234375,
      "calib/ece": 0.18460806451612904,
      "calib/final_conf_rate": 0.96875,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.016129032258064516,
      "calib/gap": 0.009849599374633644,
      "calib/mean_conf": 0.38434354838709683,
      "calib/mu_c": 0.38906976744186045,
      "calib/mu_w": 0.3792201680672268,
      "calib/nonempty_final_conf_rate": 0.96875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.024395161290322577,
      "calib/std_conf": 0.16708054086516494,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3036.0,
      "completions/max_terminated_length": 3036.0,
      "completions/mean_length": 474.2265625,
      "completions/mean_terminated_length": 477.96063232421875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.13653333333333334,
      "grad_norm": 0.10523315519094467,
      "learning_rate": 2.0000000000000003e-06,
      "loss": 0.0545,
      "num_tokens": 27791752.0,
      "reward": 1.1542980670928955,
      "reward_std": 0.30696868896484375,
      "rewards/accuracy_reward_step": 0.50390625,
      "rewards/final_brier_reward_step": 0.6796922087669373,
      "rewards/format_reward_step": 0.9609375,
      "step": 128
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.1682828749207119e-07,
      "aux_brier/mean_r": 0.8982313759624958,
      "aux_brier/n_active_tok": 208.5,
      "aux_brier/n_step_records": 52.125,
      "aux_brier/std_r": 0.1205826576333493,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5295098039215687,
      "calib/avg_num_step_conf": 6.5234375,
      "calib/ece": 0.24430830039525697,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.007905138339920948,
      "calib/gap": 0.017050326797385607,
      "calib/mean_conf": 0.4024110671936759,
      "calib/mu_c": 0.4091503267973856,
      "calib/mu_w": 0.3921,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.020988142292490117,
      "calib/std_conf": 0.1788879708716506,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2589.0,
      "completions/max_terminated_length": 2589.0,
      "completions/mean_length": 421.234375,
      "completions/mean_terminated_length": 422.88629150390625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.1376,
      "grad_norm": 0.08267068862915039,
      "learning_rate": 1.9722222222222224e-06,
      "loss": 0.0427,
      "num_tokens": 28001972.0,
      "reward": 1.2638030052185059,
      "reward_std": 0.23557443916797638,
      "rewards/accuracy_reward_step": 0.59765625,
      "rewards/final_brier_reward_step": 0.688024640083313,
      "rewards/format_reward_step": 0.98828125,
      "step": 129
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.9647965827116494e-08,
      "aux_brier/mean_r": 0.8863687589764595,
      "aux_brier/n_active_tok": 208.0,
      "aux_brier/n_step_records": 52.0,
      "aux_brier/std_r": 0.14508607238531113,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.5415671735578736,
      "calib/avg_num_step_conf": 6.60546875,
      "calib/ece": 0.24348862745098035,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.03529411764705882,
      "calib/gap": 0.019205957018976927,
      "calib/mean_conf": 0.3954917647058824,
      "calib/mu_c": 0.40370136986301375,
      "calib/mu_w": 0.3844954128440368,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0332156862745098,
      "calib/std_conf": 0.1962376010091516,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1488.0,
      "completions/max_terminated_length": 1488.0,
      "completions/mean_length": 395.33984375,
      "completions/mean_terminated_length": 396.8902282714844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 118.0,
      "epoch": 0.13866666666666666,
      "grad_norm": 0.0783396065235138,
      "learning_rate": 1.944444444444445e-06,
      "loss": -0.012,
      "num_tokens": 28208467.0,
      "reward": 1.245288610458374,
      "reward_std": 0.22594107687473297,
      "rewards/accuracy_reward_step": 0.57421875,
      "rewards/final_brier_reward_step": 0.6920915842056274,
      "rewards/format_reward_step": 0.99609375,
      "step": 130
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -7.633915508065137e-08,
      "aux_brier/mean_r": 0.888754416257143,
      "aux_brier/n_active_tok": 198.0,
      "aux_brier/n_step_records": 49.5,
      "aux_brier/std_r": 0.1411213732790202,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.617965367965368,
      "calib/avg_num_step_conf": 6.37890625,
      "calib/ece": 0.0931796875,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.015625,
      "calib/gap": 0.07048051948051959,
      "calib/mean_conf": 0.4176015625,
      "calib/mu_c": 0.46,
      "calib/mu_w": 0.38951948051948043,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.05617187500000002,
      "calib/std_conf": 0.19151764441575247,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1827.0,
      "completions/max_terminated_length": 1827.0,
      "completions/mean_length": 394.60546875,
      "completions/mean_terminated_length": 396.1529541015625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.13973333333333332,
      "grad_norm": 0.050378166139125824,
      "learning_rate": 1.916666666666667e-06,
      "loss": 0.0227,
      "num_tokens": 28415694.0,
      "reward": 1.0798051357269287,
      "reward_std": 0.19457346200942993,
      "rewards/accuracy_reward_step": 0.3984375,
      "rewards/final_brier_reward_step": 0.7489080429077148,
      "rewards/format_reward_step": 0.98828125,
      "step": 131
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -7.26369157044715e-08,
      "aux_brier/mean_r": 0.8635993953794241,
      "aux_brier/n_active_tok": 211.125,
      "aux_brier/n_step_records": 52.78125,
      "aux_brier/std_r": 0.1567929598968476,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.4409935897435897,
      "calib/avg_num_step_conf": 6.6953125,
      "calib/ece": 0.26063908386230467,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0234375,
      "calib/gap": -0.040887022235577,
      "calib/mean_conf": 0.4440484161376954,
      "calib/mu_c": 0.4280769230769231,
      "calib/mu_w": 0.4689639453125001,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.04765625,
      "calib/std_conf": 0.19377898635015073,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1145.0,
      "completions/max_terminated_length": 1145.0,
      "completions/mean_length": 405.3515625,
      "completions/mean_terminated_length": 406.9411926269531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 86.0,
      "epoch": 0.1408,
      "grad_norm": 0.09182775765657425,
      "learning_rate": 1.888888888888889e-06,
      "loss": 0.0135,
      "num_tokens": 28625056.0,
      "reward": 1.2787785530090332,
      "reward_std": 0.24889016151428223,
      "rewards/accuracy_reward_step": 0.609375,
      "rewards/final_brier_reward_step": 0.6776144504547119,
      "rewards/format_reward_step": 1.0,
      "step": 132
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -4.4734078885610984e-08,
      "aux_brier/mean_r": 0.8646962121129036,
      "aux_brier/n_active_tok": 227.0,
      "aux_brier/n_step_records": 56.75,
      "aux_brier/std_r": 0.16395573201589286,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.5164760348583878,
      "calib/avg_num_step_conf": 7.62109375,
      "calib/ece": 0.18841526104417672,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96484375,
      "calib/frac_conf_gt_0.9": 0.05622489959839357,
      "calib/gap": 0.0016290441176470272,
      "calib/mean_conf": 0.4535927710843374,
      "calib/mu_c": 0.45459374999999996,
      "calib/mu_w": 0.45296470588235294,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.1282329317269076,
      "calib/std_conf": 0.21843736263177577,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2994.0,
      "completions/max_terminated_length": 2994.0,
      "completions/mean_length": 487.44921875,
      "completions/mean_terminated_length": 497.15936279296875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 145.0,
      "epoch": 0.14186666666666667,
      "grad_norm": 0.10583373159170151,
      "learning_rate": 1.8611111111111113e-06,
      "loss": 0.0108,
      "num_tokens": 28856187.0,
      "reward": 1.028933048248291,
      "reward_std": 0.37068942189216614,
      "rewards/accuracy_reward_step": 0.375,
      "rewards/final_brier_reward_step": 0.6860446929931641,
      "rewards/format_reward_step": 0.96484375,
      "step": 133
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 7.99102370385718e-09,
      "aux_brier/mean_r": 0.8611395936459303,
      "aux_brier/n_active_tok": 221.625,
      "aux_brier/n_step_records": 55.40625,
      "aux_brier/std_r": 0.16742704529315233,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.4701492537313433,
      "calib/avg_num_step_conf": 7.13671875,
      "calib/ece": 0.19036086956521736,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.039525691699604744,
      "calib/gap": -0.023781650570676105,
      "calib/mean_conf": 0.43343359683794463,
      "calib/mu_c": 0.42224776119402985,
      "calib/mu_w": 0.44602941176470595,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.047075098814229235,
      "calib/std_conf": 0.21354872836213862,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2367.0,
      "completions/max_terminated_length": 2367.0,
      "completions/mean_length": 494.671875,
      "completions/mean_terminated_length": 494.671875,
      "completions/min_length": 43.0,
      "completions/min_terminated_length": 43.0,
      "epoch": 0.14293333333333333,
      "grad_norm": 0.12384790182113647,
      "learning_rate": 1.8333333333333333e-06,
      "loss": 0.0724,
      "num_tokens": 29091775.0,
      "reward": 1.183688998222351,
      "reward_std": 0.35327237844467163,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.6722561120986938,
      "rewards/format_reward_step": 0.984375,
      "step": 134
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 8.770893714094008e-08,
      "aux_brier/mean_r": 0.8544477913528681,
      "aux_brier/n_active_tok": 224.75,
      "aux_brier/n_step_records": 56.1875,
      "aux_brier/std_r": 0.17348002130165696,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5098343030609663,
      "calib/avg_num_step_conf": 7.3515625,
      "calib/ece": 0.1913301587301587,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.03571428571428571,
      "calib/gap": 0.00045906906147230586,
      "calib/mean_conf": 0.44716190476190476,
      "calib/mu_c": 0.4473768656716418,
      "calib/mu_w": 0.4469177966101695,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05337301587301588,
      "calib/std_conf": 0.2107196854836779,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2850.0,
      "completions/max_terminated_length": 2850.0,
      "completions/mean_length": 470.65625,
      "completions/mean_terminated_length": 472.5019836425781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 121.0,
      "epoch": 0.144,
      "grad_norm": 0.08482138067483902,
      "learning_rate": 1.8055555555555557e-06,
      "loss": 0.0304,
      "num_tokens": 29318143.0,
      "reward": 1.1878116130828857,
      "reward_std": 0.31839847564697266,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.6887466907501221,
      "rewards/format_reward_step": 0.984375,
      "step": 135
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -7.606758112288503e-09,
      "aux_brier/mean_r": 0.8612100519239902,
      "aux_brier/n_active_tok": 235.0,
      "aux_brier/n_step_records": 58.75,
      "aux_brier/std_r": 0.15901895437855273,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.47738951695786225,
      "calib/avg_num_step_conf": 7.44921875,
      "calib/ece": 0.16964741035856576,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.03187250996015936,
      "calib/gap": -0.015089157245632034,
      "calib/mean_conf": 0.48924900398406374,
      "calib/mu_c": 0.4808928571428571,
      "calib/mu_w": 0.49598201438848916,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10634063745019923,
      "calib/std_conf": 0.18945002901827143,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2170.0,
      "completions/max_terminated_length": 2170.0,
      "completions/mean_length": 442.34375,
      "completions/mean_terminated_length": 445.8267822265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.14506666666666668,
      "grad_norm": 0.03690337762236595,
      "learning_rate": 1.777777777777778e-06,
      "loss": -0.0317,
      "num_tokens": 29539871.0,
      "reward": 1.1051082611083984,
      "reward_std": 0.26446324586868286,
      "rewards/accuracy_reward_step": 0.44140625,
      "rewards/final_brier_reward_step": 0.6938701868057251,
      "rewards/format_reward_step": 0.98046875,
      "step": 136
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 9.798647643277647e-09,
      "aux_brier/mean_r": 0.855107743293047,
      "aux_brier/n_active_tok": 225.875,
      "aux_brier/n_step_records": 56.46875,
      "aux_brier/std_r": 0.15921062580309808,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5522699004975125,
      "calib/avg_num_step_conf": 7.20703125,
      "calib/ece": 0.11597047244094486,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.01968503937007874,
      "calib/gap": 0.02715677860696525,
      "calib/mean_conf": 0.4966476377952756,
      "calib/mu_c": 0.5094776119402986,
      "calib/mu_w": 0.48232083333333337,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.042529527559055105,
      "calib/std_conf": 0.1831278727283494,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3042.0,
