zhaohq/PureRL-1.5B-v7-s2-l2-maskon-afew
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
c85ddb8dd199a0b6d3a351ba700d6dca4fe65ae3
PureRL-1.5B-v7-s2-l2-maskon…/trainer_state.json
ModelHub XC c85ddb8dd1 初始化项目，由ModelHub XC社区提供模型 
Model: zhaohq/PureRL-1.5B-v7-s2-l2-maskon-afew
Source: Original Platform
2026-06-01 21:20:49 +08:00

12243 lines
502 KiB
JSON
Raw Blame History

{
"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": [
{
"adv/mean_abs_final_conf": 0.47760647535324097,
"adv/mean_abs_reasoning": 0.4569147527217865,
"adv/mean_abs_step_conf": 0.7568557262420654,
"adv/ratio_final_to_reasoning": 1.0452857398632815,
"adv/ratio_step_to_reasoning": 1.6564484331783258,
"adv/std_final_conf": 0.7227410674095154,
"adv/std_reasoning": 0.7206857204437256,
"adv/std_step_conf": 0.9254623055458069,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5086206896551725,
"calib/avg_num_step_conf": 7.875,
"calib/ece": 0.2888991935483871,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0001713264989126051,
"calib/mean_conf": 0.9905120967741936,
"calib/mu_c": 0.9905632183908043,
"calib/mu_w": 0.9903918918918917,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2888991935483871,
"calib/std_conf": 0.0021794159006610276,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9119477557027226,
"calib/step_q_c_n": 1359.0,
"calib/step_q_gap": 0.0056311651395566376,
"calib/step_q_w": 0.9063165905631659,
"calib/step_q_w_n": 657.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2494.0,
"completions/max_terminated_length": 2494.0,
"completions/mean_length": 755.49609375,
"completions/mean_terminated_length": 776.7349243164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 397.0,
"epoch": 0.0010666666666666667,
"grad_norm": 0.02502845786511898,
"kl": 0.0005849599838256836,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.1291,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018737709149718285,
"mask/share_reasoning": 0.845859944820404,
"mask/share_step_conf": 0.10805858671665192,
"num_tokens": 300991.0,
"reward": 0.8747976422309875,
"reward_std": 0.23758892714977264,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/final_brier_reward_step": 0.6851503849029541,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7347574234008789,
"step": 1
},
{
"adv/mean_abs_final_conf": 0.437887966632843,
"adv/mean_abs_reasoning": 0.4207462966442108,
"adv/mean_abs_step_conf": 0.7215397357940674,
"adv/ratio_final_to_reasoning": 1.0407411072310102,
"adv/ratio_step_to_reasoning": 1.7149045435430461,
"adv/std_final_conf": 0.6832791566848755,
"adv/std_reasoning": 0.6817297339439392,
"adv/std_step_conf": 0.9222810864448547,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.4872611464968153,
"calib/avg_num_step_conf": 7.6953125,
"calib/ece": 0.36465737051792824,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00024481637078155316,
"calib/mean_conf": 0.9901553784860557,
"calib/mu_c": 0.990063694267516,
"calib/mu_w": 0.9903085106382975,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36465737051792824,
"calib/std_conf": 0.001222205307190084,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9075405636208369,
"calib/step_q_c_n": 1171.0,
"calib/step_q_gap": -0.003804868168900244,
"calib/step_q_w": 0.9113454317897371,
"calib/step_q_w_n": 799.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2743.0,
"completions/max_terminated_length": 2743.0,
"completions/mean_length": 840.640625,
"completions/mean_terminated_length": 850.6087036132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 466.0,
"epoch": 0.0021333333333333334,
"grad_norm": 0.021789954975247383,
"kl": 0.0016820430755615234,
"learning_rate": 5.000000000000001e-07,
"loss": -0.0097,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.01788702979683876,
"mask/share_reasoning": 0.8706268668174744,
"mask/share_step_conf": 0.09976735711097717,
"num_tokens": 619483.0,
"reward": 0.818764328956604,
"reward_std": 0.2176743745803833,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6203019618988037,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.6984766721725464,
"step": 2
},
{
"adv/mean_abs_final_conf": 0.3907315135002136,
"adv/mean_abs_reasoning": 0.37838390469551086,
"adv/mean_abs_step_conf": 0.7500925660133362,
"adv/ratio_final_to_reasoning": 1.0326324895205017,
"adv/ratio_step_to_reasoning": 1.9823585430171582,
"adv/std_final_conf": 0.6628881096839905,
"adv/std_reasoning": 0.6613768935203552,
"adv/std_step_conf": 0.9263537526130676,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.4993700904105529,
"calib/avg_num_step_conf": 7.734375,
"calib/ece": 0.30087649402390426,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -1.2598191788670832e-05,
"calib/mean_conf": 0.9901195219123505,
"calib/mu_c": 0.990115606936416,
"calib/mu_w": 0.9901282051282047,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.30087649402390426,
"calib/std_conf": 0.001086707704939113,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9121627906976743,
"calib/step_q_c_n": 1290.0,
"calib/step_q_gap": -0.00253286147623899,
"calib/step_q_w": 0.9146956521739132,
"calib/step_q_w_n": 690.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1638.0,
"completions/max_terminated_length": 1638.0,
"completions/mean_length": 786.2890625,
"completions/mean_terminated_length": 801.9522094726562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 431.0,
"epoch": 0.0032,
"grad_norm": 0.022982632741332054,
"kl": 0.0005571246147155762,
"learning_rate": 7.5e-07,
"loss": -0.0907,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.018434403464198112,
"mask/share_reasoning": 0.8618663549423218,
"mask/share_step_conf": 0.10016795992851257,
"num_tokens": 926029.0,
"reward": 0.8705581426620483,
"reward_std": 0.2126719355583191,
"rewards/accuracy_reward_step": 0.67578125,
"rewards/final_brier_reward_step": 0.6816222667694092,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7290253043174744,
"step": 3
},
{
"adv/mean_abs_final_conf": 0.49821069836616516,
"adv/mean_abs_reasoning": 0.48443976044654846,
"adv/mean_abs_step_conf": 0.776309609413147,
"adv/ratio_final_to_reasoning": 1.028426522849658,
"adv/ratio_step_to_reasoning": 1.6024894585398146,
"adv/std_final_conf": 0.7398562431335449,
"adv/std_reasoning": 0.7394483685493469,
"adv/std_step_conf": 0.9243927597999573,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5064102564102564,
"calib/avg_num_step_conf": 7.61328125,
"calib/ece": 0.30415322580645165,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9959677419354839,
"calib/gap": 0.001153846153846727,
"calib/mean_conf": 0.9896370967741935,
"calib/mu_c": 0.9900000000000001,
"calib/mu_w": 0.9888461538461534,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.30415322580645165,
"calib/std_conf": 0.005703471887480457,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9096366508688782,
"calib/step_q_c_n": 1266.0,
"calib/step_q_gap": -0.002105662454694235,
"calib/step_q_w": 0.9117423133235725,
"calib/step_q_w_n": 683.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2882.0,
"completions/max_terminated_length": 2882.0,
"completions/mean_length": 810.3046875,
"completions/mean_terminated_length": 819.9130859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 471.0,
"epoch": 0.004266666666666667,
"grad_norm": 0.027756016701459885,
"kl": 0.0005552768707275391,
"learning_rate": 1.0000000000000002e-06,
"loss": -0.0467,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018273094668984413,
"mask/share_reasoning": 0.8700456619262695,
"mask/share_step_conf": 0.0999625027179718,
"num_tokens": 1239635.0,
"reward": 0.8520616292953491,
"reward_std": 0.26222509145736694,
"rewards/accuracy_reward_step": 0.66796875,
"rewards/final_brier_reward_step": 0.6667401790618896,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7116017937660217,
"step": 4
},
{
"adv/mean_abs_final_conf": 0.43325841426849365,
"adv/mean_abs_reasoning": 0.43388810753822327,
"adv/mean_abs_step_conf": 0.7429900765419006,
"adv/ratio_final_to_reasoning": 0.9985487196842008,
"adv/ratio_step_to_reasoning": 1.7124001871299206,
"adv/std_final_conf": 0.7010958194732666,
"adv/std_reasoning": 0.7014488577842712,
"adv/std_step_conf": 0.9139078855514526,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.49477886977886976,
"calib/avg_num_step_conf": 7.58984375,
"calib/ece": 0.44691358024691363,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00010442260442244944,
"calib/mean_conf": 0.9901234567901235,
"calib/mu_c": 0.9900757575757576,
"calib/mu_w": 0.9901801801801801,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44691358024691363,
"calib/std_conf": 0.0011042310999998973,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9138218390804598,
"calib/step_q_c_n": 1044.0,
"calib/step_q_gap": 0.009327956989247377,
"calib/step_q_w": 0.9044938820912124,
"calib/step_q_w_n": 899.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3016.0,
"completions/max_terminated_length": 3016.0,
"completions/mean_length": 816.8515625,
"completions/mean_terminated_length": 833.12353515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 435.0,
"epoch": 0.005333333333333333,
"grad_norm": 0.02338205836713314,
"kl": 0.000583648681640625,
"learning_rate": 1.25e-06,
"loss": -0.076,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.018079116940498352,
"mask/share_reasoning": 0.8604636192321777,
"mask/share_step_conf": 0.1019260585308075,
"num_tokens": 1555437.0,
"reward": 0.716595470905304,
"reward_std": 0.2071603238582611,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.5240468978881836,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6161753535270691,
"step": 5
},
{
"adv/mean_abs_final_conf": 0.3801310658454895,
"adv/mean_abs_reasoning": 0.37981629371643066,
"adv/mean_abs_step_conf": 0.7121192216873169,
"adv/ratio_final_to_reasoning": 1.0008287483561562,
"adv/ratio_step_to_reasoning": 1.8749043510465675,
"adv/std_final_conf": 0.6808795928955078,
"adv/std_reasoning": 0.6816310286521912,
"adv/std_step_conf": 0.9247093796730042,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.49514563106796117,
"calib/avg_num_step_conf": 8.0703125,
"calib/ece": 0.4070445344129554,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -9.708737864055905e-05,
"calib/mean_conf": 0.9900404858299594,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9900970873786406,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4070445344129554,
"calib/std_conf": 0.0006349954308242161,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9109003496503498,
"calib/step_q_c_n": 1144.0,
"calib/step_q_gap": -0.0015399974212336742,
"calib/step_q_w": 0.9124403470715835,
"calib/step_q_w_n": 922.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2505.0,
"completions/max_terminated_length": 2505.0,
"completions/mean_length": 722.78515625,
"completions/mean_terminated_length": 743.1043701171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 474.0,
"epoch": 0.0064,
"grad_norm": 0.017047492787241936,
"kl": 0.0007506012916564941,
"learning_rate": 1.5e-06,
"loss": -0.1258,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.01940595731139183,
"mask/share_reasoning": 0.842772364616394,
"mask/share_step_conf": 0.11047791689634323,
"num_tokens": 1846422.0,
"reward": 0.7581819295883179,
"reward_std": 0.19750076532363892,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.5703726410865784,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6405224800109863,
"step": 6
},
{
"adv/mean_abs_final_conf": 0.47522249817848206,
"adv/mean_abs_reasoning": 0.44085246324539185,
"adv/mean_abs_step_conf": 0.7773077487945557,
"adv/ratio_final_to_reasoning": 1.07796266959715,
"adv/ratio_step_to_reasoning": 1.763192481839174,
"adv/std_final_conf": 0.7201507091522217,
"adv/std_reasoning": 0.7015153765678406,
"adv/std_step_conf": 0.9311604499816895,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.4850123385106692,
"calib/avg_num_step_conf": 7.453125,
"calib/ece": 0.3234859437751004,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00027710843373507377,
"calib/mean_conf": 0.9901526104417671,
"calib/mu_c": 0.9900602409638553,
"calib/mu_w": 0.9903373493975903,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3234859437751004,
"calib/std_conf": 0.001196045699004783,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9102295081967213,
"calib/step_q_c_n": 1220.0,
"calib/step_q_gap": 0.006629217499046991,
"calib/step_q_w": 0.9036002906976743,
"calib/step_q_w_n": 688.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2809.0,
"completions/max_terminated_length": 2809.0,
"completions/mean_length": 827.58984375,
"completions/mean_terminated_length": 844.0757446289062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 483.0,
"epoch": 0.007466666666666667,
"grad_norm": 0.018745925277471542,
"kl": 0.0005162954330444336,
"learning_rate": 1.75e-06,
"loss": -0.0747,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.017496969550848007,
"mask/share_reasoning": 0.8684599995613098,
"mask/share_step_conf": 0.09451176226139069,
"num_tokens": 2165709.0,
"reward": 0.8448113203048706,
"reward_std": 0.22868016362190247,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/final_brier_reward_step": 0.6545296311378479,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7116554379463196,
"step": 7
},
{
"adv/mean_abs_final_conf": 0.39101529121398926,
"adv/mean_abs_reasoning": 0.3464837372303009,
"adv/mean_abs_step_conf": 0.7708368897438049,
"adv/ratio_final_to_reasoning": 1.128524225522565,
"adv/ratio_step_to_reasoning": 2.224741905365229,
"adv/std_final_conf": 0.6386479139328003,
"adv/std_reasoning": 0.6186771988868713,
"adv/std_step_conf": 0.9308523535728455,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.48971253465635484,
"calib/avg_num_step_conf": 7.4765625,
"calib/ece": 0.35645679012345677,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00019597256675907282,
"calib/mean_conf": 0.990201646090535,
"calib/mu_c": 0.9901298701298701,
"calib/mu_w": 0.9903258426966292,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35645679012345677,
"calib/std_conf": 0.0013923946001348608,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9107012987012988,
"calib/step_q_c_n": 1155.0,
"calib/step_q_gap": 0.019510257858083913,
"calib/step_q_w": 0.8911910408432149,
"calib/step_q_w_n": 759.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2407.0,
"completions/max_terminated_length": 2407.0,
"completions/mean_length": 850.4921875,
"completions/mean_terminated_length": 867.4342651367188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 434.0,
"epoch": 0.008533333333333334,
"grad_norm": 0.025057844817638397,
"kl": 0.0005748271942138672,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.0931,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.017516685649752617,
"mask/share_reasoning": 0.8684801459312439,
"mask/share_step_conf": 0.0944719910621643,
"num_tokens": 2489947.0,
"reward": 0.8075144290924072,
"reward_std": 0.1630677580833435,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6081961393356323,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6966763734817505,
"step": 8
},
{
"adv/mean_abs_final_conf": 0.43854090571403503,
"adv/mean_abs_reasoning": 0.3960496783256531,
"adv/mean_abs_step_conf": 0.7646069526672363,
"adv/ratio_final_to_reasoning": 1.1072876199976192,
"adv/ratio_step_to_reasoning": 1.930583445742723,
"adv/std_final_conf": 0.6832506060600281,
"adv/std_reasoning": 0.6614897847175598,
"adv/std_step_conf": 0.9325852990150452,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.4974292072749597,
"calib/avg_num_step_conf": 8.27734375,
"calib/ece": 0.3319291666666666,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.9958333333333333,
"calib/gap": 0.011749597114573018,
"calib/mean_conf": 0.9860958333333334,
"calib/mu_c": 0.9901592356687897,
"calib/mu_w": 0.9784096385542167,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3319291666666666,
"calib/std_conf": 0.06380075743520255,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9100164338537388,
"calib/step_q_c_n": 1217.0,
"calib/step_q_gap": -0.005450306722758036,
"calib/step_q_w": 0.9154667405764968,
"calib/step_q_w_n": 902.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2320.0,
"completions/max_terminated_length": 2320.0,
"completions/mean_length": 760.39453125,
"completions/mean_terminated_length": 801.0740356445312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 460.0,
"epoch": 0.0096,
"grad_norm": 0.01747042126953602,
"kl": 0.0006932616233825684,
"learning_rate": 2.25e-06,
"loss": -0.2195,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.017814621329307556,
"mask/share_reasoning": 0.8285001516342163,
"mask/share_step_conf": 0.10290395468473434,
"num_tokens": 2792144.0,
"reward": 0.79695063829422,
"reward_std": 0.23099634051322937,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.623283863067627,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.6604610681533813,
"step": 9
},
{
"adv/mean_abs_final_conf": 0.4888319671154022,
"adv/mean_abs_reasoning": 0.4622969925403595,
"adv/mean_abs_step_conf": 0.7920699119567871,
"adv/ratio_final_to_reasoning": 1.0573981120431497,
"adv/ratio_step_to_reasoning": 1.713335636479698,
"adv/std_final_conf": 0.7395802140235901,
"adv/std_reasoning": 0.7206360697746277,
"adv/std_step_conf": 0.9307116866111755,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5014992503748127,
"calib/avg_num_step_conf": 7.44921875,
"calib/ece": 0.3411612903225807,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 3.312629399632172e-05,
"calib/mean_conf": 0.9903548387096774,
"calib/mu_c": 0.9903664596273293,
"calib/mu_w": 0.990333333333333,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3411612903225807,
"calib/std_conf": 0.0018302720949977284,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9100284280936453,
"calib/step_q_c_n": 1196.0,
"calib/step_q_gap": 0.006634616560593054,
"calib/step_q_w": 0.9033938115330522,
"calib/step_q_w_n": 711.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3047.0,
"completions/max_terminated_length": 3047.0,
"completions/mean_length": 827.7421875,
"completions/mean_terminated_length": 840.8809814453125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 442.0,
"epoch": 0.010666666666666666,
"grad_norm": 0.025133531540632248,
"kl": 0.0028550028800964355,
"learning_rate": 2.5e-06,
"loss": -0.0486,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.017763039097189903,
"mask/share_reasoning": 0.8690170645713806,
"mask/share_step_conf": 0.09759492427110672,
"num_tokens": 3110846.0,
"reward": 0.8297994136810303,
"reward_std": 0.23124775290489197,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.6353831887245178,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7046843767166138,
"step": 10
},
{
"adv/mean_abs_final_conf": 0.4417019486427307,
"adv/mean_abs_reasoning": 0.4121132791042328,
"adv/mean_abs_step_conf": 0.7525558471679688,
"adv/ratio_final_to_reasoning": 1.07179741842537,
"adv/ratio_step_to_reasoning": 1.8260897799840863,
"adv/std_final_conf": 0.7013117074966431,
"adv/std_reasoning": 0.6817113161087036,
"adv/std_step_conf": 0.9160208702087402,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.4804754993434239,
"calib/avg_num_step_conf": 7.4140625,
"calib/ece": 0.35407200000000005,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0003705853894533373,
"calib/mean_conf": 0.9900720000000001,
"calib/mu_c": 0.9899371069182389,
"calib/mu_w": 0.9903076923076922,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35407200000000005,
"calib/std_conf": 0.0012044982357811913,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9084249384741593,
"calib/step_q_c_n": 1219.0,
"calib/step_q_gap": 0.002796072494777757,
"calib/step_q_w": 0.9056288659793815,
"calib/step_q_w_n": 679.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2120.0,
"completions/max_terminated_length": 2120.0,
"completions/mean_length": 829.9453125,
"completions/mean_terminated_length": 843.1190795898438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 450.0,
"epoch": 0.011733333333333333,
"grad_norm": 0.028363078832626343,
"kl": 0.0006840825080871582,
"learning_rate": 2.7500000000000004e-06,
"loss": -0.0067,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017736785113811493,
"mask/share_reasoning": 0.8707066178321838,
"mask/share_step_conf": 0.09593162685632706,
"num_tokens": 3427792.0,
"reward": 0.8359595537185669,
"reward_std": 0.22266630828380585,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.6278089284896851,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7253601551055908,
"step": 11
},
{
"adv/mean_abs_final_conf": 0.44267383217811584,
"adv/mean_abs_reasoning": 0.43362054228782654,
"adv/mean_abs_step_conf": 0.7366559505462646,
"adv/ratio_final_to_reasoning": 1.0208783694668229,
"adv/ratio_step_to_reasoning": 1.6988492903486356,
"adv/std_final_conf": 0.7020049095153809,
"adv/std_reasoning": 0.7015119194984436,
"adv/std_step_conf": 0.9088842272758484,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5025559668605676,
"calib/avg_num_step_conf": 8.26171875,
"calib/ece": 0.23716599190283394,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 5.111933721113626e-05,
"calib/mean_conf": 0.9902024291497975,
"calib/mu_c": 0.9902150537634409,
"calib/mu_w": 0.9901639344262297,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23716599190283394,
"calib/std_conf": 0.0014083017919778221,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9120886889460155,
"calib/step_q_c_n": 1556.0,
"calib/step_q_gap": 0.004432159429020643,
"calib/step_q_w": 0.9076565295169948,
"calib/step_q_w_n": 559.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2495.0,
"completions/max_terminated_length": 2495.0,
"completions/mean_length": 760.328125,
"completions/mean_terminated_length": 775.47412109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 329.0,
"epoch": 0.0128,
"grad_norm": 0.022897174581885338,
"kl": 0.0008499026298522949,
"learning_rate": 3e-06,
"loss": -0.0844,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.019012587144970894,
"mask/share_reasoning": 0.8476526141166687,
"mask/share_step_conf": 0.11380352079868317,
"num_tokens": 3726612.0,
"reward": 0.9202487468719482,
"reward_std": 0.21560242772102356,
"rewards/accuracy_reward_step": 0.7265625,
"rewards/final_brier_reward_step": 0.7311554551124573,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7710606455802917,
"step": 12
},
{
"adv/mean_abs_final_conf": 0.5370169281959534,
"adv/mean_abs_reasoning": 0.5062729120254517,
"adv/mean_abs_step_conf": 0.7291202545166016,
"adv/ratio_final_to_reasoning": 1.0607261724659607,
"adv/ratio_step_to_reasoning": 1.4401723600016483,
"adv/std_final_conf": 0.8014257550239563,
"adv/std_reasoning": 0.7929041981697083,
"adv/std_step_conf": 0.9328606724739075,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.49700598802395207,
"calib/avg_num_step_conf": 8.265625,
"calib/ece": 0.31101626016260164,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00017964071856246822,
"calib/mean_conf": 0.9898780487804878,
"calib/mu_c": 0.9898203592814371,
"calib/mu_w": 0.9899999999999995,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.31101626016260164,
"calib/std_conf": 0.0019088385173778716,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9110812883435584,
"calib/step_q_c_n": 1304.0,
"calib/step_q_gap": 0.009049268639124808,
"calib/step_q_w": 0.9020320197044336,
"calib/step_q_w_n": 812.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2578.0,
"completions/max_terminated_length": 2578.0,
"completions/mean_length": 782.8828125,
"completions/mean_terminated_length": 808.1370849609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 360.0,
"epoch": 0.013866666666666666,
"grad_norm": 3275.6826171875,
"kl": 6432.037212371826,
"learning_rate": 3.2500000000000002e-06,
"loss": 74.9262,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.01816052757203579,
"mask/share_reasoning": 0.8450461626052856,
"mask/share_step_conf": 0.1055433452129364,
"num_tokens": 4031622.0,
"reward": 0.8452422022819519,
"reward_std": 0.2676650285720825,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.6583359241485596,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7102735042572021,
"step": 13
},
{
"adv/mean_abs_final_conf": 0.39153605699539185,
"adv/mean_abs_reasoning": 0.3841407597064972,
"adv/mean_abs_step_conf": 0.7506622076034546,
"adv/ratio_final_to_reasoning": 1.0192515298156464,
"adv/ratio_step_to_reasoning": 1.954133188514018,
"adv/std_final_conf": 0.6616212129592896,
"adv/std_reasoning": 0.6613654494285583,
"adv/std_step_conf": 0.9337601661682129,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.5099337748344371,
"calib/avg_num_step_conf": 7.6953125,
"calib/ece": 0.36615289256198347,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0001920529801324733,
"calib/mean_conf": 0.9901198347107438,
"calib/mu_c": 0.9901920529801324,
"calib/mu_w": 0.9899999999999999,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36615289256198347,
"calib/std_conf": 0.001070885926181841,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9089094827586207,
"calib/step_q_c_n": 1160.0,
"calib/step_q_gap": 0.007625532141336766,
"calib/step_q_w": 0.901283950617284,
"calib/step_q_w_n": 810.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2683.0,
"completions/max_terminated_length": 2683.0,
"completions/mean_length": 808.359375,
"completions/mean_terminated_length": 837.8137817382812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 494.0,
"epoch": 0.014933333333333333,
"grad_norm": 0.015768852084875107,
"kl": 0.0010614991188049316,
"learning_rate": 3.5e-06,
"loss": -0.1338,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.01759328506886959,
"mask/share_reasoning": 0.8488487601280212,
"mask/share_step_conf": 0.09840168058872223,
"num_tokens": 4343962.0,
"reward": 0.8007533550262451,
"reward_std": 0.1906447410583496,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.5968597531318665,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.6976156234741211,
"step": 14
},
{
"adv/mean_abs_final_conf": 0.4488527774810791,
"adv/mean_abs_reasoning": 0.4263514280319214,
"adv/mean_abs_step_conf": 0.7604366540908813,
"adv/ratio_final_to_reasoning": 1.0527765312128214,
"adv/ratio_step_to_reasoning": 1.7835911975271879,
"adv/std_final_conf": 0.7022488117218018,
"adv/std_reasoning": 0.7015137076377869,
"adv/std_step_conf": 0.9223707318305969,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5033783783783784,
"calib/avg_num_step_conf": 7.6171875,
"calib/ece": 0.38841463414634136,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 6.75675675674281e-05,
"calib/mean_conf": 0.990040650406504,
"calib/mu_c": 0.9900675675675674,
"calib/mu_w": 0.99,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.38841463414634136,
"calib/std_conf": 0.0006362795057926237,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9109266609145816,
"calib/step_q_c_n": 1159.0,
"calib/step_q_gap": 0.0021655989676788723,
"calib/step_q_w": 0.9087610619469028,
"calib/step_q_w_n": 791.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2554.0,
"completions/max_terminated_length": 2554.0,
"completions/mean_length": 765.7421875,
"completions/mean_terminated_length": 787.26904296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 438.0,
"epoch": 0.016,
"grad_norm": 0.021051140502095222,
"kl": 0.0009889006614685059,
"learning_rate": 3.7500000000000005e-06,
"loss": -0.1004,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.01849520020186901,
"mask/share_reasoning": 0.8517429828643799,
"mask/share_step_conf": 0.10241805016994476,
"num_tokens": 4647872.0,
"reward": 0.7852047681808472,
"reward_std": 0.22741200029850006,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.5856855511665344,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.6761302947998047,
"step": 15
},
{
"adv/mean_abs_final_conf": 0.44075024127960205,
"adv/mean_abs_reasoning": 0.4255930781364441,
"adv/mean_abs_step_conf": 0.7281934022903442,
"adv/ratio_final_to_reasoning": 1.0356142144264353,
"adv/ratio_step_to_reasoning": 1.7110085659261762,
"adv/std_final_conf": 0.7203139662742615,
"adv/std_reasoning": 0.7205998301506042,
"adv/std_step_conf": 0.9330427646636963,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.49390243902439024,
"calib/avg_num_step_conf": 7.68359375,
"calib/ece": 0.32336991869918696,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00010975609756103122,
"calib/mean_conf": 0.9900365853658536,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9901097560975609,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.32336991869918696,
"calib/std_conf": 0.0005726515552133617,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9064437450826123,
"calib/step_q_c_n": 1271.0,
"calib/step_q_gap": 0.007529951979163951,
"calib/step_q_w": 0.8989137931034483,
"calib/step_q_w_n": 696.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3015.0,
"completions/max_terminated_length": 3015.0,
"completions/mean_length": 916.58984375,
"completions/mean_terminated_length": 931.138916015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 472.0,
"epoch": 0.017066666666666667,
"grad_norm": 0.024245938286185265,
"kl": 0.0011619329452514648,
"learning_rate": 4.000000000000001e-06,
"loss": -0.0161,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.016265347599983215,
"mask/share_reasoning": 0.8771912455558777,
"mask/share_step_conf": 0.0909183919429779,
"num_tokens": 4991367.0,
"reward": 0.8399591445922852,
"reward_std": 0.2094450294971466,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.64686518907547,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7127405405044556,
"step": 16
},
{
"adv/mean_abs_final_conf": 0.446701318025589,
"adv/mean_abs_reasoning": 0.4303044080734253,
"adv/mean_abs_step_conf": 0.7567791938781738,
"adv/ratio_final_to_reasoning": 1.0381053729511547,
"adv/ratio_step_to_reasoning": 1.7587065799917168,
"adv/std_final_conf": 0.6828343272209167,
"adv/std_reasoning": 0.6818340420722961,
"adv/std_step_conf": 0.9191504120826721,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 7.921875,
"calib/ece": 0.2509236947791166,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00046153846153840217,
"calib/mean_conf": 0.9898795180722892,
"calib/mu_c": 0.9900000000000001,
"calib/mu_w": 0.9895384615384617,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2509236947791166,
"calib/std_conf": 0.002610132820766576,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9065742024965325,
"calib/step_q_c_n": 1442.0,
"calib/step_q_gap": 0.006215840721788268,
"calib/step_q_w": 0.9003583617747443,
"calib/step_q_w_n": 586.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2965.0,
"completions/max_terminated_length": 2965.0,
"completions/mean_length": 813.078125,
"completions/mean_terminated_length": 832.592041015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 469.0,
"epoch": 0.018133333333333335,
"grad_norm": 0.01629556715488434,
"kl": 0.0016171932220458984,
"learning_rate": 4.25e-06,
"loss": -0.1074,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.017606137320399284,
"mask/share_reasoning": 0.8571971654891968,
"mask/share_step_conf": 0.1017591655254364,
"num_tokens": 5303043.0,
"reward": 0.9147970080375671,
"reward_std": 0.23791097104549408,
"rewards/accuracy_reward_step": 0.71875,
"rewards/final_brier_reward_step": 0.7239562273025513,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7673563957214355,
"step": 17
},
{
"adv/mean_abs_final_conf": 0.38064736127853394,
"adv/mean_abs_reasoning": 0.3477587401866913,
"adv/mean_abs_step_conf": 0.7276158332824707,
"adv/ratio_final_to_reasoning": 1.0945730970677736,
"adv/ratio_step_to_reasoning": 2.0923006360439897,
"adv/std_final_conf": 0.6817070841789246,
"adv/std_reasoning": 0.6612668037414551,
"adv/std_step_conf": 0.9329667687416077,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.4954954954954955,
"calib/avg_num_step_conf": 7.2421875,
"calib/ece": 0.4376209677419356,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -9.00900900897561e-05,
"calib/mean_conf": 0.9900403225806452,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9900900900900897,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.4376209677419356,
"calib/std_conf": 0.0006337190986089406,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.9095242718446602,
"calib/step_q_c_n": 1030.0,
"calib/step_q_gap": 0.009682038834951445,
"calib/step_q_w": 0.8998422330097088,
"calib/step_q_w_n": 824.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1674.0,
"completions/max_terminated_length": 1674.0,
"completions/mean_length": 813.5625,
"completions/mean_terminated_length": 829.7689208984375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.0192,
"grad_norm": 0.022068968042731285,
"kl": 0.0018869638442993164,
"learning_rate": 4.5e-06,
"loss": -0.0985,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.01774272322654724,
"mask/share_reasoning": 0.8674094080924988,
"mask/share_step_conf": 0.09531664848327637,
"num_tokens": 5622035.0,
"reward": 0.7283351421356201,
"reward_std": 0.20351342856884003,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.5358418226242065,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.6216096878051758,
"step": 18
},
{
"adv/mean_abs_final_conf": 0.4503156542778015,
"adv/mean_abs_reasoning": 0.4281848669052124,
"adv/mean_abs_step_conf": 0.7243156433105469,
"adv/ratio_final_to_reasoning": 1.0516851226726989,
"adv/ratio_step_to_reasoning": 1.6915956151035323,
"adv/std_final_conf": 0.7379180788993835,
"adv/std_reasoning": 0.7205255627632141,
"adv/std_step_conf": 0.9321519732475281,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5071428571428571,
"calib/avg_num_step_conf": 7.796875,
"calib/ece": 0.4278273092369479,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0001520314547837076,
"calib/mean_conf": 0.9900763052208836,
"calib/mu_c": 0.990142857142857,
"calib/mu_w": 0.9899908256880733,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4278273092369479,
"calib/std_conf": 0.001234624637367013,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9074858223062381,
"calib/step_q_c_n": 1058.0,
"calib/step_q_gap": 0.011288594161248722,
"calib/step_q_w": 0.8961972281449894,
"calib/step_q_w_n": 938.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2846.0,
"completions/max_terminated_length": 2846.0,
"completions/mean_length": 823.5078125,
"completions/mean_terminated_length": 833.2727661132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 412.0,
"epoch": 0.020266666666666665,
"grad_norm": 0.01661011576652527,
"kl": 0.0021848678588867188,
"learning_rate": 4.75e-06,
"loss": -0.061,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.018002573400735855,
"mask/share_reasoning": 0.8686628937721252,
"mask/share_step_conf": 0.1016157791018486,
"num_tokens": 5937613.0,
"reward": 0.7750768661499023,
"reward_std": 0.21773789823055267,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.5513952970504761,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.6956332921981812,
"step": 19
},
{
"adv/mean_abs_final_conf": 0.4298800826072693,
"adv/mean_abs_reasoning": 0.4013995826244354,
"adv/mean_abs_step_conf": 0.7212347984313965,
"adv/ratio_final_to_reasoning": 1.0709529885323306,
"adv/ratio_step_to_reasoning": 1.7968000706822134,
"adv/std_final_conf": 0.7196903228759766,
"adv/std_reasoning": 0.7013648748397827,
"adv/std_step_conf": 0.9234609603881836,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 8.44140625,
"calib/ece": 0.3956533864541833,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9960159362549801,
"calib/gap": 0.0017745098039217977,
"calib/mean_conf": 0.9892788844621515,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9882254901960782,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3956533864541833,
"calib/std_conf": 0.011984465058551374,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9077379095163807,
"calib/step_q_c_n": 1282.0,
"calib/step_q_gap": 0.0017947923377685804,
"calib/step_q_w": 0.9059431171786121,
"calib/step_q_w_n": 879.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2765.0,
"completions/max_terminated_length": 2765.0,
"completions/mean_length": 816.8046875,
"completions/mean_terminated_length": 823.2362060546875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 461.0,
"epoch": 0.021333333333333333,
"grad_norm": 0.015205039642751217,
"kl": 0.003265380859375,
"learning_rate": 5e-06,
"loss": -0.0495,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.01863781549036503,
"mask/share_reasoning": 0.8605009317398071,
"mask/share_step_conf": 0.1130487322807312,
"num_tokens": 6251587.0,
"reward": 0.7963709831237793,
"reward_std": 0.21171826124191284,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.5911605358123779,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.6890814304351807,
"step": 20
},
{
"adv/mean_abs_final_conf": 0.5546455383300781,
"adv/mean_abs_reasoning": 0.5302027463912964,
"adv/mean_abs_step_conf": 0.7633264064788818,
"adv/ratio_final_to_reasoning": 1.046100839924248,
"adv/ratio_step_to_reasoning": 1.4396877641134231,
"adv/std_final_conf": 0.7927553653717041,
"adv/std_reasoning": 0.7929099798202515,
"adv/std_step_conf": 0.9229240417480469,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.503030303030303,
"calib/avg_num_step_conf": 8.16796875,
"calib/ece": 0.3220242914979756,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 6.06060606060943e-05,
"calib/mean_conf": 0.9900404858299594,
"calib/mu_c": 0.990060606060606,
"calib/mu_w": 0.9899999999999999,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3220242914979756,
"calib/std_conf": 0.0006349954308242161,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9071587537091987,
"calib/step_q_c_n": 1348.0,
"calib/step_q_gap": 0.005395631232751885,
"calib/step_q_w": 0.9017631224764469,
"calib/step_q_w_n": 743.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 3032.0,
"completions/max_terminated_length": 3032.0,
"completions/mean_length": 809.734375,
"completions/mean_terminated_length": 829.1680297851562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 459.0,
"epoch": 0.0224,
"grad_norm": 0.011656345799565315,
"kl": 0.004233837127685547,
"learning_rate": 4.9722222222222224e-06,
"loss": -0.1312,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.017971325665712357,
"mask/share_reasoning": 0.8535507917404175,
"mask/share_step_conf": 0.10504040122032166,
"num_tokens": 6561839.0,
"reward": 0.8428006768226624,
"reward_std": 0.28343454003334045,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.6469355225563049,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7175719738006592,
"step": 21
},
{
"adv/mean_abs_final_conf": 0.4129585325717926,
"adv/mean_abs_reasoning": 0.39449843764305115,
"adv/mean_abs_step_conf": 0.724773108959198,
"adv/ratio_final_to_reasoning": 1.0467938353293162,
"adv/ratio_step_to_reasoning": 1.837201468501087,
"adv/std_final_conf": 0.6997731924057007,
"adv/std_reasoning": 0.6815393567085266,
"adv/std_step_conf": 0.9256146550178528,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.49444444444444446,
"calib/avg_num_step_conf": 8.41796875,
"calib/ece": 0.34860557768924305,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00011111111111106187,
"calib/mean_conf": 0.9900398406374502,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9901111111111109,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34860557768924305,
"calib/std_conf": 0.0006299357888781636,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9074658573596358,
"calib/step_q_c_n": 1318.0,
"calib/step_q_gap": 0.00268688483872781,
"calib/step_q_w": 0.904778972520908,
"calib/step_q_w_n": 837.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2572.0,
"completions/max_terminated_length": 2572.0,
"completions/mean_length": 812.296875,
"completions/mean_terminated_length": 825.1905517578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 441.0,
"epoch": 0.023466666666666667,
"grad_norm": 0.016305429860949516,
"kl": 0.0047855377197265625,
"learning_rate": 4.944444444444445e-06,
"loss": -0.0414,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.017927877604961395,
"mask/share_reasoning": 0.8585070967674255,
"mask/share_step_conf": 0.10794000327587128,
"num_tokens": 6871603.0,
"reward": 0.8370054364204407,
"reward_std": 0.19273146986961365,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.6357617378234863,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7163741588592529,
"step": 22
},
{
"adv/mean_abs_final_conf": 0.5349942445755005,
"adv/mean_abs_reasoning": 0.5196344256401062,
"adv/mean_abs_step_conf": 0.7811074256896973,
"adv/ratio_final_to_reasoning": 1.0295588940560922,
"adv/ratio_step_to_reasoning": 1.5031864463704427,
"adv/std_final_conf": 0.7593243718147278,
"adv/std_reasoning": 0.7576791644096375,
"adv/std_step_conf": 0.9311987161636353,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5031847133757962,
"calib/avg_num_step_conf": 8.15234375,
"calib/ece": 0.36702380952380953,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 6.369426751606788e-05,
"calib/mean_conf": 0.9900396825396826,
"calib/mu_c": 0.9900636942675157,
"calib/mu_w": 0.9899999999999997,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36702380952380953,
"calib/std_conf": 0.0006286896634029713,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.906015625,
"calib/step_q_c_n": 1280.0,
"calib/step_q_gap": -0.0004899512081784074,
"calib/step_q_w": 0.9065055762081784,
"calib/step_q_w_n": 807.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 3015.0,
"completions/max_terminated_length": 3015.0,
"completions/mean_length": 824.48046875,
"completions/mean_terminated_length": 834.2569580078125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 494.0,
"epoch": 0.024533333333333334,
"grad_norm": 0.018598461523652077,
"kl": 0.0061092376708984375,
"learning_rate": 4.9166666666666665e-06,
"loss": -0.0258,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.01792784407734871,
"mask/share_reasoning": 0.8628264665603638,
"mask/share_step_conf": 0.10752691328525543,
"num_tokens": 7186606.0,
"reward": 0.8155762553215027,
"reward_std": 0.2864585518836975,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6206049919128418,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.6910161972045898,
"step": 23
},
{
"adv/mean_abs_final_conf": 0.6040255427360535,
"adv/mean_abs_reasoning": 0.5883716344833374,
"adv/mean_abs_step_conf": 0.7876837253570557,
"adv/ratio_final_to_reasoning": 1.0266054774487252,
"adv/ratio_step_to_reasoning": 1.3387520390046315,
"adv/std_final_conf": 0.8085963726043701,
"adv/std_reasoning": 0.8101164102554321,
"adv/std_step_conf": 0.9350883364677429,
"calib/answer_extract_rate": 0.91015625,
"calib/auroc": 0.5038461538461538,
"calib/avg_num_step_conf": 8.375,
"calib/ece": 0.43210300429184556,
"calib/final_conf_rate": 0.91015625,
"calib/format_rate": 0.90625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 7.692307692352962e-05,
"calib/mean_conf": 0.9900429184549356,
"calib/mu_c": 0.9900769230769232,
"calib/mu_w": 0.9899999999999997,
"calib/nonempty_final_conf_rate": 0.91015625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.43210300429184556,
"calib/std_conf": 0.0006537144296878898,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9081593927893739,
"calib/step_q_c_n": 1054.0,
"calib/step_q_gap": 0.0001777414132270172,
"calib/step_q_w": 0.9079816513761468,
"calib/step_q_w_n": 1090.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2672.0,
"completions/max_terminated_length": 2672.0,
"completions/mean_length": 863.80859375,
"completions/mean_terminated_length": 917.5726928710938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 492.0,
"epoch": 0.0256,
"grad_norm": 0.021909039467573166,
"kl": 0.0059413909912109375,
"learning_rate": 4.888888888888889e-06,
"loss": -0.3015,
"mask/has_final_conf_rate": 0.91015625,
"mask/share_final_conf": 0.015839194878935814,
"mask/share_reasoning": 0.831562340259552,
"mask/share_step_conf": 0.09400475025177002,
"num_tokens": 7512253.0,
"reward": 0.7043561935424805,
"reward_std": 0.2977864742279053,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.5118628740310669,
"rewards/format_reward_step": 0.90625,
"rewards/step_l2_reward": 0.6140369176864624,
"step": 24
},
{
"adv/mean_abs_final_conf": 0.3722795844078064,
"adv/mean_abs_reasoning": 0.3595278859138489,
"adv/mean_abs_step_conf": 0.7372299432754517,
"adv/ratio_final_to_reasoning": 1.0354678983009766,
"adv/ratio_step_to_reasoning": 2.050550102397645,
"adv/std_final_conf": 0.6617670655250549,
"adv/std_reasoning": 0.6612728238105774,
"adv/std_step_conf": 0.9230087399482727,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.50625,
"calib/avg_num_step_conf": 7.9375,
"calib/ece": 0.35007600000000005,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00011875000000005631,
"calib/mean_conf": 0.9900760000000001,
"calib/mu_c": 0.99011875,
"calib/mu_w": 0.99,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35007600000000005,
"calib/std_conf": 0.0008474809732377484,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.90592277992278,
"calib/step_q_c_n": 1295.0,
"calib/step_q_gap": 0.0039960499363485,
"calib/step_q_w": 0.9019267299864315,
"calib/step_q_w_n": 737.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2920.0,
"completions/max_terminated_length": 2920.0,
"completions/mean_length": 801.48046875,
"completions/mean_terminated_length": 804.6235961914062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 437.0,
"epoch": 0.02666666666666667,
"grad_norm": 0.0172025877982378,
"kl": 0.008014678955078125,
"learning_rate": 4.861111111111111e-06,
"loss": -0.0154,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.018549812957644463,
"mask/share_reasoning": 0.8690973520278931,
"mask/share_step_conf": 0.10844656825065613,
"num_tokens": 7820656.0,
"reward": 0.823807954788208,
"reward_std": 0.194789856672287,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.631934404373169,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.6953690052032471,
"step": 25
},
{
"adv/mean_abs_final_conf": 0.44211018085479736,
"adv/mean_abs_reasoning": 0.41643744707107544,
"adv/mean_abs_step_conf": 0.7278153300285339,
"adv/ratio_final_to_reasoning": 1.0616484755736681,
"adv/ratio_step_to_reasoning": 1.747718259122634,
"adv/std_final_conf": 0.7366110682487488,
"adv/std_reasoning": 0.7205003499984741,
"adv/std_step_conf": 0.9267618656158447,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.4888888888888889,
"calib/avg_num_step_conf": 8.203125,
"calib/ece": 0.35008000000000006,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00022222222222212373,
"calib/mean_conf": 0.9900800000000001,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9902222222222221,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35008000000000006,
"calib/std_conf": 0.0008908422980528043,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9057789716039908,
"calib/step_q_c_n": 1303.0,
"calib/step_q_gap": 0.0012243919302140638,
"calib/step_q_w": 0.9045545796737767,
"calib/step_q_w_n": 797.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2958.0,
"completions/max_terminated_length": 2958.0,
"completions/mean_length": 834.125,
"completions/mean_terminated_length": 840.6929321289062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 494.0,
"epoch": 0.027733333333333332,
"grad_norm": 0.018005995079874992,
"kl": 0.00823354721069336,
"learning_rate": 4.833333333333333e-06,
"loss": -0.0557,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.01745178923010826,
"mask/share_reasoning": 0.8718188405036926,
"mask/share_step_conf": 0.10291685163974762,
"num_tokens": 8139432.0,
"reward": 0.8200647830963135,
"reward_std": 0.19963213801383972,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.6317781209945679,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.6880388855934143,
"step": 26
},
{
"adv/mean_abs_final_conf": 0.4970061182975769,
"adv/mean_abs_reasoning": 0.46902090311050415,
"adv/mean_abs_step_conf": 0.8002022504806519,
"adv/ratio_final_to_reasoning": 1.0596673090718929,
"adv/ratio_step_to_reasoning": 1.7061121267171313,
"adv/std_final_conf": 0.7211012244224548,
"adv/std_reasoning": 0.7208249568939209,
"adv/std_step_conf": 0.9283366799354553,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 8.91015625,
"calib/ece": 0.4511836734693877,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 8.849557522139229e-05,
"calib/mean_conf": 0.9899591836734694,
"calib/mu_c": 0.9900000000000001,
"calib/mu_w": 0.9899115044247787,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4511836734693877,
"calib/std_conf": 0.0014279882191457894,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9062722419928826,
"calib/step_q_c_n": 1124.0,
"calib/step_q_gap": -0.0009619844548270917,
"calib/step_q_w": 0.9072342264477097,
"calib/step_q_w_n": 1157.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3013.0,
"completions/max_terminated_length": 3013.0,
"completions/mean_length": 828.1015625,
"completions/mean_terminated_length": 858.2753295898438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 488.0,
"epoch": 0.0288,
"grad_norm": 0.017703456804156303,
"kl": 0.0074367523193359375,
"learning_rate": 4.805555555555556e-06,
"loss": -0.1344,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016773667186498642,
"mask/share_reasoning": 0.847740888595581,
"mask/share_step_conf": 0.10032917559146881,
"num_tokens": 8456642.0,
"reward": 0.7276268005371094,
"reward_std": 0.2555812895298004,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.5244327783584595,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6362894773483276,
"step": 27
},
{
"adv/mean_abs_final_conf": 0.4407339096069336,
"adv/mean_abs_reasoning": 0.4065035879611969,
"adv/mean_abs_step_conf": 0.7146235704421997,
"adv/ratio_final_to_reasoning": 1.0842066900747853,
"adv/ratio_step_to_reasoning": 1.75797604647567,
"adv/std_final_conf": 0.7218397855758667,
"adv/std_reasoning": 0.7207350134849548,
"adv/std_step_conf": 0.9135732650756836,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 7.890625,
"calib/ece": 0.2768852459016393,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 2.220446049250313e-16,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9899999999999998,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2768852459016393,
"calib/std_conf": 0.0,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9043002915451895,
"calib/step_q_c_n": 1372.0,
"calib/step_q_gap": -0.004866375121477251,
"calib/step_q_w": 0.9091666666666668,
"calib/step_q_w_n": 648.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2613.0,
"completions/max_terminated_length": 2613.0,
"completions/mean_length": 850.53515625,
"completions/mean_terminated_length": 881.5263671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 468.0,
"epoch": 0.029866666666666666,
"grad_norm": 0.015081276185810566,
"kl": 0.0077877044677734375,
"learning_rate": 4.777777777777778e-06,
"loss": -0.1394,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.016542209312319756,
"mask/share_reasoning": 0.8544609546661377,
"mask/share_step_conf": 0.09384053945541382,
"num_tokens": 8781323.0,
"reward": 0.8726210594177246,
"reward_std": 0.2400265485048294,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/final_brier_reward_step": 0.6850608587265015,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7336186170578003,
"step": 28
},
{
"adv/mean_abs_final_conf": 0.48541906476020813,
"adv/mean_abs_reasoning": 0.47797638177871704,
"adv/mean_abs_step_conf": 0.7424254417419434,
"adv/ratio_final_to_reasoning": 1.0155712358710995,
"adv/ratio_step_to_reasoning": 1.5532680484736903,
"adv/std_final_conf": 0.7575644850730896,
"adv/std_reasoning": 0.757592499256134,
"adv/std_step_conf": 0.9264086484909058,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5134751773049645,
"calib/avg_num_step_conf": 7.828125,
"calib/ece": 0.4047261410788382,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0006638297872341159,
"calib/mean_conf": 0.9897883817427386,
"calib/mu_c": 0.990063829787234,
"calib/mu_w": 0.9893999999999998,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4047261410788382,
"calib/std_conf": 0.002785713008604025,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9064064748201439,
"calib/step_q_c_n": 1112.0,
"calib/step_q_gap": -0.0006450946865825813,
"calib/step_q_w": 0.9070515695067265,
"calib/step_q_w_n": 892.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2696.0,
"completions/max_terminated_length": 2696.0,
"completions/mean_length": 887.25,
"completions/mean_terminated_length": 927.085693359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 361.0,
"epoch": 0.030933333333333334,
"grad_norm": 0.020906547084450722,
"kl": 0.0071125030517578125,
"learning_rate": 4.75e-06,
"loss": -0.1638,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.015757828950881958,
"mask/share_reasoning": 0.8528798222541809,
"mask/share_step_conf": 0.08839359879493713,
"num_tokens": 9115587.0,
"reward": 0.7560781836509705,
"reward_std": 0.2232230305671692,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/final_brier_reward_step": 0.5589569807052612,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.654761791229248,
"step": 29
},
{
"adv/mean_abs_final_conf": 0.6214361190795898,
"adv/mean_abs_reasoning": 0.5938949584960938,
"adv/mean_abs_step_conf": 0.7599336504936218,
"adv/ratio_final_to_reasoning": 1.0463737908353994,
"adv/ratio_step_to_reasoning": 1.2795758570134759,
"adv/std_final_conf": 0.8352395296096802,
"adv/std_reasoning": 0.8266966938972473,
"adv/std_step_conf": 0.9322589039802551,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.49640287769784175,
"calib/avg_num_step_conf": 8.03125,
"calib/ece": 0.41549586776859515,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0002158273381293041,
"calib/mean_conf": 0.9898760330578513,
"calib/mu_c": 0.9897841726618704,
"calib/mu_w": 0.9899999999999997,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.41549586776859515,
"calib/std_conf": 0.0019244844664785174,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9041459074733096,
"calib/step_q_c_n": 1124.0,
"calib/step_q_gap": 0.0051920448123654594,
"calib/step_q_w": 0.8989538626609441,
"calib/step_q_w_n": 932.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2493.0,
"completions/max_terminated_length": 2493.0,
"completions/mean_length": 853.328125,
"completions/mean_terminated_length": 877.3172607421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 518.0,
"epoch": 0.032,
"grad_norm": 0.02036236599087715,
"kl": 0.008612632751464844,
"learning_rate": 4.722222222222222e-06,
"loss": -0.1238,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016304276883602142,
"mask/share_reasoning": 0.8609112501144409,
"mask/share_step_conf": 0.09544073045253754,
"num_tokens": 9441023.0,
"reward": 0.7334448099136353,
"reward_std": 0.3001938462257385,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.550837516784668,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.6191771030426025,
"step": 30
},
{
"adv/mean_abs_final_conf": 0.4535819888114929,
"adv/mean_abs_reasoning": 0.4273011088371277,
"adv/mean_abs_step_conf": 0.7394912838935852,
"adv/ratio_final_to_reasoning": 1.0615043570700926,
"adv/ratio_step_to_reasoning": 1.730609325835973,
"adv/std_final_conf": 0.718247652053833,
"adv/std_reasoning": 0.7207338213920593,
"adv/std_step_conf": 0.9325244426727295,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.49167305414256846,
"calib/avg_num_step_conf": 8.5078125,
"calib/ece": 0.5190791666666668,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0003363528673956395,
"calib/mean_conf": 0.9899125000000001,
"calib/mu_c": 0.9897345132743365,
"calib/mu_w": 0.9900708661417321,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.5190791666666668,
"calib/std_conf": 0.0020198623096637073,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9051240560949299,
"calib/step_q_c_n": 927.0,
"calib/step_q_gap": 0.013545319084538154,
"calib/step_q_w": 0.8915787370103917,
"calib/step_q_w_n": 1251.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 3058.0,
"completions/max_terminated_length": 3058.0,
"completions/mean_length": 950.38671875,
"completions/mean_terminated_length": 989.020263671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 511.0,
"epoch": 0.03306666666666667,
"grad_norm": 0.014488758519291878,
"kl": 0.007906913757324219,
"learning_rate": 4.694444444444445e-06,
"loss": -0.1558,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.01523461751639843,
"mask/share_reasoning": 0.8540059924125671,
"mask/share_step_conf": 0.09169687330722809,
"num_tokens": 9790234.0,
"reward": 0.6493246555328369,
"reward_std": 0.2197844237089157,
"rewards/accuracy_reward_step": 0.44140625,
"rewards/final_brier_reward_step": 0.4511585831642151,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.5717094540596008,
"step": 31
},
{
"adv/mean_abs_final_conf": 0.4951266944408417,
"adv/mean_abs_reasoning": 0.46537792682647705,
"adv/mean_abs_step_conf": 0.7736719250679016,
"adv/ratio_final_to_reasoning": 1.0639238904544281,
"adv/ratio_step_to_reasoning": 1.6624594345154158,
"adv/std_final_conf": 0.7361029982566833,
"adv/std_reasoning": 0.7206937670707703,
"adv/std_step_conf": 0.9204629063606262,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.49647139431312093,
"calib/avg_num_step_conf": 8.72265625,
"calib/ece": 0.41995491803278684,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9959016393442623,
"calib/gap": 0.0006994861253852847,
"calib/mean_conf": 0.9896270491803278,
"calib/mu_c": 0.9899280575539567,
"calib/mu_w": 0.9892285714285715,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.41995491803278684,
"calib/std_conf": 0.005813727915569303,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9043189964157706,
"calib/step_q_c_n": 1116.0,
"calib/step_q_gap": -0.007287986574381566,
"calib/step_q_w": 0.9116069829901522,
"calib/step_q_w_n": 1117.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2998.0,
"completions/max_terminated_length": 2998.0,
"completions/mean_length": 839.625,
"completions/mean_terminated_length": 870.2186279296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 453.0,
"epoch": 0.034133333333333335,
"grad_norm": 0.017103182151913643,
"kl": 0.010321617126464844,
"learning_rate": 4.666666666666667e-06,
"loss": -0.1508,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01689147762954235,
"mask/share_reasoning": 0.8503791093826294,
"mask/share_step_conf": 0.0975731611251831,
"num_tokens": 10111882.0,
"reward": 0.7398710250854492,
"reward_std": 0.22319459915161133,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.5509277582168579,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6303766965866089,
"step": 32
},
{
"adv/mean_abs_final_conf": 0.41008493304252625,
"adv/mean_abs_reasoning": 0.4098625183105469,
"adv/mean_abs_step_conf": 0.741827130317688,
"adv/ratio_final_to_reasoning": 1.0005426569204137,
"adv/ratio_step_to_reasoning": 1.8099413758923335,
"adv/std_final_conf": 0.7017787098884583,
"adv/std_reasoning": 0.7012677788734436,
"adv/std_step_conf": 0.9291035532951355,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.49137931034482757,
"calib/avg_num_step_conf": 8.2578125,
"calib/ece": 0.4467637795275591,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0001551724137930366,
"calib/mean_conf": 0.9900708661417323,
"calib/mu_c": 0.99,
"calib/mu_w": 0.990155172413793,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4467637795275591,
"calib/std_conf": 0.000795470468054303,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9031728288907998,
"calib/step_q_c_n": 1163.0,
"calib/step_q_gap": 0.01485316537870729,
"calib/step_q_w": 0.8883196635120925,
"calib/step_q_w_n": 951.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2178.0,
"completions/max_terminated_length": 2178.0,
"completions/mean_length": 828.76171875,
"completions/mean_terminated_length": 835.2874145507812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 378.0,
"epoch": 0.0352,
"grad_norm": 0.022355087101459503,
"kl": 0.011810302734375,
"learning_rate": 4.638888888888889e-06,
"loss": -0.0232,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.017790237441658974,
"mask/share_reasoning": 0.8689032793045044,
"mask/share_step_conf": 0.1054939404129982,
"num_tokens": 10430917.0,
"reward": 0.759544849395752,
"reward_std": 0.20857521891593933,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.5478081703186035,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.6658127307891846,
"step": 33
},
{
"adv/mean_abs_final_conf": 0.5660641193389893,
"adv/mean_abs_reasoning": 0.524277925491333,
"adv/mean_abs_step_conf": 0.7607817649841309,
"adv/ratio_final_to_reasoning": 1.0797023712346765,
"adv/ratio_step_to_reasoning": 1.4511039431445747,
"adv/std_final_conf": 0.7932949066162109,
"adv/std_reasoning": 0.7755244374275208,
"adv/std_step_conf": 0.9272559881210327,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5062840803298311,
"calib/avg_num_step_conf": 8.11328125,
"calib/ece": 0.40353413654618486,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00012568160659687244,
"calib/mean_conf": 0.9898795180722892,
"calib/mu_c": 0.9899315068493151,
"calib/mu_w": 0.9898058252427182,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40353413654618486,
"calib/std_conf": 0.0010910102576069185,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9026694915254238,
"calib/step_q_c_n": 1180.0,
"calib/step_q_gap": 0.008622668782948817,
"calib/step_q_w": 0.8940468227424749,
"calib/step_q_w_n": 897.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2749.0,
"completions/max_terminated_length": 2749.0,
"completions/mean_length": 801.87890625,
"completions/mean_terminated_length": 817.8526000976562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 525.0,
"epoch": 0.03626666666666667,
"grad_norm": 0.017718534916639328,
"kl": 0.012401580810546875,
"learning_rate": 4.611111111111112e-06,
"loss": -0.0627,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01772000640630722,
"mask/share_reasoning": 0.8578258752822876,
"mask/share_step_conf": 0.10492290556430817,
"num_tokens": 10741310.0,
"reward": 0.8029850721359253,
"reward_std": 0.2710467576980591,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.5745089054107666,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7236486673355103,
"step": 34
},
{
"adv/mean_abs_final_conf": 0.5090737342834473,
"adv/mean_abs_reasoning": 0.48734915256500244,
"adv/mean_abs_step_conf": 0.7572380900382996,
"adv/ratio_final_to_reasoning": 1.0445770380518866,
"adv/ratio_step_to_reasoning": 1.5537896927753443,
"adv/std_final_conf": 0.7659239172935486,
"adv/std_reasoning": 0.7575402855873108,
"adv/std_step_conf": 0.9266098141670227,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.501396462295518,
"calib/avg_num_step_conf": 7.76171875,
"calib/ece": 0.4034939759036146,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00010839207341373225,
"calib/mean_conf": 0.989839357429719,
"calib/mu_c": 0.9897945205479451,
"calib/mu_w": 0.9899029126213589,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4034939759036146,
"calib/std_conf": 0.001997563070754361,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8986625874125875,
"calib/step_q_c_n": 1144.0,
"calib/step_q_gap": 0.011473975312943585,
"calib/step_q_w": 0.887188612099644,
"calib/step_q_w_n": 843.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2988.0,
"completions/max_terminated_length": 2988.0,
"completions/mean_length": 904.2890625,
"completions/mean_terminated_length": 922.3027954101562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 496.0,
"epoch": 0.037333333333333336,
"grad_norm": 0.03180506452918053,
"kl": 0.010568618774414062,
"learning_rate": 4.583333333333333e-06,
"loss": -0.1109,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01625356636941433,
"mask/share_reasoning": 0.8752022385597229,
"mask/share_step_conf": 0.08901292085647583,
"num_tokens": 11082064.0,
"reward": 0.7653804421424866,
"reward_std": 0.23321697115898132,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.5783331990242004,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.6438338756561279,
"step": 35
},
{
"adv/mean_abs_final_conf": 0.38228654861450195,
"adv/mean_abs_reasoning": 0.3659414052963257,
"adv/mean_abs_step_conf": 0.7873794436454773,
"adv/ratio_final_to_reasoning": 1.0446660123222202,
"adv/ratio_step_to_reasoning": 2.151654424040611,
"adv/std_final_conf": 0.6414028406143188,
"adv/std_reasoning": 0.6404587030410767,
"adv/std_step_conf": 0.9225838780403137,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 8.484375,
"calib/ece": 0.24592000000000003,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0003124999999997158,
"calib/mean_conf": 0.98992,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9896875000000002,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.24592000000000003,
"calib/std_conf": 0.0019983993594874894,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9021782799748902,
"calib/step_q_c_n": 1593.0,
"calib/step_q_gap": -0.003296676847216906,
"calib/step_q_w": 0.9054749568221071,
"calib/step_q_w_n": 579.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2509.0,
"completions/max_terminated_length": 2509.0,
"completions/mean_length": 794.8125,
"completions/mean_terminated_length": 807.4285888671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 388.0,
"epoch": 0.0384,
"grad_norm": 0.019437743350863457,
"kl": 0.01312255859375,
"learning_rate": 4.555555555555556e-06,
"loss": -0.0454,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.01833474636077881,
"mask/share_reasoning": 0.853883683681488,
"mask/share_step_conf": 0.11215655505657196,
"num_tokens": 11388248.0,
"reward": 0.9233396053314209,
"reward_std": 0.20475012063980103,
"rewards/accuracy_reward_step": 0.7265625,
"rewards/final_brier_reward_step": 0.7316156029701233,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7744386196136475,
"step": 36
},
{
"adv/mean_abs_final_conf": 0.42273592948913574,
"adv/mean_abs_reasoning": 0.3929756283760071,
"adv/mean_abs_step_conf": 0.7315104007720947,
"adv/ratio_final_to_reasoning": 1.0757306534151105,
"adv/ratio_step_to_reasoning": 1.8614650577571459,
"adv/std_final_conf": 0.6996673941612244,
"adv/std_reasoning": 0.6817983388900757,
"adv/std_step_conf": 0.9243629574775696,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 8.83984375,
"calib/ece": 0.517337552742616,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00016800000000016801,
"calib/mean_conf": 0.9899113924050632,
"calib/mu_c": 0.9900000000000001,
"calib/mu_w": 0.9898319999999999,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.517337552742616,
"calib/std_conf": 0.0020325819049791453,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9004334038054967,
"calib/step_q_c_n": 946.0,
"calib/step_q_gap": 0.003611459993803412,
"calib/step_q_w": 0.8968219438116933,
"calib/step_q_w_n": 1317.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2960.0,
"completions/max_terminated_length": 2960.0,
"completions/mean_length": 845.92578125,
"completions/mean_terminated_length": 894.8635864257812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 517.0,
"epoch": 0.039466666666666664,
"grad_norm": 0.017507895827293396,
"kl": 0.010918617248535156,
"learning_rate": 4.527777777777778e-06,
"loss": -0.1952,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.01621920056641102,
"mask/share_reasoning": 0.8316308856010437,
"mask/share_step_conf": 0.09746243804693222,
"num_tokens": 11711901.0,
"reward": 0.6471365094184875,
"reward_std": 0.20595934987068176,
"rewards/accuracy_reward_step": 0.44140625,
"rewards/final_brier_reward_step": 0.44733160734176636,
"rewards/format_reward_step": 0.92578125,
"rewards/step_l2_reward": 0.5735039710998535,
"step": 37
},
{
"adv/mean_abs_final_conf": 0.4951311945915222,
"adv/mean_abs_reasoning": 0.46970370411872864,
"adv/mean_abs_step_conf": 0.7426836490631104,
"adv/ratio_final_to_reasoning": 1.054135171278884,
"adv/ratio_step_to_reasoning": 1.581174775822887,
"adv/std_final_conf": 0.7573981285095215,
"adv/std_reasoning": 0.7577217817306519,
"adv/std_step_conf": 0.9252141714096069,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 7.953125,
"calib/ece": 0.4544351464435147,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 3.3306690738754696e-16,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.9900000000000001,
"calib/mu_w": 0.9899999999999998,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4544351464435147,
"calib/std_conf": 0.0,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9032015065913371,
"calib/step_q_c_n": 1062.0,
"calib/step_q_gap": 0.008704586673472625,
"calib/step_q_w": 0.8944969199178645,
"calib/step_q_w_n": 974.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2911.0,
"completions/max_terminated_length": 2911.0,
"completions/mean_length": 852.28515625,
"completions/mean_terminated_length": 890.5509643554688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 507.0,
"epoch": 0.04053333333333333,
"grad_norm": 0.023108594119548798,
"kl": 0.011475563049316406,
"learning_rate": 4.5e-06,
"loss": -0.116,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.016291677951812744,
"mask/share_reasoning": 0.8450714349746704,
"mask/share_step_conf": 0.09566812217235565,
"num_tokens": 12036974.0,
"reward": 0.6965200901031494,
"reward_std": 0.24555054306983948,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.5085784792900085,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.5977429747581482,
"step": 38
},
{
"adv/mean_abs_final_conf": 0.4412452280521393,
"adv/mean_abs_reasoning": 0.42349773645401,
"adv/mean_abs_step_conf": 0.745875895023346,
"adv/ratio_final_to_reasoning": 1.0419069337813487,
"adv/ratio_step_to_reasoning": 1.7612275835725624,
"adv/std_final_conf": 0.6999874711036682,
"adv/std_reasoning": 0.7014147639274597,
"adv/std_step_conf": 0.9331080317497253,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.4903186768858411,
"calib/avg_num_step_conf": 8.2265625,
"calib/ece": 0.44319591836734695,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0001325803415352933,
"calib/mean_conf": 0.990134693877551,
"calib/mu_c": 0.9900746268656716,
"calib/mu_w": 0.9902072072072069,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44319591836734695,
"calib/std_conf": 0.001073911602870015,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.902592429577465,
"calib/step_q_c_n": 1136.0,
"calib/step_q_gap": 0.008604800711485505,
"calib/step_q_w": 0.8939876288659795,
"calib/step_q_w_n": 970.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3003.0,
"completions/max_terminated_length": 3003.0,
"completions/mean_length": 890.3828125,
"completions/mean_terminated_length": 904.5159301757812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 472.0,
"epoch": 0.0416,
"grad_norm": 0.019972793757915497,
"kl": 0.013368606567382812,
"learning_rate": 4.472222222222223e-06,
"loss": -0.1042,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016698993742465973,
"mask/share_reasoning": 0.8672369122505188,
"mask/share_step_conf": 0.10043909400701523,
"num_tokens": 12371000.0,
"reward": 0.735222578048706,
"reward_std": 0.21249210834503174,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/final_brier_reward_step": 0.5318354368209839,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6425158977508545,
"step": 39
},
{
"adv/mean_abs_final_conf": 0.43904614448547363,
"adv/mean_abs_reasoning": 0.4147336483001709,
"adv/mean_abs_step_conf": 0.7419914603233337,
"adv/ratio_final_to_reasoning": 1.0586219523902871,
"adv/ratio_step_to_reasoning": 1.7890794811669202,
"adv/std_final_conf": 0.7193395495414734,
"adv/std_reasoning": 0.7012978792190552,
"adv/std_step_conf": 0.9300381541252136,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.49990397541770687,
"calib/avg_num_step_conf": 8.4140625,
"calib/ece": 0.48207600000000006,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 5.569425772788961e-06,
"calib/mean_conf": 0.9900760000000001,
"calib/mu_c": 0.9900787401574801,
"calib/mu_w": 0.9900731707317073,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.48207600000000006,
"calib/std_conf": 0.0008474809732377484,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9001939864209505,
"calib/step_q_c_n": 1031.0,
"calib/step_q_gap": -0.005035755342184034,
"calib/step_q_w": 0.9052297417631345,
"calib/step_q_w_n": 1123.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2676.0,
"completions/max_terminated_length": 2676.0,
"completions/mean_length": 884.18359375,
"completions/mean_terminated_length": 898.2183227539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 478.0,
"epoch": 0.042666666666666665,
"grad_norm": 0.021310077980160713,
"kl": 0.014337539672851562,
"learning_rate": 4.444444444444444e-06,
"loss": -0.0803,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.016493458300828934,
"mask/share_reasoning": 0.8709720969200134,
"mask/share_step_conf": 0.09690943360328674,
"num_tokens": 12704111.0,
"reward": 0.6947786808013916,
"reward_std": 0.2072494924068451,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.5055359601974487,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.5894901156425476,
"step": 40
},
{
"adv/mean_abs_final_conf": 0.43155983090400696,
"adv/mean_abs_reasoning": 0.4150729179382324,
"adv/mean_abs_step_conf": 0.7384018898010254,
"adv/ratio_final_to_reasoning": 1.0397205219932657,
"adv/ratio_step_to_reasoning": 1.778969086850682,
"adv/std_final_conf": 0.712813675403595,
"adv/std_reasoning": 0.7012646794319153,
"adv/std_step_conf": 0.9279786944389343,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.4975369458128079,
"calib/avg_num_step_conf": 8.4375,
"calib/ece": 0.19380392156862758,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00014778325123177005,
"calib/mean_conf": 0.9898823529411765,
"calib/mu_c": 0.9898522167487684,
"calib/mu_w": 0.9900000000000002,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.19380392156862758,
"calib/std_conf": 0.0018749855824128521,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9017281780902168,
"calib/step_q_c_n": 1707.0,
"calib/step_q_gap": -0.0025985327265601876,
"calib/step_q_w": 0.904326710816777,
"calib/step_q_w_n": 453.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1557.0,
"completions/max_terminated_length": 1557.0,
"completions/mean_length": 798.296875,
"completions/mean_terminated_length": 801.427490234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 445.0,
"epoch": 0.04373333333333333,
"grad_norm": 0.015647921711206436,
"kl": 0.014879226684570312,
"learning_rate": 4.416666666666667e-06,
"loss": -0.0233,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.018516331911087036,
"mask/share_reasoning": 0.8660091161727905,
"mask/share_step_conf": 0.11156836152076721,
"num_tokens": 13015723.0,
"reward": 0.9741901159286499,
"reward_std": 0.2122625857591629,
"rewards/accuracy_reward_step": 0.79296875,
"rewards/final_brier_reward_step": 0.7969257831573486,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7936419248580933,
"step": 41
},
{
"adv/mean_abs_final_conf": 0.37356019020080566,
"adv/mean_abs_reasoning": 0.35552409291267395,
"adv/mean_abs_step_conf": 0.7282037734985352,
"adv/ratio_final_to_reasoning": 1.0507310127433807,
"adv/ratio_step_to_reasoning": 2.0482543602956356,
"adv/std_final_conf": 0.6794334650039673,
"adv/std_reasoning": 0.6613250374794006,
"adv/std_step_conf": 0.9256712198257446,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.49843260188087773,
"calib/avg_num_step_conf": 8.3984375,
"calib/ece": 0.3958151639344262,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -2.2633228840107655e-05,
"calib/mean_conf": 0.9900774590163934,
"calib/mu_c": 0.9900682758620689,
"calib/mu_w": 0.990090909090909,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3958151639344262,
"calib/std_conf": 0.0008530226502051216,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8990497147514263,
"calib/step_q_c_n": 1227.0,
"calib/step_q_gap": -0.005915615692777321,
"calib/step_q_w": 0.9049653304442036,
"calib/step_q_w_n": 923.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2788.0,
"completions/max_terminated_length": 2788.0,
"completions/mean_length": 804.41796875,
"completions/mean_terminated_length": 827.0321044921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 476.0,
"epoch": 0.0448,
"grad_norm": 0.017200641334056854,
"kl": 0.01670074462890625,
"learning_rate": 4.388888888888889e-06,
"loss": -0.094,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01786448247730732,
"mask/share_reasoning": 0.847210168838501,
"mask/share_step_conf": 0.10758156329393387,
"num_tokens": 13326022.0,
"reward": 0.7614730596542358,
"reward_std": 0.188564732670784,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.5739758014678955,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.645064115524292,
"step": 42
},
{
"adv/mean_abs_final_conf": 0.48687943816185,
"adv/mean_abs_reasoning": 0.4713999032974243,
"adv/mean_abs_step_conf": 0.7664150595664978,
"adv/ratio_final_to_reasoning": 1.0328373738648373,
"adv/ratio_step_to_reasoning": 1.6258277827497498,
"adv/std_final_conf": 0.7410199642181396,
"adv/std_reasoning": 0.7394036054611206,
"adv/std_step_conf": 0.930462121963501,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5029940119760479,
"calib/avg_num_step_conf": 8.07421875,
"calib/ece": 0.3246932270916336,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 4.7904191616821024e-05,
"calib/mean_conf": 0.9900318725099603,
"calib/mu_c": 0.9900479041916166,
"calib/mu_w": 0.9899999999999998,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3246932270916336,
"calib/std_conf": 0.0010211801233396883,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8998834178131789,
"calib/step_q_c_n": 1381.0,
"calib/step_q_gap": -0.000626786268453694,
"calib/step_q_w": 0.9005102040816326,
"calib/step_q_w_n": 686.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2775.0,
"completions/max_terminated_length": 2775.0,
"completions/mean_length": 845.00390625,
"completions/mean_terminated_length": 851.657470703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 491.0,
"epoch": 0.04586666666666667,
"grad_norm": 0.019190490245819092,
"kl": 0.016111373901367188,
"learning_rate": 4.361111111111112e-06,
"loss": -0.0029,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.017595965415239334,
"mask/share_reasoning": 0.8708884716033936,
"mask/share_step_conf": 0.10370305180549622,
"num_tokens": 13647567.0,
"reward": 0.8473519086837769,
"reward_std": 0.25601252913475037,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.6588077545166016,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7093334197998047,
"step": 43
},
{
"adv/mean_abs_final_conf": 0.5013235211372375,
"adv/mean_abs_reasoning": 0.4585108757019043,
"adv/mean_abs_step_conf": 0.7387799024581909,
"adv/ratio_final_to_reasoning": 1.0933732386822759,
"adv/ratio_step_to_reasoning": 1.6112592778246342,
"adv/std_final_conf": 0.7603207230567932,
"adv/std_reasoning": 0.739250659942627,
"adv/std_step_conf": 0.9254706501960754,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5048664122137405,
"calib/avg_num_step_conf": 7.95703125,
"calib/ece": 0.4678884462151396,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 9.732824427455267e-05,
"calib/mean_conf": 0.9898007968127491,
"calib/mu_c": 0.9898473282442747,
"calib/mu_w": 0.9897500000000001,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4678884462151396,
"calib/std_conf": 0.0013972651726494184,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9010283018867925,
"calib/step_q_c_n": 1060.0,
"calib/step_q_gap": 0.004262692879627772,
"calib/step_q_w": 0.8967656090071647,
"calib/step_q_w_n": 977.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2593.0,
"completions/max_terminated_length": 2593.0,
"completions/mean_length": 854.3046875,
"completions/mean_terminated_length": 861.031494140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 498.0,
"epoch": 0.046933333333333334,
"grad_norm": 0.024990690872073174,
"kl": 0.018238067626953125,