      "completions/max_terminated_length": 3042.0,
      "completions/mean_length": 453.9140625,
      "completions/mean_terminated_length": 455.69415283203125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 149.0,
      "epoch": 0.14613333333333334,
      "grad_norm": 0.04309280961751938,
      "learning_rate": 1.75e-06,
      "loss": -0.0056,
      "num_tokens": 29763057.0,
      "reward": 1.2005572319030762,
      "reward_std": 0.33588626980781555,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.7241036891937256,
      "rewards/format_reward_step": 0.9921875,
      "step": 137
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.0094727063980624e-07,
      "aux_brier/mean_r": 0.8415694739669561,
      "aux_brier/n_active_tok": 224.25,
      "aux_brier/n_step_records": 56.0625,
      "aux_brier/std_r": 0.16858775913715363,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.4650232318017553,
      "calib/avg_num_step_conf": 7.45703125,
      "calib/ece": 0.1920541501976285,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.03557312252964427,
      "calib/gap": -0.028633763551884295,
      "calib/mean_conf": 0.5221750988142292,
      "calib/mu_c": 0.5104046979865773,
      "calib/mu_w": 0.5390384615384616,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.06264822134387354,
      "calib/std_conf": 0.20784797132111368,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2717.0,
      "completions/max_terminated_length": 2717.0,
      "completions/mean_length": 446.45703125,
      "completions/mean_terminated_length": 446.45703125,
      "completions/min_length": 117.0,
      "completions/min_terminated_length": 117.0,
      "epoch": 0.1472,
      "grad_norm": 0.008997076191008091,
      "learning_rate": 1.7222222222222224e-06,
      "loss": -0.0032,
      "num_tokens": 29981686.0,
      "reward": 1.2454711198806763,
      "reward_std": 0.3116518259048462,
      "rewards/accuracy_reward_step": 0.58203125,
      "rewards/final_brier_reward_step": 0.685009777545929,
      "rewards/format_reward_step": 0.984375,
      "step": 138
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.5731516884066634e-08,
      "aux_brier/mean_r": 0.866764098405838,
      "aux_brier/n_active_tok": 201.75,
      "aux_brier/n_step_records": 50.4375,
      "aux_brier/std_r": 0.1419970259303227,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5766108891108891,
      "calib/avg_num_step_conf": 6.32421875,
      "calib/ece": 0.13423098039215686,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.03137254901960784,
      "calib/gap": 0.046206150099900145,
      "calib/mean_conf": 0.4853768627450981,
      "calib/mu_c": 0.5056713286713287,
      "calib/mu_w": 0.45946517857142855,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.029411764705882353,
      "calib/std_conf": 0.18573747859509537,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2483.0,
      "completions/max_terminated_length": 2483.0,
      "completions/mean_length": 407.8984375,
      "completions/mean_terminated_length": 407.8984375,
      "completions/min_length": 112.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.14826666666666666,
      "grad_norm": 0.019953807815909386,
      "learning_rate": 1.6944444444444446e-06,
      "loss": 0.0184,
      "num_tokens": 30189204.0,
      "reward": 1.2343586683273315,
      "reward_std": 0.29369819164276123,
      "rewards/accuracy_reward_step": 0.55859375,
      "rewards/final_brier_reward_step": 0.7264971733093262,
      "rewards/format_reward_step": 0.98828125,
      "step": 139
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.680944515860318e-08,
      "aux_brier/mean_r": 0.8431448042392731,
      "aux_brier/n_active_tok": 215.125,
      "aux_brier/n_step_records": 53.78125,
      "aux_brier/std_r": 0.17906808014959097,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.47558922558922556,
      "calib/avg_num_step_conf": 6.8125,
      "calib/ece": 0.19137843137254903,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.050980392156862744,
      "calib/gap": -0.008792637917637847,
      "calib/mean_conf": 0.49364117647058825,
      "calib/mu_c": 0.4902275641025641,
      "calib/mu_w": 0.4990202020202019,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.036627450980392134,
      "calib/std_conf": 0.21302567660433228,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1491.0,
      "completions/max_terminated_length": 1491.0,
      "completions/mean_length": 440.4921875,
      "completions/mean_terminated_length": 442.2196350097656,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.14933333333333335,
      "grad_norm": 0.03961111977696419,
      "learning_rate": 1.6666666666666667e-06,
      "loss": -0.0209,
      "num_tokens": 30406986.0,
      "reward": 1.281484603881836,
      "reward_std": 0.31075361371040344,
      "rewards/accuracy_reward_step": 0.609375,
      "rewards/final_brier_reward_step": 0.6962511539459229,
      "rewards/format_reward_step": 0.99609375,
      "step": 140
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.8728322965253525e-08,
      "aux_brier/mean_r": 0.834965918213129,
      "aux_brier/n_active_tok": 203.875,
      "aux_brier/n_step_records": 50.96875,
      "aux_brier/std_r": 0.18984848959371448,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5758879113717823,
      "calib/avg_num_step_conf": 6.60546875,
      "calib/ece": 0.17545157480314968,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.07874015748031496,
      "calib/gap": 0.05084051482567603,
      "calib/mean_conf": 0.5069893700787402,
      "calib/mu_c": 0.5268051612903225,
      "calib/mu_w": 0.4759646464646465,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.036102362204724424,
      "calib/std_conf": 0.21634318720677176,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1456.0,
      "completions/max_terminated_length": 1456.0,
      "completions/mean_length": 442.4921875,
      "completions/mean_terminated_length": 445.97637939453125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 159.0,
      "epoch": 0.1504,
      "grad_norm": 0.04783289507031441,
      "learning_rate": 1.638888888888889e-06,
      "loss": -0.0162,
      "num_tokens": 30627360.0,
      "reward": 1.2777183055877686,
      "reward_std": 0.3144700527191162,
      "rewards/accuracy_reward_step": 0.60546875,
      "rewards/final_brier_reward_step": 0.7202483415603638,
      "rewards/format_reward_step": 0.984375,
      "step": 141
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.755059033161423e-08,
      "aux_brier/mean_r": 0.8158276583999395,
      "aux_brier/n_active_tok": 228.75,
      "aux_brier/n_step_records": 57.1875,
      "aux_brier/std_r": 0.19122428877744824,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.45670807453416146,
      "calib/avg_num_step_conf": 7.1953125,
      "calib/ece": 0.18248039215686274,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.0784313725490196,
      "calib/gap": -0.03583757763975137,
      "calib/mean_conf": 0.5617549019607844,
      "calib/mu_c": 0.5455928571428572,
      "calib/mu_w": 0.5814304347826086,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.09760784313725493,
      "calib/std_conf": 0.21130024573700762,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2377.0,
      "completions/max_terminated_length": 2377.0,
      "completions/mean_length": 465.90234375,
      "completions/mean_terminated_length": 465.90234375,
      "completions/min_length": 138.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.15146666666666667,
      "grad_norm": 0.02992737479507923,
      "learning_rate": 1.6111111111111113e-06,
      "loss": 0.0106,
      "num_tokens": 30851791.0,
      "reward": 1.2167097330093384,
      "reward_std": 0.2924765348434448,
      "rewards/accuracy_reward_step": 0.546875,
      "rewards/final_brier_reward_step": 0.6871516704559326,
      "rewards/format_reward_step": 0.99609375,
      "step": 142
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 6.07010374464867e-08,
      "aux_brier/mean_r": 0.8592647183686495,
      "aux_brier/n_active_tok": 217.0,
      "aux_brier/n_step_records": 54.25,
      "aux_brier/std_r": 0.14784136065281928,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5375318461442864,
      "calib/avg_num_step_conf": 6.96484375,
      "calib/ece": 0.12099960629921258,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.01968503937007874,
      "calib/gap": 0.021490778599391025,
      "calib/mean_conf": 0.5034885826771653,
      "calib/mu_c": 0.5137263157894737,
      "calib/mu_w": 0.4922355371900827,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.05043307086614174,
      "calib/std_conf": 0.1865499698407842,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3029.0,
      "completions/max_terminated_length": 3029.0,
      "completions/mean_length": 451.6953125,
      "completions/mean_terminated_length": 451.6953125,
      "completions/min_length": 124.0,
      "completions/min_terminated_length": 124.0,
      "epoch": 0.15253333333333333,
      "grad_norm": 0.023315150290727615,
      "learning_rate": 1.5833333333333333e-06,
      "loss": 0.0414,
      "num_tokens": 31074761.0,
      "reward": 1.1957252025604248,
      "reward_std": 0.2707880139350891,
      "rewards/accuracy_reward_step": 0.51953125,
      "rewards/final_brier_reward_step": 0.7204006910324097,
      "rewards/format_reward_step": 0.9921875,
      "step": 143
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.952828720952482e-08,
      "aux_brier/mean_r": 0.820482898503542,
      "aux_brier/n_active_tok": 215.25,
      "aux_brier/n_step_records": 53.8125,
      "aux_brier/std_r": 0.18170885206200182,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.49164983164983167,
      "calib/avg_num_step_conf": 6.97265625,
      "calib/ece": 0.22038235294117642,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.99609375,
      "calib/frac_conf_gt_0.9": 0.06274509803921569,
      "calib/gap": 0.0008270707070706651,
      "calib/mean_conf": 0.5251862745098039,
      "calib/mu_c": 0.5254781818181818,
      "calib/mu_w": 0.5246511111111112,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.0492549019607843,
      "calib/std_conf": 0.22322515057857306,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2441.0,
      "completions/max_terminated_length": 2441.0,
      "completions/mean_length": 452.41015625,
      "completions/mean_terminated_length": 452.41015625,
      "completions/min_length": 65.0,
      "completions/min_terminated_length": 65.0,
      "epoch": 0.1536,
      "grad_norm": 0.024755064398050308,
      "learning_rate": 1.5555555555555558e-06,
      "loss": -0.0216,
      "num_tokens": 31294706.0,
      "reward": 1.3187178373336792,
      "reward_std": 0.226515993475914,
      "rewards/accuracy_reward_step": 0.64453125,
      "rewards/final_brier_reward_step": 0.7045587301254272,
      "rewards/format_reward_step": 0.99609375,
      "step": 144
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.076441402154572e-08,
      "aux_brier/mean_r": 0.8027276508510113,
      "aux_brier/n_active_tok": 219.0,
      "aux_brier/n_step_records": 54.75,
      "aux_brier/std_r": 0.19599402940366417,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.48295861844248944,
      "calib/avg_num_step_conf": 7.109375,
      "calib/ece": 0.19662244094488193,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.047244094488188976,
      "calib/gap": -0.009343023786249405,
      "calib/mean_conf": 0.5370783464566928,
      "calib/mu_c": 0.5334367741935485,
      "calib/mu_w": 0.5427797979797979,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.061732283464566926,
      "calib/std_conf": 0.2230492077951706,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2651.0,
      "completions/max_terminated_length": 2651.0,
      "completions/mean_length": 442.28515625,
      "completions/mean_terminated_length": 442.28515625,
      "completions/min_length": 125.0,
      "completions/min_terminated_length": 125.0,
      "epoch": 0.15466666666666667,
      "grad_norm": 0.028061792254447937,
      "learning_rate": 1.527777777777778e-06,
      "loss": 0.0462,
      "num_tokens": 31510635.0,
      "reward": 1.2733901739120483,
      "reward_std": 0.3333938717842102,
      "rewards/accuracy_reward_step": 0.60546875,
      "rewards/final_brier_reward_step": 0.6951233148574829,
      "rewards/format_reward_step": 0.98828125,
      "step": 145
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.299429776615348e-08,
      "aux_brier/mean_r": 0.8234522491693497,
      "aux_brier/n_active_tok": 214.5,
      "aux_brier/n_step_records": 53.625,
      "aux_brier/std_r": 0.18366397521458566,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5631024860476916,
      "calib/avg_num_step_conf": 6.9296875,
      "calib/ece": 0.15656746062992125,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.05905511811023622,
      "calib/gap": 0.04451860857432777,
      "calib/mean_conf": 0.5206698228346457,
      "calib/mu_c": 0.5462592592592593,
      "calib/mu_w": 0.5017406506849316,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.12602021653543305,
      "calib/std_conf": 0.22569164079354553,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1224.0,
      "completions/max_terminated_length": 1224.0,
      "completions/mean_length": 444.71484375,
      "completions/mean_terminated_length": 446.4588623046875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.15573333333333333,
      "grad_norm": 0.025134719908237457,
      "learning_rate": 1.5e-06,