"learning_rate": 4.333333333333334e-06,
"loss": -0.037,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.01713503897190094,
"mask/share_reasoning": 0.8744046688079834,
"mask/share_step_conf": 0.10064785182476044,
"num_tokens": 13972589.0,
"reward": 0.7188483476638794,
"reward_std": 0.22061403095722198,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.5212242007255554,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.6180349588394165,
"step": 44
},
{
"adv/mean_abs_final_conf": 0.4545263648033142,
"adv/mean_abs_reasoning": 0.4324095547199249,
"adv/mean_abs_step_conf": 0.746199369430542,
"adv/ratio_final_to_reasoning": 1.0511478292789218,
"adv/ratio_step_to_reasoning": 1.7256773382675625,
"adv/std_final_conf": 0.7177832722663879,
"adv/std_reasoning": 0.7207047939300537,
"adv/std_step_conf": 0.9280015230178833,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.4965034965034965,
"calib/avg_num_step_conf": 9.17578125,
"calib/ece": 0.40356557377049174,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9959016393442623,
"calib/gap": -0.0006293706293705181,
"calib/mean_conf": 0.9896311475409836,
"calib/mu_c": 0.9893706293706293,
"calib/mu_w": 0.9899999999999998,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.40356557377049174,
"calib/std_conf": 0.005749840795617992,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8993268450932683,
"calib/step_q_c_n": 1233.0,
"calib/step_q_gap": -0.011890000784868016,
"calib/step_q_w": 0.9112168458781363,
"calib/step_q_w_n": 1116.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2809.0,
"completions/max_terminated_length": 2809.0,
"completions/mean_length": 866.65234375,
"completions/mean_terminated_length": 894.6088256835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 483.0,
"epoch": 0.048,
"grad_norm": 0.024536319077014923,
"kl": 0.018016815185546875,
"learning_rate": 4.305555555555556e-06,
"loss": -0.1097,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.016593432053923607,
"mask/share_reasoning": 0.849403977394104,
"mask/share_step_conf": 0.10275260359048843,
"num_tokens": 14299500.0,
"reward": 0.7554113268852234,
"reward_std": 0.21753597259521484,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.5662726163864136,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6429873704910278,
"step": 45
},
{
"adv/mean_abs_final_conf": 0.4345059394836426,
"adv/mean_abs_reasoning": 0.4139566719532013,
"adv/mean_abs_step_conf": 0.7318849563598633,
"adv/ratio_final_to_reasoning": 1.0496411072044864,
"adv/ratio_step_to_reasoning": 1.7680230950417064,
"adv/std_final_conf": 0.7010535597801208,
"adv/std_reasoning": 0.7015208005905151,
"adv/std_step_conf": 0.9241446256637573,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.49860270144387525,
"calib/avg_num_step_conf": 8.37109375,
"calib/ece": 0.449479674796748,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -5.4028877503609074e-05,
"calib/mean_conf": 0.990130081300813,
"calib/mu_c": 0.9901052631578947,
"calib/mu_w": 0.9901592920353983,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.449479674796748,
"calib/std_conf": 0.0010356204659467386,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9008309608540925,
"calib/step_q_c_n": 1124.0,
"calib/step_q_gap": 0.004519871550854093,
"calib/step_q_w": 0.8963110893032384,
"calib/step_q_w_n": 1019.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2791.0,
"completions/max_terminated_length": 2791.0,
"completions/mean_length": 925.25,
"completions/mean_terminated_length": 943.6812744140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 432.0,
"epoch": 0.04906666666666667,
"grad_norm": 0.018395164981484413,
"kl": 0.017238616943359375,
"learning_rate": 4.277777777777778e-06,
"loss": -0.1028,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.016764050349593163,
"mask/share_reasoning": 0.8601210713386536,
"mask/share_step_conf": 0.10358359664678574,
"num_tokens": 14641132.0,
"reward": 0.7202885746955872,
"reward_std": 0.2138737291097641,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5280463695526123,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6172181963920593,
"step": 46
},
{
"adv/mean_abs_final_conf": 0.43521004915237427,
"adv/mean_abs_reasoning": 0.432475209236145,
"adv/mean_abs_step_conf": 0.7342606782913208,
"adv/ratio_final_to_reasoning": 1.0063236917581002,
"adv/ratio_step_to_reasoning": 1.697809868889829,
"adv/std_final_conf": 0.6987658739089966,
"adv/std_reasoning": 0.7014555335044861,
"adv/std_step_conf": 0.9218693375587463,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.49151234567901236,
"calib/avg_num_step_conf": 7.94140625,
"calib/ece": 0.397320987654321,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00029924242424239633,
"calib/mean_conf": 0.9899135802469136,
"calib/mu_c": 0.9897916666666666,
"calib/mu_w": 0.990090909090909,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.397320987654321,
"calib/std_conf": 0.002007378555059689,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8949606299212599,
"calib/step_q_c_n": 1143.0,
"calib/step_q_gap": 0.002522427674068961,
"calib/step_q_w": 0.892438202247191,
"calib/step_q_w_n": 890.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2475.0,
"completions/max_terminated_length": 2475.0,
"completions/mean_length": 868.7578125,
"completions/mean_terminated_length": 896.7822265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 524.0,
"epoch": 0.050133333333333335,
"grad_norm": 0.023173121735453606,
"kl": 0.019071578979492188,
"learning_rate": 4.25e-06,
"loss": -0.1861,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.016180016100406647,
"mask/share_reasoning": 0.8588327169418335,
"mask/share_step_conf": 0.09373725205659866,
"num_tokens": 14969510.0,
"reward": 0.7658748626708984,
"reward_std": 0.20331431925296783,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.5700636506080627,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6593424081802368,
"step": 47
},
{
"adv/mean_abs_final_conf": 0.6294046640396118,
"adv/mean_abs_reasoning": 0.5752204656600952,
"adv/mean_abs_step_conf": 0.7964880466461182,
"adv/ratio_final_to_reasoning": 1.0941972715058694,
"adv/ratio_step_to_reasoning": 1.3846656963641009,
"adv/std_final_conf": 0.8177751302719116,
"adv/std_reasoning": 0.7930737137794495,
"adv/std_step_conf": 0.9295130968093872,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.5138781533272433,
"calib/avg_num_step_conf": 8.65625,
"calib/ece": 0.43153749999999996,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.9958333333333333,
"calib/gap": 0.009281357599606466,
"calib/mean_conf": 0.9857041666666667,
"calib/mu_c": 0.9898421052631579,
"calib/mu_w": 0.9805607476635514,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43153749999999996,
"calib/std_conf": 0.06379824957869577,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8976549053356283,
"calib/step_q_c_n": 1162.0,
"calib/step_q_gap": 0.0032526282957801023,
"calib/step_q_w": 0.8944022770398482,
"calib/step_q_w_n": 1054.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2969.0,
"completions/max_terminated_length": 2969.0,
"completions/mean_length": 828.91796875,
"completions/mean_terminated_length": 859.1214599609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 475.0,
"epoch": 0.0512,
"grad_norm": 0.02995246835052967,
"kl": 0.020765304565429688,
"learning_rate": 4.222222222222223e-06,
"loss": -0.1281,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.017236867919564247,
"mask/share_reasoning": 0.8404216766357422,
"mask/share_step_conf": 0.1071852594614029,
"num_tokens": 15285401.0,
"reward": 0.727776288986206,
"reward_std": 0.30010148882865906,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.531619131565094,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.6333083510398865,
"step": 48
},
{
"adv/mean_abs_final_conf": 0.44170159101486206,
"adv/mean_abs_reasoning": 0.4147529602050781,
"adv/mean_abs_step_conf": 0.7751125693321228,
"adv/ratio_final_to_reasoning": 1.0649751379627501,
"adv/ratio_step_to_reasoning": 1.868853615773741,
"adv/std_final_conf": 0.6997136473655701,
"adv/std_reasoning": 0.6818037033081055,
"adv/std_step_conf": 0.925434947013855,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.4942528735632184,
"calib/avg_num_step_conf": 8.40234375,
"calib/ece": 0.3465983606557377,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00011494252873545108,
"calib/mean_conf": 0.9900409836065573,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9901149425287353,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3465983606557377,
"calib/std_conf": 0.0006388711995131111,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8977125658389766,
"calib/step_q_c_n": 1329.0,
"calib/step_q_gap": 0.006617675328027706,
"calib/step_q_w": 0.8910948905109489,
"calib/step_q_w_n": 822.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 3055.0,
"completions/max_terminated_length": 3055.0,
"completions/mean_length": 813.41796875,
"completions/mean_terminated_length": 843.0567016601562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 466.0,
"epoch": 0.05226666666666667,
"grad_norm": 0.025971531867980957,
"kl": 0.018663406372070312,
"learning_rate": 4.194444444444445e-06,
"loss": -0.1239,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01712140068411827,
"mask/share_reasoning": 0.845194935798645,
"mask/share_step_conf": 0.1025274246931076,
"num_tokens": 15598172.0,
"reward": 0.7944972515106201,
"reward_std": 0.21691881120204926,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6159992218017578,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6604952216148376,
"step": 49
},
{
"adv/mean_abs_final_conf": 0.46414607763290405,
"adv/mean_abs_reasoning": 0.43221813440322876,
"adv/mean_abs_step_conf": 0.7735408544540405,
"adv/ratio_final_to_reasoning": 1.0738699760336496,
"adv/ratio_step_to_reasoning": 1.7897001372283514,
"adv/std_final_conf": 0.721068799495697,
"adv/std_reasoning": 0.7014055252075195,
"adv/std_step_conf": 0.9300537705421448,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5085791264889393,
"calib/avg_num_step_conf": 8.359375,
"calib/ece": 0.33388000000000007,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00017158252977889088,
"calib/mean_conf": 0.9898800000000001,
"calib/mu_c": 0.9899390243902437,
"calib/mu_w": 0.9897674418604648,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33388000000000007,
"calib/std_conf": 0.001088852607105297,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8999010600706715,
"calib/step_q_c_n": 1415.0,
"calib/step_q_gap": 0.004038991105154266,
"calib/step_q_w": 0.8958620689655172,
"calib/step_q_w_n": 725.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2104.0,
"completions/max_terminated_length": 2104.0,
"completions/mean_length": 836.765625,
"completions/mean_terminated_length": 846.6878051757812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 458.0,
"epoch": 0.05333333333333334,
"grad_norm": 0.03273485600948334,
"kl": 0.02405548095703125,
"learning_rate": 4.166666666666667e-06,
"loss": -0.064,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017580296844244003,
"mask/share_reasoning": 0.8621413707733154,
"mask/share_step_conf": 0.10855959355831146,
"num_tokens": 15917744.0,
"reward": 0.82176673412323,
"reward_std": 0.22579185664653778,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6473987698554993,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.6726971864700317,
"step": 50
},
{
"adv/mean_abs_final_conf": 0.43641749024391174,
"adv/mean_abs_reasoning": 0.40805771946907043,
"adv/mean_abs_step_conf": 0.7661856412887573,
"adv/ratio_final_to_reasoning": 1.0694994100632151,
"adv/ratio_step_to_reasoning": 1.8776403551087137,
"adv/std_final_conf": 0.6907274127006531,
"adv/std_reasoning": 0.6817567348480225,
"adv/std_step_conf": 0.925344705581665,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.4966442953020134,
"calib/avg_num_step_conf": 8.12890625,
"calib/ece": 0.37670781893004124,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0002013422818791799,
"calib/mean_conf": 0.9898765432098766,
"calib/mu_c": 0.9897986577181206,
"calib/mu_w": 0.9899999999999998,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.37670781893004124,
"calib/std_conf": 0.0019205369365560565,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8977551020408163,
"calib/step_q_c_n": 1274.0,
"calib/step_q_gap": 0.006379637356801582,
"calib/step_q_w": 0.8913754646840147,
"calib/step_q_w_n": 807.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 3061.0,
"completions/max_terminated_length": 3061.0,
"completions/mean_length": 784.70703125,
"completions/mean_terminated_length": 810.0201416015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 71.0,
"epoch": 0.0544,
"grad_norm": 0.028437655419111252,
"kl": 0.023008346557617188,
"learning_rate": 4.138888888888889e-06,
"loss": -0.1319,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.017778389155864716,
"mask/share_reasoning": 0.8434497714042664,
"mask/share_step_conf": 0.10752184689044952,
"num_tokens": 16227925.0,
"reward": 0.7702333927154541,
"reward_std": 0.208187073469162,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.5891964435577393,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.6458014249801636,
"step": 51
},
{
"adv/mean_abs_final_conf": 0.4756985306739807,
"adv/mean_abs_reasoning": 0.41691356897354126,
"adv/mean_abs_step_conf": 0.7791311740875244,
"adv/ratio_final_to_reasoning": 1.1410003561293782,
"adv/ratio_step_to_reasoning": 1.8688074269345039,
"adv/std_final_conf": 0.7390729188919067,
"adv/std_reasoning": 0.7014615535736084,
"adv/std_step_conf": 0.9296015501022339,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5012301587301587,
"calib/avg_num_step_conf": 7.96484375,
"calib/ece": 0.28135222672064764,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 9.920634920335125e-06,
"calib/mean_conf": 0.9898542510121456,
"calib/mu_c": 0.9898571428571427,
"calib/mu_w": 0.9898472222222223,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.28135222672064764,
"calib/std_conf": 0.0015595507455950335,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8934075967859751,
"calib/step_q_c_n": 1369.0,
"calib/step_q_gap": -0.007855089781189095,
"calib/step_q_w": 0.9012626865671642,
"calib/step_q_w_n": 670.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2799.0,
"completions/max_terminated_length": 2799.0,
"completions/mean_length": 875.765625,
"completions/mean_terminated_length": 896.7840576171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 530.0,
"epoch": 0.055466666666666664,
"grad_norm": 0.03539959341287613,
"kl": 0.02320098876953125,
"learning_rate": 4.111111111111111e-06,
"loss": -0.0999,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.01645956188440323,
"mask/share_reasoning": 0.8659517168998718,
"mask/share_step_conf": 0.09415123611688614,
"num_tokens": 16560073.0,
"reward": 0.8869292736053467,
"reward_std": 0.22455470263957977,
"rewards/accuracy_reward_step": 0.68359375,
"rewards/final_brier_reward_step": 0.6892030239105225,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.754967987537384,
"step": 52
},
{
"adv/mean_abs_final_conf": 0.5189765691757202,
"adv/mean_abs_reasoning": 0.49131283164024353,
"adv/mean_abs_step_conf": 0.7682706713676453,
"adv/ratio_final_to_reasoning": 1.056305750133009,
"adv/ratio_step_to_reasoning": 1.5637097626837477,
"adv/std_final_conf": 0.7610706686973572,
"adv/std_reasoning": 0.7577739357948303,
"adv/std_step_conf": 0.9258795976638794,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.4946581196581197,
"calib/avg_num_step_conf": 8.875,
"calib/ece": 0.3068695121951218,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00010860805860757328,
"calib/mean_conf": 0.9897963414634146,
"calib/mu_c": 0.989761904761905,
"calib/mu_w": 0.9898705128205125,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3068695121951218,
"calib/std_conf": 0.001672113101696819,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.890580693815988,
"calib/step_q_c_n": 1326.0,
"calib/step_q_gap": -0.01226645206139032,
"calib/step_q_w": 0.9028471458773784,
"calib/step_q_w_n": 946.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2791.0,
"completions/max_terminated_length": 2791.0,
"completions/mean_length": 825.0234375,
"completions/mean_terminated_length": 855.0850219726562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 424.0,
"epoch": 0.05653333333333333,
"grad_norm": 0.031232628971338272,
"kl": 0.039867401123046875,
"learning_rate": 4.083333333333334e-06,
"loss": -0.1512,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.01677112840116024,
"mask/share_reasoning": 0.8481372594833374,
"mask/share_step_conf": 0.09993541985750198,
"num_tokens": 16877103.0,
"reward": 0.8562500476837158,
"reward_std": 0.2774738669395447,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.6623198986053467,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7267426252365112,
"step": 53
},
{
"adv/mean_abs_final_conf": 0.34591519832611084,
"adv/mean_abs_reasoning": 0.3050154447555542,
"adv/mean_abs_step_conf": 0.6924338936805725,
"adv/ratio_final_to_reasoning": 1.1340907625295322,
"adv/ratio_step_to_reasoning": 2.2701601036482058,
"adv/std_final_conf": 0.6442092657089233,
"adv/std_reasoning": 0.6185151934623718,
"adv/std_step_conf": 0.9229856729507446,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.49490647229296564,
"calib/avg_num_step_conf": 8.42578125,
"calib/ece": 0.22404761904761905,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00010362694300536379,
"calib/mean_conf": 0.989920634920635,
"calib/mu_c": 0.9898963730569948,
"calib/mu_w": 0.9900000000000002,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.22404761904761905,
"calib/std_conf": 0.0012573793268059424,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.893818407960199,
"calib/step_q_c_n": 1608.0,
"calib/step_q_gap": 0.00039399994562694385,
"calib/step_q_w": 0.893424408014572,
"calib/step_q_w_n": 549.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2961.0,
"completions/max_terminated_length": 2961.0,
"completions/mean_length": 804.60546875,
"completions/mean_terminated_length": 810.94091796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 416.0,
"epoch": 0.0576,
"grad_norm": 0.02178231254220009,
"kl": 0.02564239501953125,
"learning_rate": 4.055555555555556e-06,
"loss": 0.0001,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.018358973786234856,
"mask/share_reasoning": 0.8616392612457275,
"mask/share_step_conf": 0.11218926310539246,
"num_tokens": 17189314.0,
"reward": 0.9394882321357727,
"reward_std": 0.15851876139640808,
"rewards/accuracy_reward_step": 0.75390625,
"rewards/final_brier_reward_step": 0.7584140300750732,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7729063034057617,
"step": 54
},
{
"adv/mean_abs_final_conf": 0.5040771961212158,
"adv/mean_abs_reasoning": 0.3714796006679535,
"adv/mean_abs_step_conf": 0.7354352474212646,
"adv/ratio_final_to_reasoning": 1.3569444868973695,
"adv/ratio_step_to_reasoning": 1.979745983625713,
"adv/std_final_conf": 0.7724276185035706,
"adv/std_reasoning": 0.681528627872467,
"adv/std_step_conf": 0.9326515793800354,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5210697977821266,
"calib/avg_num_step_conf": 8.36328125,
"calib/ece": 0.40023904382470127,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9840637450199203,
"calib/gap": 0.017532289628180098,
"calib/mean_conf": 0.9819123505976095,
"calib/mu_c": 0.9892465753424659,
"calib/mu_w": 0.9717142857142858,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40023904382470127,
"calib/std_conf": 0.06495116374757216,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.893628318584071,
"calib/step_q_c_n": 1243.0,
"calib/step_q_gap": 0.009686225042868313,
"calib/step_q_w": 0.8839420935412027,
"calib/step_q_w_n": 898.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1995.0,
"completions/max_terminated_length": 1995.0,
"completions/mean_length": 783.578125,
"completions/mean_terminated_length": 799.187255859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 477.0,
"epoch": 0.058666666666666666,
"grad_norm": 0.033670056611299515,
"kl": 0.026943206787109375,
"learning_rate": 4.027777777777779e-06,
"loss": -0.1008,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.018136566504836082,
"mask/share_reasoning": 0.8516228795051575,
"mask/share_step_conf": 0.1107092797756195,
"num_tokens": 17497734.0,
"reward": 0.7946262955665588,
"reward_std": 0.20714420080184937,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.5890586376190186,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.6900376081466675,
"step": 55
},
{
"adv/mean_abs_final_conf": 0.5695196390151978,
"adv/mean_abs_reasoning": 0.4986175298690796,
"adv/mean_abs_step_conf": 0.741671085357666,
"adv/ratio_final_to_reasoning": 1.1421973855687237,
"adv/ratio_step_to_reasoning": 1.4874548946411976,
"adv/std_final_conf": 0.7978363037109375,
"adv/std_reasoning": 0.7577388882637024,
"adv/std_step_conf": 0.9346494674682617,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.504842875591907,
"calib/avg_num_step_conf": 8.1328125,
"calib/ece": 0.4117154811715481,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.99581589958159,
"calib/gap": 0.0008788922370498442,
"calib/mean_conf": 0.9891213389121338,
"calib/mu_c": 0.9894927536231884,
"calib/mu_w": 0.9886138613861386,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4117154811715481,
"calib/std_conf": 0.006243148938441668,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8859311562224185,
"calib/step_q_c_n": 1133.0,
"calib/step_q_gap": -0.003584123018888108,
"calib/step_q_w": 0.8895152792413066,
"calib/step_q_w_n": 949.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2915.0,
"completions/max_terminated_length": 2915.0,
"completions/mean_length": 831.46875,
"completions/mean_terminated_length": 861.7651977539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 459.0,
"epoch": 0.05973333333333333,
"grad_norm": 0.019158584997057915,
"kl": 0.02632904052734375,
"learning_rate": 4.000000000000001e-06,
"loss": -0.1299,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.01669345796108246,
"mask/share_reasoning": 0.8484475612640381,
"mask/share_step_conf": 0.09970273077487946,
"num_tokens": 17817430.0,
"reward": 0.7434085011482239,
"reward_std": 0.2679101228713989,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.543994128704071,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.6490728259086609,
"step": 56
},
{
"adv/mean_abs_final_conf": 0.5240992307662964,
"adv/mean_abs_reasoning": 0.4981672465801239,
"adv/mean_abs_step_conf": 0.73002690076828,
"adv/ratio_final_to_reasoning": 1.0520547755080114,
"adv/ratio_step_to_reasoning": 1.4654253280998562,
"adv/std_final_conf": 0.7817376852035522,
"adv/std_reasoning": 0.7754046320915222,
"adv/std_step_conf": 0.9265749454498291,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5551038843721771,
"calib/avg_num_step_conf": 8.35546875,
"calib/ece": 0.3196285714285714,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9959183673469387,
"calib/gap": 0.0020347034025894883,
"calib/mean_conf": 0.9890163265306122,
"calib/mu_c": 0.9896890243902438,
"calib/mu_w": 0.9876543209876543,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3196285714285714,
"calib/std_conf": 0.006306438140172847,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.894915191740413,
"calib/step_q_c_n": 1356.0,
"calib/step_q_gap": 0.016677643847692547,
"calib/step_q_w": 0.8782375478927205,
"calib/step_q_w_n": 783.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2789.0,
"completions/max_terminated_length": 2789.0,
"completions/mean_length": 792.07421875,
"completions/mean_terminated_length": 824.2723388671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 519.0,
"epoch": 0.0608,
"grad_norm": 0.02918510138988495,
"kl": 0.024547576904296875,
"learning_rate": 3.972222222222223e-06,
"loss": -0.1605,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01758004166185856,
"mask/share_reasoning": 0.8421605229377747,
"mask/share_step_conf": 0.10119695961475372,
"num_tokens": 18126993.0,
"reward": 0.83964604139328,
"reward_std": 0.261970579624176,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6482839584350586,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7114768028259277,
"step": 57
},
{
"adv/mean_abs_final_conf": 0.5979112386703491,
"adv/mean_abs_reasoning": 0.5257660746574402,
"adv/mean_abs_step_conf": 0.7536758184432983,
"adv/ratio_final_to_reasoning": 1.1372191312646307,
"adv/ratio_step_to_reasoning": 1.4334812662348961,
"adv/std_final_conf": 0.8220943808555603,
"adv/std_reasoning": 0.7928545475006104,
"adv/std_step_conf": 0.9348974823951721,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.45082572502685286,
"calib/avg_num_step_conf": 8.4375,
"calib/ece": 0.44456734693877564,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9918367346938776,
"calib/gap": -0.0006325187969927759,
"calib/mean_conf": 0.9874244897959185,
"calib/mu_c": 0.9871353383458645,
"calib/mu_w": 0.9877678571428573,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.44456734693877564,
"calib/std_conf": 0.009137465611476537,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8876113657195234,
"calib/step_q_c_n": 1091.0,
"calib/step_q_gap": 0.007789101921581443,
"calib/step_q_w": 0.8798222637979419,
"calib/step_q_w_n": 1069.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2950.0,
"completions/max_terminated_length": 2950.0,
"completions/mean_length": 906.1640625,
"completions/mean_terminated_length": 924.2151489257812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 486.0,
"epoch": 0.06186666666666667,
"grad_norm": 0.032789744436740875,
"kl": 0.025289535522460938,
"learning_rate": 3.944444444444445e-06,
"loss": -0.072,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01625051349401474,
"mask/share_reasoning": 0.8657860159873962,
"mask/share_step_conf": 0.09843222051858902,
"num_tokens": 18465291.0,
"reward": 0.723969578742981,
"reward_std": 0.2588580250740051,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5299254059791565,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6234825849533081,
"step": 58
},
{
"adv/mean_abs_final_conf": 0.6285378932952881,
"adv/mean_abs_reasoning": 0.5240025520324707,
"adv/mean_abs_step_conf": 0.7756651639938354,
"adv/ratio_final_to_reasoning": 1.199493954480473,
"adv/ratio_step_to_reasoning": 1.4802698211778367,
"adv/std_final_conf": 0.7962230443954468,
"adv/std_reasoning": 0.7577937245368958,
"adv/std_step_conf": 0.9284113049507141,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.49198454161597366,
"calib/avg_num_step_conf": 8.18359375,
"calib/ece": 0.35284552845528466,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.991869918699187,
"calib/gap": -0.00509697273312848,
"calib/mean_conf": 0.9829268292682928,
"calib/mu_c": 0.9810828025477705,
"calib/mu_w": 0.986179775280899,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34878048780487814,
"calib/std_conf": 0.06326138455744801,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8827112135176651,
"calib/step_q_c_n": 1302.0,
"calib/step_q_gap": -0.002559908802637456,
"calib/step_q_w": 0.8852711223203026,
"calib/step_q_w_n": 793.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 3015.0,
"completions/max_terminated_length": 3015.0,
"completions/mean_length": 826.66796875,
"completions/mean_terminated_length": 846.508056640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 367.0,
"epoch": 0.06293333333333333,
"grad_norm": 0.03274433687329292,
"kl": 0.02954864501953125,
"learning_rate": 3.916666666666667e-06,
"loss": -0.142,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.017488282173871994,
"mask/share_reasoning": 0.8532015085220337,
"mask/share_step_conf": 0.10587271302938461,
"num_tokens": 18783166.0,
"reward": 0.8164454698562622,
"reward_std": 0.2749021649360657,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6187648177146912,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.6992822289466858,
"step": 59
},
{
"adv/mean_abs_final_conf": 0.6482319831848145,
"adv/mean_abs_reasoning": 0.5000711679458618,
"adv/mean_abs_step_conf": 0.7497437000274658,
"adv/ratio_final_to_reasoning": 1.2962794592768696,
"adv/ratio_step_to_reasoning": 1.4992739995532671,
"adv/std_final_conf": 0.8127201795578003,
"adv/std_reasoning": 0.7576885223388672,
"adv/std_step_conf": 0.9321392178535461,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.44521059782608696,
"calib/avg_num_step_conf": 8.8984375,
"calib/ece": 0.45851851851851855,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9876543209876543,
"calib/gap": -0.0008953804347823224,
"calib/mean_conf": 0.9852674897119343,
"calib/mu_c": 0.9848437500000002,
"calib/mu_w": 0.9857391304347826,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.45851851851851855,
"calib/std_conf": 0.012998514186810313,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8873805147058824,
"calib/step_q_c_n": 1088.0,
"calib/step_q_gap": -0.004892594537815098,
"calib/step_q_w": 0.8922731092436975,
"calib/step_q_w_n": 1190.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2874.0,
"completions/max_terminated_length": 2874.0,
"completions/mean_length": 825.25390625,
"completions/mean_terminated_length": 855.3239135742188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 439.0,
"epoch": 0.064,
"grad_norm": 0.037229057401418686,
"kl": 0.03716278076171875,
"learning_rate": 3.88888888888889e-06,
"loss": -0.1014,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01703135296702385,
"mask/share_reasoning": 0.8419076204299927,
"mask/share_step_conf": 0.10590481013059616,
"num_tokens": 19103287.0,
"reward": 0.7158203125,
"reward_std": 0.2458280324935913,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.5124461054801941,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6293507218360901,
"step": 60
},
{
"adv/mean_abs_final_conf": 0.6403580904006958,
"adv/mean_abs_reasoning": 0.41955268383026123,
"adv/mean_abs_step_conf": 0.7369972467422485,
"adv/ratio_final_to_reasoning": 1.5262876751368035,
"adv/ratio_step_to_reasoning": 1.7566262239438233,
"adv/std_final_conf": 0.7925984859466553,
"adv/std_reasoning": 0.7014154195785522,
"adv/std_step_conf": 0.9274880290031433,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.4565770348837209,
"calib/avg_num_step_conf": 8.46484375,
"calib/ece": 0.30206349206349203,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9920634920634921,
"calib/gap": 0.0010668604651163927,
"calib/mean_conf": 0.9846031746031746,
"calib/mu_c": 0.9849418604651164,
"calib/mu_w": 0.983875,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.30206349206349203,
"calib/std_conf": 0.009098298606934282,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8922908093278464,
"calib/step_q_c_n": 1458.0,
"calib/step_q_gap": 0.014759074490046653,
"calib/step_q_w": 0.8775317348377998,
"calib/step_q_w_n": 709.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2080.0,
"completions/max_terminated_length": 2080.0,
"completions/mean_length": 786.16015625,
"completions/mean_terminated_length": 795.4822387695312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 380.0,
"epoch": 0.06506666666666666,
"grad_norm": 0.025057543069124222,
"kl": 0.03397369384765625,
"learning_rate": 3.861111111111112e-06,
"loss": -0.0575,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.01846371404826641,
"mask/share_reasoning": 0.8561471700668335,
"mask/share_step_conf": 0.11367038637399673,
"num_tokens": 19408608.0,
"reward": 0.8597561120986938,
"reward_std": 0.2305433750152588,
"rewards/accuracy_reward_step": 0.671875,
"rewards/final_brier_reward_step": 0.6777324080467224,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7113110423088074,
"step": 61
},
{
"adv/mean_abs_final_conf": 0.7482225298881531,
"adv/mean_abs_reasoning": 0.5922585129737854,
"adv/mean_abs_step_conf": 0.7756792306900024,
"adv/ratio_final_to_reasoning": 1.2633377376565813,
"adv/ratio_step_to_reasoning": 1.3096970557590544,
"adv/std_final_conf": 0.8975844979286194,
"adv/std_reasoning": 0.826655924320221,
"adv/std_step_conf": 0.934055745601654,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.4916534181240063,
"calib/avg_num_step_conf": 8.140625,
"calib/ece": 0.43056680161943334,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.9878542510121457,
"calib/gap": 0.00998608903020648,
"calib/mean_conf": 0.9811740890688261,
"calib/mu_c": 0.9856617647058824,
"calib/mu_w": 0.9756756756756759,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43056680161943334,
"calib/std_conf": 0.06321864616836151,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8915182755388941,
"calib/step_q_c_n": 1067.0,
"calib/step_q_gap": 0.009945020868294296,
"calib/step_q_w": 0.8815732546705998,
"calib/step_q_w_n": 1017.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1956.0,
"completions/max_terminated_length": 1956.0,
"completions/mean_length": 803.88671875,
"completions/mean_terminated_length": 816.6468505859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 481.0,
"epoch": 0.06613333333333334,
"grad_norm": 0.03266182541847229,
"kl": 0.036937713623046875,
"learning_rate": 3.833333333333334e-06,
"loss": -0.0886,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.0176240187138319,
"mask/share_reasoning": 0.8623931407928467,
"mask/share_step_conf": 0.10435787588357925,
"num_tokens": 19721483.0,
"reward": 0.735969603061676,
"reward_std": 0.31463518738746643,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.5481457114219666,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6245747804641724,
"step": 62
},
{
"adv/mean_abs_final_conf": 0.6453641653060913,
"adv/mean_abs_reasoning": 0.4795626997947693,
"adv/mean_abs_step_conf": 0.735736608505249,
"adv/ratio_final_to_reasoning": 1.3457346986791872,
"adv/ratio_step_to_reasoning": 1.5341823057133308,
"adv/std_final_conf": 0.8454790711402893,
"adv/std_reasoning": 0.7753329873085022,
"adv/std_step_conf": 0.9310802221298218,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.4635060934779697,
"calib/avg_num_step_conf": 8.34765625,
"calib/ece": 0.44995918367346965,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9959183673469387,
"calib/gap": -0.001305075666264921,
"calib/mean_conf": 0.98465306122449,
"calib/mu_c": 0.9840458015267177,
"calib/mu_w": 0.9853508771929826,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44995918367346965,
"calib/std_conf": 0.007366640251029523,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8937358490566037,
"calib/step_q_c_n": 1060.0,
"calib/step_q_gap": -0.0007488306091345542,
"calib/step_q_w": 0.8944846796657383,
"calib/step_q_w_n": 1077.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2105.0,
"completions/max_terminated_length": 2105.0,
"completions/mean_length": 838.203125,
"completions/mean_terminated_length": 858.320068359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 439.0,
"epoch": 0.0672,
"grad_norm": 0.02464851178228855,
"kl": 0.03795623779296875,
"learning_rate": 3.8055555555555556e-06,
"loss": -0.1238,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01703377068042755,
"mask/share_reasoning": 0.8587255477905273,
"mask/share_step_conf": 0.10080315172672272,
"num_tokens": 20044703.0,
"reward": 0.7243086695671082,
"reward_std": 0.24416929483413696,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.5244882702827454,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6303789615631104,
"step": 63
},
{
"adv/mean_abs_final_conf": 0.5989560484886169,
"adv/mean_abs_reasoning": 0.3883668780326843,
"adv/mean_abs_step_conf": 0.7467473745346069,
"adv/ratio_final_to_reasoning": 1.5422428697387778,
"adv/ratio_step_to_reasoning": 1.9227885197556984,
"adv/std_final_conf": 0.7767327427864075,
"adv/std_reasoning": 0.6816226243972778,
"adv/std_step_conf": 0.9268690347671509,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.44446606475716055,
"calib/avg_num_step_conf": 8.51953125,
"calib/ece": 0.27610441767068294,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9879518072289156,
"calib/gap": 0.0006592465753426646,
"calib/mean_conf": 0.9829317269076308,
"calib/mu_c": 0.9831250000000001,
"calib/mu_w": 0.9824657534246575,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.27610441767068294,
"calib/std_conf": 0.010596627363252014,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8877248322147652,
"calib/step_q_c_n": 1490.0,
"calib/step_q_gap": 0.001545382142406182,
"calib/step_q_w": 0.886179450072359,
"calib/step_q_w_n": 691.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2988.0,
"completions/max_terminated_length": 2988.0,
"completions/mean_length": 829.1953125,
"completions/mean_terminated_length": 835.7244262695312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 476.0,
"epoch": 0.06826666666666667,
"grad_norm": 0.04055461660027504,
"kl": 0.04143524169921875,
"learning_rate": 3.777777777777778e-06,
"loss": -0.0142,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.017893537878990173,
"mask/share_reasoning": 0.8629002571105957,
"mask/share_step_conf": 0.11139366030693054,
"num_tokens": 20360753.0,
"reward": 0.8846273422241211,
"reward_std": 0.19556494057178497,
"rewards/accuracy_reward_step": 0.6875,
"rewards/final_brier_reward_step": 0.6971074342727661,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7401160001754761,
"step": 64
},
{
"adv/mean_abs_final_conf": 0.6421319246292114,
"adv/mean_abs_reasoning": 0.34305283427238464,
"adv/mean_abs_step_conf": 0.7524635791778564,
"adv/ratio_final_to_reasoning": 1.8718164098284562,
"adv/ratio_step_to_reasoning": 2.193433500626317,
"adv/std_final_conf": 0.8032649755477905,
"adv/std_reasoning": 0.6401863694190979,
"adv/std_step_conf": 0.9258484840393066,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.5481290322580645,
"calib/avg_num_step_conf": 8.38671875,
"calib/ece": 0.3761960784313728,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.996078431372549,
"calib/gap": 0.0017096774193545716,
"calib/mean_conf": 0.9840392156862747,
"calib/mu_c": 0.9847096774193548,
"calib/mu_w": 0.9830000000000002,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3761960784313728,
"calib/std_conf": 0.007232716475857886,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8924881516587677,
"calib/step_q_c_n": 1266.0,
"calib/step_q_gap": 0.0016708985373147245,
"calib/step_q_w": 0.890817253121453,
"calib/step_q_w_n": 881.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1474.0,
"completions/max_terminated_length": 1474.0,
"completions/mean_length": 737.390625,
"completions/mean_terminated_length": 740.2824096679688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 381.0,
"epoch": 0.06933333333333333,
"grad_norm": 0.04465280845761299,
"kl": 0.046863555908203125,
"learning_rate": 3.7500000000000005e-06,
"loss": -0.0059,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.019956696778535843,
"mask/share_reasoning": 0.855435311794281,
"mask/share_step_conf": 0.12070174515247345,
"num_tokens": 20654549.0,
"reward": 0.8108848929405212,
"reward_std": 0.18426105380058289,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.6184441447257996,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.6830130219459534,
"step": 65
},
{
"adv/mean_abs_final_conf": 0.683743417263031,
"adv/mean_abs_reasoning": 0.46175944805145264,
"adv/mean_abs_step_conf": 0.7483415603637695,
"adv/ratio_final_to_reasoning": 1.4807350886880897,
"adv/ratio_step_to_reasoning": 1.6206307494554693,
"adv/std_final_conf": 0.8428446054458618,
"adv/std_reasoning": 0.7392820715904236,
"adv/std_step_conf": 0.9290730953216553,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5531841414560021,
"calib/avg_num_step_conf": 8.11328125,
"calib/ece": 0.45605809128630714,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.995850622406639,
"calib/gap": 0.0014200856471887624,
"calib/mean_conf": 0.9830290456431536,
"calib/mu_c": 0.983700787401575,
"calib/mu_w": 0.9822807017543862,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.45605809128630714,
"calib/std_conf": 0.008420681091429592,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8887060998151571,
"calib/step_q_c_n": 1082.0,
"calib/step_q_gap": 0.021178461624202338,
"calib/step_q_w": 0.8675276381909548,
"calib/step_q_w_n": 995.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2664.0,
"completions/max_terminated_length": 2664.0,
"completions/mean_length": 886.8671875,
"completions/mean_terminated_length": 904.5338745117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 447.0,
"epoch": 0.0704,
"grad_norm": 0.028026781976222992,
"kl": 0.038791656494140625,
"learning_rate": 3.7222222222222225e-06,
"loss": -0.0837,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016637243330478668,
"mask/share_reasoning": 0.8638635873794556,
"mask/share_step_conf": 0.09996792674064636,
"num_tokens": 20987939.0,
"reward": 0.7157400250434875,
"reward_std": 0.21129858493804932,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.5115370750427246,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.6324428915977478,
"step": 66
},
{
"adv/mean_abs_final_conf": 0.5312223434448242,
"adv/mean_abs_reasoning": 0.30277425050735474,
"adv/mean_abs_step_conf": 0.7451149821281433,
"adv/ratio_final_to_reasoning": 1.7545162528010956,
"adv/ratio_step_to_reasoning": 2.460958885636953,
"adv/std_final_conf": 0.7096737623214722,
"adv/std_reasoning": 0.5961373448371887,
"adv/std_step_conf": 0.9236457347869873,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5587776517300056,
"calib/avg_num_step_conf": 8.58984375,
"calib/ece": 0.3273880000000001,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.992,
"calib/gap": 0.0028957742484401194,
"calib/mean_conf": 0.9833880000000002,
"calib/mu_c": 0.9843841463414635,
"calib/mu_w": 0.9814883720930234,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3273880000000001,
"calib/std_conf": 0.00901451363080671,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8887572953736657,
"calib/step_q_c_n": 1405.0,
"calib/step_q_gap": 0.0033353810159830743,
"calib/step_q_w": 0.8854219143576826,
"calib/step_q_w_n": 794.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2551.0,
"completions/max_terminated_length": 2551.0,
"completions/mean_length": 846.890625,
"completions/mean_terminated_length": 860.3333740234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 491.0,
"epoch": 0.07146666666666666,
"grad_norm": 0.05231475085020065,
"kl": 0.04485321044921875,
"learning_rate": 3.694444444444445e-06,
"loss": -0.0351,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017291639000177383,
"mask/share_reasoning": 0.8610464930534363,
"mask/share_step_conf": 0.10603685677051544,
"num_tokens": 21309751.0,
"reward": 0.8452394008636475,
"reward_std": 0.17514097690582275,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6527136564254761,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7143276333808899,
"step": 67
},
{
"adv/mean_abs_final_conf": 0.6500144004821777,
"adv/mean_abs_reasoning": 0.41077685356140137,
"adv/mean_abs_step_conf": 0.763282835483551,
"adv/ratio_final_to_reasoning": 1.5824026958836814,
"adv/ratio_step_to_reasoning": 1.858144705248,
"adv/std_final_conf": 0.8139645457267761,
"adv/std_reasoning": 0.6816565990447998,
"adv/std_step_conf": 0.9272443652153015,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5556972789115646,
"calib/avg_num_step_conf": 8.61328125,
"calib/ece": 0.37884773662551463,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.9958847736625515,
"calib/gap": 0.0009204931972788266,
"calib/mean_conf": 0.9837860082304529,
"calib/mu_c": 0.9841496598639456,
"calib/mu_w": 0.9832291666666668,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.37884773662551463,
"calib/std_conf": 0.007675752112042767,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8871908396946565,
"calib/step_q_c_n": 1310.0,
"calib/step_q_gap": 0.026710392767282265,
"calib/step_q_w": 0.8604804469273742,
"calib/step_q_w_n": 895.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2809.0,
"completions/max_terminated_length": 2809.0,
"completions/mean_length": 812.13671875,
"completions/mean_terminated_length": 828.3147583007812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 387.0,
"epoch": 0.07253333333333334,
"grad_norm": 0.02631233073771,
"kl": 0.04804229736328125,
"learning_rate": 3.6666666666666666e-06,
"loss": -0.0813,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.018240146338939667,
"mask/share_reasoning": 0.8457576632499695,
"mask/share_step_conf": 0.11647099256515503,
"num_tokens": 21621746.0,
"reward": 0.7804622054100037,
"reward_std": 0.1851496696472168,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.582507848739624,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.6752915382385254,
"step": 68
},
{
"adv/mean_abs_final_conf": 0.6945899724960327,
"adv/mean_abs_reasoning": 0.5294466614723206,
"adv/mean_abs_step_conf": 0.7615878582000732,
"adv/ratio_final_to_reasoning": 1.3119168049232204,
"adv/ratio_step_to_reasoning": 1.4384600255712237,
"adv/std_final_conf": 0.8556989431381226,
"adv/std_reasoning": 0.7754108309745789,
"adv/std_step_conf": 0.9333360195159912,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5152447552447552,
"calib/avg_num_step_conf": 8.15625,
"calib/ece": 0.3955967078189303,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9958847736625515,
"calib/gap": -0.00021398601398614492,
"calib/mean_conf": 0.9840740740740743,
"calib/mu_c": 0.9839860139860143,
"calib/mu_w": 0.9842000000000004,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3955967078189303,
"calib/std_conf": 0.007332959212304939,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8818712955122779,
"calib/step_q_c_n": 1181.0,
"calib/step_q_gap": 0.0006915821715943737,
"calib/step_q_w": 0.8811797133406836,
"calib/step_q_w_n": 907.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2943.0,
"completions/max_terminated_length": 2943.0,
"completions/mean_length": 840.9140625,
"completions/mean_terminated_length": 868.040283203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 514.0,
"epoch": 0.0736,
"grad_norm": 0.029118498787283897,
"kl": 0.043987274169921875,
"learning_rate": 3.638888888888889e-06,
"loss": -0.1294,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.016698405146598816,
"mask/share_reasoning": 0.8528545498847961,
"mask/share_step_conf": 0.09919705986976624,
"num_tokens": 21941516.0,
"reward": 0.769717812538147,
"reward_std": 0.2492179572582245,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.5706456899642944,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6672273874282837,
"step": 69
},
{
"adv/mean_abs_final_conf": 0.6649423241615295,
"adv/mean_abs_reasoning": 0.4383116364479065,
"adv/mean_abs_step_conf": 0.742194652557373,
"adv/ratio_final_to_reasoning": 1.5170537783350824,
"adv/ratio_step_to_reasoning": 1.693303555826502,
"adv/std_final_conf": 0.8065817356109619,
"adv/std_reasoning": 0.7205896973609924,
"adv/std_step_conf": 0.9320774078369141,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5444521497919557,
"calib/avg_num_step_conf": 9.32421875,
"calib/ece": 0.40720164609053516,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9876543209876543,
"calib/gap": 0.005280166435506439,
"calib/mean_conf": 0.9833333333333335,
"calib/mu_c": 0.9855714285714287,
"calib/mu_w": 0.9802912621359222,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40720164609053516,
"calib/std_conf": 0.020507952888751305,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8900559552358114,
"calib/step_q_c_n": 1251.0,
"calib/step_q_gap": -0.00013770673601964933,
"calib/step_q_w": 0.890193661971831,
"calib/step_q_w_n": 1136.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2640.0,
"completions/max_terminated_length": 2640.0,
"completions/mean_length": 817.83984375,
"completions/mean_terminated_length": 854.5591430664062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 460.0,
"epoch": 0.07466666666666667,
"grad_norm": 0.04475827515125275,
"kl": 0.048248291015625,
"learning_rate": 3.6111111111111115e-06,
"loss": -0.2043,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.017057053744792938,
"mask/share_reasoning": 0.832466721534729,
"mask/share_step_conf": 0.10750746726989746,
"num_tokens": 22257875.0,
"reward": 0.7592463493347168,
"reward_std": 0.2055116891860962,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.5620715022087097,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6572023630142212,
"step": 70
},
{
"adv/mean_abs_final_conf": 0.6414079666137695,
"adv/mean_abs_reasoning": 0.4914470911026001,
"adv/mean_abs_step_conf": 0.7306813597679138,
"adv/ratio_final_to_reasoning": 1.3051414449818401,
"adv/ratio_step_to_reasoning": 1.486795573717962,
"adv/std_final_conf": 0.8314691781997681,
"adv/std_reasoning": 0.7753694653511047,
"adv/std_step_conf": 0.9239351749420166,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.48824191838897724,
"calib/avg_num_step_conf": 8.6953125,
"calib/ece": 0.4347368421052634,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9919028340080972,
"calib/gap": -0.00011128775834678972,
"calib/mean_conf": 0.9853441295546561,
"calib/mu_c": 0.9852941176470589,
"calib/mu_w": 0.9854054054054057,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4347368421052634,
"calib/std_conf": 0.009120371154252042,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8855157715260017,
"calib/step_q_c_n": 1173.0,
"calib/step_q_gap": -0.001283848606951743,
"calib/step_q_w": 0.8867996201329534,
"calib/step_q_w_n": 1053.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2967.0,
"completions/max_terminated_length": 2967.0,
"completions/mean_length": 847.86328125,
"completions/mean_terminated_length": 861.3214721679688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 498.0,
"epoch": 0.07573333333333333,
"grad_norm": 0.026600291952490807,
"kl": 0.044338226318359375,
"learning_rate": 3.5833333333333335e-06,
"loss": -0.0945,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.017498448491096497,
"mask/share_reasoning": 0.8599810600280762,
"mask/share_step_conf": 0.10689544677734375,
"num_tokens": 22579336.0,
"reward": 0.7447740435600281,
"reward_std": 0.24545089900493622,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.5397129058837891,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.6513975858688354,
"step": 71
},
{
"adv/mean_abs_final_conf": 0.6619936227798462,
"adv/mean_abs_reasoning": 0.5136440396308899,
"adv/mean_abs_step_conf": 0.7351285219192505,
"adv/ratio_final_to_reasoning": 1.2888178810671334,
"adv/ratio_step_to_reasoning": 1.431202282513629,
"adv/std_final_conf": 0.867818295955658,
"adv/std_reasoning": 0.7926632761955261,
"adv/std_step_conf": 0.9319378733634949,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.4561651583710407,
"calib/avg_num_step_conf": 8.15234375,
"calib/ece": 0.44885375494071145,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.9960474308300395,
"calib/gap": -0.00144419306183996,
"calib/mean_conf": 0.986403162055336,
"calib/mu_c": 0.985735294117647,
"calib/mu_w": 0.987179487179487,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44885375494071145,
"calib/std_conf": 0.0071756277962406025,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8852765583845479,
"calib/step_q_c_n": 1139.0,
"calib/step_q_gap": -0.005419644147097835,
"calib/step_q_w": 0.8906962025316457,
"calib/step_q_w_n": 948.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1663.0,
"completions/max_terminated_length": 1663.0,
"completions/mean_length": 786.3046875,
"completions/mean_terminated_length": 795.6284790039062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 490.0,
"epoch": 0.0768,
"grad_norm": 0.02710404060781002,
"kl": 0.046268463134765625,
"learning_rate": 3.555555555555556e-06,
"loss": -0.0352,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.018414214253425598,
"mask/share_reasoning": 0.8599804639816284,
"mask/share_step_conf": 0.10988658666610718,
"num_tokens": 22885038.0,
"reward": 0.7634243965148926,
"reward_std": 0.26252222061157227,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.5388292670249939,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.6848944425582886,
"step": 72
},
{
"adv/mean_abs_final_conf": 0.6329753994941711,
"adv/mean_abs_reasoning": 0.48065903782844543,
"adv/mean_abs_step_conf": 0.7773274183273315,
"adv/ratio_final_to_reasoning": 1.3168906640221956,
"adv/ratio_step_to_reasoning": 1.6172116971714399,
"adv/std_final_conf": 0.7763375639915466,
"adv/std_reasoning": 0.720600962638855,
"adv/std_step_conf": 0.9217331409454346,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.4776785714285714,
"calib/avg_num_step_conf": 8.37890625,
"calib/ece": 0.3250393700787403,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.9960629921259843,
"calib/gap": 0.00044435215946891926,
"calib/mean_conf": 0.986456692913386,
"calib/mu_c": 0.986607142857143,
"calib/mu_w": 0.9861627906976741,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3250393700787403,
"calib/std_conf": 0.0071571294838786385,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8882424677187949,
"calib/step_q_c_n": 1394.0,
"calib/step_q_gap": 0.0009455303020172723,
"calib/step_q_w": 0.8872969374167776,
"calib/step_q_w_n": 751.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1805.0,
"completions/max_terminated_length": 1805.0,
"completions/mean_length": 782.203125,
"completions/mean_terminated_length": 788.3621826171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 460.0,
"epoch": 0.07786666666666667,
"grad_norm": 0.02680094726383686,
"kl": 0.044742584228515625,
"learning_rate": 3.5277777777777784e-06,
"loss": -0.0179,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.018433989956974983,
"mask/share_reasoning": 0.863990068435669,
"mask/share_step_conf": 0.10976342111825943,
"num_tokens": 23192314.0,
"reward": 0.8560742139816284,
"reward_std": 0.23352429270744324,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.6653140783309937,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.7171468138694763,
"step": 73
},
{
"adv/mean_abs_final_conf": 0.5732172131538391,
"adv/mean_abs_reasoning": 0.3150753974914551,
"adv/mean_abs_step_conf": 0.7501686215400696,
"adv/ratio_final_to_reasoning": 1.8193017218025882,
"adv/ratio_step_to_reasoning": 2.3809177978119167,
"adv/std_final_conf": 0.7667513489723206,
"adv/std_reasoning": 0.6185932159423828,
"adv/std_step_conf": 0.9282718896865845,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5352107993084186,
"calib/avg_num_step_conf": 8.390625,
"calib/ece": 0.4000803212851406,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9959839357429718,
"calib/gap": 0.001520813938023835,
"calib/mean_conf": 0.986425702811245,
"calib/mu_c": 0.987054794520548,
"calib/mu_w": 0.9855339805825242,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4000803212851406,
"calib/std_conf": 0.007258529258466635,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8932904263877715,
"calib/step_q_c_n": 1243.0,
"calib/step_q_gap": 0.018273851802136032,
"calib/step_q_w": 0.8750165745856354,
"calib/step_q_w_n": 905.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2657.0,
"completions/max_terminated_length": 2657.0,
"completions/mean_length": 815.6015625,
"completions/mean_terminated_length": 828.5476684570312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 431.0,
"epoch": 0.07893333333333333,
"grad_norm": 0.02454979531466961,
"kl": 0.043239593505859375,
"learning_rate": 3.5e-06,
"loss": -0.0846,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.018031660467386246,
"mask/share_reasoning": 0.8548358082771301,
"mask/share_step_conf": 0.11150752007961273,
"num_tokens": 23505036.0,
"reward": 0.8020550012588501,
"reward_std": 0.17195594310760498,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.5817226767539978,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7137935161590576,
"step": 74
},
{
"adv/mean_abs_final_conf": 0.602027416229248,
"adv/mean_abs_reasoning": 0.387719064950943,
"adv/mean_abs_step_conf": 0.7657424807548523,
"adv/ratio_final_to_reasoning": 1.552741329099772,
"adv/ratio_step_to_reasoning": 1.974993107062557,
"adv/std_final_conf": 0.7324143648147583,
"adv/std_reasoning": 0.6404212713241577,
"adv/std_step_conf": 0.9212512969970703,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.4856928838951311,
"calib/avg_num_step_conf": 8.65625,
"calib/ece": 0.2824901185770751,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.9920948616600791,
"calib/gap": 0.0030996254681646684,
"calib/mean_conf": 0.9860474308300397,
"calib/mu_c": 0.9869662921348314,
"calib/mu_w": 0.9838666666666668,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2824901185770751,
"calib/std_conf": 0.013694497153698134,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8953473132372216,
"calib/step_q_c_n": 1526.0,
"calib/step_q_gap": 0.025448762512583878,
"calib/step_q_w": 0.8698985507246377,
"calib/step_q_w_n": 690.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1944.0,
"completions/max_terminated_length": 1944.0,
"completions/mean_length": 770.1796875,
"completions/mean_terminated_length": 779.312255859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 460.0,
"epoch": 0.08,
"grad_norm": 0.03222297132015228,
"kl": 0.044467926025390625,
"learning_rate": 3.4722222222222224e-06,
"loss": -0.0331,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.01898212358355522,
"mask/share_reasoning": 0.8517194986343384,
"mask/share_step_conf": 0.11757959425449371,
"num_tokens": 23806954.0,
"reward": 0.897788941860199,
"reward_std": 0.20893093943595886,
"rewards/accuracy_reward_step": 0.6953125,
"rewards/final_brier_reward_step": 0.7043101787567139,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7553301453590393,
"step": 75
},
{
"adv/mean_abs_final_conf": 0.5657827258110046,
"adv/mean_abs_reasoning": 0.45047199726104736,
"adv/mean_abs_step_conf": 0.720933198928833,
"adv/ratio_final_to_reasoning": 1.2559775729702793,
"adv/ratio_step_to_reasoning": 1.60039514844927,
"adv/std_final_conf": 0.7905355095863342,
"adv/std_reasoning": 0.739392876625061,
"adv/std_step_conf": 0.924590528011322,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.463395979020979,
"calib/avg_num_step_conf": 8.546875,
"calib/ece": 0.3495040983606558,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9918032786885246,
"calib/gap": -0.002392482517482475,
"calib/mean_conf": 0.9872090163934427,
"calib/mu_c": 0.9863461538461539,
"calib/mu_w": 0.9887386363636363,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3486844262295083,
"calib/std_conf": 0.013801927307197861,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8927719033232628,
"calib/step_q_c_n": 1324.0,
"calib/step_q_gap": 0.006402690360299945,
"calib/step_q_w": 0.8863692129629629,
"calib/step_q_w_n": 864.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2912.0,
"completions/max_terminated_length": 2912.0,
"completions/mean_length": 872.39453125,
"completions/mean_terminated_length": 889.77294921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 479.0,
"epoch": 0.08106666666666666,
"grad_norm": 0.025584116578102112,
"kl": 0.03765106201171875,
"learning_rate": 3.444444444444445e-06,
"loss": -0.0472,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01710599660873413,
"mask/share_reasoning": 0.8587711453437805,
"mask/share_step_conf": 0.10459160804748535,
"num_tokens": 24133343.0,
"reward": 0.8029855489730835,
"reward_std": 0.23085039854049683,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6128737926483154,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.680597186088562,
"step": 76
},
{
"adv/mean_abs_final_conf": 0.543570876121521,
"adv/mean_abs_reasoning": 0.48223191499710083,
"adv/mean_abs_step_conf": 0.7708549499511719,
"adv/ratio_final_to_reasoning": 1.1271980539172504,
"adv/ratio_step_to_reasoning": 1.5985150007249236,
"adv/std_final_conf": 0.7704401612281799,
"adv/std_reasoning": 0.7395368814468384,
"adv/std_step_conf": 0.9321821928024292,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.4929743339831059,
"calib/avg_num_step_conf": 8.58203125,
"calib/ece": 0.2843621399176955,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9958847736625515,
"calib/gap": 0.0015935672514618693,
"calib/mean_conf": 0.9880658436213993,
"calib/mu_c": 0.9885380116959064,
"calib/mu_w": 0.9869444444444445,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2843621399176955,
"calib/std_conf": 0.009556190009526783,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8938732394366199,
"calib/step_q_c_n": 1420.0,
"calib/step_q_gap": 0.03588096144434194,
"calib/step_q_w": 0.857992277992278,
"calib/step_q_w_n": 777.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2766.0,
"completions/max_terminated_length": 2766.0,
"completions/mean_length": 813.8046875,
"completions/mean_terminated_length": 843.45751953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 450.0,
"epoch": 0.08213333333333334,
"grad_norm": 0.01862136833369732,
"kl": 0.036975860595703125,
"learning_rate": 3.416666666666667e-06,
"loss": -0.1461,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.017553173005580902,
"mask/share_reasoning": 0.8386209607124329,
"mask/share_step_conf": 0.10866963863372803,
"num_tokens": 24446341.0,
"reward": 0.8542374968528748,
"reward_std": 0.26453351974487305,
"rewards/accuracy_reward_step": 0.66796875,
"rewards/final_brier_reward_step": 0.6750906109809875,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.709946870803833,
"step": 77
},
{
"adv/mean_abs_final_conf": 0.5738725662231445,
"adv/mean_abs_reasoning": 0.4448193609714508,
"adv/mean_abs_step_conf": 0.7464429140090942,
"adv/ratio_final_to_reasoning": 1.2901249733596387,
"adv/ratio_step_to_reasoning": 1.6780809908519294,
"adv/std_final_conf": 0.789870023727417,
"adv/std_reasoning": 0.7205432057380676,
"adv/std_step_conf": 0.9322795271873474,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5005253572429252,
"calib/avg_num_step_conf": 8.08203125,
"calib/ece": 0.3290079365079366,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.9920634920634921,
"calib/gap": 0.0004497057999441889,
"calib/mean_conf": 0.9877380952380953,
"calib/mu_c": 0.9878915662650601,
"calib/mu_w": 0.987441860465116,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3290079365079366,
"calib/std_conf": 0.00863977381706126,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8860716952949963,
"calib/step_q_c_n": 1339.0,
"calib/step_q_gap": 0.00035936652787293966,
"calib/step_q_w": 0.8857123287671234,
"calib/step_q_w_n": 730.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2601.0,
"completions/max_terminated_length": 2601.0,
"completions/mean_length": 881.0703125,
"completions/mean_terminated_length": 881.0703125,
"completions/min_length": 456.0,
"completions/min_terminated_length": 456.0,
"epoch": 0.0832,
"grad_norm": 0.02497708797454834,
"kl": 0.039813995361328125,
"learning_rate": 3.3888888888888893e-06,
"loss": 0.0272,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.017089759930968285,
"mask/share_reasoning": 0.8821457624435425,
"mask/share_step_conf": 0.1007644385099411,
"num_tokens": 24779919.0,
"reward": 0.8477307558059692,
"reward_std": 0.221160426735878,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/final_brier_reward_step": 0.6566535234451294,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7122455835342407,
"step": 78
},
{
"adv/mean_abs_final_conf": 0.4717371463775635,
"adv/mean_abs_reasoning": 0.3694051206111908,
"adv/mean_abs_step_conf": 0.7264292240142822,
"adv/ratio_final_to_reasoning": 1.277018427890392,
"adv/ratio_step_to_reasoning": 1.9664839047503873,
"adv/std_final_conf": 0.7244833111763,
"adv/std_reasoning": 0.640355110168457,
"adv/std_step_conf": 0.930465817451477,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5197294645360248,
"calib/avg_num_step_conf": 8.15234375,
"calib/ece": 0.3344223107569721,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9920318725099602,
"calib/gap": 0.0013961312026915529,
"calib/mean_conf": 0.9878087649402391,
"calib/mu_c": 0.9882926829268291,
"calib/mu_w": 0.9868965517241376,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3344223107569721,
"calib/std_conf": 0.01087685885878777,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8910483870967743,
"calib/step_q_c_n": 1364.0,
"calib/step_q_gap": 0.011179784053897412,
"calib/step_q_w": 0.8798686030428768,
"calib/step_q_w_n": 723.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1490.0,
"completions/max_terminated_length": 1490.0,
"completions/mean_length": 813.0234375,
"completions/mean_terminated_length": 829.2191772460938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 454.0,
"epoch": 0.08426666666666667,
"grad_norm": 0.024105433374643326,
"kl": 0.0389556884765625,
"learning_rate": 3.3611111111111117e-06,
"loss": -0.0774,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.017385980114340782,
"mask/share_reasoning": 0.859592080116272,
"mask/share_step_conf": 0.1034906804561615,
"num_tokens": 25094429.0,
"reward": 0.8523935079574585,
"reward_std": 0.19485636055469513,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6492695212364197,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7312988042831421,
"step": 79
},
{
"adv/mean_abs_final_conf": 0.48533910512924194,
"adv/mean_abs_reasoning": 0.4345003366470337,
"adv/mean_abs_step_conf": 0.7359863519668579,
"adv/ratio_final_to_reasoning": 1.1170051302480513,
"adv/ratio_step_to_reasoning": 1.6938683123846148,
"adv/std_final_conf": 0.736708402633667,
"adv/std_reasoning": 0.7206791639328003,
"adv/std_step_conf": 0.9287194013595581,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.4836915957605613,
"calib/avg_num_step_conf": 8.76171875,
"calib/ece": 0.29513904382470113,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.9920318725099602,
"calib/gap": -0.0012328481862962093,
"calib/mean_conf": 0.9883661354581673,
"calib/mu_c": 0.9879879310344827,
"calib/mu_w": 0.9892207792207789,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.29513904382470113,
"calib/std_conf": 0.008474936601766516,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8913844816484223,
"calib/step_q_c_n": 1553.0,
"calib/step_q_gap": 0.00932651063392953,
"calib/step_q_w": 0.8820579710144928,
"calib/step_q_w_n": 690.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3072.0,
"completions/max_terminated_length": 3072.0,
"completions/mean_length": 797.9453125,
"completions/mean_terminated_length": 807.4071655273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 473.0,
"epoch": 0.08533333333333333,
"grad_norm": 0.026427265256643295,
"kl": 0.04227447509765625,
"learning_rate": 3.3333333333333333e-06,
"loss": -0.0398,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.018396034836769104,
"mask/share_reasoning": 0.8537461757659912,
"mask/share_step_conf": 0.11613905429840088,
"num_tokens": 25400863.0,
"reward": 0.8714803457260132,
"reward_std": 0.24076946079730988,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/final_brier_reward_step": 0.6858910322189331,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7258197069168091,
"step": 80
},
{
"adv/mean_abs_final_conf": 0.4667571783065796,
"adv/mean_abs_reasoning": 0.38079017400741577,
"adv/mean_abs_step_conf": 0.7052954435348511,
"adv/ratio_final_to_reasoning": 1.2257595131577361,
"adv/ratio_step_to_reasoning": 1.8521891888972841,
"adv/std_final_conf": 0.722691535949707,
"adv/std_reasoning": 0.7015290856361389,
"adv/std_step_conf": 0.9257702827453613,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5446386946386946,
"calib/avg_num_step_conf": 8.7265625,
"calib/ece": 0.3096707818930041,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0008927738927744233,
"calib/mean_conf": 0.9886831275720165,
"calib/mu_c": 0.988969696969697,
"calib/mu_w": 0.9880769230769226,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3096707818930041,
"calib/std_conf": 0.003381504293691219,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8905312954876274,
"calib/step_q_c_n": 1374.0,
"calib/step_q_gap": 0.03453943502251111,
"calib/step_q_w": 0.8559918604651163,
"calib/step_q_w_n": 860.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2985.0,
"completions/max_terminated_length": 2985.0,
"completions/mean_length": 807.69921875,
"completions/mean_terminated_length": 837.1295776367188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 471.0,
"epoch": 0.0864,
"grad_norm": 0.023954054340720177,
"kl": 0.041957855224609375,
"learning_rate": 3.3055555555555558e-06,
"loss": -0.1408,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.017638327553868294,
"mask/share_reasoning": 0.8397368788719177,
"mask/share_step_conf": 0.10746851563453674,
"num_tokens": 25713882.0,
"reward": 0.8331348896026611,
"reward_std": 0.22398020327091217,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.6516644954681396,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6958553194999695,
"step": 81
},
{
"adv/mean_abs_final_conf": 0.5066346526145935,
"adv/mean_abs_reasoning": 0.42569518089294434,
"adv/mean_abs_step_conf": 0.7474675178527832,
"adv/ratio_final_to_reasoning": 1.1901348085544905,
"adv/ratio_step_to_reasoning": 1.7558749814477217,
"adv/std_final_conf": 0.7515454292297363,
"adv/std_reasoning": 0.7013655304908752,
"adv/std_step_conf": 0.9292082786560059,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5395319855995568,
"calib/avg_num_step_conf": 8.40625,
"calib/ece": 0.35685943775100404,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9919678714859438,
"calib/gap": 0.00501827748546102,
"calib/mean_conf": 0.9873815261044177,
"calib/mu_c": 0.9892356687898087,
"calib/mu_w": 0.9842173913043477,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35685943775100404,
"calib/std_conf": 0.01948707274314714,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8947453874538747,
"calib/step_q_c_n": 1355.0,
"calib/step_q_gap": 0.02878428331334759,
"calib/step_q_w": 0.8659611041405271,
"calib/step_q_w_n": 797.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2186.0,
"completions/max_terminated_length": 2186.0,
"completions/mean_length": 776.33984375,
"completions/mean_terminated_length": 788.6627197265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.08746666666666666,
"grad_norm": 0.016892658546566963,
"kl": 0.04087066650390625,
"learning_rate": 3.277777777777778e-06,
"loss": -0.1035,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.018185051158070564,
"mask/share_reasoning": 0.853351891040802,
"mask/share_step_conf": 0.11283804476261139,
"num_tokens": 26018177.0,
"reward": 0.8254947066307068,
"reward_std": 0.19936254620552063,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6241007447242737,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7097011208534241,
"step": 82
},
{
"adv/mean_abs_final_conf": 0.41412049531936646,
"adv/mean_abs_reasoning": 0.26981937885284424,
"adv/mean_abs_step_conf": 0.7399859428405762,
"adv/ratio_final_to_reasoning": 1.5348063474166622,
"adv/ratio_step_to_reasoning": 2.742523335375975,
"adv/std_final_conf": 0.6758379936218262,
"adv/std_reasoning": 0.5727855563163757,
"adv/std_step_conf": 0.9238594770431519,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5109394229454397,
"calib/avg_num_step_conf": 8.1640625,
"calib/ece": 0.40860816326530613,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9918367346938776,
"calib/gap": 0.001752153698892478,
"calib/mean_conf": 0.9882,
"calib/mu_c": 0.9889366197183097,
"calib/mu_w": 0.9871844660194172,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40860816326530613,
"calib/std_conf": 0.008615884713002783,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8929226713532513,
"calib/step_q_c_n": 1138.0,
"calib/step_q_gap": 0.027365948664175743,
"calib/step_q_w": 0.8655567226890756,
"calib/step_q_w_n": 952.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3038.0,
"completions/max_terminated_length": 3038.0,
"completions/mean_length": 890.20703125,
"completions/mean_terminated_length": 897.216552734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 485.0,
"epoch": 0.08853333333333334,
"grad_norm": 0.021719301119446754,
"kl": 0.0386505126953125,
"learning_rate": 3.2500000000000002e-06,
"loss": -0.0612,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016932372003793716,
"mask/share_reasoning": 0.8745453953742981,
"mask/share_step_conf": 0.10070972144603729,
"num_tokens": 26353334.0,
"reward": 0.7632154226303101,
"reward_std": 0.13492730259895325,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.564795732498169,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6592912673950195,
"step": 83
},
{
"adv/mean_abs_final_conf": 0.4269552230834961,
"adv/mean_abs_reasoning": 0.35676440596580505,
"adv/mean_abs_step_conf": 0.7534521818161011,
"adv/ratio_final_to_reasoning": 1.1967427690205694,
"adv/ratio_step_to_reasoning": 2.111904016255247,
"adv/std_final_conf": 0.689056932926178,
"adv/std_reasoning": 0.6404105424880981,
"adv/std_step_conf": 0.9300892353057861,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5342475640562974,
"calib/avg_num_step_conf": 8.37109375,
"calib/ece": 0.33128629032258056,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.9959677419354839,
"calib/gap": 0.002797185131721669,
"calib/mean_conf": 0.9885443548387097,
"calib/mu_c": 0.9895030674846625,
"calib/mu_w": 0.9867058823529409,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33128629032258056,
"calib/std_conf": 0.012289438177264208,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.893919770773639,
"calib/step_q_c_n": 1396.0,
"calib/step_q_gap": 0.008210266088230722,
"calib/step_q_w": 0.8857095046854083,
"calib/step_q_w_n": 747.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2809.0,
"completions/max_terminated_length": 2809.0,
"completions/mean_length": 775.32421875,
"completions/mean_terminated_length": 793.9320678710938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 464.0,
"epoch": 0.0896,
"grad_norm": 0.01733134128153324,
"kl": 0.04576873779296875,
"learning_rate": 3.2222222222222227e-06,
"loss": -0.0971,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.01821688935160637,
"mask/share_reasoning": 0.8478014469146729,
"mask/share_step_conf": 0.11054414510726929,
"num_tokens": 26657737.0,
"reward": 0.8509284257888794,
"reward_std": 0.20014074444770813,
"rewards/accuracy_reward_step": 0.63671875,
"rewards/final_brier_reward_step": 0.6452734470367432,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7354897260665894,
"step": 84
},
{
"adv/mean_abs_final_conf": 0.5446729063987732,
"adv/mean_abs_reasoning": 0.47163212299346924,
"adv/mean_abs_step_conf": 0.7699087858200073,
"adv/ratio_final_to_reasoning": 1.1548681267546217,
"adv/ratio_step_to_reasoning": 1.632435002377199,
"adv/std_final_conf": 0.7841033935546875,
"adv/std_reasoning": 0.7394751906394958,
"adv/std_step_conf": 0.9309518933296204,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.5223388305847078,
"calib/avg_num_step_conf": 8.890625,
"calib/ece": 0.37514767932489457,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 0.9873417721518988,
"calib/gap": 0.005516491754123276,
"calib/mean_conf": 0.9869620253164557,
"calib/mu_c": 0.9891034482758622,
"calib/mu_w": 0.9835869565217389,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.37514767932489457,
"calib/std_conf": 0.01857958798167976,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8873767752715122,
"calib/step_q_c_n": 1197.0,
"calib/step_q_gap": 0.008933772491160008,
"calib/step_q_w": 0.8784430027803521,
"calib/step_q_w_n": 1079.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2759.0,
"completions/max_terminated_length": 2759.0,
"completions/mean_length": 814.48828125,
"completions/mean_terminated_length": 851.05712890625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 487.0,
"epoch": 0.09066666666666667,
"grad_norm": 0.01905909925699234,
"kl": 0.044300079345703125,
"learning_rate": 3.1944444444444443e-06,
"loss": -0.1392,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.0170908086001873,
"mask/share_reasoning": 0.8357868194580078,
"mask/share_step_conf": 0.104153573513031,
"num_tokens": 26974070.0,
"reward": 0.7635809183120728,
"reward_std": 0.2349778413772583,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.5776484608650208,
"rewards/format_reward_step": 0.921875,
"rewards/step_l2_reward": 0.6518570184707642,
"step": 85
},
{
"adv/mean_abs_final_conf": 0.525831937789917,
"adv/mean_abs_reasoning": 0.43927913904190063,
"adv/mean_abs_step_conf": 0.7811049222946167,
"adv/ratio_final_to_reasoning": 1.1970337105850148,
"adv/ratio_step_to_reasoning": 1.7781516417972014,
"adv/std_final_conf": 0.751258373260498,
"adv/std_reasoning": 0.7014529705047607,
"adv/std_step_conf": 0.9302042722702026,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.4905166775670373,
"calib/avg_num_step_conf": 8.46875,
"calib/ece": 0.42939759036144587,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9919678714859438,
"calib/gap": 0.0025951602354481462,
"calib/mean_conf": 0.9876305220883534,
"calib/mu_c": 0.9887769784172662,
"calib/mu_w": 0.986181818181818,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42939759036144587,
"calib/std_conf": 0.01641602802241287,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8897205757832345,
"calib/step_q_c_n": 1181.0,
"calib/step_q_gap": 0.007218042855169693,
"calib/step_q_w": 0.8825025329280648,
"calib/step_q_w_n": 987.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1973.0,
"completions/max_terminated_length": 1973.0,
"completions/mean_length": 849.42578125,
"completions/mean_terminated_length": 856.1141967773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 444.0,
"epoch": 0.09173333333333333,
"grad_norm": 0.018472064286470413,
"kl": 0.046112060546875,
"learning_rate": 3.1666666666666667e-06,
"loss": -0.0582,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01759396679699421,
"mask/share_reasoning": 0.8653203248977661,
"mask/share_step_conf": 0.10927321016788483,
"num_tokens": 27297035.0,
"reward": 0.754798412322998,
"reward_std": 0.2171429842710495,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.55443274974823,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.6520390510559082,
"step": 86
},
{
"adv/mean_abs_final_conf": 0.5290802121162415,
"adv/mean_abs_reasoning": 0.44977080821990967,
"adv/mean_abs_step_conf": 0.7352198362350464,
"adv/ratio_final_to_reasoning": 1.1763329287870423,
"adv/ratio_step_to_reasoning": 1.6346544124214706,
"adv/std_final_conf": 0.769879937171936,
"adv/std_reasoning": 0.7393056750297546,
"adv/std_step_conf": 0.9318410158157349,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5486928104575164,
"calib/avg_num_step_conf": 8.63671875,
"calib/ece": 0.2612096774193548,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9879032258064516,
"calib/gap": 0.004196078431372374,
"calib/mean_conf": 0.9870161290322582,
"calib/mu_c": 0.9881666666666666,
"calib/mu_w": 0.9839705882352943,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2612096774193548,
"calib/std_conf": 0.020179830185260015,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8923865110246433,
"calib/step_q_c_n": 1542.0,
"calib/step_q_gap": 0.009021937033611871,
"calib/step_q_w": 0.8833645739910314,
"calib/step_q_w_n": 669.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3017.0,
"completions/max_terminated_length": 3017.0,
"completions/mean_length": 773.546875,
"completions/mean_terminated_length": 795.2931518554688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.0928,
"grad_norm": 0.019448574632406235,
"kl": 0.04779815673828125,
"learning_rate": 3.138888888888889e-06,
"loss": -0.0334,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.01838276907801628,
"mask/share_reasoning": 0.840671956539154,
"mask/share_step_conf": 0.11360155045986176,
"num_tokens": 27600559.0,
"reward": 0.9018452167510986,
"reward_std": 0.2351362109184265,
"rewards/accuracy_reward_step": 0.703125,
"rewards/final_brier_reward_step": 0.7070223093032837,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7638556361198425,
"step": 87
},
{
"adv/mean_abs_final_conf": 0.45828092098236084,
"adv/mean_abs_reasoning": 0.39458200335502625,
"adv/mean_abs_step_conf": 0.7401527762413025,