      "loss": -0.0001,
      "num_tokens": 31731698.0,
      "reward": 1.0931053161621094,
      "reward_std": 0.30833274126052856,
      "rewards/accuracy_reward_step": 0.421875,
      "rewards/final_brier_reward_step": 0.708358645439148,
      "rewards/format_reward_step": 0.98828125,
      "step": 146
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.911516463841249e-08,
      "aux_brier/mean_r": 0.8122192062437534,
      "aux_brier/n_active_tok": 216.5,
      "aux_brier/n_step_records": 54.125,
      "aux_brier/std_r": 0.1872737722005695,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5760673868312757,
      "calib/avg_num_step_conf": 6.81640625,
      "calib/ece": 0.17974920634920633,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.06349206349206349,
      "calib/gap": 0.050612962962962915,
      "calib/mean_conf": 0.5747746031746032,
      "calib/mu_c": 0.6036962962962963,
      "calib/mu_w": 0.5530833333333334,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.16297619047619044,
      "calib/std_conf": 0.2062286354250713,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2591.0,
      "completions/max_terminated_length": 2591.0,
      "completions/mean_length": 460.56640625,
      "completions/mean_terminated_length": 464.1929016113281,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 114.0,
      "epoch": 0.1568,
      "grad_norm": 0.0515858419239521,
      "learning_rate": 1.4722222222222225e-06,
      "loss": 0.0014,
      "num_tokens": 31953283.0,
      "reward": 1.0902622938156128,
      "reward_std": 0.24324165284633636,
      "rewards/accuracy_reward_step": 0.421875,
      "rewards/final_brier_reward_step": 0.7047991752624512,
      "rewards/format_reward_step": 0.984375,
      "step": 147
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.7710615885245815e-09,
      "aux_brier/mean_r": 0.8038588259369135,
      "aux_brier/n_active_tok": 222.5,
      "aux_brier/n_step_records": 55.625,
      "aux_brier/std_r": 0.18959468672983348,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.481797385620915,
      "calib/avg_num_step_conf": 7.21875,
      "calib/ece": 0.1828735177865613,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.07905138339920949,
      "calib/gap": -0.013629823529411778,
      "calib/mean_conf": 0.5986284584980237,
      "calib/mu_c": 0.5932411764705883,
      "calib/mu_w": 0.606871,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08837944664031622,
      "calib/std_conf": 0.20025337798810655,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2421.0,
      "completions/max_terminated_length": 2421.0,
      "completions/mean_length": 447.41796875,
      "completions/mean_terminated_length": 449.1725769042969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 94.0,
      "epoch": 0.15786666666666666,
      "grad_norm": 0.05226176977157593,
      "learning_rate": 1.4444444444444445e-06,
      "loss": 0.0016,
      "num_tokens": 32172934.0,
      "reward": 1.2658051252365112,
      "reward_std": 0.31609511375427246,
      "rewards/accuracy_reward_step": 0.59765625,
      "rewards/final_brier_reward_step": 0.7038455605506897,
      "rewards/format_reward_step": 0.984375,
      "step": 148
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.1584773019233126e-08,
      "aux_brier/mean_r": 0.7744041997939348,
      "aux_brier/n_active_tok": 228.75,
      "aux_brier/n_step_records": 57.1875,
      "aux_brier/std_r": 0.21435644826851785,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.4474770938185572,
      "calib/avg_num_step_conf": 7.7890625,
      "calib/ece": 0.2526072289156627,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.07228915662650602,
      "calib/gap": -0.019915911730546076,
      "calib/mean_conf": 0.5614570281124499,
      "calib/mu_c": 0.5516190476190475,
      "calib/mu_w": 0.5715349593495935,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.15402008032128514,
      "calib/std_conf": 0.2302332410010566,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2652.0,
      "completions/max_terminated_length": 2652.0,
      "completions/mean_length": 523.11328125,
      "completions/mean_terminated_length": 529.3162231445312,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 111.0,
      "epoch": 0.15893333333333334,
      "grad_norm": 0.1554420441389084,
      "learning_rate": 1.4166666666666667e-06,
      "loss": 0.0156,
      "num_tokens": 32411307.0,
      "reward": 1.1419100761413574,
      "reward_std": 0.28279829025268555,
      "rewards/accuracy_reward_step": 0.4921875,
      "rewards/final_brier_reward_step": 0.6613902449607849,
      "rewards/format_reward_step": 0.96875,
      "step": 149
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.219894197439068e-09,
      "aux_brier/mean_r": 0.8038320560008287,
      "aux_brier/n_active_tok": 205.375,
      "aux_brier/n_step_records": 51.34375,
      "aux_brier/std_r": 0.19461392844095826,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5002602641681306,
      "calib/avg_num_step_conf": 6.94921875,
      "calib/ece": 0.16590600000000003,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.064,
      "calib/gap": -0.005148578306981566,
      "calib/mean_conf": 0.563454,
      "calib/mu_c": 0.561209219858156,
      "calib/mu_w": 0.5663577981651375,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.08267999999999998,
      "calib/std_conf": 0.2013730291871282,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2373.0,
      "completions/max_terminated_length": 2373.0,
      "completions/mean_length": 439.04296875,
      "completions/mean_terminated_length": 442.5,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 127.0,
      "epoch": 0.16,
      "grad_norm": 0.0734696164727211,
      "learning_rate": 1.3888888888888892e-06,
      "loss": 0.002,
      "num_tokens": 32628662.0,
      "reward": 1.2126495838165283,
      "reward_std": 0.27915123105049133,
      "rewards/accuracy_reward_step": 0.55078125,
      "rewards/final_brier_reward_step": 0.6943480968475342,
      "rewards/format_reward_step": 0.9765625,
      "step": 150
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.5514008961113888e-09,
      "aux_brier/mean_r": 0.7861856613308191,
      "aux_brier/n_active_tok": 219.0,
      "aux_brier/n_step_records": 54.75,
      "aux_brier/std_r": 0.20428610290400684,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.4747511612475116,
      "calib/avg_num_step_conf": 7.78515625,
      "calib/ece": 0.2190323886639676,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.9453125,
      "calib/frac_conf_gt_0.9": 0.05668016194331984,
      "calib/gap": -0.018078666224286777,
      "calib/mean_conf": 0.5513319838056682,
      "calib/mu_c": 0.5413045454545454,
      "calib/mu_w": 0.5593832116788322,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9765625,
      "calib/pce": 0.16251012145748986,
      "calib/std_conf": 0.21728345614212916,
      "calib/step_conf_rate": 0.9765625,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2956.0,
      "completions/max_terminated_length": 2956.0,
      "completions/mean_length": 537.99609375,
      "completions/mean_terminated_length": 540.1058959960938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.16106666666666666,
      "grad_norm": 0.11959611624479294,
      "learning_rate": 1.3611111111111112e-06,
      "loss": 0.0422,
      "num_tokens": 32873413.0,
      "reward": 1.0640509128570557,
      "reward_std": 0.3209001421928406,
      "rewards/accuracy_reward_step": 0.4296875,
      "rewards/final_brier_reward_step": 0.646828830242157,
      "rewards/format_reward_step": 0.9453125,
      "step": 151
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.702060511976633e-08,
      "aux_brier/mean_r": 0.8102346882224083,
      "aux_brier/n_active_tok": 223.125,
      "aux_brier/n_step_records": 55.78125,
      "aux_brier/std_r": 0.18743508565239608,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.431973667226023,
      "calib/avg_num_step_conf": 7.31640625,
      "calib/ece": 0.20082530120481928,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.0321285140562249,
      "calib/gap": -0.04694078352910791,
      "calib/mean_conf": 0.5519457831325301,
      "calib/mu_c": 0.5280040983606558,
      "calib/mu_w": 0.5749448818897637,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.13140562248995985,
      "calib/std_conf": 0.19334658954074496,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3001.0,
      "completions/max_terminated_length": 3001.0,
      "completions/mean_length": 466.2109375,
      "completions/mean_terminated_length": 469.88189697265625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 187.0,
      "epoch": 0.16213333333333332,
      "grad_norm": 0.10813432186841965,
      "learning_rate": 1.3333333333333334e-06,
      "loss": 0.0248,
      "num_tokens": 33098155.0,
      "reward": 1.1269245147705078,
      "reward_std": 0.3119822144508362,
      "rewards/accuracy_reward_step": 0.4765625,
      "rewards/final_brier_reward_step": 0.6639482975006104,
      "rewards/format_reward_step": 0.96875,
      "step": 152
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 6.942694269485017e-08,
      "aux_brier/mean_r": 0.8132796436548233,
      "aux_brier/n_active_tok": 226.125,
      "aux_brier/n_step_records": 56.53125,
      "aux_brier/std_r": 0.18426152970641851,
      "calib/answer_extract_rate": 0.97265625,
      "calib/auroc": 0.510104643557881,
      "calib/avg_num_step_conf": 7.45703125,
      "calib/ece": 0.15398072289156622,
      "calib/final_conf_rate": 0.97265625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.08433734939759036,
      "calib/gap": 0.004058862001308006,
      "calib/mean_conf": 0.6040112449799196,
      "calib/mu_c": 0.6058043165467626,
      "calib/mu_w": 0.6017454545454546,
      "calib/nonempty_final_conf_rate": 0.97265625,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.09987951807228912,
      "calib/std_conf": 0.19901518991606024,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1489.0,
      "completions/max_terminated_length": 1489.0,
      "completions/mean_length": 455.08984375,
      "completions/mean_terminated_length": 464.1553955078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 150.0,
      "epoch": 0.1632,
      "grad_norm": 0.14978143572807312,
      "learning_rate": 1.3055555555555556e-06,
      "loss": 0.0096,
      "num_tokens": 33321978.0,
      "reward": 1.202840805053711,
      "reward_std": 0.29197293519973755,
      "rewards/accuracy_reward_step": 0.54296875,
      "rewards/final_brier_reward_step": 0.6941753625869751,
      "rewards/format_reward_step": 0.97265625,
      "step": 153
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.0014349349773966e-08,
      "aux_brier/mean_r": 0.8017580434679985,
      "aux_brier/n_active_tok": 221.625,
      "aux_brier/n_step_records": 55.40625,
      "aux_brier/std_r": 0.19344893470406532,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.5956202563145452,
      "calib/avg_num_step_conf": 6.96875,
      "calib/ece": 0.1741133333333333,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.06666666666666667,
      "calib/gap": 0.058423236282194924,
      "calib/mean_conf": 0.6001611764705881,
      "calib/mu_c": 0.6324657894736843,
      "calib/mu_w": 0.5740425531914893,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.16360784313725485,
      "calib/std_conf": 0.19478726943994779,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2833.0,
      "completions/max_terminated_length": 2833.0,
      "completions/mean_length": 453.81640625,
      "completions/mean_terminated_length": 453.81640625,
      "completions/min_length": 160.0,
      "completions/min_terminated_length": 160.0,
      "epoch": 0.16426666666666667,
      "grad_norm": 0.03800573572516441,
      "learning_rate": 1.2777777777777779e-06,
      "loss": 0.0285,
      "num_tokens": 33542595.0,
      "reward": 1.1202408075332642,
      "reward_std": 0.33582866191864014,
      "rewards/accuracy_reward_step": 0.4453125,
      "rewards/final_brier_reward_step": 0.7153380513191223,
      "rewards/format_reward_step": 0.9921875,
      "step": 154
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -4.634872838304993e-08,
      "aux_brier/mean_r": 0.8097844570875168,
      "aux_brier/n_active_tok": 206.0,
      "aux_brier/n_step_records": 51.5,
      "aux_brier/std_r": 0.17847078037448227,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.5928487978847691,
      "calib/avg_num_step_conf": 6.4921875,
      "calib/ece": 0.17345703125000003,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.078125,
      "calib/gap": 0.05238203283527032,
      "calib/mean_conf": 0.58669921875,
      "calib/mu_c": 0.6151410256410257,
      "calib/mu_w": 0.5627589928057554,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.15156250000000002,
      "calib/std_conf": 0.18830996243564666,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 924.0,
      "completions/max_terminated_length": 924.0,
      "completions/mean_length": 404.1875,
      "completions/mean_terminated_length": 405.7725830078125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 104.0,
      "epoch": 0.16533333333333333,
      "grad_norm": 0.062165211886167526,
      "learning_rate": 1.25e-06,
      "loss": 0.0342,
      "num_tokens": 33753283.0,
      "reward": 1.1384236812591553,
      "reward_std": 0.24906030297279358,
      "rewards/accuracy_reward_step": 0.45703125,
      "rewards/final_brier_reward_step": 0.725569486618042,