"adv/ratio_final_to_reasoning": 1.1614339150942505,
"adv/ratio_step_to_reasoning": 1.875789493560222,
"adv/std_final_conf": 0.7244842648506165,
"adv/std_reasoning": 0.6816993355751038,
"adv/std_step_conf": 0.9296145439147949,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5261466549810109,
"calib/avg_num_step_conf": 8.265625,
"calib/ece": 0.32947368421052625,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0005229330996203707,
"calib/mean_conf": 0.9893927125506072,
"calib/mu_c": 0.9895705521472392,
"calib/mu_w": 0.9890476190476188,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.32947368421052625,
"calib/std_conf": 0.0023883208427129777,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8906811377245512,
"calib/step_q_c_n": 1336.0,
"calib/step_q_gap": 0.00586062490403827,
"calib/step_q_w": 0.8848205128205129,
"calib/step_q_w_n": 780.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2735.0,
"completions/max_terminated_length": 2735.0,
"completions/mean_length": 803.74609375,
"completions/mean_terminated_length": 823.0360107421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 423.0,
"epoch": 0.09386666666666667,
"grad_norm": 0.017657170072197914,
"kl": 0.04468536376953125,
"learning_rate": 3.1111111111111116e-06,
"loss": -0.1192,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.017656173557043076,
"mask/share_reasoning": 0.854995846748352,
"mask/share_step_conf": 0.10391046106815338,
"num_tokens": 27916166.0,
"reward": 0.8516751527786255,
"reward_std": 0.21041029691696167,
"rewards/accuracy_reward_step": 0.63671875,
"rewards/final_brier_reward_step": 0.6437921524047852,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7392456531524658,
"step": 88
},
{
"adv/mean_abs_final_conf": 0.5156100392341614,
"adv/mean_abs_reasoning": 0.4167075455188751,
"adv/mean_abs_step_conf": 0.7418398857116699,
"adv/ratio_final_to_reasoning": 1.237342699403571,
"adv/ratio_step_to_reasoning": 1.780241067600413,
"adv/std_final_conf": 0.763513445854187,
"adv/std_reasoning": 0.7014049291610718,
"adv/std_step_conf": 0.928699791431427,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5216002747252748,
"calib/avg_num_step_conf": 8.4453125,
"calib/ece": 0.41422131147540997,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9918032786885246,
"calib/gap": 0.0003543956043953189,
"calib/mean_conf": 0.9879918032786886,
"calib/mu_c": 0.988142857142857,
"calib/mu_w": 0.9877884615384617,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.41422131147540997,
"calib/std_conf": 0.009124813855401486,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8837217099748533,
"calib/step_q_c_n": 1193.0,
"calib/step_q_gap": 0.010646374577536477,
"calib/step_q_w": 0.8730753353973169,
"calib/step_q_w_n": 969.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2964.0,
"completions/max_terminated_length": 2964.0,
"completions/mean_length": 859.01171875,
"completions/mean_terminated_length": 883.1605834960938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 427.0,
"epoch": 0.09493333333333333,
"grad_norm": 0.025277912616729736,
"kl": 0.044895172119140625,
"learning_rate": 3.0833333333333336e-06,
"loss": -0.0676,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.016975190490484238,
"mask/share_reasoning": 0.8520951271057129,
"mask/share_step_conf": 0.10358592122793198,
"num_tokens": 28244961.0,
"reward": 0.7633959650993347,
"reward_std": 0.19842402637004852,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.5565800666809082,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6702117919921875,
"step": 89
},
{
"adv/mean_abs_final_conf": 0.5149222612380981,
"adv/mean_abs_reasoning": 0.4274728298187256,
"adv/mean_abs_step_conf": 0.7600886821746826,
"adv/ratio_final_to_reasoning": 1.204573075337809,
"adv/ratio_step_to_reasoning": 1.7780982302360744,
"adv/std_final_conf": 0.7509579658508301,
"adv/std_reasoning": 0.7014234066009521,
"adv/std_step_conf": 0.9276285171508789,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5043859649122807,
"calib/avg_num_step_conf": 9.07421875,
"calib/ece": 0.29223200000000016,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.992,
"calib/gap": 0.00067362371445856,
"calib/mean_conf": 0.9882320000000001,
"calib/mu_c": 0.9884367816091953,
"calib/mu_w": 0.9877631578947368,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.29223200000000016,
"calib/std_conf": 0.008574507332785948,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8846670918367348,
"calib/step_q_c_n": 1568.0,
"calib/step_q_gap": 0.0035809991215028925,
"calib/step_q_w": 0.8810860927152319,
"calib/step_q_w_n": 755.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2763.0,
"completions/max_terminated_length": 2763.0,
"completions/mean_length": 829.625,
"completions/mean_terminated_length": 842.793701171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 399.0,
"epoch": 0.096,
"grad_norm": 0.01628192327916622,
"kl": 0.04920196533203125,
"learning_rate": 3.055555555555556e-06,
"loss": -0.0194,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017995685338974,
"mask/share_reasoning": 0.8512020111083984,
"mask/share_step_conf": 0.11517727375030518,
"num_tokens": 28560665.0,
"reward": 0.8831427693367004,
"reward_std": 0.22143936157226562,
"rewards/accuracy_reward_step": 0.68359375,
"rewards/final_brier_reward_step": 0.6867460608482361,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7475081086158752,
"step": 90
},
{
"adv/mean_abs_final_conf": 0.5486507415771484,
"adv/mean_abs_reasoning": 0.4548686742782593,
"adv/mean_abs_step_conf": 0.7140244245529175,
"adv/ratio_final_to_reasoning": 1.2061739411879555,
"adv/ratio_step_to_reasoning": 1.569737519704695,
"adv/std_final_conf": 0.79872065782547,
"adv/std_reasoning": 0.7394763231277466,
"adv/std_step_conf": 0.9286575317382812,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5275906183368869,
"calib/avg_num_step_conf": 8.875,
"calib/ece": 0.26574380165247946,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.987603305785124,
"calib/gap": 0.0005057569290661812,
"calib/mean_conf": 0.9872314049582647,
"calib/mu_c": 0.9873714285708571,
"calib/mu_w": 0.986865671641791,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.26491735537148775,
"calib/std_conf": 0.01499931697064191,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.887391186440678,
"calib/step_q_c_n": 1475.0,
"calib/step_q_gap": 0.02412267953979974,
"calib/step_q_w": 0.8632685069008783,
"calib/step_q_w_n": 797.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 3070.0,
"completions/max_terminated_length": 3070.0,
"completions/mean_length": 850.0078125,
"completions/mean_terminated_length": 873.903564453125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 528.0,
"epoch": 0.09706666666666666,
"grad_norm": 0.01996271312236786,
"kl": 0.048755645751953125,
"learning_rate": 3.0277777777777776e-06,
"loss": -0.1626,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016846606507897377,
"mask/share_reasoning": 0.8526731133460999,
"mask/share_step_conf": 0.10313647985458374,
"num_tokens": 28885979.0,
"reward": 0.8746767640113831,
"reward_std": 0.24369308352470398,
"rewards/accuracy_reward_step": 0.6875,
"rewards/final_brier_reward_step": 0.6861194968223572,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.7382339835166931,
"step": 91
},
{
"adv/mean_abs_final_conf": 0.537227988243103,
"adv/mean_abs_reasoning": 0.3748323321342468,
"adv/mean_abs_step_conf": 0.7305430173873901,
"adv/ratio_final_to_reasoning": 1.433248794692273,
"adv/ratio_step_to_reasoning": 1.9489861326203444,
"adv/std_final_conf": 0.7719976902008057,
"adv/std_reasoning": 0.661332905292511,
"adv/std_step_conf": 0.9242905974388123,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5755091723241369,
"calib/avg_num_step_conf": 9.01171875,
"calib/ece": 0.33553846153846156,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.9757085020242915,
"calib/gap": 0.008520511339015324,
"calib/mean_conf": 0.9792631578947368,
"calib/mu_c": 0.9822298136645964,
"calib/mu_w": 0.9737093023255811,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33148987854251016,
"calib/std_conf": 0.08567871193649285,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8842680339462518,
"calib/step_q_c_n": 1414.0,
"calib/step_q_gap": 0.02231618624188436,
"calib/step_q_w": 0.8619518477043674,
"calib/step_q_w_n": 893.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2816.0,
"completions/max_terminated_length": 2816.0,
"completions/mean_length": 805.90625,
"completions/mean_terminated_length": 818.698486328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 434.0,
"epoch": 0.09813333333333334,
"grad_norm": 0.022326858714222908,
"kl": 0.0508575439453125,
"learning_rate": 3e-06,
"loss": -0.0959,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.01844978705048561,
"mask/share_reasoning": 0.8456693291664124,
"mask/share_step_conf": 0.12025590240955353,
"num_tokens": 29199011.0,
"reward": 0.8462902307510376,
"reward_std": 0.18844886124134064,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.639072597026825,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.734757661819458,
"step": 92
},
{
"adv/mean_abs_final_conf": 0.5825698375701904,
"adv/mean_abs_reasoning": 0.4962431788444519,
"adv/mean_abs_step_conf": 0.7400534152984619,
"adv/ratio_final_to_reasoning": 1.1739603936254763,
"adv/ratio_step_to_reasoning": 1.491312015656809,
"adv/std_final_conf": 0.8054587244987488,
"adv/std_reasoning": 0.7754384875297546,
"adv/std_step_conf": 0.9276998043060303,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5200179600389134,
"calib/avg_num_step_conf": 9.0859375,
"calib/ece": 0.32852459016393454,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9959016393442623,
"calib/gap": 0.0013499962583252145,
"calib/mean_conf": 0.988360655737705,
"calib/mu_c": 0.9888198757763974,
"calib/mu_w": 0.9874698795180722,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.32852459016393454,
"calib/std_conf": 0.006573750051069972,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8862597864768684,
"calib/step_q_c_n": 1405.0,
"calib/step_q_gap": 0.015358592122905357,
"calib/step_q_w": 0.8709011943539631,
"calib/step_q_w_n": 921.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2945.0,
"completions/max_terminated_length": 2945.0,
"completions/mean_length": 818.04296875,
"completions/mean_terminated_length": 844.431396484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 410.0,
"epoch": 0.0992,
"grad_norm": 6.2837347984313965,
"kl": 3.0508956909179688,
"learning_rate": 2.9722222222222225e-06,
"loss": -0.0992,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017331324517726898,
"mask/share_reasoning": 0.8419540524482727,
"mask/share_step_conf": 0.1094646155834198,
"num_tokens": 29514206.0,
"reward": 0.8271012902259827,
"reward_std": 0.2480258047580719,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.6368609666824341,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7009353637695312,
"step": 93
},
{
"adv/mean_abs_final_conf": 0.5216161608695984,
"adv/mean_abs_reasoning": 0.43027183413505554,
"adv/mean_abs_step_conf": 0.7488489747047424,
"adv/ratio_final_to_reasoning": 1.212294460124646,
"adv/ratio_step_to_reasoning": 1.7404090049493934,
"adv/std_final_conf": 0.7995529770851135,
"adv/std_reasoning": 0.7396001815795898,
"adv/std_step_conf": 0.9273629784584045,
"calib/answer_extract_rate": 0.9140625,
"calib/auroc": 0.5579310344827586,
"calib/avg_num_step_conf": 9.29296875,
"calib/ece": 0.36985957446808526,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 0.9787234042553191,
"calib/gap": 0.0030482758620687944,
"calib/mean_conf": 0.98688085106383,
"calib/mu_c": 0.9880482758620689,
"calib/mu_w": 0.9850000000000001,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36985957446808526,
"calib/std_conf": 0.013457365906745298,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8902324455205811,
"calib/step_q_c_n": 1239.0,
"calib/step_q_gap": 0.03742542797672144,
"calib/step_q_w": 0.8528070175438597,
"calib/step_q_w_n": 1140.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2935.0,
"completions/max_terminated_length": 2935.0,
"completions/mean_length": 828.6015625,
"completions/mean_terminated_length": 869.3524169921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 465.0,
"epoch": 0.10026666666666667,
"grad_norm": 0.0222991481423378,
"kl": 0.054683685302734375,
"learning_rate": 2.944444444444445e-06,
"loss": -0.1839,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.016976788640022278,
"mask/share_reasoning": 0.8307620286941528,
"mask/share_step_conf": 0.10538618266582489,
"num_tokens": 29835008.0,
"reward": 0.767528235912323,
"reward_std": 0.2512366771697998,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.5765511393547058,
"rewards/format_reward_step": 0.9140625,
"rewards/step_l2_reward": 0.6624114513397217,
"step": 94
},
{
"adv/mean_abs_final_conf": 0.527319073677063,
"adv/mean_abs_reasoning": 0.3920542001724243,
"adv/mean_abs_step_conf": 0.7362141609191895,
"adv/ratio_final_to_reasoning": 1.3450157489580512,
"adv/ratio_step_to_reasoning": 1.8778377086520297,
"adv/std_final_conf": 0.7650555968284607,
"adv/std_reasoning": 0.6817232370376587,
"adv/std_step_conf": 0.9281631112098694,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5381760339342524,
"calib/avg_num_step_conf": 8.74609375,
"calib/ece": 0.3236625514403294,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9794238683127572,
"calib/gap": 0.004409180427207837,
"calib/mean_conf": 0.9862139917695475,
"calib/mu_c": 0.9877018633540372,
"calib/mu_w": 0.9832926829268294,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3236625514403294,
"calib/std_conf": 0.014560343067339289,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8967602591792656,
"calib/step_q_c_n": 1389.0,
"calib/step_q_gap": 0.050474376826324496,
"calib/step_q_w": 0.8462858823529411,
"calib/step_q_w_n": 850.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2133.0,
"completions/max_terminated_length": 2133.0,
"completions/mean_length": 782.41015625,
"completions/mean_terminated_length": 797.9960327148438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 417.0,
"epoch": 0.10133333333333333,
"grad_norm": 0.0220172256231308,
"kl": 0.0579681396484375,
"learning_rate": 2.916666666666667e-06,
"loss": -0.0616,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.0183150302618742,
"mask/share_reasoning": 0.8456125259399414,
"mask/share_step_conf": 0.11654117703437805,
"num_tokens": 30141433.0,
"reward": 0.8251522779464722,
"reward_std": 0.2152584344148636,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.6392277479171753,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6954517364501953,
"step": 95
},
{
"adv/mean_abs_final_conf": 0.5503877997398376,
"adv/mean_abs_reasoning": 0.3534834384918213,
"adv/mean_abs_step_conf": 0.7321844100952148,
"adv/ratio_final_to_reasoning": 1.5570398491316368,
"adv/ratio_step_to_reasoning": 2.071340069620138,
"adv/std_final_conf": 0.7819110155105591,
"adv/std_reasoning": 0.6613574624061584,
"adv/std_step_conf": 0.9222843050956726,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5541159880133968,
"calib/avg_num_step_conf": 8.9453125,
"calib/ece": 0.23044129554655873,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9838056680161943,
"calib/gap": 0.016938039837828267,
"calib/mean_conf": 0.9822631578947368,
"calib/mu_c": 0.9864462365591397,
"calib/mu_w": 0.9695081967213114,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.22983400809716603,
"calib/std_conf": 0.06409430977468505,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8916709102482495,
"calib/step_q_c_n": 1571.0,
"calib/step_q_gap": 0.011142398426274513,
"calib/step_q_w": 0.880528511821975,
"calib/step_q_w_n": 719.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2074.0,
"completions/max_terminated_length": 2074.0,
"completions/mean_length": 743.26171875,
"completions/mean_terminated_length": 767.2378540039062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 422.0,
"epoch": 0.1024,
"grad_norm": 0.020050369203090668,
"kl": 0.062469482421875,
"learning_rate": 2.888888888888889e-06,
"loss": -0.148,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.01858597621321678,
"mask/share_reasoning": 0.8389642238616943,
"mask/share_step_conf": 0.1111997589468956,
"num_tokens": 30437524.0,
"reward": 0.9249247312545776,
"reward_std": 0.19512362778186798,
"rewards/accuracy_reward_step": 0.7265625,
"rewards/final_brier_reward_step": 0.7367488145828247,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7756006121635437,
"step": 96
},
{
"adv/mean_abs_final_conf": 0.5968998074531555,
"adv/mean_abs_reasoning": 0.5177270770072937,
"adv/mean_abs_step_conf": 0.7378278970718384,
"adv/ratio_final_to_reasoning": 1.1529236811478307,
"adv/ratio_step_to_reasoning": 1.4251290493377922,
"adv/std_final_conf": 0.8343783617019653,
"adv/std_reasoning": 0.7928089499473572,
"adv/std_step_conf": 0.9322978854179382,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.4990490278951817,
"calib/avg_num_step_conf": 8.9140625,
"calib/ece": 0.3497813765182187,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.97165991902834,
"calib/gap": 0.01189743589743597,
"calib/mean_conf": 0.9813603238866397,
"calib/mu_c": 0.9857435897435897,
"calib/mu_w": 0.9738461538461537,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3497813765182187,
"calib/std_conf": 0.06528024082990255,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8857207407407407,
"calib/step_q_c_n": 1350.0,
"calib/step_q_gap": 0.007405290096964001,
"calib/step_q_w": 0.8783154506437767,
"calib/step_q_w_n": 932.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2562.0,
"completions/max_terminated_length": 2562.0,
"completions/mean_length": 770.0859375,
"completions/mean_terminated_length": 791.7349243164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 422.0,
"epoch": 0.10346666666666667,
"grad_norm": 0.021622126922011375,
"kl": 0.0661468505859375,
"learning_rate": 2.861111111111111e-06,
"loss": -0.1201,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.018322018906474113,
"mask/share_reasoning": 0.8374599814414978,
"mask/share_step_conf": 0.11687424778938293,
"num_tokens": 30739738.0,
"reward": 0.8113015294075012,
"reward_std": 0.2594345211982727,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.6235218048095703,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6842373609542847,
"step": 97
},
{
"adv/mean_abs_final_conf": 0.5326055288314819,
"adv/mean_abs_reasoning": 0.4282838702201843,
"adv/mean_abs_step_conf": 0.7558221817016602,
"adv/ratio_final_to_reasoning": 1.2435806386023012,
"adv/ratio_step_to_reasoning": 1.7647691969185897,
"adv/std_final_conf": 0.7673627734184265,
"adv/std_reasoning": 0.7014768123626709,
"adv/std_step_conf": 0.9310451149940491,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5135707344162154,
"calib/avg_num_step_conf": 8.60546875,
"calib/ece": 0.3856204081632655,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9795918367346939,
"calib/gap": 0.0012210884353739404,
"calib/mean_conf": 0.9856204081632655,
"calib/mu_c": 0.9861088435374148,
"calib/mu_w": 0.9848877551020409,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3856204081632655,
"calib/std_conf": 0.014391848310505025,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8865980629539952,
"calib/step_q_c_n": 1239.0,
"calib/step_q_gap": 0.0038791832859453867,
"calib/step_q_w": 0.8827188796680498,
"calib/step_q_w_n": 964.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3041.0,
"completions/max_terminated_length": 3041.0,
"completions/mean_length": 808.16015625,
"completions/mean_terminated_length": 834.2297973632812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 442.0,
"epoch": 0.10453333333333334,
"grad_norm": 0.016606997698545456,
"kl": 0.0679779052734375,
"learning_rate": 2.8333333333333335e-06,
"loss": -0.0813,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01783130317926407,
"mask/share_reasoning": 0.8402774930000305,
"mask/share_step_conf": 0.1106412261724472,
"num_tokens": 31052811.0,
"reward": 0.7877363562583923,
"reward_std": 0.22248929738998413,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.585392951965332,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6838297843933105,
"step": 98
},
{
"adv/mean_abs_final_conf": 0.6818572282791138,
"adv/mean_abs_reasoning": 0.5733513236045837,
"adv/mean_abs_step_conf": 0.760155200958252,
"adv/ratio_final_to_reasoning": 1.1892485465846103,
"adv/ratio_step_to_reasoning": 1.3258104929090544,
"adv/std_final_conf": 0.8731258511543274,
"adv/std_reasoning": 0.8100834488868713,
"adv/std_step_conf": 0.93084317445755,
"calib/answer_extract_rate": 0.8984375,
"calib/auroc": 0.5508882866606444,
"calib/avg_num_step_conf": 9.7890625,
"calib/ece": 0.5188268398268399,
"calib/final_conf_rate": 0.90234375,
"calib/format_rate": 0.89453125,
"calib/frac_conf_gt_0.9": 0.9783549783549783,
"calib/gap": -0.0005745257452575148,
"calib/mean_conf": 0.9841948051948052,
"calib/mu_c": 0.9838888888888888,
"calib/mu_w": 0.9844634146341463,
"calib/nonempty_final_conf_rate": 0.90234375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.5177445887445887,
"calib/std_conf": 0.02099229231528474,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8921641791044775,
"calib/step_q_c_n": 938.0,
"calib/step_q_gap": 0.008407163798354933,
"calib/step_q_w": 0.8837570153061226,
"calib/step_q_w_n": 1568.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2717.0,
"completions/max_terminated_length": 2717.0,
"completions/mean_length": 807.765625,
"completions/mean_terminated_length": 865.2217407226562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 396.0,
"epoch": 0.1056,
"grad_norm": 0.028910474851727486,
"kl": 0.06258392333984375,
"learning_rate": 2.805555555555556e-06,
"loss": -0.2888,
"mask/has_final_conf_rate": 0.90234375,
"mask/share_final_conf": 0.016516810283064842,
"mask/share_reasoning": 0.8145354986190796,
"mask/share_step_conf": 0.10254141688346863,
"num_tokens": 31365399.0,
"reward": 0.6236565113067627,
"reward_std": 0.2722751200199127,
"rewards/accuracy_reward_step": 0.421875,
"rewards/final_brier_reward_step": 0.4360261559486389,
"rewards/format_reward_step": 0.89453125,
"rewards/step_l2_reward": 0.5480057001113892,
"step": 99
},
{
"adv/mean_abs_final_conf": 0.5936969518661499,
"adv/mean_abs_reasoning": 0.4137655794620514,
"adv/mean_abs_step_conf": 0.7432897686958313,
"adv/ratio_final_to_reasoning": 1.434863075459376,
"adv/ratio_step_to_reasoning": 1.7964030977690408,
"adv/std_final_conf": 0.8098013401031494,
"adv/std_reasoning": 0.7014086246490479,
"adv/std_step_conf": 0.9335170984268188,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5957593330916999,
"calib/avg_num_step_conf": 9.30859375,
"calib/ece": 0.34782377049180346,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9631147540983607,
"calib/gap": 0.006088075389633518,
"calib/mean_conf": 0.9830696721311477,
"calib/mu_c": 0.985290322580645,
"calib/mu_w": 0.9792022471910115,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34782377049180346,
"calib/std_conf": 0.018483689705748783,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8866344005956813,
"calib/step_q_c_n": 1343.0,
"calib/step_q_gap": -0.004342522481241895,
"calib/step_q_w": 0.8909769230769232,
"calib/step_q_w_n": 1040.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2693.0,
"completions/max_terminated_length": 2693.0,
"completions/mean_length": 797.84765625,
"completions/mean_terminated_length": 826.9190673828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 412.0,
"epoch": 0.10666666666666667,
"grad_norm": 0.023713137954473495,
"kl": 0.06809234619140625,
"learning_rate": 2.7777777777777783e-06,
"loss": -0.1413,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01752588339149952,
"mask/share_reasoning": 0.8371009826660156,
"mask/share_step_conf": 0.11021691560745239,
"num_tokens": 31677056.0,
"reward": 0.8176565170288086,
"reward_std": 0.2034880518913269,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.619330883026123,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7042633891105652,
"step": 100
},
{
"adv/mean_abs_final_conf": 0.701155424118042,
"adv/mean_abs_reasoning": 0.5449827909469604,
"adv/mean_abs_step_conf": 0.7575080990791321,
"adv/ratio_final_to_reasoning": 1.2865643388476844,
"adv/ratio_step_to_reasoning": 1.3899670075139223,
"adv/std_final_conf": 0.8829103112220764,
"adv/std_reasoning": 0.7928749918937683,
"adv/std_step_conf": 0.932668924331665,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.49327324973876696,
"calib/avg_num_step_conf": 8.59375,
"calib/ece": 0.515241935483871,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.9717741935483871,
"calib/gap": -0.0023667711598743812,
"calib/mean_conf": 0.9829838709677421,
"calib/mu_c": 0.9817241379310345,
"calib/mu_w": 0.9840909090909089,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.515241935483871,
"calib/std_conf": 0.02107895444512084,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8749799599198397,
"calib/step_q_c_n": 998.0,
"calib/step_q_gap": -0.00976472486478086,
"calib/step_q_w": 0.8847446847846205,
"calib/step_q_w_n": 1202.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2546.0,
"completions/max_terminated_length": 2546.0,
"completions/mean_length": 858.80859375,
"completions/mean_terminated_length": 875.9163818359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 426.0,
"epoch": 0.10773333333333333,
"grad_norm": 0.031017888337373734,
"kl": 0.07028961181640625,
"learning_rate": 2.7500000000000004e-06,
"loss": -0.0571,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.017236188054084778,
"mask/share_reasoning": 0.8573594093322754,
"mask/share_step_conf": 0.1058732271194458,
"num_tokens": 32003903.0,
"reward": 0.6754347085952759,
"reward_std": 0.2761334776878357,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.4688202738761902,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.5976740717887878,
"step": 101
},
{
"adv/mean_abs_final_conf": 0.6097878813743591,
"adv/mean_abs_reasoning": 0.32521986961364746,
"adv/mean_abs_step_conf": 0.7349889874458313,
"adv/ratio_final_to_reasoning": 1.8750019243866338,
"adv/ratio_step_to_reasoning": 2.2599756537599585,
"adv/std_final_conf": 0.8272060751914978,
"adv/std_reasoning": 0.6403584480285645,
"adv/std_step_conf": 0.9255380630493164,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5119204329384233,
"calib/avg_num_step_conf": 9.3984375,
"calib/ece": 0.33240408163265317,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9551020408163265,
"calib/gap": -0.0018097118619277985,
"calib/mean_conf": 0.9813836734693878,
"calib/mu_c": 0.9807484276729559,
"calib/mu_w": 0.9825581395348837,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33240408163265317,
"calib/std_conf": 0.02426822350770464,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8855560026827632,
"calib/step_q_c_n": 1491.0,
"calib/step_q_gap": 0.006915565524293288,
"calib/step_q_w": 0.87864043715847,
"calib/step_q_w_n": 915.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2415.0,
"completions/max_terminated_length": 2415.0,
"completions/mean_length": 752.43359375,
"completions/mean_terminated_length": 773.5863037109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 380.0,
"epoch": 0.1088,
"grad_norm": 0.031540971249341965,
"kl": 0.0884246826171875,
"learning_rate": 2.7222222222222224e-06,
"loss": -0.1251,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01865723729133606,
"mask/share_reasoning": 0.8307026028633118,
"mask/share_step_conf": 0.12329642474651337,
"num_tokens": 32303222.0,
"reward": 0.8231228590011597,
"reward_std": 0.17860785126686096,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.6319171786308289,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6987035274505615,
"step": 102
},
{
"adv/mean_abs_final_conf": 0.5567460060119629,
"adv/mean_abs_reasoning": 0.3415173292160034,
"adv/mean_abs_step_conf": 0.7269463539123535,
"adv/ratio_final_to_reasoning": 1.63021304743172,
"adv/ratio_step_to_reasoning": 2.12857823519864,
"adv/std_final_conf": 0.7949619889259338,
"adv/std_reasoning": 0.6612794995307922,
"adv/std_step_conf": 0.9273793697357178,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5904428409734013,
"calib/avg_num_step_conf": 8.5625,
"calib/ece": 0.35893877551020414,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9224489795918367,
"calib/gap": 0.010909734012450545,
"calib/mean_conf": 0.9773061224489797,
"calib/mu_c": 0.9814473684210526,
"calib/mu_w": 0.9705376344086021,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3579183673469389,
"calib/std_conf": 0.029569844044251104,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8903230890464934,
"calib/step_q_c_n": 1269.0,
"calib/step_q_gap": 0.019304670844976668,
"calib/step_q_w": 0.8710184182015167,
"calib/step_q_w_n": 923.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2749.0,
"completions/max_terminated_length": 2749.0,
"completions/mean_length": 844.40625,
"completions/mean_terminated_length": 868.14453125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 341.0,
"epoch": 0.10986666666666667,
"grad_norm": 0.02460324950516224,
"kl": 0.06529998779296875,
"learning_rate": 2.6944444444444444e-06,
"loss": -0.1313,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.017562564462423325,
"mask/share_reasoning": 0.8493900299072266,
"mask/share_step_conf": 0.10570365190505981,
"num_tokens": 32623942.0,
"reward": 0.812166154384613,
"reward_std": 0.17410849034786224,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6138265132904053,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7003495693206787,
"step": 103
},
{
"adv/mean_abs_final_conf": 0.6447658538818359,
"adv/mean_abs_reasoning": 0.45572638511657715,
"adv/mean_abs_step_conf": 0.7367246747016907,
"adv/ratio_final_to_reasoning": 1.4148091375418201,
"adv/ratio_step_to_reasoning": 1.6165942959682547,
"adv/std_final_conf": 0.8271310925483704,
"adv/std_reasoning": 0.7392915487289429,
"adv/std_step_conf": 0.9331515431404114,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.4921793534932221,
"calib/avg_num_step_conf": 9.33984375,
"calib/ece": 0.42979919678714884,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9558232931726908,
"calib/gap": 0.004381517205422458,
"calib/mean_conf": 0.9800000000000002,
"calib/mu_c": 0.9819708029197083,
"calib/mu_w": 0.9775892857142858,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42979919678714884,
"calib/std_conf": 0.02084557873052321,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8894085173501577,
"calib/step_q_c_n": 1268.0,
"calib/step_q_gap": 0.013023833467699975,
"calib/step_q_w": 0.8763846838824577,
"calib/step_q_w_n": 1123.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2769.0,
"completions/max_terminated_length": 2769.0,
"completions/mean_length": 790.4609375,
"completions/mean_terminated_length": 806.2072143554688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 408.0,
"epoch": 0.11093333333333333,
"grad_norm": 0.02433106303215027,
"kl": 0.0781707763671875,
"learning_rate": 2.666666666666667e-06,
"loss": -0.0497,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01839319057762623,
"mask/share_reasoning": 0.8394039869308472,
"mask/share_step_conf": 0.12267155945301056,
"num_tokens": 32932980.0,
"reward": 0.7579681873321533,
"reward_std": 0.2207770049571991,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.5539538860321045,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.6604199409484863,
"step": 104
},
{
"adv/mean_abs_final_conf": 0.7216190695762634,
"adv/mean_abs_reasoning": 0.5883845090866089,
"adv/mean_abs_step_conf": 0.7586150169372559,
"adv/ratio_final_to_reasoning": 1.226441312495606,
"adv/ratio_step_to_reasoning": 1.2893184732461904,
"adv/std_final_conf": 0.8924412131309509,
"adv/std_reasoning": 0.8267587423324585,
"adv/std_step_conf": 0.9350519180297852,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.582585658777817,
"calib/avg_num_step_conf": 9.62890625,
"calib/ece": 0.34719262295081976,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9098360655737705,
"calib/gap": 0.036468032497350866,
"calib/mean_conf": 0.9578483606557378,
"calib/mu_c": 0.9720469798657717,
"calib/mu_w": 0.9355789473684208,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34719262295081976,
"calib/std_conf": 0.10714046604586741,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8859402654867257,
"calib/step_q_c_n": 1356.0,
"calib/step_q_gap": 0.04099977856156789,
"calib/step_q_w": 0.8449404869251578,
"calib/step_q_w_n": 1109.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 3010.0,
"completions/max_terminated_length": 3010.0,
"completions/mean_length": 839.49609375,
"completions/mean_terminated_length": 866.5765991210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 450.0,
"epoch": 0.112,
"grad_norm": 0.02181658335030079,
"kl": 0.07389068603515625,
"learning_rate": 2.6388888888888893e-06,
"loss": -0.1146,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01731002703309059,
"mask/share_reasoning": 0.838695764541626,
"mask/share_step_conf": 0.1127442717552185,
"num_tokens": 33253651.0,
"reward": 0.8030605316162109,
"reward_std": 0.31802046298980713,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.617209255695343,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6818804740905762,
"step": 105
},
{
"adv/mean_abs_final_conf": 0.6939954161643982,
"adv/mean_abs_reasoning": 0.4305272102355957,
"adv/mean_abs_step_conf": 0.7503472566604614,
"adv/ratio_final_to_reasoning": 1.6119664440828858,
"adv/ratio_step_to_reasoning": 1.7428567552091583,
"adv/std_final_conf": 0.8775339722633362,
"adv/std_reasoning": 0.7013723254203796,
"adv/std_step_conf": 0.9299072623252869,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5432639271924986,
"calib/avg_num_step_conf": 8.765625,
"calib/ece": 0.3660121951219513,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.8861788617886179,
"calib/gap": 0.013204771097628298,
"calib/mean_conf": 0.9676382113821139,
"calib/mu_c": 0.9728986486486487,
"calib/mu_w": 0.9596938775510204,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3660121951219513,
"calib/std_conf": 0.04002889204595677,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.889991889699919,
"calib/step_q_c_n": 1233.0,
"calib/step_q_gap": 0.020328190392302825,
"calib/step_q_w": 0.8696636993076162,
"calib/step_q_w_n": 1011.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2921.0,
"completions/max_terminated_length": 2921.0,
"completions/mean_length": 760.08203125,
"completions/mean_terminated_length": 784.6007690429688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 410.0,
"epoch": 0.11306666666666666,
"grad_norm": 0.02546885423362255,
"kl": 0.0869293212890625,
"learning_rate": 2.6111111111111113e-06,
"loss": -0.0712,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.018718719482421875,
"mask/share_reasoning": 0.835868239402771,
"mask/share_step_conf": 0.11416309326887131,
"num_tokens": 33552816.0,
"reward": 0.7969148755073547,
"reward_std": 0.19696387648582458,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.6064382791519165,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.679578959941864,
"step": 106
},
{
"adv/mean_abs_final_conf": 0.6982734799385071,
"adv/mean_abs_reasoning": 0.5172214508056641,
"adv/mean_abs_step_conf": 0.7512728571891785,
"adv/ratio_final_to_reasoning": 1.3500474097716217,
"adv/ratio_step_to_reasoning": 1.4525168204430383,
"adv/std_final_conf": 0.8845649361610413,
"adv/std_reasoning": 0.7755434513092041,
"adv/std_step_conf": 0.9340380430221558,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5586052322163434,
"calib/avg_num_step_conf": 9.26953125,
"calib/ece": 0.3012449799196789,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9437751004016064,
"calib/gap": 0.013053350970017474,
"calib/mean_conf": 0.974979919678715,
"calib/mu_c": 0.9792261904761904,
"calib/mu_w": 0.9661728395061729,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.30076305220883554,
"calib/std_conf": 0.04145084181948238,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8911875843454791,
"calib/step_q_c_n": 1482.0,
"calib/step_q_gap": 0.0021303452882401874,
"calib/step_q_w": 0.8890572390572389,
"calib/step_q_w_n": 891.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2458.0,
"completions/max_terminated_length": 2458.0,
"completions/mean_length": 757.328125,
"completions/mean_terminated_length": 775.5040283203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 440.0,
"epoch": 0.11413333333333334,
"grad_norm": 0.018556181341409683,
"kl": 0.08465576171875,
"learning_rate": 2.5833333333333337e-06,
"loss": -0.1297,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01875002309679985,
"mask/share_reasoning": 0.8351476192474365,
"mask/share_step_conf": 0.12266480922698975,
"num_tokens": 33851308.0,
"reward": 0.8634672164916992,
"reward_std": 0.2831929624080658,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.6712472438812256,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7322496175765991,
"step": 107
},
{
"adv/mean_abs_final_conf": 0.6516192555427551,
"adv/mean_abs_reasoning": 0.37682825326919556,
"adv/mean_abs_step_conf": 0.7557332515716553,
"adv/ratio_final_to_reasoning": 1.7292208052066005,
"adv/ratio_step_to_reasoning": 2.0055111181692116,
"adv/std_final_conf": 0.879697859287262,
"adv/std_reasoning": 0.6613324284553528,
"adv/std_step_conf": 0.9278945922851562,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6354972375690608,
"calib/avg_num_step_conf": 9.08984375,
"calib/ece": 0.21908298755186717,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.8672199170124482,
"calib/gap": 0.024716390423572765,
"calib/mean_conf": 0.9638962655601662,
"calib/mu_c": 0.9700497237569061,
"calib/mu_w": 0.9453333333333334,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.21597095435684643,
"calib/std_conf": 0.0629978069380198,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8831625766871165,
"calib/step_q_c_n": 1630.0,
"calib/step_q_gap": 0.014439477691420621,
"calib/step_q_w": 0.8687230989956959,
"calib/step_q_w_n": 697.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2792.0,
"completions/max_terminated_length": 2792.0,
"completions/mean_length": 815.3203125,
"completions/mean_terminated_length": 841.6209716796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 405.0,
"epoch": 0.1152,
"grad_norm": 0.03747807815670967,
"kl": 0.081787109375,
"learning_rate": 2.5555555555555557e-06,
"loss": -0.1129,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01746443659067154,
"mask/share_reasoning": 0.8368449211120605,
"mask/share_step_conf": 0.11444061994552612,
"num_tokens": 34163262.0,
"reward": 0.9091541767120361,
"reward_std": 0.18528440594673157,
"rewards/accuracy_reward_step": 0.7109375,
"rewards/final_brier_reward_step": 0.7276929616928101,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.7601466774940491,
"step": 108
},
{
"adv/mean_abs_final_conf": 0.6833958625793457,
"adv/mean_abs_reasoning": 0.43323150277137756,
"adv/mean_abs_step_conf": 0.7291561365127563,
"adv/ratio_final_to_reasoning": 1.5774380630394356,
"adv/ratio_step_to_reasoning": 1.6830635164994971,
"adv/std_final_conf": 0.879310667514801,
"adv/std_reasoning": 0.7205653786659241,
"adv/std_step_conf": 0.9332891702651978,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6254714439655172,
"calib/avg_num_step_conf": 9.68359375,
"calib/ece": 0.43293852459016396,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.8442622950819673,
"calib/gap": 0.032313038793103654,
"calib/mean_conf": 0.9575286885245904,
"calib/mu_c": 0.9728906250000002,
"calib/mu_w": 0.9405775862068966,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43293852459016396,
"calib/std_conf": 0.05920507562909826,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8847472150814053,
"calib/step_q_c_n": 1167.0,
"calib/step_q_gap": 0.0233790748375029,
"calib/step_q_w": 0.8613681402439024,
"calib/step_q_w_n": 1312.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2984.0,
"completions/max_terminated_length": 2984.0,
"completions/mean_length": 840.765625,
"completions/mean_terminated_length": 860.9440307617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 483.0,
"epoch": 0.11626666666666667,
"grad_norm": 0.03594356030225754,
"kl": 0.0813751220703125,
"learning_rate": 2.5277777777777778e-06,
"loss": -0.0475,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017205409705638885,
"mask/share_reasoning": 0.8475127220153809,
"mask/share_step_conf": 0.11184436082839966,
"num_tokens": 34483098.0,
"reward": 0.7405705451965332,
"reward_std": 0.23001110553741455,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.5487913489341736,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6409435272216797,
"step": 109
},
{
"adv/mean_abs_final_conf": 0.7198961973190308,
"adv/mean_abs_reasoning": 0.47418078780174255,
"adv/mean_abs_step_conf": 0.7653753757476807,
"adv/ratio_final_to_reasoning": 1.518189297918209,
"adv/ratio_step_to_reasoning": 1.614100350408309,
"adv/std_final_conf": 0.8994455337524414,
"adv/std_reasoning": 0.7393791675567627,
"adv/std_step_conf": 0.9331598281860352,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5265456989247311,
"calib/avg_num_step_conf": 9.43359375,
"calib/ece": 0.35470119521912363,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.8884462151394422,
"calib/gap": 0.004077284946236603,
"calib/mean_conf": 0.9682470119521913,
"calib/mu_c": 0.9698064516129034,
"calib/mu_w": 0.9657291666666667,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3527091633466137,
"calib/std_conf": 0.046478744028319606,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8881674842326559,
"calib/step_q_c_n": 1427.0,
"calib/step_q_gap": 0.010019710953303673,
"calib/step_q_w": 0.8781477732793522,
"calib/step_q_w_n": 988.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2747.0,
"completions/max_terminated_length": 2747.0,
"completions/mean_length": 755.0390625,
"completions/mean_terminated_length": 763.9921264648438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 306.0,
"epoch": 0.11733333333333333,
"grad_norm": 0.020071037113666534,
"kl": 0.09243011474609375,
"learning_rate": 2.5e-06,
"loss": -0.0492,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.019367707893252373,
"mask/share_reasoning": 0.8397272825241089,
"mask/share_step_conf": 0.1291862428188324,
"num_tokens": 34781308.0,
"reward": 0.8190223574638367,
"reward_std": 0.2533964216709137,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.6280652284622192,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.6927918791770935,
"step": 110
},
{
"adv/mean_abs_final_conf": 0.712738037109375,
"adv/mean_abs_reasoning": 0.506923496723175,
"adv/mean_abs_step_conf": 0.7473180294036865,
"adv/ratio_final_to_reasoning": 1.4060071030769221,
"adv/ratio_step_to_reasoning": 1.4742225093815056,
"adv/std_final_conf": 0.9178887605667114,
"adv/std_reasoning": 0.7756215333938599,
"adv/std_step_conf": 0.9329038262367249,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.614922206506365,
"calib/avg_num_step_conf": 9.5625,
"calib/ece": 0.37982987551867237,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.8008298755186722,
"calib/gap": 0.021407355021216268,
"calib/mean_conf": 0.9607427385892118,
"calib/mu_c": 0.9697142857142856,
"calib/mu_w": 0.9483069306930694,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.37982987551867237,
"calib/std_conf": 0.05120085133630464,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8879902755267425,
"calib/step_q_c_n": 1234.0,
"calib/step_q_gap": 0.039388957569576055,
"calib/step_q_w": 0.8486013179571664,
"calib/step_q_w_n": 1214.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2456.0,
"completions/max_terminated_length": 2456.0,
"completions/mean_length": 791.375,
"completions/mean_terminated_length": 816.9031982421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 360.0,
"epoch": 0.1184,
"grad_norm": 0.031903259456157684,
"kl": 0.0840606689453125,
"learning_rate": 2.4722222222222226e-06,
"loss": -0.1046,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.018131183460354805,
"mask/share_reasoning": 0.8329363465309143,
"mask/share_step_conf": 0.11768247932195663,
"num_tokens": 35091308.0,
"reward": 0.7689356803894043,
"reward_std": 0.2724992632865906,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.5790972709655762,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.6626803278923035,
"step": 111
},
{
"adv/mean_abs_final_conf": 0.7234814167022705,
"adv/mean_abs_reasoning": 0.458984911441803,
"adv/mean_abs_step_conf": 0.7522188425064087,
"adv/ratio_final_to_reasoning": 1.5762640528413197,
"adv/ratio_step_to_reasoning": 1.6388748818419194,
"adv/std_final_conf": 0.8879250884056091,
"adv/std_reasoning": 0.7393825650215149,
"adv/std_step_conf": 0.9295467138290405,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.6773631479599246,
"calib/avg_num_step_conf": 9.70703125,
"calib/ece": 0.359121338912134,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.7364016736401674,
"calib/gap": 0.03483229272542465,
"calib/mean_conf": 0.9438912133891215,
"calib/mu_c": 0.9580281690140846,
"calib/mu_w": 0.92319587628866,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35443514644351476,
"calib/std_conf": 0.08277309741101296,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8867267032106498,
"calib/step_q_c_n": 1277.0,
"calib/step_q_gap": 0.02705782903846443,
"calib/step_q_w": 0.8596688741721854,
"calib/step_q_w_n": 1208.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3048.0,
"completions/max_terminated_length": 3048.0,
"completions/mean_length": 826.0078125,
"completions/mean_terminated_length": 856.1052856445312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 489.0,
"epoch": 0.11946666666666667,
"grad_norm": 0.040970560163259506,
"kl": 0.0791778564453125,
"learning_rate": 2.4444444444444447e-06,
"loss": -0.1348,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.017311958596110344,
"mask/share_reasoning": 0.8367998600006104,
"mask/share_step_conf": 0.11073192209005356,
"num_tokens": 35410686.0,
"reward": 0.7834147214889526,
"reward_std": 0.21543002128601074,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.6035550832748413,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.6656181216239929,
"step": 112
},
{
"adv/mean_abs_final_conf": 0.7123876810073853,
"adv/mean_abs_reasoning": 0.5197149515151978,
"adv/mean_abs_step_conf": 0.7662708163261414,
"adv/ratio_final_to_reasoning": 1.3707277016573445,
"adv/ratio_step_to_reasoning": 1.4744059490536587,
"adv/std_final_conf": 0.8939816355705261,
"adv/std_reasoning": 0.7754083871841431,
"adv/std_step_conf": 0.9343150854110718,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5899899564780717,
"calib/avg_num_step_conf": 10.49609375,
"calib/ece": 0.3649193548387099,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.7983870967741935,
"calib/gap": 0.037502510880482065,
"calib/mean_conf": 0.9495967741935485,
"calib/mu_c": 0.9651724137931036,
"calib/mu_w": 0.9276699029126215,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3649193548387099,
"calib/std_conf": 0.09619284388872504,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8905063291139242,
"calib/step_q_c_n": 1343.0,
"calib/step_q_gap": 0.02546168625678147,
"calib/step_q_w": 0.8650446428571428,
"calib/step_q_w_n": 1344.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2983.0,
"completions/max_terminated_length": 2983.0,
"completions/mean_length": 777.46875,
"completions/mean_terminated_length": 792.9561767578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 417.0,
"epoch": 0.12053333333333334,
"grad_norm": 0.034543026238679886,
"kl": 0.09393310546875,
"learning_rate": 2.4166666666666667e-06,
"loss": -0.0128,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.01877540722489357,
"mask/share_reasoning": 0.833499550819397,
"mask/share_step_conf": 0.12819385528564453,
"num_tokens": 35714918.0,
"reward": 0.8004428148269653,
"reward_std": 0.25604891777038574,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.6131843328475952,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.680669903755188,
"step": 113
},
{
"adv/mean_abs_final_conf": 0.7102067470550537,
"adv/mean_abs_reasoning": 0.4649397134780884,
"adv/mean_abs_step_conf": 0.7417643666267395,
"adv/ratio_final_to_reasoning": 1.5275243788967583,
"adv/ratio_step_to_reasoning": 1.5953990272799041,
"adv/std_final_conf": 0.9010103344917297,
"adv/std_reasoning": 0.7395108342170715,
"adv/std_step_conf": 0.9281858205795288,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.6205092287409362,
"calib/avg_num_step_conf": 10.296875,
"calib/ece": 0.26420168067226923,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.7857142857142857,
"calib/gap": 0.04142880685563621,
"calib/mean_conf": 0.9532773109243698,
"calib/mu_c": 0.966158536585366,
"calib/mu_w": 0.9247297297297298,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.26420168067226923,
"calib/std_conf": 0.0815769909447045,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8923828382838284,
"calib/step_q_c_n": 1515.0,
"calib/step_q_gap": 0.03360183917588899,
"calib/step_q_w": 0.8587809991079394,
"calib/step_q_w_n": 1121.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2915.0,
"completions/max_terminated_length": 2915.0,
"completions/mean_length": 782.15625,
"completions/mean_terminated_length": 820.6229248046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 388.0,
"epoch": 0.1216,
"grad_norm": 0.05427626147866249,
"kl": 0.07926177978515625,
"learning_rate": 2.388888888888889e-06,
"loss": -0.1523,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.018315622583031654,
"mask/share_reasoning": 0.8093851804733276,
"mask/share_step_conf": 0.12542423605918884,
"num_tokens": 36020174.0,
"reward": 0.8564561009407043,
"reward_std": 0.24829316139221191,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6759241819381714,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.7229254245758057,
"step": 114
},
{
"adv/mean_abs_final_conf": 0.7221448421478271,
"adv/mean_abs_reasoning": 0.5368207693099976,
"adv/mean_abs_step_conf": 0.7720561027526855,
"adv/ratio_final_to_reasoning": 1.3452252286662378,
"adv/ratio_step_to_reasoning": 1.4382008798673116,
"adv/std_final_conf": 0.9052088856697083,
"adv/std_reasoning": 0.792934775352478,
"adv/std_step_conf": 0.9331409335136414,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5650809716599191,
"calib/avg_num_step_conf": 9.82421875,
"calib/ece": 0.4268852459016395,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.8278688524590164,
"calib/gap": 0.0170013495276653,
"calib/mean_conf": 0.9596721311475411,
"calib/mu_c": 0.9676153846153848,
"calib/mu_w": 0.9506140350877195,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4268852459016395,
"calib/std_conf": 0.05791365005481284,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.889716494845361,
"calib/step_q_c_n": 1164.0,
"calib/step_q_gap": 0.019390810167344674,
"calib/step_q_w": 0.8703256846780163,
"calib/step_q_w_n": 1351.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2484.0,
"completions/max_terminated_length": 2484.0,
"completions/mean_length": 783.9375,
"completions/mean_terminated_length": 815.8048706054688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 420.0,
"epoch": 0.12266666666666666,
"grad_norm": 0.03652571141719818,
"kl": 0.07973480224609375,
"learning_rate": 2.361111111111111e-06,
"loss": -0.1839,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017669178545475006,
"mask/share_reasoning": 0.8261470794677734,
"mask/share_step_conf": 0.11712120473384857,
"num_tokens": 36326126.0,
"reward": 0.7415735721588135,
"reward_std": 0.26013612747192383,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.5470499992370605,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6439095735549927,
"step": 115
},
{
"adv/mean_abs_final_conf": 0.7074733972549438,
"adv/mean_abs_reasoning": 0.4718044102191925,
"adv/mean_abs_step_conf": 0.7679315805435181,
"adv/ratio_final_to_reasoning": 1.4995056890762497,
"adv/ratio_step_to_reasoning": 1.627648160784995,
"adv/std_final_conf": 0.8698757290840149,
"adv/std_reasoning": 0.7208515405654907,
"adv/std_step_conf": 0.9321540594100952,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.49127605068199126,
"calib/avg_num_step_conf": 9.60546875,
"calib/ece": 0.3838114754098362,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.8934426229508197,
"calib/gap": 0.00012808973205025165,
"calib/mean_conf": 0.969877049180328,
"calib/mu_c": 0.96993006993007,
"calib/mu_w": 0.9698019801980198,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3838114754098362,
"calib/std_conf": 0.0437343091991714,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8948684210526318,
"calib/step_q_c_n": 1368.0,
"calib/step_q_gap": 0.028828091080129514,
"calib/step_q_w": 0.8660403299725022,
"calib/step_q_w_n": 1091.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2561.0,
"completions/max_terminated_length": 2561.0,
"completions/mean_length": 815.3984375,
"completions/mean_terminated_length": 841.7015991210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 391.0,
"epoch": 0.12373333333333333,
"grad_norm": 0.03004090115427971,
"kl": 0.07421112060546875,
"learning_rate": 2.3333333333333336e-06,
"loss": -0.1704,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01729145646095276,
"mask/share_reasoning": 0.8337869644165039,
"mask/share_step_conf": 0.11767159402370453,
"num_tokens": 36639388.0,
"reward": 0.7805377244949341,
"reward_std": 0.25898727774620056,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.5796562433242798,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6798566579818726,
"step": 116
},
{
"adv/mean_abs_final_conf": 0.6913862228393555,
"adv/mean_abs_reasoning": 0.5055603384971619,
"adv/mean_abs_step_conf": 0.780232310295105,
"adv/ratio_final_to_reasoning": 1.3675642058761632,
"adv/ratio_step_to_reasoning": 1.5433020569106315,
"adv/std_final_conf": 0.865074634552002,
"adv/std_reasoning": 0.7577515840530396,
"adv/std_step_conf": 0.934880256652832,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6497433535140829,
"calib/avg_num_step_conf": 9.4296875,
"calib/ece": 0.4290283400809717,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.8663967611336032,
"calib/gap": 0.03380165833113968,
"calib/mean_conf": 0.9593927125506073,
"calib/mu_c": 0.9752671755725191,
"calib/mu_w": 0.9414655172413794,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4290283400809717,
"calib/std_conf": 0.08955601026861996,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8945716709075489,
"calib/step_q_c_n": 1179.0,
"calib/step_q_gap": 0.018442116251678242,
"calib/step_q_w": 0.8761295546558706,
"calib/step_q_w_n": 1235.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2718.0,
"completions/max_terminated_length": 2718.0,
"completions/mean_length": 789.875,
"completions/mean_terminated_length": 808.83203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 416.0,
"epoch": 0.1248,
"grad_norm": 0.03556828573346138,
"kl": 0.07696533203125,
"learning_rate": 2.305555555555556e-06,
"loss": -0.0637,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.01829782873392105,
"mask/share_reasoning": 0.8373548984527588,
"mask/share_step_conf": 0.12090975791215897,
"num_tokens": 36948196.0,
"reward": 0.7476822137832642,
"reward_std": 0.26584988832473755,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.5554363131523132,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6446155905723572,
"step": 117
},
{
"adv/mean_abs_final_conf": 0.625870943069458,
"adv/mean_abs_reasoning": 0.3719392418861389,
"adv/mean_abs_step_conf": 0.7527309656143188,
"adv/ratio_final_to_reasoning": 1.6827236080162113,
"adv/ratio_step_to_reasoning": 2.023800881555679,
"adv/std_final_conf": 0.846591055393219,
"adv/std_reasoning": 0.6613606214523315,
"adv/std_step_conf": 0.9336501955986023,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6724489795918368,
"calib/avg_num_step_conf": 10.08203125,
"calib/ece": 0.35510080645161307,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.8266129032258065,
"calib/gap": 0.040221088435373975,
"calib/mean_conf": 0.9599395161290325,
"calib/mu_c": 0.9758333333333333,
"calib/mu_w": 0.9356122448979594,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.35510080645161307,
"calib/std_conf": 0.08253845860053945,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8710762483130904,
"calib/step_q_c_n": 1482.0,
"calib/step_q_gap": 0.03379690345412767,
"calib/step_q_w": 0.8372793448589627,
"calib/step_q_w_n": 1099.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2620.0,
"completions/max_terminated_length": 2620.0,
"completions/mean_length": 815.83984375,
"completions/mean_terminated_length": 832.0916748046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.12586666666666665,
"grad_norm": 0.029230020940303802,
"kl": 0.07103729248046875,
"learning_rate": 2.277777777777778e-06,
"loss": -0.1422,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018248315900564194,
"mask/share_reasoning": 0.8373794555664062,
"mask/share_step_conf": 0.12484100461006165,
"num_tokens": 37261059.0,
"reward": 0.8076679110527039,
"reward_std": 0.19901418685913086,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.6259959936141968,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6791835427284241,
"step": 118
},
{
"adv/mean_abs_final_conf": 0.6673562526702881,
"adv/mean_abs_reasoning": 0.5177932977676392,
"adv/mean_abs_step_conf": 0.7421085834503174,
"adv/ratio_final_to_reasoning": 1.2888468343399948,
"adv/ratio_step_to_reasoning": 1.433213961342814,
"adv/std_final_conf": 0.8697852492332458,
"adv/std_reasoning": 0.792900025844574,
"adv/std_step_conf": 0.9320849776268005,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.6407605552342395,
"calib/avg_num_step_conf": 8.83984375,
"calib/ece": 0.39105485232067516,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.810126582278481,
"calib/gap": 0.03139965297860037,
"calib/mean_conf": 0.9522362869198314,
"calib/mu_c": 0.9660150375939851,
"calib/mu_w": 0.9346153846153847,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.39105485232067516,
"calib/std_conf": 0.08091029485832527,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.885243578387954,
"calib/step_q_c_n": 1129.0,
"calib/step_q_gap": 0.07308308456079349,
"calib/step_q_w": 0.8121604938271605,
"calib/step_q_w_n": 1134.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3053.0,
"completions/max_terminated_length": 3053.0,
"completions/mean_length": 853.38671875,
"completions/mean_terminated_length": 877.3775024414062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.12693333333333334,
"grad_norm": 0.03167301416397095,
"kl": 0.070526123046875,
"learning_rate": 2.25e-06,
"loss": -0.0566,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.017237205058336258,
"mask/share_reasoning": 0.846515417098999,
"mask/share_step_conf": 0.10890362411737442,
"num_tokens": 37584590.0,
"reward": 0.7539892792701721,
"reward_std": 0.266414999961853,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5644835829734802,
"rewards/format_reward_step": 0.92578125,
"rewards/step_l2_reward": 0.6544324159622192,
"step": 119
},
{
"adv/mean_abs_final_conf": 0.6957762241363525,
"adv/mean_abs_reasoning": 0.5754411220550537,
"adv/mean_abs_step_conf": 0.7391219139099121,
"adv/ratio_final_to_reasoning": 1.209118009591581,
"adv/ratio_step_to_reasoning": 1.2844440301212932,
"adv/std_final_conf": 0.8899182081222534,
"adv/std_reasoning": 0.8429538011550903,
"adv/std_step_conf": 0.9355378746986389,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6211656441717792,
"calib/avg_num_step_conf": 9.72265625,
"calib/ece": 0.27576763485477174,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.8008298755186722,
"calib/gap": 0.05051753971999384,
"calib/mean_conf": 0.9521161825726142,
"calib/mu_c": 0.9684662576687117,
"calib/mu_w": 0.9179487179487179,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.27576763485477174,
"calib/std_conf": 0.07791954428304795,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8885578231292517,
"calib/step_q_c_n": 1470.0,
"calib/step_q_gap": 0.04782180742758335,
"calib/step_q_w": 0.8407360157016683,
"calib/step_q_w_n": 1019.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2620.0,
"completions/max_terminated_length": 2620.0,
"completions/mean_length": 816.375,
"completions/mean_terminated_length": 846.1214599609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 472.0,
"epoch": 0.128,
"grad_norm": 0.034337036311626434,
"kl": 0.07117462158203125,
"learning_rate": 2.222222222222222e-06,
"loss": -0.1318,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01761537417769432,
"mask/share_reasoning": 0.832998514175415,
"mask/share_step_conf": 0.11422987282276154,
"num_tokens": 37900270.0,
"reward": 0.8556807637214661,
"reward_std": 0.29194173216819763,
"rewards/accuracy_reward_step": 0.63671875,
"rewards/final_brier_reward_step": 0.6786898374557495,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.7178279161453247,
"step": 120
},
{
"adv/mean_abs_final_conf": 0.7192496061325073,
"adv/mean_abs_reasoning": 0.5807757377624512,
"adv/mean_abs_step_conf": 0.7497106790542603,
"adv/ratio_final_to_reasoning": 1.2384291549498143,
"adv/ratio_step_to_reasoning": 1.2908780968410682,
"adv/std_final_conf": 0.9046310782432556,
"adv/std_reasoning": 0.8265811800956726,
"adv/std_step_conf": 0.9339022636413574,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5918091809180919,
"calib/avg_num_step_conf": 9.27734375,
"calib/ece": 0.36618852459016404,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.7704918032786885,
"calib/gap": 0.024492833898774657,
"calib/mean_conf": 0.9440573770491805,
"calib/mu_c": 0.9541958041958043,
"calib/mu_w": 0.9297029702970296,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3620901639344263,
"calib/std_conf": 0.10597871725564956,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8727382875605815,
"calib/step_q_c_n": 1238.0,
"calib/step_q_gap": 0.0370126938930353,
"calib/step_q_w": 0.8357255936675462,
"calib/step_q_w_n": 1137.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2797.0,
"completions/max_terminated_length": 2797.0,
"completions/mean_length": 835.12109375,
"completions/mean_terminated_length": 865.5506591796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.12906666666666666,
"grad_norm": 0.03736064210534096,
"kl": 0.069427490234375,
"learning_rate": 2.1944444444444445e-06,
"loss": -0.1096,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017177481204271317,
"mask/share_reasoning": 0.8382337093353271,
"mask/share_step_conf": 0.10943257063627243,
"num_tokens": 38219117.0,
"reward": 0.7833296060562134,
"reward_std": 0.27490270137786865,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.5992906093597412,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6658061742782593,
"step": 121
},
{
"adv/mean_abs_final_conf": 0.6296117901802063,
"adv/mean_abs_reasoning": 0.4547215402126312,
"adv/mean_abs_step_conf": 0.7533493041992188,
"adv/ratio_final_to_reasoning": 1.3846095566218286,
"adv/ratio_step_to_reasoning": 1.6567266724311036,
"adv/std_final_conf": 0.8445394039154053,
"adv/std_reasoning": 0.7206957936286926,
"adv/std_step_conf": 0.9326170682907104,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6663586635127143,
"calib/avg_num_step_conf": 9.625,
"calib/ece": 0.32323651452282165,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.8008298755186722,
"calib/gap": 0.06292652276759325,
"calib/mean_conf": 0.9456431535269711,
"calib/mu_c": 0.9688815789473685,
"calib/mu_w": 0.9059550561797752,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.31908713692946067,
"calib/std_conf": 0.11622208871556197,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.88203611898017,
"calib/step_q_c_n": 1412.0,
"calib/step_q_gap": 0.04434600491172891,
"calib/step_q_w": 0.8376901140684411,
"calib/step_q_w_n": 1052.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2919.0,
"completions/max_terminated_length": 2919.0,
"completions/mean_length": 809.8984375,
"completions/mean_terminated_length": 836.024169921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 465.0,
"epoch": 0.13013333333333332,
"grad_norm": 0.034300535917282104,
"kl": 0.07027435302734375,
"learning_rate": 2.166666666666667e-06,
"loss": -0.0697,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.017662089318037033,
"mask/share_reasoning": 0.8331197500228882,
"mask/share_step_conf": 0.11796815693378448,
"num_tokens": 38533795.0,
"reward": 0.8321857452392578,
"reward_std": 0.2278578281402588,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6436429619789124,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.7136971950531006,
"step": 122
},
{
"adv/mean_abs_final_conf": 0.6647226810455322,
"adv/mean_abs_reasoning": 0.47725650668144226,
"adv/mean_abs_step_conf": 0.748039960861206,
"adv/ratio_final_to_reasoning": 1.392799619784376,
"adv/ratio_step_to_reasoning": 1.5673750915679092,
"adv/std_final_conf": 0.8729745745658875,
"adv/std_reasoning": 0.7577446699142456,
"adv/std_step_conf": 0.9337034225463867,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5375591615956727,
"calib/avg_num_step_conf": 8.8984375,
"calib/ece": 0.3648987854251011,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.7611336032388664,
"calib/gap": 0.007183231913455268,
"calib/mean_conf": 0.944412955465587,
"calib/mu_c": 0.9473793103448276,
"calib/mu_w": 0.9401960784313723,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36113360323886634,
"calib/std_conf": 0.0868403971029498,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8690188383045526,
"calib/step_q_c_n": 1274.0,
"calib/step_q_gap": 0.010562663005747819,
"calib/step_q_w": 0.8584561752988048,
"calib/step_q_w_n": 1004.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2915.0,
"completions/max_terminated_length": 2915.0,
"completions/mean_length": 899.86328125,
"completions/mean_terminated_length": 899.86328125,
"completions/min_length": 458.0,
"completions/min_terminated_length": 458.0,
"epoch": 0.1312,
"grad_norm": 0.026797449216246605,
"kl": 0.0637359619140625,
"learning_rate": 2.138888888888889e-06,
"loss": 0.0474,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.017092389985919,
"mask/share_reasoning": 0.8718984723091125,
"mask/share_step_conf": 0.1110091581940651,
"num_tokens": 38869448.0,
"reward": 0.7914524078369141,
"reward_std": 0.2560619115829468,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.6038050651550293,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6728497743606567,
"step": 123
},
{
"adv/mean_abs_final_conf": 0.6712331771850586,
"adv/mean_abs_reasoning": 0.43861687183380127,
"adv/mean_abs_step_conf": 0.7770863771438599,
"adv/ratio_final_to_reasoning": 1.5303405324531139,
"adv/ratio_step_to_reasoning": 1.7716746140999107,
"adv/std_final_conf": 0.8607048392295837,
"adv/std_reasoning": 0.7207204699516296,
"adv/std_step_conf": 0.9323819875717163,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6848791951237425,
"calib/avg_num_step_conf": 10.140625,
"calib/ece": 0.30999999999999994,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.7107438016528925,
"calib/gap": 0.04948373356833369,
"calib/mean_conf": 0.9389256198347109,
"calib/mu_c": 0.9571241830065359,
"calib/mu_w": 0.9076404494382022,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3083471074380164,
"calib/std_conf": 0.08353058184853406,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8742908827785818,
"calib/step_q_c_n": 1382.0,
"calib/step_q_gap": 0.06927440831400189,
"calib/step_q_w": 0.8050164744645799,
"calib/step_q_w_n": 1214.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2449.0,
"completions/max_terminated_length": 2449.0,
"completions/mean_length": 820.33203125,
"completions/mean_terminated_length": 846.7943115234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 393.0,
"epoch": 0.13226666666666667,
"grad_norm": 0.04688045009970665,
"kl": 0.0717620849609375,
"learning_rate": 2.1111111111111114e-06,
"loss": -0.1779,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.01713072881102562,
"mask/share_reasoning": 0.8390701413154602,
"mask/share_step_conf": 0.11254914104938507,
"num_tokens": 39186269.0,
"reward": 0.8271832466125488,
"reward_std": 0.2311488538980484,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.6517531275749207,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.6940196752548218,
"step": 124
},
{
"adv/mean_abs_final_conf": 0.7343705296516418,
"adv/mean_abs_reasoning": 0.5540212392807007,
"adv/mean_abs_step_conf": 0.7735514044761658,
"adv/ratio_final_to_reasoning": 1.3255277552266644,
"adv/ratio_step_to_reasoning": 1.396248644691829,
"adv/std_final_conf": 0.9081211090087891,
"adv/std_reasoning": 0.7930463552474976,
"adv/std_step_conf": 0.9331655502319336,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6197039842873177,
"calib/avg_num_step_conf": 10.15234375,
"calib/ece": 0.3788333333333334,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.6791666666666667,
"calib/gap": 0.051077441077441166,
"calib/mean_conf": 0.9238333333333334,
"calib/mu_c": 0.9468181818181819,
"calib/mu_w": 0.8957407407407407,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3763333333333334,
"calib/std_conf": 0.12223054264608152,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8688908145580589,
"calib/step_q_c_n": 1154.0,
"calib/step_q_gap": 0.036157250544218034,
"calib/step_q_w": 0.8327335640138409,
"calib/step_q_w_n": 1445.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2610.0,
"completions/max_terminated_length": 2610.0,
"completions/mean_length": 844.04296875,
"completions/mean_terminated_length": 885.55322265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 431.0,
"epoch": 0.13333333333333333,
"grad_norm": 0.036926452070474625,
"kl": 0.064727783203125,
"learning_rate": 2.0833333333333334e-06,
"loss": -0.1825,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.016466014087200165,
"mask/share_reasoning": 0.8317209482192993,
"mask/share_step_conf": 0.10493804514408112,
"num_tokens": 39507152.0,
"reward": 0.7655426263809204,
"reward_std": 0.2716066837310791,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.5841484069824219,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.6563118696212769,
"step": 125
},
{
"adv/mean_abs_final_conf": 0.6951326131820679,
"adv/mean_abs_reasoning": 0.46520280838012695,
"adv/mean_abs_step_conf": 0.7669442296028137,
"adv/ratio_final_to_reasoning": 1.4942571297077392,
"adv/ratio_step_to_reasoning": 1.6486233870198985,
"adv/std_final_conf": 0.8844326138496399,
"adv/std_reasoning": 0.7206853628158569,
"adv/std_step_conf": 0.9299399852752686,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.6635110104959868,
"calib/avg_num_step_conf": 9.2421875,
"calib/ece": 0.36739669421487586,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.6198347107438017,
"calib/gap": 0.08087192152020317,
"calib/mean_conf": 0.9004545454545455,
"calib/mu_c": 0.9382170542635659,
"calib/mu_w": 0.8573451327433628,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36739669421487586,
"calib/std_conf": 0.14376935730249876,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8738846480067853,
"calib/step_q_c_n": 1179.0,
"calib/step_q_gap": 0.09074648456954859,
"calib/step_q_w": 0.7831381634372367,
"calib/step_q_w_n": 1187.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3049.0,
"completions/max_terminated_length": 3049.0,
"completions/mean_length": 849.1796875,
"completions/mean_terminated_length": 876.5725708007812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 435.0,
"epoch": 0.1344,
"grad_norm": 0.044366173446178436,
"kl": 0.06201934814453125,
"learning_rate": 2.0555555555555555e-06,
"loss": -0.1284,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.017499610781669617,
"mask/share_reasoning": 0.8362378478050232,
"mask/share_step_conf": 0.115012526512146,
"num_tokens": 39830006.0,
"reward": 0.7772961258888245,
"reward_std": 0.23498398065567017,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.6009371280670166,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.6638113260269165,
"step": 126
},
{
"adv/mean_abs_final_conf": 0.6320117712020874,
"adv/mean_abs_reasoning": 0.43845459818840027,
"adv/mean_abs_step_conf": 0.7619009613990784,
"adv/ratio_final_to_reasoning": 1.4414531716930865,
"adv/ratio_step_to_reasoning": 1.7376963647937291,
"adv/std_final_conf": 0.8390277028083801,
"adv/std_reasoning": 0.7206178903579712,
"adv/std_step_conf": 0.9327162504196167,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.72906454248366,
"calib/avg_num_step_conf": 9.70703125,
"calib/ece": 0.3370901639344262,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.6557377049180327,
"calib/gap": 0.1311111111111113,
"calib/mean_conf": 0.8944672131147541,
"calib/mu_c": 0.9525000000000001,
"calib/mu_w": 0.8213888888888888,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3370901639344262,
"calib/std_conf": 0.1769925294536388,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8744261006289308,
"calib/step_q_c_n": 1272.0,
"calib/step_q_gap": 0.07998257218705107,
"calib/step_q_w": 0.7944435284418797,
"calib/step_q_w_n": 1213.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2990.0,
"completions/max_terminated_length": 2990.0,
"completions/mean_length": 803.41015625,
"completions/mean_terminated_length": 832.6842041015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.13546666666666668,
"grad_norm": 0.039158936589956284,
"kl": 0.06972503662109375,
"learning_rate": 2.027777777777778e-06,
"loss": -0.1332,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017885904759168625,
"mask/share_reasoning": 0.828772783279419,
"mask/share_step_conf": 0.11818505823612213,
"num_tokens": 40139351.0,
"reward": 0.8012007474899292,
"reward_std": 0.21756742894649506,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.6414800882339478,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6640462875366211,
"step": 127
},
{
"adv/mean_abs_final_conf": 0.7376161813735962,
"adv/mean_abs_reasoning": 0.6009324789047241,
"adv/mean_abs_step_conf": 0.7559952139854431,
"adv/ratio_final_to_reasoning": 1.2274526794058387,
"adv/ratio_step_to_reasoning": 1.2580368685735552,
"adv/std_final_conf": 0.9184370040893555,
"adv/std_reasoning": 0.8267238140106201,
"adv/std_step_conf": 0.9349108934402466,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.6201575366988901,
"calib/avg_num_step_conf": 9.3828125,
"calib/ece": 0.3379411764705882,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.5882352941176471,
"calib/gap": 0.06202005012531342,
"calib/mean_conf": 0.8854201680672269,
"calib/mu_c": 0.9127819548872181,
"calib/mu_w": 0.8507619047619047,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33226890756302513,
"calib/std_conf": 0.1614054342799455,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8687878787878787,
"calib/step_q_c_n": 1188.0,
"calib/step_q_gap": 0.06706629723927893,
"calib/step_q_w": 0.8017215815485997,
"calib/step_q_w_n": 1214.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2986.0,
"completions/max_terminated_length": 2986.0,
"completions/mean_length": 856.234375,
"completions/mean_terminated_length": 891.0405883789062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 335.0,
"epoch": 0.13653333333333334,
"grad_norm": 0.04041381925344467,
"kl": 0.066192626953125,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.1361,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.017022691667079926,
"mask/share_reasoning": 0.8357027769088745,
"mask/share_step_conf": 0.10821200907230377,
"num_tokens": 40465211.0,
"reward": 0.7812422513961792,
"reward_std": 0.30478304624557495,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.6055277585983276,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.6671130657196045,
"step": 128
},
{
"adv/mean_abs_final_conf": 0.6258381605148315,
"adv/mean_abs_reasoning": 0.3574847877025604,
"adv/mean_abs_step_conf": 0.7670409679412842,
"adv/ratio_final_to_reasoning": 1.7506707475215708,
"adv/ratio_step_to_reasoning": 2.14566044298223,
"adv/std_final_conf": 0.8416181206703186,
"adv/std_reasoning": 0.6404501795768738,
"adv/std_step_conf": 0.9318187236785889,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.6007423117709437,
"calib/avg_num_step_conf": 9.50390625,
"calib/ece": 0.24427983539094644,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.6419753086419753,
"calib/gap": 0.06941221027117139,
"calib/mean_conf": 0.8911111111111112,
"calib/mu_c": 0.9145341614906833,
"calib/mu_w": 0.845121951219512,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23641975308641971,
"calib/std_conf": 0.1703059459512129,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.864235924932976,
"calib/step_q_c_n": 1492.0,
"calib/step_q_gap": 0.018688634816078964,
"calib/step_q_w": 0.845547290116897,
"calib/step_q_w_n": 941.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2888.0,
"completions/max_terminated_length": 2888.0,
"completions/mean_length": 815.046875,
"completions/mean_terminated_length": 831.2828979492188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 418.0,
"epoch": 0.1376,
"grad_norm": 0.053492747247219086,
"kl": 0.070281982421875,
"learning_rate": 1.9722222222222224e-06,
"loss": -0.066,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.018132753670215607,
"mask/share_reasoning": 0.8407728672027588,
"mask/share_step_conf": 0.1215631514787674,
"num_tokens": 40776247.0,
"reward": 0.8621791005134583,
"reward_std": 0.18524247407913208,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.6893390417098999,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.7193941473960876,
"step": 129
},
{
"adv/mean_abs_final_conf": 0.6381160020828247,
"adv/mean_abs_reasoning": 0.3434726595878601,
"adv/mean_abs_step_conf": 0.7246782779693604,
"adv/ratio_final_to_reasoning": 1.8578363787339383,
"adv/ratio_step_to_reasoning": 2.109857241152518,
"adv/std_final_conf": 0.8502756357192993,
"adv/std_reasoning": 0.6401984691619873,
"adv/std_step_conf": 0.9270868301391602,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6616675839295543,
"calib/avg_num_step_conf": 9.28125,
"calib/ece": 0.28260000000000013,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.716,
"calib/gap": 0.11069345074298276,
"calib/mean_conf": 0.8914800000000002,
"calib/mu_c": 0.9322151898734177,
"calib/mu_w": 0.8215217391304349,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.2710400000000001,
"calib/std_conf": 0.20137082608958032,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.8756263577118031,
"calib/step_q_c_n": 1381.0,
"calib/step_q_gap": 0.00785751349069752,
"calib/step_q_w": 0.8677688442211056,
"calib/step_q_w_n": 995.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 1719.0,
"completions/max_terminated_length": 1719.0,
"completions/mean_length": 740.9140625,
"completions/mean_terminated_length": 758.696044921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 428.0,
"epoch": 0.13866666666666666,
"grad_norm": 0.03951597586274147,
"kl": 0.06847381591796875,
"learning_rate": 1.944444444444445e-06,
"loss": -0.1244,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.019231311976909637,
"mask/share_reasoning": 0.8349332213401794,
"mask/share_step_conf": 0.12239795178174973,
"num_tokens": 41071209.0,
"reward": 0.8673378229141235,
"reward_std": 0.18943777680397034,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/final_brier_reward_step": 0.6826508045196533,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7356186509132385,
"step": 130
},
{
"adv/mean_abs_final_conf": 0.6092958450317383,
"adv/mean_abs_reasoning": 0.4230729341506958,
"adv/mean_abs_step_conf": 0.7680660486221313,
"adv/ratio_final_to_reasoning": 1.4401673939621746,
"adv/ratio_step_to_reasoning": 1.815445958895946,
"adv/std_final_conf": 0.8299340605735779,
"adv/std_reasoning": 0.7013399004936218,
"adv/std_step_conf": 0.9331469535827637,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6704498125780926,
"calib/avg_num_step_conf": 9.55859375,
"calib/ece": 0.4599999999999999,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.5795918367346938,