      "rewards/format_reward_step": 1.0,
      "step": 155
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.007677693873557e-08,
      "aux_brier/mean_r": 0.7784804869443178,
      "aux_brier/n_active_tok": 219.875,
      "aux_brier/n_step_records": 54.96875,
      "aux_brier/std_r": 0.2090508914552629,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5300525920360631,
      "calib/avg_num_step_conf": 7.1171875,
      "calib/ece": 0.19717707509881421,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.06324110671936758,
      "calib/gap": 0.02168250688705231,
      "calib/mean_conf": 0.5817952569169962,
      "calib/mu_c": 0.5921651515151515,
      "calib/mu_w": 0.5704826446280992,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.12861660079051385,
      "calib/std_conf": 0.2393334609098778,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3069.0,
      "completions/max_terminated_length": 3069.0,
      "completions/mean_length": 459.0859375,
      "completions/mean_terminated_length": 460.88629150390625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 119.0,
      "epoch": 0.1664,
      "grad_norm": 0.01808425970375538,
      "learning_rate": 1.2222222222222223e-06,
      "loss": -0.0063,
      "num_tokens": 33975569.0,
      "reward": 1.1721856594085693,
      "reward_std": 0.29748740792274475,
      "rewards/accuracy_reward_step": 0.515625,
      "rewards/final_brier_reward_step": 0.6809302568435669,
      "rewards/format_reward_step": 0.97265625,
      "step": 156
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.992033432125396e-08,
      "aux_brier/mean_r": 0.7912592738866806,
      "aux_brier/n_active_tok": 223.25,
      "aux_brier/n_step_records": 55.8125,
      "aux_brier/std_r": 0.19565992313437164,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5183146766169154,
      "calib/avg_num_step_conf": 7.21484375,
      "calib/ece": 0.179605905511811,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.07480314960629922,
      "calib/gap": 0.010377922885572155,
      "calib/mean_conf": 0.5900791338582678,
      "calib/mu_c": 0.5949820895522389,
      "calib/mu_w": 0.5846041666666667,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12106299212598423,
      "calib/std_conf": 0.20903415423195865,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1496.0,
      "completions/max_terminated_length": 1496.0,
      "completions/mean_length": 432.34765625,
      "completions/mean_terminated_length": 434.04315185546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 90.0,
      "epoch": 0.16746666666666668,
      "grad_norm": 0.07360821962356567,
      "learning_rate": 1.1944444444444446e-06,
      "loss": 0.0128,
      "num_tokens": 34189978.0,
      "reward": 1.1952300071716309,
      "reward_std": 0.28814953565597534,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.7027948498725891,
      "rewards/format_reward_step": 0.9921875,
      "step": 157
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.6169309630684765e-08,
      "aux_brier/mean_r": 0.7900642603635788,
      "aux_brier/n_active_tok": 215.375,
      "aux_brier/n_step_records": 53.84375,
      "aux_brier/std_r": 0.20204701973125339,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.395147868390731,
      "calib/avg_num_step_conf": 7.2421875,
      "calib/ece": 0.25689288537549415,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.08300395256916997,
      "calib/gap": -0.06423064268339507,
      "calib/mean_conf": 0.6048462450592885,
      "calib/mu_c": 0.5776815068493151,
      "calib/mu_w": 0.6419121495327101,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1423320158102767,
      "calib/std_conf": 0.2096806007803486,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2807.0,
      "completions/max_terminated_length": 2807.0,
      "completions/mean_length": 457.0625,
      "completions/mean_terminated_length": 458.85491943359375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 92.0,
      "epoch": 0.16853333333333334,
      "grad_norm": 0.030399853363633156,
      "learning_rate": 1.1666666666666668e-06,
      "loss": 0.026,
      "num_tokens": 34412226.0,
      "reward": 1.2299730777740479,
      "reward_std": 0.31046023964881897,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6698920130729675,
      "rewards/format_reward_step": 0.984375,
      "step": 158
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 9.161260805523597e-09,
      "aux_brier/mean_r": 0.8144844230264425,
      "aux_brier/n_active_tok": 200.875,
      "aux_brier/n_step_records": 50.21875,
      "aux_brier/std_r": 0.18931242334656417,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.5227889478228691,
      "calib/avg_num_step_conf": 6.46484375,
      "calib/ece": 0.15896627450980394,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.054901960784313725,
      "calib/gap": 0.015614999383249062,
      "calib/mean_conf": 0.532327843137255,
      "calib/mu_c": 0.5397373134328359,
      "calib/mu_w": 0.5241223140495869,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.08290196078431374,
      "calib/std_conf": 0.21394798444549348,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2738.0,
      "completions/max_terminated_length": 2738.0,
      "completions/mean_length": 418.04296875,
      "completions/mean_terminated_length": 418.04296875,
      "completions/min_length": 150.0,
      "completions/min_terminated_length": 150.0,
      "epoch": 0.1696,
      "grad_norm": 0.016105517745018005,
      "learning_rate": 1.138888888888889e-06,
      "loss": 0.0391,
      "num_tokens": 34624029.0,
      "reward": 1.1962106227874756,
      "reward_std": 0.2835184335708618,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.7067175507545471,
      "rewards/format_reward_step": 0.9921875,
      "step": 159
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.0736986780511835e-08,
      "aux_brier/mean_r": 0.7846236266195774,
      "aux_brier/n_active_tok": 218.75,
      "aux_brier/n_step_records": 54.6875,
      "aux_brier/std_r": 0.19788970216177404,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5085429369918699,
      "calib/avg_num_step_conf": 7.390625,
      "calib/ece": 0.17402191235059758,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.05179282868525897,
      "calib/gap": 0.010204363567073216,
      "calib/mean_conf": 0.5824721115537849,
      "calib/mu_c": 0.5874726562500001,
      "calib/mu_w": 0.5772682926829269,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.12326693227091633,
      "calib/std_conf": 0.21788352461333965,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2803.0,
      "completions/max_terminated_length": 2803.0,
      "completions/mean_length": 445.61328125,
      "completions/mean_terminated_length": 449.1220397949219,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 146.0,
      "epoch": 0.17066666666666666,
      "grad_norm": 0.070772185921669,
      "learning_rate": 1.111111111111111e-06,
      "loss": 0.0123,
      "num_tokens": 34842946.0,
      "reward": 1.166327714920044,
      "reward_std": 0.3040813207626343,
      "rewards/accuracy_reward_step": 0.50390625,
      "rewards/final_brier_reward_step": 0.6887481212615967,
      "rewards/format_reward_step": 0.98046875,
      "step": 160
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.201992906740283e-08,
      "aux_brier/mean_r": 0.800326943397522,
      "aux_brier/n_active_tok": 197.875,
      "aux_brier/n_step_records": 49.46875,
      "aux_brier/std_r": 0.18986242730170488,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.5327109673625404,
      "calib/avg_num_step_conf": 6.33203125,
      "calib/ece": 0.18333529411764704,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.050980392156862744,
      "calib/gap": 0.029023228462216943,
      "calib/mean_conf": 0.568978431372549,
      "calib/mu_c": 0.5776284916201118,
      "calib/mu_w": 0.5486052631578948,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.02517647058823529,
      "calib/std_conf": 0.20709775234171518,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1354.0,
      "completions/max_terminated_length": 1354.0,
      "completions/mean_length": 413.76953125,
      "completions/mean_terminated_length": 415.3921813964844,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 122.0,
      "epoch": 0.17173333333333332,
      "grad_norm": 0.1003478467464447,
      "learning_rate": 1.0833333333333335e-06,
      "loss": 0.0251,
      "num_tokens": 35052791.0,
      "reward": 1.3792005777359009,
      "reward_std": 0.24497488141059875,
      "rewards/accuracy_reward_step": 0.69921875,
      "rewards/final_brier_reward_step": 0.7355523109436035,
      "rewards/format_reward_step": 0.9921875,
      "step": 161
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.6508984868334196e-09,
      "aux_brier/mean_r": 0.7857399228960276,
      "aux_brier/n_active_tok": 209.0,
      "aux_brier/n_step_records": 52.25,
      "aux_brier/std_r": 0.20222072442993522,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.4963493959909731,
      "calib/avg_num_step_conf": 6.61328125,
      "calib/ece": 0.16089294117647057,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.06666666666666667,
      "calib/gap": 0.0019580446037434918,
      "calib/mean_conf": 0.5973815686274511,
      "calib/mu_c": 0.5980956790123457,
      "calib/mu_w": 0.5961376344086022,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.06149019607843135,
      "calib/std_conf": 0.20633723836184856,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1275.0,
      "completions/max_terminated_length": 1275.0,
      "completions/mean_length": 407.96484375,
      "completions/mean_terminated_length": 409.5647277832031,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 131.0,
      "epoch": 0.1728,
      "grad_norm": 0.02969902940094471,
      "learning_rate": 1.0555555555555557e-06,
      "loss": -0.0291,
      "num_tokens": 35261374.0,
      "reward": 1.3031649589538574,
      "reward_std": 0.3088793158531189,
      "rewards/accuracy_reward_step": 0.6328125,
      "rewards/final_brier_reward_step": 0.7126595377922058,
      "rewards/format_reward_step": 0.984375,
      "step": 162
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.6231071363624903e-08,
      "aux_brier/mean_r": 0.7842877190560102,
      "aux_brier/n_active_tok": 216.75,
      "aux_brier/n_step_records": 54.1875,
      "aux_brier/std_r": 0.19400051701813936,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.4706771653543307,
      "calib/avg_num_step_conf": 6.93359375,
      "calib/ece": 0.17219166666666663,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.09523809523809523,
      "calib/gap": -0.023430834645669307,
      "calib/mean_conf": 0.6063003968253968,
      "calib/mu_c": 0.5944919999999999,
      "calib/mu_w": 0.6179228346456692,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.14123015873015868,
      "calib/std_conf": 0.20429980935629172,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2313.0,
      "completions/max_terminated_length": 2313.0,
      "completions/mean_length": 458.52734375,
      "completions/mean_terminated_length": 460.3255310058594,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 113.0,
      "epoch": 0.17386666666666667,
      "grad_norm": 0.03033628687262535,
      "learning_rate": 1.0277777777777777e-06,
      "loss": 0.0218,
      "num_tokens": 35483589.0,
      "reward": 1.1464180946350098,
      "reward_std": 0.3199278712272644,
      "rewards/accuracy_reward_step": 0.48828125,
      "rewards/final_brier_reward_step": 0.67160964012146,
      "rewards/format_reward_step": 0.98046875,
      "step": 163
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 1.783250752618315e-08,
      "aux_brier/mean_r": 0.7870215680450201,
      "aux_brier/n_active_tok": 228.625,
      "aux_brier/n_step_records": 57.15625,
      "aux_brier/std_r": 0.1935525459703058,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5013118440779609,
      "calib/avg_num_step_conf": 7.4453125,
      "calib/ece": 0.14960866141732285,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.05511811023622047,
      "calib/gap": 0.0026458020989506004,
      "calib/mean_conf": 0.5933047244094489,
      "calib/mu_c": 0.5945130434782608,
      "calib/mu_w": 0.5918672413793102,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.09980314960629923,
      "calib/std_conf": 0.20247135399922225,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2803.0,
      "completions/max_terminated_length": 2803.0,
      "completions/mean_length": 497.72265625,
      "completions/mean_terminated_length": 499.6745300292969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 121.0,
      "epoch": 0.17493333333333333,
      "grad_norm": 0.11999011784791946,
      "learning_rate": 1.0000000000000002e-06,
      "loss": 0.0088,
      "num_tokens": 35717142.0,
      "reward": 1.208479642868042,
      "reward_std": 0.28109174966812134,
      "rewards/accuracy_reward_step": 0.5390625,
      "rewards/final_brier_reward_step": 0.7011061906814575,
      "rewards/format_reward_step": 0.98828125,
      "step": 164
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.5884240820173154e-08,
      "aux_brier/mean_r": 0.7734075896441936,
      "aux_brier/n_active_tok": 228.875,
      "aux_brier/n_step_records": 57.21875,
      "aux_brier/std_r": 0.21174128586426377,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.48524795982423097,
      "calib/avg_num_step_conf": 7.46875,
      "calib/ece": 0.23406916996047436,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.1225296442687747,