"calib/gap": 0.11802721088435364,
"calib/mean_conf": 0.86,
"calib/mu_c": 0.9308163265306122,
"calib/mu_w": 0.8127891156462586,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4599999999999999,
"calib/std_conf": 0.1908712569631527,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8648687350835322,
"calib/step_q_c_n": 838.0,
"calib/step_q_gap": 0.054533122902053144,
"calib/step_q_w": 0.8103356121814791,
"calib/step_q_w_n": 1609.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2588.0,
"completions/max_terminated_length": 2588.0,
"completions/mean_length": 762.38671875,
"completions/mean_terminated_length": 796.6162719726562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 467.0,
"epoch": 0.13973333333333332,
"grad_norm": 0.04783172532916069,
"kl": 0.0788421630859375,
"learning_rate": 1.916666666666667e-06,
"loss": -0.1903,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01806638203561306,
"mask/share_reasoning": 0.8219820261001587,
"mask/share_step_conf": 0.11698280274868011,
"num_tokens": 41372588.0,
"reward": 0.6918598413467407,
"reward_std": 0.20048105716705322,
"rewards/accuracy_reward_step": 0.3828125,
"rewards/final_brier_reward_step": 0.5441882610321045,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.5715625286102295,
"step": 131
},
{
"adv/mean_abs_final_conf": 0.635727047920227,
"adv/mean_abs_reasoning": 0.5477725267410278,
"adv/mean_abs_step_conf": 0.7592942714691162,
"adv/ratio_final_to_reasoning": 1.1605676022170819,
"adv/ratio_step_to_reasoning": 1.3861488745821862,
"adv/std_final_conf": 0.8477063775062561,
"adv/std_reasoning": 0.7928697466850281,
"adv/std_step_conf": 0.9318060874938965,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7335164835164835,
"calib/avg_num_step_conf": 9.31640625,
"calib/ece": 0.24838056680161935,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.6639676113360324,
"calib/gap": 0.1624450549450549,
"calib/mean_conf": 0.8799595141700406,
"calib/mu_c": 0.9398076923076922,
"calib/mu_w": 0.7773626373626373,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.24838056680161935,
"calib/std_conf": 0.18466842586293722,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8826722925457101,
"calib/step_q_c_n": 1422.0,
"calib/step_q_gap": 0.05299420324145254,
"calib/step_q_w": 0.8296780893042576,
"calib/step_q_w_n": 963.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2868.0,
"completions/max_terminated_length": 2868.0,
"completions/mean_length": 777.60546875,
"completions/mean_terminated_length": 799.4658203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 405.0,
"epoch": 0.1408,
"grad_norm": 0.04476737231016159,
"kl": 0.0642242431640625,
"learning_rate": 1.888888888888889e-06,
"loss": -0.1192,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.018550019711256027,
"mask/share_reasoning": 0.8334195613861084,
"mask/share_step_conf": 0.12068665027618408,
"num_tokens": 41677247.0,
"reward": 0.8809730410575867,
"reward_std": 0.24961604177951813,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.7208496332168579,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7262526750564575,
"step": 132
},
{
"adv/mean_abs_final_conf": 0.7479898929595947,
"adv/mean_abs_reasoning": 0.5849982500076294,
"adv/mean_abs_step_conf": 0.779026985168457,
"adv/ratio_final_to_reasoning": 1.2786190265523694,
"adv/ratio_step_to_reasoning": 1.331674043739955,
"adv/std_final_conf": 0.908273458480835,
"adv/std_reasoning": 0.8268160820007324,
"adv/std_step_conf": 0.9347463250160217,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.6537166900420758,
"calib/avg_num_step_conf": 9.4453125,
"calib/ece": 0.37397489539748946,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.5941422594142259,
"calib/gap": 0.11778892005610075,
"calib/mean_conf": 0.8417573221757323,
"calib/mu_c": 0.9028695652173911,
"calib/mu_w": 0.7850806451612904,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36728033472803334,
"calib/std_conf": 0.2256989323259645,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.867897435897436,
"calib/step_q_c_n": 975.0,
"calib/step_q_gap": 0.03466805266805284,
"calib/step_q_w": 0.8332293832293831,
"calib/step_q_w_n": 1443.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2521.0,
"completions/max_terminated_length": 2521.0,
"completions/mean_length": 816.62890625,
"completions/mean_terminated_length": 871.0708618164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 407.0,
"epoch": 0.14186666666666667,
"grad_norm": 0.045341163873672485,
"kl": 0.06060791015625,
"learning_rate": 1.8611111111111113e-06,
"loss": -0.2734,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.016313698142766953,
"mask/share_reasoning": 0.8198044300079346,
"mask/share_step_conf": 0.10138189792633057,
"num_tokens": 41992648.0,
"reward": 0.7376042604446411,
"reward_std": 0.3090021014213562,
"rewards/accuracy_reward_step": 0.44921875,
"rewards/final_brier_reward_step": 0.5864866971969604,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.612159252166748,
"step": 133
},
{
"adv/mean_abs_final_conf": 0.6931519508361816,
"adv/mean_abs_reasoning": 0.5527679920196533,
"adv/mean_abs_step_conf": 0.7484570741653442,
"adv/ratio_final_to_reasoning": 1.253965426441582,
"adv/ratio_step_to_reasoning": 1.3540166669757776,
"adv/std_final_conf": 0.8895355463027954,
"adv/std_reasoning": 0.8099076151847839,
"adv/std_step_conf": 0.9333269596099854,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6481481481481481,
"calib/avg_num_step_conf": 8.66796875,
"calib/ece": 0.2990688259109312,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.5910931174089069,
"calib/gap": 0.13690013227513242,
"calib/mean_conf": 0.8456275303643724,
"calib/mu_c": 0.9077037037037038,
"calib/mu_w": 0.7708035714285714,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2990688259109312,
"calib/std_conf": 0.21186350480245592,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.8718744313011829,
"calib/step_q_c_n": 1099.0,
"calib/step_q_gap": 0.039347645586897184,
"calib/step_q_w": 0.8325267857142857,
"calib/step_q_w_n": 1120.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2675.0,
"completions/max_terminated_length": 2675.0,
"completions/mean_length": 896.453125,
"completions/mean_terminated_length": 907.0830688476562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 427.0,
"epoch": 0.14293333333333333,
"grad_norm": 0.06819811463356018,
"kl": 0.0545196533203125,
"learning_rate": 1.8333333333333333e-06,
"loss": -0.0256,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.016849879175424576,
"mask/share_reasoning": 0.8710737228393555,
"mask/share_step_conf": 0.10035759210586548,
"num_tokens": 42331092.0,
"reward": 0.8029290437698364,
"reward_std": 0.26459646224975586,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.6504179239273071,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6601275205612183,
"step": 134
},
{
"adv/mean_abs_final_conf": 0.6313650012016296,
"adv/mean_abs_reasoning": 0.5012447237968445,
"adv/mean_abs_step_conf": 0.7458860278129578,
"adv/ratio_final_to_reasoning": 1.2595943083832304,
"adv/ratio_step_to_reasoning": 1.4880675893463706,
"adv/std_final_conf": 0.8750061988830566,
"adv/std_reasoning": 0.7929680943489075,
"adv/std_step_conf": 0.9351843595504761,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.5803836492349851,
"calib/avg_num_step_conf": 9.984375,
"calib/ece": 0.28176470588235286,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.6470588235294118,
"calib/gap": 0.0517896018877978,
"calib/mean_conf": 0.8650420168067227,
"calib/mu_c": 0.8839735099337745,
"calib/mu_w": 0.8321839080459768,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.25617647058823523,
"calib/std_conf": 0.20826362555757613,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8771782945736434,
"calib/step_q_c_n": 1290.0,
"calib/step_q_gap": 0.051862338807450725,
"calib/step_q_w": 0.8253159557661927,
"calib/step_q_w_n": 1266.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2998.0,
"completions/max_terminated_length": 2998.0,
"completions/mean_length": 795.2421875,
"completions/mean_terminated_length": 834.3524169921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 429.0,
"epoch": 0.144,
"grad_norm": 0.06078675389289856,
"kl": 0.0625457763671875,
"learning_rate": 1.8055555555555557e-06,
"loss": -0.2651,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.01735134981572628,
"mask/share_reasoning": 0.8256722688674927,
"mask/share_step_conf": 0.11010138690471649,
"num_tokens": 42640554.0,
"reward": 0.8158265352249146,
"reward_std": 0.2599346935749054,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.646649181842804,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.6810975074768066,
"step": 135
},
{
"adv/mean_abs_final_conf": 0.6751922965049744,
"adv/mean_abs_reasoning": 0.5883626341819763,
"adv/mean_abs_step_conf": 0.7381198406219482,
"adv/ratio_final_to_reasoning": 1.147578478439102,
"adv/ratio_step_to_reasoning": 1.2545321503092821,
"adv/std_final_conf": 0.8754206299781799,
"adv/std_reasoning": 0.8267235159873962,
"adv/std_step_conf": 0.9343062043190002,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6577624135763671,
"calib/avg_num_step_conf": 9.9765625,
"calib/ece": 0.31429166666666664,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.5708333333333333,
"calib/gap": 0.129585166561911,
"calib/mean_conf": 0.822625,
"calib/mu_c": 0.8825581395348838,
"calib/mu_w": 0.7529729729729728,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2997083333333333,
"calib/std_conf": 0.2318660375626409,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.870106907894737,
"calib/step_q_c_n": 1216.0,
"calib/step_q_gap": 0.11727432194555909,
"calib/step_q_w": 0.7528325859491779,
"calib/step_q_w_n": 1338.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2462.0,
"completions/max_terminated_length": 2462.0,
"completions/mean_length": 796.77734375,
"completions/mean_terminated_length": 835.9630737304688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 472.0,
"epoch": 0.14506666666666668,
"grad_norm": 0.03421531990170479,
"kl": 0.06191253662109375,
"learning_rate": 1.777777777777778e-06,
"loss": -0.2477,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.017344415187835693,
"mask/share_reasoning": 0.822014570236206,
"mask/share_step_conf": 0.11376601457595825,
"num_tokens": 42953017.0,
"reward": 0.7648307085037231,
"reward_std": 0.2806330621242523,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.6361578106880188,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.6067847609519958,
"step": 136
},
{
"adv/mean_abs_final_conf": 0.6866587400436401,
"adv/mean_abs_reasoning": 0.4152359366416931,
"adv/mean_abs_step_conf": 0.753272294998169,
"adv/ratio_final_to_reasoning": 1.6536592318987016,
"adv/ratio_step_to_reasoning": 1.8140826179217895,
"adv/std_final_conf": 0.905501663684845,
"adv/std_reasoning": 0.6816924810409546,
"adv/std_step_conf": 0.9320009350776672,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6449728827701294,
"calib/avg_num_step_conf": 9.71484375,
"calib/ece": 0.23781376518218636,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.5870445344129555,
"calib/gap": 0.12466624947851479,
"calib/mean_conf": 0.8264777327935222,
"calib/mu_c": 0.8739215686274511,
"calib/mu_w": 0.7492553191489363,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.22242914979757097,
"calib/std_conf": 0.23045137567110754,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.878910447761194,
"calib/step_q_c_n": 1407.0,
"calib/step_q_gap": 0.06743822553897172,
"calib/step_q_w": 0.8114722222222223,
"calib/step_q_w_n": 1080.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2302.0,
"completions/max_terminated_length": 2302.0,
"completions/mean_length": 790.46484375,
"completions/mean_terminated_length": 812.6867065429688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 489.0,
"epoch": 0.14613333333333334,
"grad_norm": 0.05176422744989395,
"kl": 0.05794525146484375,
"learning_rate": 1.75e-06,
"loss": -0.1748,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.017869368195533752,
"mask/share_reasoning": 0.8374629020690918,
"mask/share_step_conf": 0.11732391268014908,
"num_tokens": 43262360.0,
"reward": 0.8528109788894653,
"reward_std": 0.22672004997730255,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6977546811103821,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.6953673362731934,
"step": 137
},
{
"adv/mean_abs_final_conf": 0.6333698630332947,
"adv/mean_abs_reasoning": 0.5395751595497131,
"adv/mean_abs_step_conf": 0.7648397088050842,
"adv/ratio_final_to_reasoning": 1.173830655143308,
"adv/ratio_step_to_reasoning": 1.4174850255214846,
"adv/std_final_conf": 0.8593973517417908,
"adv/std_reasoning": 0.8100116848945618,
"adv/std_step_conf": 0.9284457564353943,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.6983270794246403,
"calib/avg_num_step_conf": 9.375,
"calib/ece": 0.17621848739495807,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.5714285714285714,
"calib/gap": 0.1657113821138213,
"calib/mean_conf": 0.8287394957983194,
"calib/mu_c": 0.8858333333333334,
"calib/mu_w": 0.7201219512195121,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.17474789915966393,
"calib/std_conf": 0.2115948031237759,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8783309352517986,
"calib/step_q_c_n": 1390.0,
"calib/step_q_gap": 0.051776479806253994,
"calib/step_q_w": 0.8265544554455446,
"calib/step_q_w_n": 1010.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2828.0,
"completions/max_terminated_length": 2828.0,
"completions/mean_length": 810.23828125,
"completions/mean_terminated_length": 850.0859985351562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 407.0,
"epoch": 0.1472,
"grad_norm": 0.05331015586853027,
"kl": 0.0602569580078125,
"learning_rate": 1.7222222222222224e-06,
"loss": -0.2486,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.017089052125811577,
"mask/share_reasoning": 0.8257807493209839,
"mask/share_step_conf": 0.11025522649288177,
"num_tokens": 43574117.0,
"reward": 0.8627781271934509,
"reward_std": 0.25776857137680054,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.7197796702384949,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.6979640126228333,
"step": 138
},
{
"adv/mean_abs_final_conf": 0.6163228750228882,
"adv/mean_abs_reasoning": 0.399943470954895,
"adv/mean_abs_step_conf": 0.7604862451553345,
"adv/ratio_final_to_reasoning": 1.5410249692322044,
"adv/ratio_step_to_reasoning": 1.9014843356227733,
"adv/std_final_conf": 0.8437338471412659,
"adv/std_reasoning": 0.7014421820640564,
"adv/std_step_conf": 0.9300811886787415,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7263708513708513,
"calib/avg_num_step_conf": 8.6640625,
"calib/ece": 0.19538152610441772,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.642570281124498,
"calib/gap": 0.1927705627705626,
"calib/mean_conf": 0.8467871485943775,
"calib/mu_c": 0.9118181818181817,
"calib/mu_w": 0.7190476190476192,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.18975903614457834,
"calib/std_conf": 0.2147047072627546,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8946872803935347,
"calib/step_q_c_n": 1423.0,
"calib/step_q_gap": 0.023190425047622698,
"calib/step_q_w": 0.871496855345912,
"calib/step_q_w_n": 795.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2639.0,
"completions/max_terminated_length": 2639.0,
"completions/mean_length": 779.59375,
"completions/mean_terminated_length": 791.9683227539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 406.0,
"epoch": 0.14826666666666666,
"grad_norm": 0.050895802676677704,
"kl": 0.05617523193359375,
"learning_rate": 1.6944444444444446e-06,
"loss": -0.0886,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01869884878396988,
"mask/share_reasoning": 0.8509533405303955,
"mask/share_step_conf": 0.11472281068563461,
"num_tokens": 43876789.0,
"reward": 0.9184540510177612,
"reward_std": 0.2043038308620453,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.761236310005188,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7522342801094055,
"step": 139
},
{
"adv/mean_abs_final_conf": 0.646989643573761,
"adv/mean_abs_reasoning": 0.4953835606575012,
"adv/mean_abs_step_conf": 0.736619234085083,
"adv/ratio_final_to_reasoning": 1.306037775486614,
"adv/ratio_step_to_reasoning": 1.4869674583213865,
"adv/std_final_conf": 0.8599854707717896,
"adv/std_reasoning": 0.7575750350952148,
"adv/std_step_conf": 0.9302241206169128,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6341016057585825,
"calib/avg_num_step_conf": 8.796875,
"calib/ece": 0.21019685039370073,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.610236220472441,
"calib/gap": 0.10283914728682175,
"calib/mean_conf": 0.8447637795275592,
"calib/mu_c": 0.8795833333333334,
"calib/mu_w": 0.7767441860465116,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.19677165354330706,
"calib/std_conf": 0.20869425161364966,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8876229508196721,
"calib/step_q_c_n": 1464.0,
"calib/step_q_gap": 0.06901888990596661,
"calib/step_q_w": 0.8186040609137055,
"calib/step_q_w_n": 788.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1631.0,
"completions/max_terminated_length": 1631.0,
"completions/mean_length": 771.68359375,
"completions/mean_terminated_length": 777.7598266601562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 398.0,
"epoch": 0.14933333333333335,
"grad_norm": 0.07742151618003845,
"kl": 0.06075286865234375,
"learning_rate": 1.6666666666666667e-06,
"loss": -0.0299,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.018998388200998306,
"mask/share_reasoning": 0.8548575639724731,
"mask/share_step_conf": 0.11833152920007706,
"num_tokens": 44179356.0,
"reward": 0.905888020992279,
"reward_std": 0.22323060035705566,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.7361972332000732,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7466724514961243,
"step": 140
},
{
"adv/mean_abs_final_conf": 0.5693894624710083,
"adv/mean_abs_reasoning": 0.4516020119190216,
"adv/mean_abs_step_conf": 0.7310876846313477,
"adv/ratio_final_to_reasoning": 1.2608213591685848,
"adv/ratio_step_to_reasoning": 1.6188760575372318,
"adv/std_final_conf": 0.8129839301109314,
"adv/std_reasoning": 0.739275336265564,
"adv/std_step_conf": 0.9309452176094055,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7155354449472096,
"calib/avg_num_step_conf": 8.515625,
"calib/ece": 0.16854838709677428,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.6411290322580645,
"calib/gap": 0.19733031674208135,
"calib/mean_conf": 0.8391129032258066,
"calib/mu_c": 0.9011764705882352,
"calib/mu_w": 0.7038461538461539,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.16108870967741942,
"calib/std_conf": 0.2282843165623637,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8867753366406802,
"calib/step_q_c_n": 1411.0,
"calib/step_q_gap": 0.05806272285654501,
"calib/step_q_w": 0.8287126137841352,
"calib/step_q_w_n": 769.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2123.0,
"completions/max_terminated_length": 2123.0,
"completions/mean_length": 781.87109375,
"completions/mean_terminated_length": 803.8513793945312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 429.0,
"epoch": 0.1504,
"grad_norm": 0.041482020169496536,
"kl": 0.054931640625,
"learning_rate": 1.638888888888889e-06,
"loss": -0.1472,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018007446080446243,
"mask/share_reasoning": 0.8456077575683594,
"mask/share_step_conf": 0.10904105007648468,
"num_tokens": 44486611.0,
"reward": 0.9167698621749878,
"reward_std": 0.2230062186717987,
"rewards/accuracy_reward_step": 0.6640625,
"rewards/final_brier_reward_step": 0.7688156366348267,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7389428615570068,
"step": 141
},
{
"adv/mean_abs_final_conf": 0.6314700841903687,
"adv/mean_abs_reasoning": 0.42496341466903687,
"adv/mean_abs_step_conf": 0.7561836242675781,
"adv/ratio_final_to_reasoning": 1.4859398771589785,
"adv/ratio_step_to_reasoning": 1.7794087635908535,
"adv/std_final_conf": 0.8592272400856018,
"adv/std_reasoning": 0.7013537883758545,
"adv/std_step_conf": 0.9269928932189941,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7295842732839187,
"calib/avg_num_step_conf": 8.57421875,
"calib/ece": 0.23301587301587293,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.5634920634920635,
"calib/gap": 0.21156815984531108,
"calib/mean_conf": 0.8028571428571429,
"calib/mu_c": 0.8926896551724138,
"calib/mu_w": 0.6811214953271028,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23023809523809516,
"calib/std_conf": 0.24811465969132396,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.892886345053148,
"calib/step_q_c_n": 1223.0,
"calib/step_q_gap": 0.04764971953874475,
"calib/step_q_w": 0.8452366255144033,
"calib/step_q_w_n": 972.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2785.0,
"completions/max_terminated_length": 2785.0,
"completions/mean_length": 858.42578125,
"completions/mean_terminated_length": 861.792236328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 305.0,
"epoch": 0.15146666666666667,
"grad_norm": 0.0425986647605896,
"kl": 0.0515899658203125,
"learning_rate": 1.6111111111111113e-06,
"loss": 0.0064,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.018087752163410187,
"mask/share_reasoning": 0.8661826848983765,
"mask/share_step_conf": 0.11182329803705215,
"num_tokens": 44811528.0,
"reward": 0.8665826320648193,
"reward_std": 0.1983642578125,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.7341116666793823,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.6888971924781799,
"step": 142
},
{
"adv/mean_abs_final_conf": 0.6143150925636292,
"adv/mean_abs_reasoning": 0.43332940340042114,
"adv/mean_abs_step_conf": 0.7397428750991821,
"adv/ratio_final_to_reasoning": 1.4176630705024347,
"adv/ratio_step_to_reasoning": 1.7071144244869472,
"adv/std_final_conf": 0.8298029899597168,
"adv/std_reasoning": 0.7206790447235107,
"adv/std_step_conf": 0.9270122647285461,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7693678870149457,
"calib/avg_num_step_conf": 8.921875,
"calib/ece": 0.26355102040816325,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6571428571428571,
"calib/gap": 0.24391128479363777,
"calib/mean_conf": 0.8357959183673469,
"calib/mu_c": 0.9373426573426571,
"calib/mu_w": 0.6934313725490193,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2578367346938775,
"calib/std_conf": 0.24057897677264514,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8957201646090537,
"calib/step_q_c_n": 1215.0,
"calib/step_q_gap": 0.08715140876246807,
"calib/step_q_w": 0.8085687558465856,
"calib/step_q_w_n": 1069.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 3060.0,
"completions/max_terminated_length": 3060.0,
"completions/mean_length": 793.13671875,
"completions/mean_terminated_length": 825.3779907226562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 403.0,
"epoch": 0.15253333333333333,
"grad_norm": 0.05409567803144455,
"kl": 0.052845001220703125,
"learning_rate": 1.5833333333333333e-06,
"loss": -0.1822,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.017870299518108368,
"mask/share_reasoning": 0.8328666687011719,
"mask/share_step_conf": 0.11020059883594513,
"num_tokens": 45121907.0,
"reward": 0.8470816016197205,
"reward_std": 0.2363058477640152,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.7216949462890625,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6693432331085205,
"step": 143
},
{
"adv/mean_abs_final_conf": 0.5318283438682556,
"adv/mean_abs_reasoning": 0.3669816851615906,
"adv/mean_abs_step_conf": 0.7277644872665405,
"adv/ratio_final_to_reasoning": 1.449195873723451,
"adv/ratio_step_to_reasoning": 1.983108467514091,
"adv/std_final_conf": 0.7739172577857971,
"adv/std_reasoning": 0.6612470746040344,
"adv/std_step_conf": 0.9174556732177734,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6250771604938272,
"calib/avg_num_step_conf": 8.6953125,
"calib/ece": 0.18960317460317455,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.6904761904761905,
"calib/gap": 0.11563888888888885,
"calib/mean_conf": 0.8591269841269841,
"calib/mu_c": 0.8921666666666666,
"calib/mu_w": 0.7765277777777777,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1672222222222222,
"calib/std_conf": 0.22965819958568048,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8850234113712375,
"calib/step_q_c_n": 1495.0,
"calib/step_q_gap": 0.027362672657147202,
"calib/step_q_w": 0.8576607387140903,
"calib/step_q_w_n": 731.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2923.0,
"completions/max_terminated_length": 2923.0,
"completions/mean_length": 819.66015625,
"completions/mean_terminated_length": 826.1141967773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 436.0,
"epoch": 0.1536,
"grad_norm": 0.06142843887209892,
"kl": 0.050872802734375,
"learning_rate": 1.5555555555555558e-06,
"loss": 0.0024,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.018356967717409134,
"mask/share_reasoning": 0.8610062003135681,
"mask/share_step_conf": 0.11282435059547424,
"num_tokens": 45435868.0,
"reward": 0.9260880947113037,
"reward_std": 0.1764068305492401,
"rewards/accuracy_reward_step": 0.703125,
"rewards/final_brier_reward_step": 0.7573742270469666,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.757301926612854,
"step": 144
},
{
"adv/mean_abs_final_conf": 0.6020931601524353,
"adv/mean_abs_reasoning": 0.4983627200126648,
"adv/mean_abs_step_conf": 0.7491400241851807,
"adv/ratio_final_to_reasoning": 1.2081424552324749,
"adv/ratio_step_to_reasoning": 1.5032023747003846,
"adv/std_final_conf": 0.8259950876235962,
"adv/std_reasoning": 0.7576925754547119,
"adv/std_step_conf": 0.9257418513298035,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6714731427607723,
"calib/avg_num_step_conf": 8.71484375,
"calib/ece": 0.16372540983606557,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.7008196721311475,
"calib/gap": 0.16040500885403486,
"calib/mean_conf": 0.8539713114754097,
"calib/mu_c": 0.8980169491525425,
"calib/mu_w": 0.7376119402985076,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.14614344262295081,
"calib/std_conf": 0.23767791264771557,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8855218525766471,
"calib/step_q_c_n": 1533.0,
"calib/step_q_gap": 0.03201182392335189,
"calib/step_q_w": 0.8535100286532952,
"calib/step_q_w_n": 698.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2981.0,
"completions/max_terminated_length": 2981.0,
"completions/mean_length": 812.96484375,
"completions/mean_terminated_length": 829.1593627929688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 444.0,
"epoch": 0.15466666666666667,
"grad_norm": 0.044827286154031754,
"kl": 0.05010986328125,
"learning_rate": 1.527777777777778e-06,
"loss": -0.0536,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.018157798796892166,
"mask/share_reasoning": 0.8492351770401001,
"mask/share_step_conf": 0.11307579278945923,
"num_tokens": 45746691.0,
"reward": 0.9063776731491089,
"reward_std": 0.23287507891654968,
"rewards/accuracy_reward_step": 0.6953125,
"rewards/final_brier_reward_step": 0.754582405090332,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7284855246543884,
"step": 145
},
{
"adv/mean_abs_final_conf": 0.6597222089767456,
"adv/mean_abs_reasoning": 0.5190834999084473,
"adv/mean_abs_step_conf": 0.7232632637023926,
"adv/ratio_final_to_reasoning": 1.2709365816734752,
"adv/ratio_step_to_reasoning": 1.393346665478593,
"adv/std_final_conf": 0.8603639602661133,
"adv/std_reasoning": 0.7928233742713928,
"adv/std_step_conf": 0.927862823009491,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6576347826086957,
"calib/avg_num_step_conf": 8.91015625,
"calib/ece": 0.3995833333333334,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.7333333333333333,
"calib/gap": 0.1361878260869568,
"calib/mean_conf": 0.8754166666666667,
"calib/mu_c": 0.9463478260869566,
"calib/mu_w": 0.8101599999999998,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.39791666666666675,
"calib/std_conf": 0.21381025791315272,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8873650793650794,
"calib/step_q_c_n": 945.0,
"calib/step_q_gap": 0.04676627696986979,
"calib/step_q_w": 0.8405988023952096,
"calib/step_q_w_n": 1336.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2517.0,
"completions/max_terminated_length": 2517.0,
"completions/mean_length": 791.56640625,
"completions/mean_terminated_length": 830.495849609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 452.0,
"epoch": 0.15573333333333333,
"grad_norm": 0.05562068521976471,
"kl": 0.047908782958984375,
"learning_rate": 1.5e-06,
"loss": -0.1804,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.017334166914224625,
"mask/share_reasoning": 0.8294928073883057,
"mask/share_step_conf": 0.10629801452159882,
"num_tokens": 46056548.0,
"reward": 0.7173329591751099,
"reward_std": 0.26008641719818115,
"rewards/accuracy_reward_step": 0.44921875,
"rewards/final_brier_reward_step": 0.5772007703781128,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.5801214575767517,
"step": 146
},
{
"adv/mean_abs_final_conf": 0.6195129156112671,
"adv/mean_abs_reasoning": 0.4309051036834717,
"adv/mean_abs_step_conf": 0.752669095993042,
"adv/ratio_final_to_reasoning": 1.4377015039170673,
"adv/ratio_step_to_reasoning": 1.7467165962042708,
"adv/std_final_conf": 0.8434000611305237,
"adv/std_reasoning": 0.7393536567687988,
"adv/std_step_conf": 0.9289423227310181,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6683632998413539,
"calib/avg_num_step_conf": 9.57421875,
"calib/ece": 0.3651626016260162,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7560975609756098,
"calib/gap": 0.18607879428873597,
"calib/mean_conf": 0.861910569105691,
"calib/mu_c": 0.9541935483870967,
"calib/mu_w": 0.7681147540983607,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36150406504065036,
"calib/std_conf": 0.25249166477846935,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8937033582089552,
"calib/step_q_c_n": 1072.0,
"calib/step_q_gap": 0.05780778170424161,
"calib/step_q_w": 0.8358955765047136,
"calib/step_q_w_n": 1379.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2775.0,
"completions/max_terminated_length": 2775.0,
"completions/mean_length": 816.8359375,
"completions/mean_terminated_length": 843.1854858398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 447.0,
"epoch": 0.1568,
"grad_norm": 0.06812643259763718,
"kl": 0.04569244384765625,
"learning_rate": 1.4722222222222225e-06,
"loss": -0.1596,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.017271030694246292,
"mask/share_reasoning": 0.8407407999038696,
"mask/share_step_conf": 0.11073816567659378,
"num_tokens": 46369338.0,
"reward": 0.7616201639175415,
"reward_std": 0.2501278519630432,
"rewards/accuracy_reward_step": 0.484375,
"rewards/final_brier_reward_step": 0.6258050203323364,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.608372688293457,
"step": 147
},
{
"adv/mean_abs_final_conf": 0.6027733087539673,
"adv/mean_abs_reasoning": 0.46370819211006165,
"adv/mean_abs_step_conf": 0.739616334438324,
"adv/ratio_final_to_reasoning": 1.2998979078008144,
"adv/ratio_step_to_reasoning": 1.5950038127917636,
"adv/std_final_conf": 0.8275401592254639,
"adv/std_reasoning": 0.7396039366722107,
"adv/std_step_conf": 0.9280713200569153,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.645531028095091,
"calib/avg_num_step_conf": 8.54296875,
"calib/ece": 0.24541666666666673,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.7958333333333333,
"calib/gap": 0.16023155294844071,
"calib/mean_conf": 0.8865833333333335,
"calib/mu_c": 0.9413291139240504,
"calib/mu_w": 0.7810975609756097,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2368333333333334,
"calib/std_conf": 0.2259258574301657,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8914390065741417,
"calib/step_q_c_n": 1369.0,
"calib/step_q_gap": 0.06576663493599988,
"calib/step_q_w": 0.8256723716381418,
"calib/step_q_w_n": 818.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2929.0,
"completions/max_terminated_length": 2929.0,
"completions/mean_length": 807.80078125,
"completions/mean_terminated_length": 837.23486328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 365.0,
"epoch": 0.15786666666666666,
"grad_norm": 0.039087146520614624,
"kl": 0.045024871826171875,
"learning_rate": 1.4444444444444445e-06,
"loss": -0.181,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.018149036914110184,
"mask/share_reasoning": 0.8342307806015015,
"mask/share_step_conf": 0.11246395111083984,
"num_tokens": 46681247.0,
"reward": 0.8411773443222046,
"reward_std": 0.27904221415519714,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/final_brier_reward_step": 0.6974323987960815,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.6747660636901855,
"step": 148
},
{
"adv/mean_abs_final_conf": 0.5810620188713074,
"adv/mean_abs_reasoning": 0.4962088465690613,
"adv/mean_abs_step_conf": 0.7387970685958862,
"adv/ratio_final_to_reasoning": 1.1710029413803213,
"adv/ratio_step_to_reasoning": 1.4888833073093186,
"adv/std_final_conf": 0.7941713929176331,
"adv/std_reasoning": 0.7576331496238708,
"adv/std_step_conf": 0.9306121468544006,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7055350805350806,
"calib/avg_num_step_conf": 8.40234375,
"calib/ece": 0.29906882591093126,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.8137651821862348,
"calib/gap": 0.2109991809991809,
"calib/mean_conf": 0.8921862348178138,
"calib/mu_c": 0.9767567567567566,
"calib/mu_w": 0.7657575757575756,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2960323886639677,
"calib/std_conf": 0.225261156878797,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8979166666666668,
"calib/step_q_c_n": 1224.0,
"calib/step_q_gap": 0.05478829557713061,
"calib/step_q_w": 0.8431283710895362,
"calib/step_q_w_n": 927.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2358.0,
"completions/max_terminated_length": 2358.0,
"completions/mean_length": 807.56640625,
"completions/mean_terminated_length": 830.26904296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.15893333333333334,
"grad_norm": 0.11413798481225967,
"kl": 0.041881561279296875,
"learning_rate": 1.4166666666666667e-06,
"loss": -0.1408,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.017695922404527664,
"mask/share_reasoning": 0.8483445644378662,
"mask/share_step_conf": 0.10661575198173523,
"num_tokens": 46992440.0,
"reward": 0.8469992876052856,
"reward_std": 0.26091471314430237,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.6991230249404907,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6862818002700806,
"step": 149
},
{
"adv/mean_abs_final_conf": 0.5980448722839355,
"adv/mean_abs_reasoning": 0.520313024520874,
"adv/mean_abs_step_conf": 0.7211383581161499,
"adv/ratio_final_to_reasoning": 1.1493943916446072,
"adv/ratio_step_to_reasoning": 1.3859702220220305,
"adv/std_final_conf": 0.8278406262397766,
"adv/std_reasoning": 0.7754331827163696,
"adv/std_step_conf": 0.9302837252616882,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6322846283783783,
"calib/avg_num_step_conf": 9.05078125,
"calib/ece": 0.28569672131147544,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.8114754098360656,
"calib/gap": 0.1797494369369369,
"calib/mean_conf": 0.8795491803278688,
"calib/mu_c": 0.9502702702702702,
"calib/mu_w": 0.7705208333333333,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.279344262295082,
"calib/std_conf": 0.25590766144813387,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8999468892261002,
"calib/step_q_c_n": 1318.0,
"calib/step_q_gap": 0.08838532766453866,
"calib/step_q_w": 0.8115615615615616,
"calib/step_q_w_n": 999.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3035.0,
"completions/max_terminated_length": 3035.0,
"completions/mean_length": 794.09375,
"completions/mean_terminated_length": 816.4176635742188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 382.0,
"epoch": 0.16,
"grad_norm": 0.03058474138379097,
"kl": 0.044178009033203125,
"learning_rate": 1.3888888888888892e-06,
"loss": -0.1441,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.018817484378814697,
"mask/share_reasoning": 0.8333878517150879,
"mask/share_step_conf": 0.12045087665319443,
"num_tokens": 47300688.0,
"reward": 0.8491833209991455,
"reward_std": 0.2430640459060669,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.6739871501922607,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7181295156478882,
"step": 150
},
{
"adv/mean_abs_final_conf": 0.6749951839447021,
"adv/mean_abs_reasoning": 0.6116622090339661,
"adv/mean_abs_step_conf": 0.7771894335746765,
"adv/ratio_final_to_reasoning": 1.103542402939625,
"adv/ratio_step_to_reasoning": 1.2706186880535537,
"adv/std_final_conf": 0.8708445429801941,
"adv/std_reasoning": 0.8267552852630615,
"adv/std_step_conf": 0.9338359832763672,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.6486131801092431,
"calib/avg_num_step_conf": 8.515625,
"calib/ece": 0.37487394957983183,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.6302521008403361,
"calib/gap": 0.1708292544513017,
"calib/mean_conf": 0.7660504201680672,
"calib/mu_c": 0.857207207207207,
"calib/mu_w": 0.6863779527559053,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3372689075630251,
"calib/std_conf": 0.3341632431732412,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8933407079646017,
"calib/step_q_c_n": 904.0,
"calib/step_q_gap": 0.08486892113074584,
"calib/step_q_w": 0.8084717868338559,
"calib/step_q_w_n": 1276.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2328.0,
"completions/max_terminated_length": 2328.0,
"completions/mean_length": 790.69921875,
"completions/mean_terminated_length": 843.4125366210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 448.0,
"epoch": 0.16106666666666666,
"grad_norm": 0.05278412252664566,
"kl": 0.04607391357421875,
"learning_rate": 1.3611111111111112e-06,
"loss": -0.2397,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.01703723706305027,
"mask/share_reasoning": 0.8188620805740356,
"mask/share_step_conf": 0.10160071402788162,
"num_tokens": 47610131.0,
"reward": 0.7299396395683289,
"reward_std": 0.29607152938842773,
"rewards/accuracy_reward_step": 0.43359375,
"rewards/final_brier_reward_step": 0.590067982673645,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.5971550345420837,
"step": 151
},
{
"adv/mean_abs_final_conf": 0.6524473428726196,
"adv/mean_abs_reasoning": 0.5343358516693115,
"adv/mean_abs_step_conf": 0.7500141859054565,
"adv/ratio_final_to_reasoning": 1.2210435456170825,
"adv/ratio_step_to_reasoning": 1.4036381492320742,
"adv/std_final_conf": 0.8601976037025452,
"adv/std_reasoning": 0.792835533618927,
"adv/std_step_conf": 0.9345920085906982,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6524177814938684,
"calib/avg_num_step_conf": 8.8359375,
"calib/ece": 0.27661157024793376,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.6446280991735537,
"calib/gap": 0.21769091415830555,
"calib/mean_conf": 0.7511570247933883,
"calib/mu_c": 0.8447101449275364,
"calib/mu_w": 0.6270192307692308,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.22876033057851228,
"calib/std_conf": 0.3624006049715013,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8907277397260275,
"calib/step_q_c_n": 1168.0,
"calib/step_q_gap": 0.057400500237910435,
"calib/step_q_w": 0.833327239488117,
"calib/step_q_w_n": 1094.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2731.0,
"completions/max_terminated_length": 2731.0,
"completions/mean_length": 833.3515625,
"completions/mean_terminated_length": 853.35205078125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 465.0,
"epoch": 0.16213333333333332,
"grad_norm": 0.053785111755132675,
"kl": 0.045932769775390625,
"learning_rate": 1.3333333333333334e-06,
"loss": -0.1069,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.017900928854942322,
"mask/share_reasoning": 0.8457293510437012,
"mask/share_step_conf": 0.11293217539787292,