      "calib/gap": -0.004246170747018163,
      "calib/mean_conf": 0.6206343873517788,
      "calib/mu_c": 0.6183686440677968,
      "calib/mu_w": 0.622614814814815,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.19415019762845856,
      "calib/std_conf": 0.22241601702167657,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2555.0,
      "completions/max_terminated_length": 2555.0,
      "completions/mean_length": 465.515625,
      "completions/mean_terminated_length": 469.18109130859375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 140.0,
      "epoch": 0.176,
      "grad_norm": 0.1529640108346939,
      "learning_rate": 9.722222222222224e-07,
      "loss": 0.0211,
      "num_tokens": 35941890.0,
      "reward": 1.1220381259918213,
      "reward_std": 0.26641708612442017,
      "rewards/accuracy_reward_step": 0.4609375,
      "rewards/final_brier_reward_step": 0.6678400039672852,
      "rewards/format_reward_step": 0.98828125,
      "step": 165
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.7232448110713356e-08,
      "aux_brier/mean_r": 0.7676201425492764,
      "aux_brier/n_active_tok": 224.0,
      "aux_brier/n_step_records": 56.0,
      "aux_brier/std_r": 0.21411156747490168,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5302083333333334,
      "calib/avg_num_step_conf": 7.3984375,
      "calib/ece": 0.14111832669322705,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.11155378486055777,
      "calib/gap": 0.013518366935483694,
      "calib/mean_conf": 0.6099573705179282,
      "calib/mu_c": 0.6151277419354838,
      "calib/mu_w": 0.6016093750000001,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.06677290836653382,
      "calib/std_conf": 0.2152603485661002,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2448.0,
      "completions/max_terminated_length": 2448.0,
      "completions/mean_length": 520.79296875,
      "completions/mean_terminated_length": 522.8353271484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 154.0,
      "epoch": 0.17706666666666668,
      "grad_norm": 0.06923183798789978,
      "learning_rate": 9.444444444444445e-07,
      "loss": 0.0105,
      "num_tokens": 36181397.0,
      "reward": 1.270603060722351,
      "reward_std": 0.2961066961288452,
      "rewards/accuracy_reward_step": 0.60546875,
      "rewards/final_brier_reward_step": 0.7074119448661804,
      "rewards/format_reward_step": 0.9765625,
      "step": 166
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.094241307095459e-09,
      "aux_brier/mean_r": 0.7531142849475145,
      "aux_brier/n_active_tok": 224.375,
      "aux_brier/n_step_records": 56.09375,
      "aux_brier/std_r": 0.22563402936793864,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5855592105263157,
      "calib/avg_num_step_conf": 7.4453125,
      "calib/ece": 0.1635956349206349,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.15079365079365079,
      "calib/gap": 0.05671534210526319,
      "calib/mean_conf": 0.610293253968254,
      "calib/mu_c": 0.6327993421052632,
      "calib/mu_w": 0.576084,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.08535714285714285,
      "calib/std_conf": 0.25165342348104297,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 0.0,
      "clip_ratio/low_mean": 0.0,
      "clip_ratio/low_min": 0.0,
      "clip_ratio/region_mean": 0.0,
      "completions/clipped_ratio": 0.0078125,
      "completions/max_length": 1343.0,
      "completions/max_terminated_length": 1343.0,
      "completions/mean_length": 471.265625,
      "completions/mean_terminated_length": 474.97637939453125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 155.0,
      "epoch": 0.17813333333333334,
      "grad_norm": 0.05178198963403702,
      "learning_rate": 9.166666666666666e-07,
      "loss": -0.0163,
      "num_tokens": 36407649.0,
      "reward": 1.2556668519973755,
      "reward_std": 0.32814353704452515,
      "rewards/accuracy_reward_step": 0.59375,
      "rewards/final_brier_reward_step": 0.7023549675941467,
      "rewards/format_reward_step": 0.97265625,
      "step": 167
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -8.276514482097319e-08,
      "aux_brier/mean_r": 0.7753052655607462,
      "aux_brier/n_active_tok": 220.75,
      "aux_brier/n_step_records": 55.1875,
      "aux_brier/std_r": 0.20657177781686187,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.4692128316632004,
      "calib/avg_num_step_conf": 7.04296875,
      "calib/ece": 0.2539781746031746,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.09523809523809523,
      "calib/gap": -0.01952363395726997,
      "calib/mean_conf": 0.6211805555555555,
      "calib/mu_c": 0.6116511627906976,
      "calib/mu_w": 0.6311747967479676,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1816269841269841,
      "calib/std_conf": 0.21944376412643188,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2588.0,
      "completions/max_terminated_length": 2588.0,
      "completions/mean_length": 520.62890625,
      "completions/mean_terminated_length": 522.6705932617188,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 148.0,
      "epoch": 0.1792,
      "grad_norm": 0.05993538722395897,
      "learning_rate": 8.88888888888889e-07,
      "loss": 0.0317,
      "num_tokens": 36645602.0,
      "reward": 1.1635085344314575,
      "reward_std": 0.375709593296051,
      "rewards/accuracy_reward_step": 0.50390625,
      "rewards/final_brier_reward_step": 0.6696591377258301,
      "rewards/format_reward_step": 0.984375,
      "step": 168
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -7.985440191732307e-08,
      "aux_brier/mean_r": 0.7509642671793699,
      "aux_brier/n_active_tok": 217.25,
      "aux_brier/n_step_records": 54.3125,
      "aux_brier/std_r": 0.22496945364400744,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5504725897920605,
      "calib/avg_num_step_conf": 7.07421875,
      "calib/ece": 0.17495652173913037,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.07905138339920949,
      "calib/gap": 0.022868985507246364,
      "calib/mean_conf": 0.5849644268774704,
      "calib/mu_c": 0.595359420289855,
      "calib/mu_w": 0.5724904347826086,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.10723320158102762,
      "calib/std_conf": 0.23726811804638298,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1772.0,
      "completions/max_terminated_length": 1772.0,
      "completions/mean_length": 487.30859375,
      "completions/mean_terminated_length": 487.30859375,
      "completions/min_length": 112.0,
      "completions/min_terminated_length": 112.0,
      "epoch": 0.18026666666666666,
      "grad_norm": 0.018199941143393517,
      "learning_rate": 8.611111111111112e-07,
      "loss": 0.0355,
      "num_tokens": 36874537.0,
      "reward": 1.2030644416809082,
      "reward_std": 0.3347136378288269,
      "rewards/accuracy_reward_step": 0.5390625,
      "rewards/final_brier_reward_step": 0.6950702667236328,
      "rewards/format_reward_step": 0.98046875,
      "step": 169
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 8.991255279200416e-08,
      "aux_brier/mean_r": 0.7418857179582119,
      "aux_brier/n_active_tok": 230.0,
      "aux_brier/n_step_records": 57.5,
      "aux_brier/std_r": 0.2211719653569162,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5159357470678225,
      "calib/avg_num_step_conf": 7.296875,
      "calib/ece": 0.2000043307086615,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.14173228346456693,
      "calib/gap": 0.008210211626721087,
      "calib/mean_conf": 0.6777122047244095,
      "calib/mu_c": 0.6811385135135136,
      "calib/mu_w": 0.6729283018867925,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.984375,
      "calib/pce": 0.14751968503937016,
      "calib/std_conf": 0.20372021759209152,
      "calib/step_conf_rate": 0.984375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1307.0,
      "completions/max_terminated_length": 1307.0,
      "completions/mean_length": 461.44921875,
      "completions/mean_terminated_length": 463.25885009765625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 151.0,
      "epoch": 0.18133333333333335,
      "grad_norm": 0.11392112821340561,
      "learning_rate": 8.333333333333333e-07,
      "loss": -0.0117,
      "num_tokens": 37096820.0,
      "reward": 1.2417938709259033,
      "reward_std": 0.3122642934322357,
      "rewards/accuracy_reward_step": 0.578125,
      "rewards/final_brier_reward_step": 0.6937382221221924,
      "rewards/format_reward_step": 0.98046875,
      "step": 170
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.966358712112708e-08,
      "aux_brier/mean_r": 0.7842296324670315,
      "aux_brier/n_active_tok": 209.5,
      "aux_brier/n_step_records": 52.375,
      "aux_brier/std_r": 0.20828879345208406,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5413246268656716,
      "calib/avg_num_step_conf": 6.83984375,
      "calib/ece": 0.1887838582677165,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.06299212598425197,
      "calib/gap": 0.04228944029850745,
      "calib/mean_conf": 0.5617673228346456,
      "calib/mu_c": 0.5840775,
      "calib/mu_w": 0.5417880597014926,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1390551181102362,
      "calib/std_conf": 0.22709650987706656,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2554.0,
      "completions/max_terminated_length": 2554.0,
      "completions/mean_length": 453.734375,
      "completions/mean_terminated_length": 455.5137634277344,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 89.0,
      "epoch": 0.1824,
      "grad_norm": 0.013566054403781891,
      "learning_rate": 8.055555555555557e-07,
      "loss": 0.0654,
      "num_tokens": 37319872.0,
      "reward": 1.1388390064239502,
      "reward_std": 0.37724727392196655,
      "rewards/accuracy_reward_step": 0.46875,
      "rewards/final_brier_reward_step": 0.7037935256958008,
      "rewards/format_reward_step": 0.98828125,
      "step": 171
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -8.25886247030061e-08,
      "aux_brier/mean_r": 0.778722807765007,
      "aux_brier/n_active_tok": 213.25,
      "aux_brier/n_step_records": 53.3125,
      "aux_brier/std_r": 0.1989540159702301,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.4723503780597206,
      "calib/avg_num_step_conf": 6.95703125,
      "calib/ece": 0.19872078431372553,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.09019607843137255,
      "calib/gap": -0.02283692810457516,
      "calib/mean_conf": 0.6222203921568628,
      "calib/mu_c": 0.6130856209150326,
      "calib/mu_w": 0.6359225490196078,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11047058823529413,
      "calib/std_conf": 0.2067231369461721,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2433.0,
      "completions/max_terminated_length": 2433.0,
      "completions/mean_length": 448.63671875,
      "completions/mean_terminated_length": 448.63671875,
      "completions/min_length": 138.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.18346666666666667,
      "grad_norm": 0.013927267864346504,
      "learning_rate": 7.777777777777779e-07,
      "loss": 0.0018,
      "num_tokens": 37538075.0,
      "reward": 1.2694181203842163,
      "reward_std": 0.33193492889404297,
      "rewards/accuracy_reward_step": 0.59765625,
      "rewards/final_brier_reward_step": 0.7026721239089966,
      "rewards/format_reward_step": 0.9921875,
      "step": 172
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.045277327002772e-08,
      "aux_brier/mean_r": 0.7391694243997335,
      "aux_brier/n_active_tok": 234.0,
      "aux_brier/n_step_records": 58.5,
      "aux_brier/std_r": 0.23577190446667373,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5733346193415638,
      "calib/avg_num_step_conf": 7.71484375,
      "calib/ece": 0.17457857142857142,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.15873015873015872,
      "calib/gap": 0.0566409722222222,
      "calib/mean_conf": 0.6194690476190476,
      "calib/mu_c": 0.6437437500000001,
      "calib/mu_w": 0.5871027777777779,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11130952380952379,
      "calib/std_conf": 0.23711986854294417,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2601.0,
      "completions/max_terminated_length": 2601.0,
      "completions/mean_length": 495.359375,
      "completions/mean_terminated_length": 497.302001953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 110.0,
      "epoch": 0.18453333333333333,
      "grad_norm": 0.04449158161878586,
      "learning_rate": 7.5e-07,
      "loss": -0.0038,
      "num_tokens": 37768047.0,
      "reward": 1.2300063371658325,
      "reward_std": 0.319230318069458,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.7090877294540405,
      "rewards/format_reward_step": 0.98046875,
      "step": 173
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.6862715496456246e-08,
      "aux_brier/mean_r": 0.7858659084886312,
      "aux_brier/n_active_tok": 226.875,
      "aux_brier/n_step_records": 56.71875,
      "aux_brier/std_r": 0.19160686992108822,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5840610687022901,
      "calib/avg_num_step_conf": 7.35546875,
      "calib/ece": 0.17277539062500002,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.05859375,
      "calib/gap": 0.05416923969465637,
      "calib/mean_conf": 0.607458984375,
      "calib/mu_c": 0.6351783999999999,
      "calib/mu_w": 0.5810091603053436,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.14597656250000002,
      "calib/std_conf": 0.20693023619968157,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1911.0,
      "completions/max_terminated_length": 1911.0,
      "completions/mean_length": 490.5234375,
      "completions/mean_terminated_length": 492.44708251953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 155.0,
      "epoch": 0.1856,
      "grad_norm": 0.02974775806069374,
      "learning_rate": 7.222222222222222e-07,
      "loss": 0.0056,
      "num_tokens": 37997853.0,
      "reward": 1.162726640701294,
      "reward_std": 0.3572935461997986,
      "rewards/accuracy_reward_step": 0.48828125,
      "rewards/final_brier_reward_step": 0.7134063243865967,
      "rewards/format_reward_step": 0.9921875,
      "step": 174
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.6371787203706134e-08,
      "aux_brier/mean_r": 0.7583804409950972,
      "aux_brier/n_active_tok": 234.0,
      "aux_brier/n_step_records": 58.5,
      "aux_brier/std_r": 0.22469458472914994,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.4853681774874098,
      "calib/avg_num_step_conf": 7.6796875,
      "calib/ece": 0.2699258964143427,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.10756972111553785,
      "calib/gap": 0.008985626786443435,
      "calib/mean_conf": 0.5952533864541832,
      "calib/mu_c": 0.6009096774193549,
      "calib/mu_w": 0.5919240506329114,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24733067729083671,
      "calib/std_conf": 0.24697494368307288,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2452.0,
      "completions/max_terminated_length": 2452.0,
      "completions/mean_length": 492.9609375,
      "completions/mean_terminated_length": 498.80633544921875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 150.0,
      "epoch": 0.18666666666666668,
      "grad_norm": 0.3764019310474396,
      "learning_rate": 6.944444444444446e-07,
      "loss": -0.0021,
      "num_tokens": 38229875.0,
      "reward": 1.0151591300964355,
      "reward_std": 0.31759583950042725,
      "rewards/accuracy_reward_step": 0.36328125,
      "rewards/final_brier_reward_step": 0.6465744972229004,
      "rewards/format_reward_step": 0.98046875,
      "step": 175
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 5.330966579286667e-08,
      "aux_brier/mean_r": 0.7834791298955679,
      "aux_brier/n_active_tok": 223.875,
      "aux_brier/n_step_records": 55.96875,
      "aux_brier/std_r": 0.19350098632276058,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5318801613212983,
      "calib/avg_num_step_conf": 7.0703125,
      "calib/ece": 0.16035680000000005,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.072,
      "calib/gap": 0.01819047436143617,
      "calib/mean_conf": 0.6228432,
      "calib/mu_c": 0.6317929133858265,
      "calib/mu_w": 0.6136024390243904,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.13760000000000006,
      "calib/std_conf": 0.18823001815268467,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2496.0,
      "completions/max_terminated_length": 2496.0,
      "completions/mean_length": 476.21484375,
      "completions/mean_terminated_length": 479.9645690917969,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 138.0,
      "epoch": 0.18773333333333334,
      "grad_norm": 0.04411868751049042,
      "learning_rate": 6.666666666666667e-07,
      "loss": 0.0043,
      "num_tokens": 38455850.0,
      "reward": 1.1590666770935059,
      "reward_std": 0.3019905388355255,
      "rewards/accuracy_reward_step": 0.5,
      "rewards/final_brier_reward_step": 0.6909544467926025,
      "rewards/format_reward_step": 0.97265625,
      "step": 176
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.4630474715348072e-08,
      "aux_brier/mean_r": 0.7609710041433573,
      "aux_brier/n_active_tok": 229.375,
      "aux_brier/n_step_records": 57.34375,
      "aux_brier/std_r": 0.21466213348321617,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5135368663594471,
      "calib/avg_num_step_conf": 7.90234375,
      "calib/ece": 0.17280240000000002,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.128,
      "calib/gap": 0.018543945212493473,
      "calib/mean_conf": 0.6188776,
      "calib/mu_c": 0.6280753968253968,
      "calib/mu_w": 0.6095314516129033,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.14384000000000002,
      "calib/std_conf": 0.23478227799014134,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2973.0,
      "completions/max_terminated_length": 2973.0,
      "completions/mean_length": 484.671875,
      "completions/mean_terminated_length": 492.3651123046875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 130.0,
      "epoch": 0.1888,
      "grad_norm": 0.10568588227033615,
      "learning_rate": 6.388888888888889e-07,
      "loss": 0.0012,
      "num_tokens": 38683758.0,
      "reward": 1.1491620540618896,
      "reward_std": 0.34768766164779663,
      "rewards/accuracy_reward_step": 0.4921875,
      "rewards/final_brier_reward_step": 0.6747732758522034,
      "rewards/format_reward_step": 0.9765625,
      "step": 177
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -5.234977790147255e-08,
      "aux_brier/mean_r": 0.7877587005496025,
      "aux_brier/n_active_tok": 218.0,
      "aux_brier/n_step_records": 54.5,
      "aux_brier/std_r": 0.1997723402455449,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5308825429642073,
      "calib/avg_num_step_conf": 7.171875,
      "calib/ece": 0.1564474103585657,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.09561752988047809,
      "calib/gap": 0.01923988241374852,
      "calib/mean_conf": 0.613273705179283,
      "calib/mu_c": 0.6216288732394366,
      "calib/mu_w": 0.6023889908256881,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10199203187250996,
      "calib/std_conf": 0.21049897811312543,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3061.0,
      "completions/max_terminated_length": 3061.0,
      "completions/mean_length": 473.12890625,
      "completions/mean_terminated_length": 476.8543395996094,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 123.0,
      "epoch": 0.18986666666666666,
      "grad_norm": 0.03318691626191139,
      "learning_rate": 6.111111111111112e-07,
      "loss": 0.0027,
      "num_tokens": 38910951.0,
      "reward": 1.2207213640213013,
      "reward_std": 0.37348127365112305,
      "rewards/accuracy_reward_step": 0.5546875,
      "rewards/final_brier_reward_step": 0.7031975984573364,
      "rewards/format_reward_step": 0.98046875,
      "step": 178
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.3590677550311412e-08,
      "aux_brier/mean_r": 0.7528609000146389,
      "aux_brier/n_active_tok": 234.875,
      "aux_brier/n_step_records": 58.71875,
      "aux_brier/std_r": 0.21921817678958178,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5900310117586252,
      "calib/avg_num_step_conf": 7.60546875,
      "calib/ece": 0.14260239043824702,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.11155378486055777,
      "calib/gap": 0.0482013503036568,
      "calib/mean_conf": 0.6175968127490039,
      "calib/mu_c": 0.6385288732394366,
      "calib/mu_w": 0.5903275229357798,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.09723107569721118,
      "calib/std_conf": 0.2299492597543088,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1997.0,
      "completions/max_terminated_length": 1997.0,
      "completions/mean_length": 471.09375,
      "completions/mean_terminated_length": 476.67987060546875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 150.0,
      "epoch": 0.19093333333333334,
      "grad_norm": 0.07718180119991302,
      "learning_rate": 5.833333333333334e-07,
      "loss": -0.0319,
      "num_tokens": 39137815.0,
      "reward": 1.2231450080871582,
      "reward_std": 0.3644871115684509,
      "rewards/accuracy_reward_step": 0.55859375,
      "rewards/final_brier_reward_step": 0.7050795555114746,
      "rewards/format_reward_step": 0.9765625,
      "step": 179
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.4701016515239473e-09,
      "aux_brier/mean_r": 0.7343279607594013,
      "aux_brier/n_active_tok": 247.25,
      "aux_brier/n_step_records": 61.8125,
      "aux_brier/std_r": 0.23598173074424267,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.5294979612640162,
      "calib/avg_num_step_conf": 8.06640625,
      "calib/ece": 0.2123094861660079,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.18181818181818182,
      "calib/gap": 0.015871839959225253,
      "calib/mean_conf": 0.6691924901185771,
      "calib/mu_c": 0.6760305555555556,
      "calib/mu_w": 0.6601587155963303,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.15616600790513835,
      "calib/std_conf": 0.23946849360566766,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2989.0,
      "completions/max_terminated_length": 2989.0,
      "completions/mean_length": 541.54296875,
      "completions/mean_terminated_length": 545.8070678710938,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 148.0,
      "epoch": 0.192,
      "grad_norm": 0.13195045292377472,
      "learning_rate": 5.555555555555555e-07,
      "loss": 0.0467,
      "num_tokens": 39380306.0,
      "reward": 1.2197463512420654,
      "reward_std": 0.3591662645339966,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.6758601665496826,
      "rewards/format_reward_step": 0.9765625,
      "step": 180
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.2559889331641436e-08,
      "aux_brier/mean_r": 0.7783240135759115,
      "aux_brier/n_active_tok": 219.5,
      "aux_brier/n_step_records": 54.875,
      "aux_brier/std_r": 0.19999481202103198,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.524643928035982,
      "calib/avg_num_step_conf": 7.24609375,
      "calib/ece": 0.21105984251968501,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.08267716535433071,
      "calib/gap": 0.0017970639680159328,
      "calib/mean_conf": 0.6065779527559055,
      "calib/mu_c": 0.6075543103448277,
      "calib/mu_w": 0.6057572463768117,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.18047244094488185,
      "calib/std_conf": 0.1986199710496047,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2662.0,
      "completions/max_terminated_length": 2662.0,
      "completions/mean_length": 449.2890625,
      "completions/mean_terminated_length": 451.0509948730469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 160.0,
      "epoch": 0.19306666666666666,
      "grad_norm": 0.06607302278280258,
      "learning_rate": 5.277777777777779e-07,
      "loss": 0.0726,
      "num_tokens": 39601588.0,
      "reward": 1.1180652379989624,
      "reward_std": 0.31925728917121887,
      "rewards/accuracy_reward_step": 0.453125,
      "rewards/final_brier_reward_step": 0.6831987500190735,
      "rewards/format_reward_step": 0.98828125,
      "step": 181
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.7544570075961374e-09,
      "aux_brier/mean_r": 0.7577127385884523,
      "aux_brier/n_active_tok": 231.125,
      "aux_brier/n_step_records": 57.78125,
      "aux_brier/std_r": 0.22663994459435344,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5823636136136136,
      "calib/avg_num_step_conf": 7.43359375,
      "calib/ece": 0.14662000000000003,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.15294117647058825,
      "calib/gap": 0.06783564189189206,
      "calib/mean_conf": 0.633615294117647,
      "calib/mu_c": 0.6631437500000001,
      "calib/mu_w": 0.5953081081081081,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10776470588235296,
      "calib/std_conf": 0.22659795086169118,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3013.0,
      "completions/max_terminated_length": 3013.0,
      "completions/mean_length": 497.4765625,
      "completions/mean_terminated_length": 497.4765625,
      "completions/min_length": 167.0,
      "completions/min_terminated_length": 167.0,
      "epoch": 0.19413333333333332,
      "grad_norm": 0.009706188924610615,
      "learning_rate": 5.000000000000001e-07,
      "loss": 0.0083,
      "num_tokens": 39835102.0,
      "reward": 1.2398512363433838,
      "reward_std": 0.2927827835083008,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.7250296473503113,
      "rewards/format_reward_step": 0.9921875,
      "step": 182
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -9.149970170430066e-09,
      "aux_brier/mean_r": 0.7672966662794352,
      "aux_brier/n_active_tok": 216.75,
      "aux_brier/n_step_records": 54.1875,
      "aux_brier/std_r": 0.20986755122430623,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.5678553427419355,
      "calib/avg_num_step_conf": 6.921875,
      "calib/ece": 0.14822738095238097,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.0873015873015873,
      "calib/gap": 0.05744493447580645,
      "calib/mean_conf": 0.5914154761904762,
      "calib/mu_c": 0.61968203125,
      "calib/mu_w": 0.5622370967741935,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1158531746031746,
      "calib/std_conf": 0.22001805093373136,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2938.0,
      "completions/max_terminated_length": 2938.0,
      "completions/mean_length": 514.3203125,
      "completions/mean_terminated_length": 514.3203125,
      "completions/min_length": 144.0,
      "completions/min_terminated_length": 144.0,
      "epoch": 0.1952,
      "grad_norm": 0.011388537473976612,
      "learning_rate": 4.7222222222222226e-07,
      "loss": 0.0331,
      "num_tokens": 40073448.0,
      "reward": 1.1702120304107666,
      "reward_std": 0.3779938220977783,
      "rewards/accuracy_reward_step": 0.5,
      "rewards/final_brier_reward_step": 0.7120984196662903,
      "rewards/format_reward_step": 0.984375,
      "step": 183
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.084384265882422e-08,
      "aux_brier/mean_r": 0.7429876867681742,
      "aux_brier/n_active_tok": 245.5,
      "aux_brier/n_step_records": 61.375,
      "aux_brier/std_r": 0.21635166136547923,
      "calib/answer_extract_rate": 0.99609375,
      "calib/auroc": 0.473652273888605,
      "calib/avg_num_step_conf": 7.9609375,
      "calib/ece": 0.2161431372549021,
      "calib/final_conf_rate": 0.99609375,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.13333333333333333,
      "calib/gap": -0.03624070643842614,
      "calib/mean_conf": 0.6745627450980393,
      "calib/mu_c": 0.6599243421052632,
      "calib/mu_w": 0.6961650485436893,
      "calib/nonempty_final_conf_rate": 0.99609375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1473137254901962,
      "calib/std_conf": 0.20710437205862595,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1243.0,
      "completions/max_terminated_length": 1243.0,
      "completions/mean_length": 493.31640625,
      "completions/mean_terminated_length": 495.2510070800781,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 147.0,
      "epoch": 0.19626666666666667,
      "grad_norm": 0.09844480454921722,
      "learning_rate": 4.444444444444445e-07,
      "loss": 0.0101,
      "num_tokens": 40305017.0,
      "reward": 1.262253999710083,
      "reward_std": 0.2725922167301178,
      "rewards/accuracy_reward_step": 0.59375,
      "rewards/final_brier_reward_step": 0.6896409392356873,
      "rewards/format_reward_step": 0.9921875,
      "step": 184
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -8.79858214064555e-10,
      "aux_brier/mean_r": 0.7613634951412678,
      "aux_brier/n_active_tok": 248.5,
      "aux_brier/n_step_records": 62.125,
      "aux_brier/std_r": 0.21437946753576398,
      "calib/answer_extract_rate": 0.96875,
      "calib/auroc": 0.5550709939148073,
      "calib/avg_num_step_conf": 8.296875,
      "calib/ece": 0.13101943319838055,
      "calib/final_conf_rate": 0.96484375,
      "calib/format_rate": 0.9609375,
      "calib/frac_conf_gt_0.9": 0.08502024291497975,
      "calib/gap": 0.036053191345503666,
      "calib/mean_conf": 0.6258226720647774,
      "calib/mu_c": 0.6407110344827586,
      "calib/mu_w": 0.6046578431372549,
      "calib/nonempty_final_conf_rate": 0.96484375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.08489878542510121,
      "calib/std_conf": 0.22071358189963147,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2835.0,
      "completions/max_terminated_length": 2835.0,
      "completions/mean_length": 497.671875,
      "completions/mean_terminated_length": 505.57147216796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 197.0,
      "epoch": 0.19733333333333333,
      "grad_norm": 0.15276029706001282,
      "learning_rate": 4.1666666666666667e-07,
      "loss": -0.0357,
      "num_tokens": 40539341.0,
      "reward": 1.225562572479248,
      "reward_std": 0.30816981196403503,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6991252899169922,
      "rewards/format_reward_step": 0.9609375,
      "step": 185
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.979132911866248e-08,
      "aux_brier/mean_r": 0.7636740412563086,
      "aux_brier/n_active_tok": 235.5,
      "aux_brier/n_step_records": 58.875,
      "aux_brier/std_r": 0.22241042833775282,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.47173006774361653,
      "calib/avg_num_step_conf": 7.6875,
      "calib/ece": 0.22383715415019764,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.11857707509881422,
      "calib/gap": -0.020000729546638785,
      "calib/mean_conf": 0.6126055335968379,
      "calib/mu_c": 0.604621052631579,
      "calib/mu_w": 0.6246217821782177,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.11782608695652177,
      "calib/std_conf": 0.24029222272348957,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 977.0,
      "completions/max_terminated_length": 977.0,
      "completions/mean_length": 469.46875,
      "completions/mean_terminated_length": 475.03558349609375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 168.0,
      "epoch": 0.1984,
      "grad_norm": 0.0938546285033226,
      "learning_rate": 3.8888888888888895e-07,
      "loss": -0.0339,
      "num_tokens": 40764565.0,
      "reward": 1.253624677658081,
      "reward_std": 0.33846354484558105,
      "rewards/accuracy_reward_step": 0.59375,
      "rewards/final_brier_reward_step": 0.6785610914230347,
      "rewards/format_reward_step": 0.98046875,
      "step": 186
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.123711770318006e-08,
      "aux_brier/mean_r": 0.7306062150746584,
      "aux_brier/n_active_tok": 256.375,
      "aux_brier/n_step_records": 64.09375,
      "aux_brier/std_r": 0.23782278783619404,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5809750996666885,
      "calib/avg_num_step_conf": 9.01953125,
      "calib/ece": 0.14691719999999994,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.9765625,
      "calib/frac_conf_gt_0.9": 0.132,
      "calib/gap": 0.06617900137245936,
      "calib/mean_conf": 0.6267628,
      "calib/mu_c": 0.6550874125874127,
      "calib/mu_w": 0.5889084112149533,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10083999999999996,
      "calib/std_conf": 0.24310063993367026,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3031.0,
      "completions/max_terminated_length": 3031.0,
      "completions/mean_length": 539.796875,
      "completions/mean_terminated_length": 546.1976318359375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 167.0,
      "epoch": 0.19946666666666665,
      "grad_norm": 0.08663739264011383,
      "learning_rate": 3.611111111111111e-07,
      "loss": 0.0235,
      "num_tokens": 41004297.0,
      "reward": 1.223996639251709,
      "reward_std": 0.38673633337020874,
      "rewards/accuracy_reward_step": 0.55859375,
      "rewards/final_brier_reward_step": 0.708486795425415,
      "rewards/format_reward_step": 0.9765625,
      "step": 187
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.242779945620498e-08,
      "aux_brier/mean_r": 0.7603856101632118,
      "aux_brier/n_active_tok": 237.75,
      "aux_brier/n_step_records": 59.4375,
      "aux_brier/std_r": 0.21150804916396737,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.48428498271668163,
      "calib/avg_num_step_conf": 7.546875,
      "calib/ece": 0.19895256916996046,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.11462450592885376,
      "calib/gap": -0.013389706823710101,
      "calib/mean_conf": 0.6331422924901186,
      "calib/mu_c": 0.6274794520547945,
      "calib/mu_w": 0.6408691588785046,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1275098814229249,
      "calib/std_conf": 0.2146966045511938,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2598.0,
      "completions/max_terminated_length": 2598.0,
      "completions/mean_length": 504.74609375,
      "completions/mean_terminated_length": 508.720458984375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 165.0,
      "epoch": 0.20053333333333334,
      "grad_norm": 0.044322092086076736,
      "learning_rate": 3.3333333333333335e-07,
      "loss": 0.0025,
      "num_tokens": 41237584.0,
      "reward": 1.234865427017212,
      "reward_std": 0.3233446776866913,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6894615888595581,
      "rewards/format_reward_step": 0.984375,
      "step": 188
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 4.70022790688418e-08,
      "aux_brier/mean_r": 0.7549281157553196,
      "aux_brier/n_active_tok": 226.75,
      "aux_brier/n_step_records": 56.6875,
      "aux_brier/std_r": 0.21891773724928498,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.4905071355759429,
      "calib/avg_num_step_conf": 7.30078125,
      "calib/ece": 0.18193083003952562,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.11067193675889328,
      "calib/gap": -0.018132186544342677,
      "calib/mean_conf": 0.640203557312253,
      "calib/mu_c": 0.6323916666666666,
      "calib/mu_w": 0.6505238532110093,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.1264822134387351,
      "calib/std_conf": 0.2101019621357078,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2203.0,
      "completions/max_terminated_length": 2203.0,
      "completions/mean_length": 469.953125,
      "completions/mean_terminated_length": 471.7961120605469,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 160.0,
      "epoch": 0.2016,
      "grad_norm": 0.07346808165311813,
      "learning_rate": 3.055555555555556e-07,
      "loss": 0.0019,
      "num_tokens": 41465660.0,
      "reward": 1.2287752628326416,
      "reward_std": 0.2613146901130676,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.6885387897491455,
      "rewards/format_reward_step": 0.98828125,
      "step": 189
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.9795711431180862e-08,
      "aux_brier/mean_r": 0.7416869290173054,
      "aux_brier/n_active_tok": 245.75,
      "aux_brier/n_step_records": 61.4375,
      "aux_brier/std_r": 0.23273182520642877,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5646343017329255,
      "calib/avg_num_step_conf": 8.0703125,
      "calib/ece": 0.17744229249011856,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.14624505928853754,
      "calib/gap": 0.05935076452599397,
      "calib/mean_conf": 0.6187632411067193,
      "calib/mu_c": 0.6443333333333333,
      "calib/mu_w": 0.5849825688073393,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1135177865612648,
      "calib/std_conf": 0.24990314841853278,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1615.0,
      "completions/max_terminated_length": 1615.0,
      "completions/mean_length": 527.42578125,
      "completions/mean_terminated_length": 531.5787353515625,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 181.0,
      "epoch": 0.20266666666666666,
      "grad_norm": 0.08770303428173065,
      "learning_rate": 2.7777777777777776e-07,
      "loss": -0.0053,
      "num_tokens": 41706289.0,
      "reward": 1.232055425643921,
      "reward_std": 0.33430615067481995,
      "rewards/accuracy_reward_step": 0.5625,
      "rewards/final_brier_reward_step": 0.7094713449478149,
      "rewards/format_reward_step": 0.984375,
      "step": 190
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.038965662976011e-08,
      "aux_brier/mean_r": 0.7504640500992537,
      "aux_brier/n_active_tok": 246.125,
      "aux_brier/n_step_records": 61.53125,
      "aux_brier/std_r": 0.2225411895196885,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.590440031152648,
      "calib/avg_num_step_conf": 7.953125,
      "calib/ece": 0.271511155378486,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.1394422310756972,
      "calib/gap": 0.050895723001038395,
      "calib/mean_conf": 0.6472139442231075,
      "calib/mu_c": 0.6764130841121495,
      "calib/mu_w": 0.6255173611111111,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.24621513944223108,
      "calib/std_conf": 0.22694470939041445,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2975.0,
      "completions/max_terminated_length": 2975.0,
      "completions/mean_length": 495.890625,
      "completions/mean_terminated_length": 497.8353271484375,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 145.0,
      "epoch": 0.20373333333333332,
      "grad_norm": 0.026141246780753136,
      "learning_rate": 2.5000000000000004e-07,
      "loss": 0.0192,
      "num_tokens": 41937405.0,
      "reward": 1.077864646911621,
      "reward_std": 0.2893039584159851,
      "rewards/accuracy_reward_step": 0.42578125,
      "rewards/final_brier_reward_step": 0.6630213260650635,
      "rewards/format_reward_step": 0.97265625,
      "step": 191
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -9.107425272270042e-09,
      "aux_brier/mean_r": 0.7753161191940308,
      "aux_brier/n_active_tok": 225.625,
      "aux_brier/n_step_records": 56.40625,
      "aux_brier/std_r": 0.20858300244435668,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.5480855423229607,
      "calib/avg_num_step_conf": 7.4140625,
      "calib/ece": 0.17907091633466135,
      "calib/final_conf_rate": 0.98046875,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.0796812749003984,
      "calib/gap": 0.04229238698937132,
      "calib/mean_conf": 0.5884988047808766,
      "calib/mu_c": 0.607707299270073,
      "calib/mu_w": 0.5654149122807017,
      "calib/nonempty_final_conf_rate": 0.98046875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.98828125,
      "calib/pce": 0.11087649402390437,
      "calib/std_conf": 0.23393689404832627,
      "calib/step_conf_rate": 0.98828125,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 3046.0,
      "completions/max_terminated_length": 3046.0,
      "completions/mean_length": 510.96875,
      "completions/mean_terminated_length": 512.9725952148438,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 159.0,
      "epoch": 0.2048,
      "grad_norm": 0.03855369985103607,
      "learning_rate": 2.2222222222222224e-07,
      "loss": 0.0562,
      "num_tokens": 42173189.0,
      "reward": 1.1960439682006836,
      "reward_std": 0.28442227840423584,
      "rewards/accuracy_reward_step": 0.53515625,
      "rewards/final_brier_reward_step": 0.6982383728027344,
      "rewards/format_reward_step": 0.97265625,
      "step": 192
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -3.864155766197808e-09,
      "aux_brier/mean_r": 0.7524569313973188,
      "aux_brier/n_active_tok": 247.75,
      "aux_brier/n_step_records": 61.9375,
      "aux_brier/std_r": 0.20505563402548432,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.4885655533399801,
      "calib/avg_num_step_conf": 8.11328125,
      "calib/ece": 0.252634251968504,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.98828125,
      "calib/frac_conf_gt_0.9": 0.12598425196850394,
      "calib/gap": 0.0020106306081757452,
      "calib/mean_conf": 0.6454759842519685,
      "calib/mu_c": 0.6465525423728815,
      "calib/mu_w": 0.6445419117647058,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.21677165354330713,
      "calib/std_conf": 0.21947669942068607,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2512.0,
      "completions/max_terminated_length": 2512.0,
      "completions/mean_length": 536.2578125,
      "completions/mean_terminated_length": 536.2578125,
      "completions/min_length": 168.0,
      "completions/min_terminated_length": 168.0,
      "epoch": 0.20586666666666667,
      "grad_norm": 0.03138832002878189,
      "learning_rate": 1.9444444444444447e-07,
      "loss": 0.0711,
      "num_tokens": 42416183.0,
      "reward": 1.1206305027008057,
      "reward_std": 0.3979037404060364,
      "rewards/accuracy_reward_step": 0.4609375,
      "rewards/final_brier_reward_step": 0.6622095108032227,
      "rewards/format_reward_step": 0.98828125,
      "step": 193
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.144103057599711e-08,
      "aux_brier/mean_r": 0.7848739847540855,
      "aux_brier/n_active_tok": 226.875,
      "aux_brier/n_step_records": 56.71875,
      "aux_brier/std_r": 0.19625219888985157,
      "calib/answer_extract_rate": 0.98046875,
      "calib/auroc": 0.48969371930731453,
      "calib/avg_num_step_conf": 7.30078125,
      "calib/ece": 0.15639246031746032,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.98046875,
      "calib/frac_conf_gt_0.9": 0.06349206349206349,
      "calib/gap": -0.031080860687516254,
      "calib/mean_conf": 0.6051154761904762,
      "calib/mu_c": 0.5920417808219177,
      "calib/mu_w": 0.623122641509434,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 0.9921875,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.09107142857142857,
      "calib/std_conf": 0.2128361685462316,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2641.0,
      "completions/max_terminated_length": 2641.0,
      "completions/mean_length": 473.24609375,
      "completions/mean_terminated_length": 476.9724426269531,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 191.0,
      "epoch": 0.20693333333333333,
      "grad_norm": 0.0721108615398407,
      "learning_rate": 1.6666666666666668e-07,
      "loss": 0.0491,
      "num_tokens": 42643278.0,
      "reward": 1.2311301231384277,
      "reward_std": 0.324305921792984,
      "rewards/accuracy_reward_step": 0.5703125,
      "rewards/final_brier_reward_step": 0.6823331117630005,
      "rewards/format_reward_step": 0.98046875,
      "step": 194
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 2.1848507131966288e-08,
      "aux_brier/mean_r": 0.7506272029131651,
      "aux_brier/n_active_tok": 242.625,
      "aux_brier/n_step_records": 60.65625,
      "aux_brier/std_r": 0.22388874553143978,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.5168831168831167,
      "calib/avg_num_step_conf": 7.890625,
      "calib/ece": 0.20170120000000002,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.97265625,
      "calib/frac_conf_gt_0.9": 0.156,
      "calib/gap": 0.005981623376623224,
      "calib/mean_conf": 0.6594588,
      "calib/mu_c": 0.6620907142857142,
      "calib/mu_w": 0.656109090909091,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 0.99609375,
      "calib/nonempty_step_conf_rate": 0.9921875,
      "calib/pce": 0.15058000000000002,
      "calib/std_conf": 0.22140749206510604,
      "calib/step_conf_rate": 0.9921875,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2367.0,
      "completions/max_terminated_length": 2367.0,
      "completions/mean_length": 499.07421875,
      "completions/mean_terminated_length": 501.0314025878906,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 155.0,
      "epoch": 0.208,
      "grad_norm": 0.02382788434624672,
      "learning_rate": 1.3888888888888888e-07,
      "loss": 0.0274,
      "num_tokens": 42877025.0,
      "reward": 1.2027819156646729,
      "reward_std": 0.31919705867767334,
      "rewards/accuracy_reward_step": 0.546875,
      "rewards/final_brier_reward_step": 0.6783152222633362,
      "rewards/format_reward_step": 0.97265625,
      "step": 195
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -1.1521586858187494e-08,
      "aux_brier/mean_r": 0.7673561163246632,
      "aux_brier/n_active_tok": 226.125,
      "aux_brier/n_step_records": 56.53125,
      "aux_brier/std_r": 0.21260583796538413,
      "calib/answer_extract_rate": 0.9921875,
      "calib/auroc": 0.4740475742170338,
      "calib/avg_num_step_conf": 7.23046875,
      "calib/ece": 0.18918661417322838,
      "calib/final_conf_rate": 0.9921875,
      "calib/format_rate": 0.9921875,
      "calib/frac_conf_gt_0.9": 0.08267716535433071,
      "calib/gap": -0.02016176489047894,
      "calib/mean_conf": 0.6183724409448819,
      "calib/mu_c": 0.6094028368794326,
      "calib/mu_w": 0.6295646017699116,
      "calib/nonempty_final_conf_rate": 0.9921875,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.12622047244094492,
      "calib/std_conf": 0.21736828165677655,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1134.0,
      "completions/max_terminated_length": 1134.0,
      "completions/mean_length": 425.83203125,
      "completions/mean_terminated_length": 429.1850280761719,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 140.0,
      "epoch": 0.20906666666666668,
      "grad_norm": 0.06819652765989304,
      "learning_rate": 1.1111111111111112e-07,
      "loss": 0.0082,
      "num_tokens": 43088582.0,
      "reward": 1.2184810638427734,
      "reward_std": 0.2531457841396332,
      "rewards/accuracy_reward_step": 0.55078125,
      "rewards/final_brier_reward_step": 0.6864246726036072,
      "rewards/format_reward_step": 0.9921875,
      "step": 196
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -7.148380794241849e-08,
      "aux_brier/mean_r": 0.7538832575082779,
      "aux_brier/n_active_tok": 242.25,
      "aux_brier/n_step_records": 60.5625,
      "aux_brier/std_r": 0.21373565820977092,
      "calib/answer_extract_rate": 0.984375,
      "calib/auroc": 0.49071955250444954,
      "calib/avg_num_step_conf": 8.04296875,
      "calib/ece": 0.25523333333333337,
      "calib/final_conf_rate": 0.984375,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.10714285714285714,
      "calib/gap": -0.013332112890923065,
      "calib/mean_conf": 0.6437746031746032,
      "calib/mu_c": 0.6364736842105263,
      "calib/mu_w": 0.6498057971014494,
      "calib/nonempty_final_conf_rate": 0.984375,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.22331349206349205,
      "calib/std_conf": 0.2132599309868712,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2674.0,
      "completions/max_terminated_length": 2674.0,
      "completions/mean_length": 504.546875,
      "completions/mean_terminated_length": 506.5255126953125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 128.0,
      "epoch": 0.21013333333333334,
      "grad_norm": 0.06600037962198257,
      "learning_rate": 8.333333333333334e-08,
      "loss": 0.0207,
      "num_tokens": 43322802.0,
      "reward": 1.1007957458496094,
      "reward_std": 0.33143335580825806,
      "rewards/accuracy_reward_step": 0.4453125,
      "rewards/final_brier_reward_step": 0.6531826853752136,
      "rewards/format_reward_step": 0.984375,
      "step": 197
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -6.852925210631966e-08,
      "aux_brier/mean_r": 0.7748613730072975,
      "aux_brier/n_active_tok": 231.875,
      "aux_brier/n_step_records": 57.96875,
      "aux_brier/std_r": 0.20690141199156642,
      "calib/answer_extract_rate": 1.0,
      "calib/auroc": 0.5847468157812985,
      "calib/avg_num_step_conf": 7.375,
      "calib/ece": 0.13784882812500004,
      "calib/final_conf_rate": 1.0,
      "calib/format_rate": 1.0,
      "calib/frac_conf_gt_0.9": 0.0703125,
      "calib/gap": 0.04433693072382727,
      "calib/mean_conf": 0.5851199218750001,
      "calib/mu_c": 0.6043441379310345,
      "calib/mu_w": 0.5600072072072072,
      "calib/nonempty_final_conf_rate": 1.0,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.07828125000000001,
      "calib/std_conf": 0.21049331907819877,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 1237.0,
      "completions/max_terminated_length": 1237.0,
      "completions/mean_length": 448.53125,
      "completions/mean_terminated_length": 450.29022216796875,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 142.0,
      "epoch": 0.2112,
      "grad_norm": 0.014602994546294212,
      "learning_rate": 5.555555555555556e-08,
      "loss": 0.0198,
      "num_tokens": 43543010.0,
      "reward": 1.249288558959961,
      "reward_std": 0.262395441532135,
      "rewards/accuracy_reward_step": 0.56640625,
      "rewards/final_brier_reward_step": 0.7315295934677124,
      "rewards/format_reward_step": 1.0,
      "step": 198
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": 3.424231964643809e-08,
      "aux_brier/mean_r": 0.746009947732091,
      "aux_brier/n_active_tok": 247.375,
      "aux_brier/n_step_records": 61.84375,
      "aux_brier/std_r": 0.21845908579416573,
      "calib/answer_extract_rate": 0.9765625,
      "calib/auroc": 0.45856922285126095,
      "calib/avg_num_step_conf": 8.15625,
      "calib/ece": 0.22041159999999999,
      "calib/final_conf_rate": 0.9765625,
      "calib/format_rate": 0.96875,
      "calib/frac_conf_gt_0.9": 0.128,
      "calib/gap": -0.02590461914565112,
      "calib/mean_conf": 0.6251884,
      "calib/mu_c": 0.613168656716418,
      "calib/mu_w": 0.6390732758620691,
      "calib/nonempty_final_conf_rate": 0.9765625,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 0.99609375,
      "calib/pce": 0.1548,
      "calib/std_conf": 0.2238250382451436,
      "calib/step_conf_rate": 0.99609375,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2801.0,
      "completions/max_terminated_length": 2801.0,
      "completions/mean_length": 556.8984375,
      "completions/mean_terminated_length": 563.5020141601562,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 116.0,
      "epoch": 0.21226666666666666,
      "grad_norm": 0.24404789507389069,
      "learning_rate": 2.777777777777778e-08,
      "loss": 0.0246,
      "num_tokens": 43789776.0,
      "reward": 1.172433853149414,
      "reward_std": 0.41650083661079407,
      "rewards/accuracy_reward_step": 0.5234375,
      "rewards/final_brier_reward_step": 0.658484935760498,
      "rewards/format_reward_step": 0.96875,
      "step": 199
    },
    {
      "aux_brier/lambda": 0.10000000000000005,
      "aux_brier/loss": -2.940847843024841e-10,
      "aux_brier/mean_r": 0.7518376670777798,
      "aux_brier/n_active_tok": 238.0,
      "aux_brier/n_step_records": 59.5,
      "aux_brier/std_r": 0.22355738934129477,
      "calib/answer_extract_rate": 0.98828125,
      "calib/auroc": 0.5108237547892721,
      "calib/avg_num_step_conf": 7.89453125,
      "calib/ece": 0.17630316205533603,
      "calib/final_conf_rate": 0.98828125,
      "calib/format_rate": 0.984375,
      "calib/frac_conf_gt_0.9": 0.11067193675889328,
      "calib/gap": -0.007147790549169808,
      "calib/mean_conf": 0.6145664031620554,
      "calib/mu_c": 0.6115151724137932,
      "calib/mu_w": 0.618662962962963,
      "calib/nonempty_final_conf_rate": 0.98828125,
      "calib/nonempty_reasoning_rate": 1.0,
      "calib/nonempty_step_conf_rate": 1.0,
      "calib/pce": 0.10887351778656132,
      "calib/std_conf": 0.22629195319484657,
      "calib/step_conf_rate": 1.0,
      "clip_ratio/high_max": 0.0,
      "clip_ratio/high_mean": 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": 2747.0,
      "completions/max_terminated_length": 2747.0,
      "completions/mean_length": 495.73828125,
      "completions/mean_terminated_length": 499.6417236328125,
      "completions/min_length": 0.0,
      "completions/min_terminated_length": 84.0,
      "epoch": 0.21333333333333335,
      "grad_norm": 0.11620119959115982,
      "learning_rate": 0.0,
      "loss": -0.0097,
      "num_tokens": 44024733.0,
      "reward": 1.2305445671081543,
      "reward_std": 0.2702827453613281,
      "rewards/accuracy_reward_step": 0.56640625,
      "rewards/final_brier_reward_step": 0.6878036260604858,
      "rewards/format_reward_step": 0.984375,
      "step": 200
    },
    {
      "epoch": 0.21333333333333335,
      "step": 200,
      "total_flos": 0.0,
      "train_loss": 0.03209023250557948,
      "train_runtime": 14667.9702,
      "train_samples_per_second": 3.491,
      "train_steps_per_second": 0.014
    }
  ],
  "logging_steps": 1,
  "max_steps": 200,
  "num_input_tokens_seen": 44024733,
  "num_train_epochs": 1,
  "save_steps": 40,
  "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": 8,
  "trial_name": null,
  "trial_params": null
}