"num_tokens": 47928861.0,
"reward": 0.8092750310897827,
"reward_std": 0.25778883695602417,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.6594210863113403,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.6622539758682251,
"step": 152
},
{
"adv/mean_abs_final_conf": 0.5960100293159485,
"adv/mean_abs_reasoning": 0.4430437684059143,
"adv/mean_abs_step_conf": 0.7543562054634094,
"adv/ratio_final_to_reasoning": 1.3452621881138553,
"adv/ratio_step_to_reasoning": 1.7026674546797198,
"adv/std_final_conf": 0.8276448845863342,
"adv/std_reasoning": 0.7207258939743042,
"adv/std_step_conf": 0.9289489388465881,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.654132082280765,
"calib/avg_num_step_conf": 8.53515625,
"calib/ece": 0.20362903225806445,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.6733870967741935,
"calib/gap": 0.23280692890653176,
"calib/mean_conf": 0.7826612903225807,
"calib/mu_c": 0.8624539877300612,
"calib/mu_w": 0.6296470588235294,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.16451612903225799,
"calib/std_conf": 0.33721011175423377,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.883667409057164,
"calib/step_q_c_n": 1347.0,
"calib/step_q_gap": 0.017247361324467025,
"calib/step_q_w": 0.866420047732697,
"calib/step_q_w_n": 838.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 3001.0,
"completions/max_terminated_length": 3001.0,
"completions/mean_length": 789.28125,
"completions/mean_terminated_length": 808.2240600585938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 412.0,
"epoch": 0.1632,
"grad_norm": 0.0818745419383049,
"kl": 0.05124664306640625,
"learning_rate": 1.3055555555555556e-06,
"loss": -0.0529,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018392134457826614,
"mask/share_reasoning": 0.845497727394104,
"mask/share_step_conf": 0.11267266422510147,
"num_tokens": 48238237.0,
"reward": 0.879777729511261,
"reward_std": 0.21167093515396118,
"rewards/accuracy_reward_step": 0.63671875,
"rewards/final_brier_reward_step": 0.7267390489578247,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7117226123809814,
"step": 153
},
{
"adv/mean_abs_final_conf": 0.6036090850830078,
"adv/mean_abs_reasoning": 0.3608783185482025,
"adv/mean_abs_step_conf": 0.7696006894111633,
"adv/ratio_final_to_reasoning": 1.6726111103357508,
"adv/ratio_step_to_reasoning": 2.132576688195713,
"adv/std_final_conf": 0.8111317753791809,
"adv/std_reasoning": 0.640428900718689,
"adv/std_step_conf": 0.9313547611236572,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.7188457950153272,
"calib/avg_num_step_conf": 8.26171875,
"calib/ece": 0.30257142857142855,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6,
"calib/gap": 0.2740690390510465,
"calib/mean_conf": 0.7428979591836734,
"calib/mu_c": 0.8804918032786887,
"calib/mu_w": 0.6064227642276422,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2737551020408163,
"calib/std_conf": 0.3528487127331878,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8891649694501019,
"calib/step_q_c_n": 982.0,
"calib/step_q_gap": 0.02997697298055202,
"calib/step_q_w": 0.8591879964695499,
"calib/step_q_w_n": 1133.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2919.0,
"completions/max_terminated_length": 2919.0,
"completions/mean_length": 771.3828125,
"completions/mean_terminated_length": 796.26611328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 402.0,
"epoch": 0.16426666666666667,
"grad_norm": 0.08071709424257278,
"kl": 0.051548004150390625,
"learning_rate": 1.2777777777777779e-06,
"loss": -0.0806,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.018711235374212265,
"mask/share_reasoning": 0.8364302515983582,
"mask/share_step_conf": 0.11360852420330048,
"num_tokens": 48540151.0,
"reward": 0.7974370121955872,
"reward_std": 0.22275030612945557,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/final_brier_reward_step": 0.6723519563674927,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6358032822608948,
"step": 154
},
{
"adv/mean_abs_final_conf": 0.753005862236023,
"adv/mean_abs_reasoning": 0.5864537358283997,
"adv/mean_abs_step_conf": 0.7431141138076782,
"adv/ratio_final_to_reasoning": 1.2839987474414478,
"adv/ratio_step_to_reasoning": 1.267131690717234,
"adv/std_final_conf": 0.9215671420097351,
"adv/std_reasoning": 0.8266345858573914,
"adv/std_step_conf": 0.9348805546760559,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6045225155279503,
"calib/avg_num_step_conf": 8.82421875,
"calib/ece": 0.2928,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.524,
"calib/gap": 0.17288819875776396,
"calib/mean_conf": 0.67472,
"calib/mu_c": 0.7521739130434784,
"calib/mu_w": 0.5792857142857144,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.20776000000000003,
"calib/std_conf": 0.3875060278240843,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8760777683854608,
"calib/step_q_c_n": 1183.0,
"calib/step_q_gap": 0.043670705188434744,
"calib/step_q_w": 0.8324070631970261,
"calib/step_q_w_n": 1076.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2896.0,
"completions/max_terminated_length": 2896.0,
"completions/mean_length": 794.5625,
"completions/mean_terminated_length": 807.1746215820312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 409.0,
"epoch": 0.16533333333333333,
"grad_norm": 0.05100875720381737,
"kl": 0.059783935546875,
"learning_rate": 1.25e-06,
"loss": -0.0423,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.01870831474661827,
"mask/share_reasoning": 0.8480067253112793,
"mask/share_step_conf": 0.11765995621681213,
"num_tokens": 48850775.0,
"reward": 0.8101447820663452,
"reward_std": 0.26722443103790283,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.6571409702301025,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.6608047485351562,
"step": 155
},
{
"adv/mean_abs_final_conf": 0.7030758857727051,
"adv/mean_abs_reasoning": 0.4742887020111084,
"adv/mean_abs_step_conf": 0.7566878795623779,
"adv/ratio_final_to_reasoning": 1.4823795776527653,
"adv/ratio_step_to_reasoning": 1.5954162018910065,
"adv/std_final_conf": 0.906621515750885,
"adv/std_reasoning": 0.7392860054969788,
"adv/std_step_conf": 0.9321083426475525,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6082010582010583,
"calib/avg_num_step_conf": 8.74609375,
"calib/ece": 0.3310526315789473,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.4493927125506073,
"calib/gap": 0.1552129629629631,
"calib/mean_conf": 0.5785829959514168,
"calib/mu_c": 0.6489629629629631,
"calib/mu_w": 0.49374999999999997,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.1815384615384615,
"calib/std_conf": 0.4222496337868507,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8693238731218699,
"calib/step_q_c_n": 1198.0,
"calib/step_q_gap": 0.02888679339084199,
"calib/step_q_w": 0.8404370797310279,
"calib/step_q_w_n": 1041.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2943.0,
"completions/max_terminated_length": 2943.0,
"completions/mean_length": 819.390625,
"completions/mean_terminated_length": 829.1067504882812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 364.0,
"epoch": 0.1664,
"grad_norm": 0.07828212529420853,
"kl": 0.05800628662109375,
"learning_rate": 1.2222222222222223e-06,
"loss": 0.0139,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.01829274743795395,
"mask/share_reasoning": 0.8516146540641785,
"mask/share_step_conf": 0.11837387084960938,
"num_tokens": 49165299.0,
"reward": 0.79437255859375,
"reward_std": 0.21283696591854095,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.6228859424591064,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6689841747283936,
"step": 156
},
{
"adv/mean_abs_final_conf": 0.74039626121521,
"adv/mean_abs_reasoning": 0.4796062111854553,
"adv/mean_abs_step_conf": 0.7580063343048096,
"adv/ratio_final_to_reasoning": 1.5437587002577655,
"adv/ratio_step_to_reasoning": 1.5804764755469394,
"adv/std_final_conf": 0.906609833240509,
"adv/std_reasoning": 0.7575482726097107,
"adv/std_step_conf": 0.933040201663971,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6703153316418993,
"calib/avg_num_step_conf": 9.07421875,
"calib/ece": 0.26471311475409837,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.5040983606557377,
"calib/gap": 0.28305255527365,
"calib/mean_conf": 0.6073360655737705,
"calib/mu_c": 0.7105806451612904,
"calib/mu_w": 0.4275280898876404,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.11840163934426229,
"calib/std_conf": 0.42741583603792177,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8700872093023256,
"calib/step_q_c_n": 1376.0,
"calib/step_q_gap": 0.11039343950295932,
"calib/step_q_w": 0.7596937697993663,
"calib/step_q_w_n": 947.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3025.0,
"completions/max_terminated_length": 3025.0,
"completions/mean_length": 815.16796875,
"completions/mean_terminated_length": 841.4636840820312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 388.0,
"epoch": 0.16746666666666668,
"grad_norm": 0.07636091113090515,
"kl": 0.06465911865234375,
"learning_rate": 1.1944444444444446e-06,
"loss": -0.0861,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017809420824050903,
"mask/share_reasoning": 0.8347960710525513,
"mask/share_step_conf": 0.11614448577165604,
"num_tokens": 49477710.0,
"reward": 0.8488029837608337,
"reward_std": 0.23808994889259338,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.6824370622634888,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7026687860488892,
"step": 157
},
{
"adv/mean_abs_final_conf": 0.7005129456520081,
"adv/mean_abs_reasoning": 0.5425291061401367,
"adv/mean_abs_step_conf": 0.7504801154136658,
"adv/ratio_final_to_reasoning": 1.2911988273511426,
"adv/ratio_step_to_reasoning": 1.3832992680392981,
"adv/std_final_conf": 0.8809415698051453,
"adv/std_reasoning": 0.7928447127342224,
"adv/std_step_conf": 0.9298225045204163,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6185819940264542,
"calib/avg_num_step_conf": 9.46875,
"calib/ece": 0.3114979757085021,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.46153846153846156,
"calib/gap": 0.20209216327691654,
"calib/mean_conf": 0.5668016194331984,
"calib/mu_c": 0.639620253164557,
"calib/mu_w": 0.4375280898876404,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.11931174089068827,
"calib/std_conf": 0.435395363535799,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8779732582688247,
"calib/step_q_c_n": 1421.0,
"calib/step_q_gap": 0.05564025727181576,
"calib/step_q_w": 0.8223330009970089,
"calib/step_q_w_n": 1003.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2820.0,
"completions/max_terminated_length": 2820.0,
"completions/mean_length": 751.07421875,
"completions/mean_terminated_length": 775.3023681640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 378.0,
"epoch": 0.16853333333333334,
"grad_norm": 0.10405830293893814,
"kl": 0.0607147216796875,
"learning_rate": 1.1666666666666668e-06,
"loss": -0.1067,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.01923152431845665,
"mask/share_reasoning": 0.8270524740219116,
"mask/share_step_conf": 0.12246599793434143,
"num_tokens": 49775225.0,
"reward": 0.8258012533187866,
"reward_std": 0.2111181616783142,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/final_brier_reward_step": 0.6408855319023132,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.695091962814331,
"step": 158
},
{
"adv/mean_abs_final_conf": 0.6621556282043457,
"adv/mean_abs_reasoning": 0.5811421275138855,
"adv/mean_abs_step_conf": 0.7509884238243103,
"adv/ratio_final_to_reasoning": 1.1394039372726847,
"adv/ratio_step_to_reasoning": 1.2922629220446018,
"adv/std_final_conf": 0.8604494333267212,
"adv/std_reasoning": 0.8266856074333191,
"adv/std_step_conf": 0.9339320063591003,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.7197873799725651,
"calib/avg_num_step_conf": 9.2890625,
"calib/ece": 0.2325102880658435,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.42386831275720166,
"calib/gap": 0.34694444444444444,
"calib/mean_conf": 0.5387654320987654,
"calib/mu_c": 0.6929629629629629,
"calib/mu_w": 0.34601851851851845,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1078600823045266,
"calib/std_conf": 0.43246460703267314,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8845020990764064,
"calib/step_q_c_n": 1191.0,
"calib/step_q_gap": 0.08339847649847876,
"calib/step_q_w": 0.8011036225779277,
"calib/step_q_w_n": 1187.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3027.0,
"completions/max_terminated_length": 3027.0,
"completions/mean_length": 792.6640625,
"completions/mean_terminated_length": 818.2338256835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 394.0,
"epoch": 0.1696,
"grad_norm": 0.04640193283557892,
"kl": 0.06124114990234375,
"learning_rate": 1.138888888888889e-06,
"loss": -0.1201,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.018069058656692505,
"mask/share_reasoning": 0.8351427316665649,
"mask/share_step_conf": 0.11553823947906494,
"num_tokens": 50082931.0,
"reward": 0.8521698117256165,
"reward_std": 0.22859559953212738,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.6974222660064697,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.7123861312866211,
"step": 159
},
{
"adv/mean_abs_final_conf": 0.6154712438583374,
"adv/mean_abs_reasoning": 0.47096890211105347,
"adv/mean_abs_step_conf": 0.7640544772148132,
"adv/ratio_final_to_reasoning": 1.3068192848818085,
"adv/ratio_step_to_reasoning": 1.6223034552600901,
"adv/std_final_conf": 0.811108410358429,
"adv/std_reasoning": 0.7395078539848328,
"adv/std_step_conf": 0.9293590188026428,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6589110707803993,
"calib/avg_num_step_conf": 9.4609375,
"calib/ece": 0.29712500000000003,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.5083333333333333,
"calib/gap": 0.2579274047186933,
"calib/mean_conf": 0.5860416666666667,
"calib/mu_c": 0.6881379310344827,
"calib/mu_w": 0.43021052631578943,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1395,
"calib/std_conf": 0.439617919255523,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.860059335443038,
"calib/step_q_c_n": 1264.0,
"calib/step_q_gap": 0.12134603665201893,
"calib/step_q_w": 0.7387132987910191,
"calib/step_q_w_n": 1158.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3043.0,
"completions/max_terminated_length": 3043.0,
"completions/mean_length": 819.359375,
"completions/mean_terminated_length": 849.214599609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 493.0,
"epoch": 0.17066666666666666,
"grad_norm": 0.15533457696437836,
"kl": 0.059192657470703125,
"learning_rate": 1.111111111111111e-06,
"loss": -0.1435,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.017590470612049103,
"mask/share_reasoning": 0.8359103798866272,
"mask/share_step_conf": 0.1113428920507431,
"num_tokens": 50397527.0,
"reward": 0.8208951354026794,
"reward_std": 0.21404847502708435,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.6474605202674866,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.693548321723938,
"step": 160
},
{
"adv/mean_abs_final_conf": 0.5511636734008789,
"adv/mean_abs_reasoning": 0.3422989547252655,
"adv/mean_abs_step_conf": 0.7528557777404785,
"adv/ratio_final_to_reasoning": 1.6101821691020106,
"adv/ratio_step_to_reasoning": 2.1994100985342837,
"adv/std_final_conf": 0.7935216426849365,
"adv/std_reasoning": 0.640374481678009,
"adv/std_step_conf": 0.9291737079620361,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6562388591800358,
"calib/avg_num_step_conf": 8.98828125,
"calib/ece": 0.271902834008097,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.5951417004048583,
"calib/gap": 0.2708288770053475,
"calib/mean_conf": 0.6700404858299596,
"calib/mu_c": 0.7358288770053476,
"calib/mu_w": 0.4650000000000001,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.09242914979757073,
"calib/std_conf": 0.420668927470327,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8642918192918193,
"calib/step_q_c_n": 1638.0,
"calib/step_q_gap": 0.05602635926165345,
"calib/step_q_w": 0.8082654600301659,
"calib/step_q_w_n": 663.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 3053.0,
"completions/max_terminated_length": 3053.0,
"completions/mean_length": 765.5859375,
"completions/mean_terminated_length": 783.9600219726562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 365.0,
"epoch": 0.17173333333333332,
"grad_norm": 0.04432344064116478,
"kl": 0.05446624755859375,
"learning_rate": 1.0833333333333335e-06,
"loss": -0.0982,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.019417721778154373,
"mask/share_reasoning": 0.8330705165863037,
"mask/share_step_conf": 0.12407425045967102,
"num_tokens": 50697437.0,
"reward": 0.9070243239402771,
"reward_std": 0.18110650777816772,
"rewards/accuracy_reward_step": 0.73046875,
"rewards/final_brier_reward_step": 0.7054632902145386,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7695227265357971,
"step": 161
},
{
"adv/mean_abs_final_conf": 0.562353253364563,
"adv/mean_abs_reasoning": 0.401167631149292,
"adv/mean_abs_step_conf": 0.7378664016723633,
"adv/ratio_final_to_reasoning": 1.401791195749008,
"adv/ratio_step_to_reasoning": 1.8392969531426901,
"adv/std_final_conf": 0.7708464860916138,
"adv/std_reasoning": 0.7014594078063965,
"adv/std_step_conf": 0.9175775051116943,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6586058651926916,
"calib/avg_num_step_conf": 8.72265625,
"calib/ece": 0.24224081632653044,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6408163265306123,
"calib/gap": 0.24810471364962405,
"calib/mean_conf": 0.7205265306122449,
"calib/mu_c": 0.7995149700598803,
"calib/mu_w": 0.5514102564102562,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.14056734693877534,
"calib/std_conf": 0.39542936146468877,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8857506925207755,
"calib/step_q_c_n": 1444.0,
"calib/step_q_gap": 0.09616894347134586,
"calib/step_q_w": 0.7895817490494297,
"calib/step_q_w_n": 789.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2954.0,
"completions/max_terminated_length": 2954.0,
"completions/mean_length": 759.67578125,
"completions/mean_terminated_length": 787.3563232421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 408.0,
"epoch": 0.1728,
"grad_norm": 0.11026094108819962,
"kl": 0.053619384765625,
"learning_rate": 1.0555555555555557e-06,
"loss": -0.1468,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01879039779305458,
"mask/share_reasoning": 0.8299312591552734,
"mask/share_step_conf": 0.11612210422754288,
"num_tokens": 50996058.0,
"reward": 0.8966192007064819,
"reward_std": 0.1913757026195526,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.7013081908226013,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7700551152229309,
"step": 162
},
{
"adv/mean_abs_final_conf": 0.5956529378890991,
"adv/mean_abs_reasoning": 0.4349844753742218,
"adv/mean_abs_step_conf": 0.7498718500137329,
"adv/ratio_final_to_reasoning": 1.3693659696168525,
"adv/ratio_step_to_reasoning": 1.7239048574517748,
"adv/std_final_conf": 0.8112250566482544,
"adv/std_reasoning": 0.7014384269714355,
"adv/std_step_conf": 0.9236242175102234,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6668549030029233,
"calib/avg_num_step_conf": 8.484375,
"calib/ece": 0.27717741935483864,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.5967741935483871,
"calib/gap": 0.2917618921073612,
"calib/mean_conf": 0.6567741935483872,
"calib/mu_c": 0.7814788732394367,
"calib/mu_w": 0.48971698113207546,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1806854838709677,
"calib/std_conf": 0.42956570038462133,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8771989966555185,
"calib/step_q_c_n": 1196.0,
"calib/step_q_gap": 0.04497563599978083,
"calib/step_q_w": 0.8322233606557377,
"calib/step_q_w_n": 976.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3052.0,
"completions/max_terminated_length": 3052.0,
"completions/mean_length": 802.96875,
"completions/mean_terminated_length": 812.4901733398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 311.0,
"epoch": 0.17386666666666667,
"grad_norm": 0.04852801561355591,
"kl": 0.05211639404296875,
"learning_rate": 1.0277777777777777e-06,
"loss": -0.0933,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.01897309347987175,
"mask/share_reasoning": 0.8520207405090332,
"mask/share_step_conf": 0.11728739738464355,
"num_tokens": 51306450.0,
"reward": 0.8401877880096436,
"reward_std": 0.2186070829629898,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.6843827962875366,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.6913052797317505,
"step": 163
},
{
"adv/mean_abs_final_conf": 0.659756064414978,
"adv/mean_abs_reasoning": 0.47427132725715637,
"adv/mean_abs_step_conf": 0.7514228820800781,
"adv/ratio_final_to_reasoning": 1.3910941406273318,
"adv/ratio_step_to_reasoning": 1.5843734143191128,
"adv/std_final_conf": 0.8602266311645508,
"adv/std_reasoning": 0.7393752932548523,
"adv/std_step_conf": 0.9299604296684265,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6942012745990616,
"calib/avg_num_step_conf": 9.1875,
"calib/ece": 0.28970833333333323,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.5916666666666667,
"calib/gap": 0.24238882274669105,
"calib/mean_conf": 0.727625,
"calib/mu_c": 0.8377099236641222,
"calib/mu_w": 0.5953211009174312,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23574999999999993,
"calib/std_conf": 0.3616314090917251,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8610126582278481,
"calib/step_q_c_n": 1106.0,
"calib/step_q_gap": 0.12315551537070513,
"calib/step_q_w": 0.737857142857143,
"calib/step_q_w_n": 1246.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2319.0,
"completions/max_terminated_length": 2319.0,
"completions/mean_length": 791.1484375,
"completions/mean_terminated_length": 840.3900756835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 414.0,
"epoch": 0.17493333333333333,
"grad_norm": 0.04501301795244217,
"kl": 0.0543060302734375,
"learning_rate": 1.0000000000000002e-06,
"loss": -0.276,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.016907479614019394,
"mask/share_reasoning": 0.8249204158782959,
"mask/share_step_conf": 0.0995783731341362,
"num_tokens": 51615120.0,
"reward": 0.8000724911689758,
"reward_std": 0.2464374154806137,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.6641730070114136,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.6453468203544617,
"step": 164
},
{
"adv/mean_abs_final_conf": 0.5247633457183838,
"adv/mean_abs_reasoning": 0.43585866689682007,
"adv/mean_abs_step_conf": 0.7637988328933716,
"adv/ratio_final_to_reasoning": 1.2039759343425191,
"adv/ratio_step_to_reasoning": 1.7524002409574297,
"adv/std_final_conf": 0.7588238716125488,
"adv/std_reasoning": 0.7014684677124023,
"adv/std_step_conf": 0.930880069732666,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.593040293040293,
"calib/avg_num_step_conf": 8.63671875,
"calib/ece": 0.340040322580645,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.7661290322580645,
"calib/gap": 0.12131401931401908,
"calib/mean_conf": 0.8384274193548388,
"calib/mu_c": 0.8897902097902096,
"calib/mu_w": 0.7684761904761905,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3009274193548386,
"calib/std_conf": 0.30262085491390733,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8641282894736844,
"calib/step_q_c_n": 1216.0,
"calib/step_q_gap": 0.026801656307855137,
"calib/step_q_w": 0.8373266331658292,
"calib/step_q_w_n": 995.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2951.0,
"completions/max_terminated_length": 2951.0,
"completions/mean_length": 816.29296875,
"completions/mean_terminated_length": 829.2500610351562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 427.0,
"epoch": 0.176,
"grad_norm": 0.05763980746269226,
"kl": 0.0528717041015625,
"learning_rate": 9.722222222222224e-07,
"loss": -0.0403,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018184591084718704,
"mask/share_reasoning": 0.8538507223129272,
"mask/share_step_conf": 0.11233967542648315,
"num_tokens": 51929667.0,
"reward": 0.8078880310058594,
"reward_std": 0.21158744394779205,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.6345081925392151,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.6757991313934326,
"step": 165
},
{
"adv/mean_abs_final_conf": 0.5661165714263916,
"adv/mean_abs_reasoning": 0.47203224897384644,
"adv/mean_abs_step_conf": 0.7055524587631226,
"adv/ratio_final_to_reasoning": 1.1993175734435002,
"adv/ratio_step_to_reasoning": 1.4947124064021622,
"adv/std_final_conf": 0.8220821619033813,
"adv/std_reasoning": 0.7753757238388062,
"adv/std_step_conf": 0.9330090284347534,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7565439219165928,
"calib/avg_num_step_conf": 8.09375,
"calib/ece": 0.21706122448979598,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6612244897959184,
"calib/gap": 0.2990476190476189,
"calib/mean_conf": 0.7803265306122449,
"calib/mu_c": 0.8828571428571428,
"calib/mu_w": 0.5838095238095239,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1701224489795919,
"calib/std_conf": 0.32980126283808553,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.874810606060606,
"calib/step_q_c_n": 1320.0,
"calib/step_q_gap": 0.08522284010315928,
"calib/step_q_w": 0.7895877659574467,
"calib/step_q_w_n": 752.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2979.0,
"completions/max_terminated_length": 2979.0,
"completions/mean_length": 818.90234375,
"completions/mean_terminated_length": 835.2151489257812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 411.0,
"epoch": 0.17706666666666668,
"grad_norm": 0.03530353680253029,
"kl": 0.0486907958984375,
"learning_rate": 9.444444444444445e-07,
"loss": -0.0762,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01848408207297325,
"mask/share_reasoning": 0.8508272767066956,
"mask/share_step_conf": 0.1111573576927185,
"num_tokens": 52245490.0,
"reward": 0.897564172744751,
"reward_std": 0.22545835375785828,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.7517530918121338,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7261877655982971,
"step": 166
},
{
"adv/mean_abs_final_conf": 0.5064303874969482,
"adv/mean_abs_reasoning": 0.43117716908454895,
"adv/mean_abs_step_conf": 0.7519251704216003,
"adv/ratio_final_to_reasoning": 1.174529691755648,
"adv/ratio_step_to_reasoning": 1.7438891117960755,
"adv/std_final_conf": 0.7763898372650146,
"adv/std_reasoning": 0.7205421924591064,
"adv/std_step_conf": 0.9301944971084595,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.518722233975037,
"calib/avg_num_step_conf": 8.34375,
"calib/ece": 0.2897999999999999,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.788,
"calib/gap": 0.027499471123334174,
"calib/mean_conf": 0.88724,
"calib/mu_c": 0.8968098159509201,
"calib/mu_w": 0.869310344827586,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2625199999999999,
"calib/std_conf": 0.219036029912889,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.869957507082153,
"calib/step_q_c_n": 1412.0,
"calib/step_q_gap": 0.0055514297340868435,
"calib/step_q_w": 0.8644060773480662,
"calib/step_q_w_n": 724.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2842.0,
"completions/max_terminated_length": 2842.0,
"completions/mean_length": 770.41015625,
"completions/mean_terminated_length": 776.4763793945312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 146.0,
"epoch": 0.17813333333333334,
"grad_norm": 0.11281926184892654,
"kl": 0.048736572265625,
"learning_rate": 9.166666666666666e-07,
"loss": -0.0473,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.018992718309164047,
"mask/share_reasoning": 0.8573728203773499,
"mask/share_step_conf": 0.1158219575881958,
"num_tokens": 52548323.0,
"reward": 0.8557096123695374,
"reward_std": 0.20991703867912292,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6662777662277222,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7217038869857788,
"step": 167
},
{
"adv/mean_abs_final_conf": 0.6514350175857544,
"adv/mean_abs_reasoning": 0.5799128413200378,
"adv/mean_abs_step_conf": 0.7642394304275513,
"adv/ratio_final_to_reasoning": 1.1233326306465516,
"adv/ratio_step_to_reasoning": 1.3178522287727523,
"adv/std_final_conf": 0.8442649841308594,
"adv/std_reasoning": 0.8100561499595642,
"adv/std_step_conf": 0.9333626627922058,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6436970672639462,
"calib/avg_num_step_conf": 8.70703125,
"calib/ece": 0.2369758064516128,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7258064516129032,
"calib/gap": 0.17539371456568909,
"calib/mean_conf": 0.848508064516129,
"calib/mu_c": 0.9128662420382165,
"calib/mu_w": 0.7374725274725275,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.22620967741935474,
"calib/std_conf": 0.25611196505560024,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8747720125786164,
"calib/step_q_c_n": 1272.0,
"calib/step_q_gap": 0.11429134382208561,
"calib/step_q_w": 0.7604806687565308,
"calib/step_q_w_n": 957.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2408.0,
"completions/max_terminated_length": 2408.0,
"completions/mean_length": 828.46875,
"completions/mean_terminated_length": 848.35205078125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 434.0,
"epoch": 0.1792,
"grad_norm": 0.09699970483779907,
"kl": 0.047489166259765625,
"learning_rate": 8.88888888888889e-07,
"loss": -0.0745,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.01726042851805687,
"mask/share_reasoning": 0.8543730974197388,
"mask/share_step_conf": 0.10492902994155884,
"num_tokens": 52865083.0,
"reward": 0.8707404136657715,
"reward_std": 0.28335040807724,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/final_brier_reward_step": 0.7139902114868164,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7118656635284424,
"step": 168
},
{
"adv/mean_abs_final_conf": 0.5136945247650146,
"adv/mean_abs_reasoning": 0.4035590589046478,
"adv/mean_abs_step_conf": 0.7587224841117859,
"adv/ratio_final_to_reasoning": 1.2729104041408457,
"adv/ratio_step_to_reasoning": 1.8800779399454774,
"adv/std_final_conf": 0.77040696144104,
"adv/std_reasoning": 0.6816996335983276,
"adv/std_step_conf": 0.932590126991272,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6487183059348008,
"calib/avg_num_step_conf": 8.1171875,
"calib/ece": 0.28783673469387766,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.7510204081632653,
"calib/gap": 0.13626288659793806,
"calib/mean_conf": 0.8735510204081632,
"calib/mu_c": 0.9274999999999999,
"calib/mu_w": 0.7912371134020618,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2786530612244899,
"calib/std_conf": 0.21838564104181507,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8780775444264943,
"calib/step_q_c_n": 1238.0,
"calib/step_q_gap": 0.05716087775982759,
"calib/step_q_w": 0.8209166666666667,
"calib/step_q_w_n": 840.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2687.0,
"completions/max_terminated_length": 2687.0,
"completions/mean_length": 827.3515625,
"completions/mean_terminated_length": 843.8327026367188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 446.0,
"epoch": 0.18026666666666666,
"grad_norm": 0.03865334019064903,
"kl": 0.0500640869140625,
"learning_rate": 8.611111111111112e-07,
"loss": -0.0522,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01773287169635296,
"mask/share_reasoning": 0.8558939099311829,
"mask/share_step_conf": 0.10684196650981903,
"num_tokens": 53181069.0,
"reward": 0.846420168876648,
"reward_std": 0.20503714680671692,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.6753827929496765,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7104262709617615,
"step": 169
},
{
"adv/mean_abs_final_conf": 0.5498205423355103,
"adv/mean_abs_reasoning": 0.4751802682876587,
"adv/mean_abs_step_conf": 0.7622960209846497,
"adv/ratio_final_to_reasoning": 1.1570778061067695,
"adv/ratio_step_to_reasoning": 1.6042249054903526,
"adv/std_final_conf": 0.7767378091812134,
"adv/std_reasoning": 0.7208435535430908,
"adv/std_step_conf": 0.929639995098114,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6231788564316568,
"calib/avg_num_step_conf": 8.61328125,
"calib/ece": 0.2904918032786884,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.8401639344262295,
"calib/gap": 0.12844424921297326,
"calib/mean_conf": 0.8982786885245901,
"calib/mu_c": 0.944076433121019,
"calib/mu_w": 0.8156321839080457,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2726639344262294,
"calib/std_conf": 0.23366885853159886,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8714438502673797,
"calib/step_q_c_n": 1309.0,
"calib/step_q_gap": 0.04695724312452254,
"calib/step_q_w": 0.8244866071428572,
"calib/step_q_w_n": 896.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2986.0,
"completions/max_terminated_length": 2986.0,
"completions/mean_length": 811.78515625,
"completions/mean_terminated_length": 831.26806640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 424.0,
"epoch": 0.18133333333333335,
"grad_norm": 0.0797303095459938,
"kl": 0.0508575439453125,
"learning_rate": 8.333333333333333e-07,
"loss": -0.0568,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01804276928305626,
"mask/share_reasoning": 0.8481639623641968,
"mask/share_step_conf": 0.11035574227571487,
"num_tokens": 53493038.0,
"reward": 0.8397831916809082,
"reward_std": 0.2692488431930542,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.676689863204956,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6895952224731445,
"step": 170
},
{
"adv/mean_abs_final_conf": 0.6254308223724365,
"adv/mean_abs_reasoning": 0.4784793257713318,
"adv/mean_abs_step_conf": 0.7460007667541504,
"adv/ratio_final_to_reasoning": 1.307121935444571,
"adv/ratio_step_to_reasoning": 1.5591076282168745,
"adv/std_final_conf": 0.8441042304039001,
"adv/std_reasoning": 0.7392579913139343,
"adv/std_step_conf": 0.9327840805053711,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6507320735827806,
"calib/avg_num_step_conf": 8.58984375,
"calib/ece": 0.34754940711462456,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.7351778656126482,
"calib/gap": 0.16773995745213388,
"calib/mean_conf": 0.8530830039525692,
"calib/mu_c": 0.9339694656488552,
"calib/mu_w": 0.7662295081967213,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3414229249011858,
"calib/std_conf": 0.25475862603103616,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8812911843276937,
"calib/step_q_c_n": 1123.0,
"calib/step_q_gap": 0.06497148172546308,
"calib/step_q_w": 0.8163197026022306,
"calib/step_q_w_n": 1076.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2273.0,
"completions/max_terminated_length": 2273.0,
"completions/mean_length": 778.046875,
"completions/mean_terminated_length": 787.2727661132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 444.0,
"epoch": 0.1824,
"grad_norm": 0.06779231131076813,
"kl": 0.052455902099609375,
"learning_rate": 8.055555555555557e-07,
"loss": -0.0545,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.01890362799167633,
"mask/share_reasoning": 0.8550175428390503,
"mask/share_step_conf": 0.11436008661985397,
"num_tokens": 53799114.0,
"reward": 0.8081221580505371,
"reward_std": 0.22378191351890564,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.6490582227706909,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.6671860218048096,
"step": 171
},
{
"adv/mean_abs_final_conf": 0.4872409701347351,
"adv/mean_abs_reasoning": 0.3581220209598541,
"adv/mean_abs_step_conf": 0.751218318939209,
"adv/ratio_final_to_reasoning": 1.3605445675437964,
"adv/ratio_step_to_reasoning": 2.097660224651255,
"adv/std_final_conf": 0.7403228282928467,
"adv/std_reasoning": 0.6403510570526123,
"adv/std_step_conf": 0.9275820255279541,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6201541425818882,
"calib/avg_num_step_conf": 7.93359375,
"calib/ece": 0.2174193548387097,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.8185483870967742,
"calib/gap": 0.11301579961464348,
"calib/mean_conf": 0.9113709677419355,
"calib/mu_c": 0.9455491329479768,
"calib/mu_w": 0.8325333333333333,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.21560483870967742,
"calib/std_conf": 0.1761841717718548,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8880857142857144,
"calib/step_q_c_n": 1400.0,
"calib/step_q_gap": 0.038529454380801686,
"calib/step_q_w": 0.8495562599049127,
"calib/step_q_w_n": 631.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2541.0,
"completions/max_terminated_length": 2541.0,
"completions/mean_length": 752.23046875,
"completions/mean_terminated_length": 761.1502075195312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 374.0,
"epoch": 0.18346666666666667,
"grad_norm": 0.038075245916843414,
"kl": 0.0482177734375,
"learning_rate": 7.777777777777779e-07,
"loss": -0.0529,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.019558429718017578,
"mask/share_reasoning": 0.853873610496521,
"mask/share_step_conf": 0.11484922468662262,
"num_tokens": 54095037.0,
"reward": 0.9087539911270142,
"reward_std": 0.18010839819908142,
"rewards/accuracy_reward_step": 0.67578125,
"rewards/final_brier_reward_step": 0.7362257838249207,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7523757815361023,
"step": 172
},
{
"adv/mean_abs_final_conf": 0.5600264072418213,
"adv/mean_abs_reasoning": 0.4762570261955261,
"adv/mean_abs_step_conf": 0.7568649053573608,
"adv/ratio_final_to_reasoning": 1.1758911185320842,
"adv/ratio_step_to_reasoning": 1.5891942033977087,
"adv/std_final_conf": 0.811123788356781,
"adv/std_reasoning": 0.7392693161964417,
"adv/std_step_conf": 0.9293073415756226,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5953743603555077,
"calib/avg_num_step_conf": 8.61328125,
"calib/ece": 0.2741269841269841,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.7896825396825397,
"calib/gap": 0.1049555615405332,
"calib/mean_conf": 0.8935714285714286,
"calib/mu_c": 0.9327215189873416,
"calib/mu_w": 0.8277659574468084,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.27035714285714285,
"calib/std_conf": 0.20683734996949568,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8779779411764708,
"calib/step_q_c_n": 1360.0,
"calib/step_q_gap": 0.016664331709015157,
"calib/step_q_w": 0.8613136094674556,
"calib/step_q_w_n": 845.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2521.0,
"completions/max_terminated_length": 2521.0,
"completions/mean_length": 797.02734375,
"completions/mean_terminated_length": 800.1530151367188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 265.0,
"epoch": 0.18453333333333333,
"grad_norm": 0.038408100605010986,
"kl": 0.04986572265625,
"learning_rate": 7.5e-07,
"loss": -0.0281,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.01891019381582737,
"mask/share_reasoning": 0.8586477041244507,
"mask/share_step_conf": 0.11853589117527008,
"num_tokens": 54402236.0,
"reward": 0.8644517660140991,
"reward_std": 0.23591762781143188,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/final_brier_reward_step": 0.6904085874557495,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.718182384967804,
"step": 173
},
{
"adv/mean_abs_final_conf": 0.6717838644981384,
"adv/mean_abs_reasoning": 0.6233100891113281,
"adv/mean_abs_step_conf": 0.7425686717033386,
"adv/ratio_final_to_reasoning": 1.077768315054744,
"adv/ratio_step_to_reasoning": 1.1913310640648238,
"adv/std_final_conf": 0.8761308789253235,
"adv/std_reasoning": 0.8590900301933289,
"adv/std_step_conf": 0.9354785680770874,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5538930084745762,
"calib/avg_num_step_conf": 8.7890625,
"calib/ece": 0.34117886178861795,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.6910569105691057,
"calib/gap": 0.05903336864406794,
"calib/mean_conf": 0.8480894308943089,
"calib/mu_c": 0.87640625,
"calib/mu_w": 0.8173728813559321,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3344715447154472,
"calib/std_conf": 0.22910396362450083,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.864655013799448,
"calib/step_q_c_n": 1087.0,
"calib/step_q_gap": 0.04448031500426719,
"calib/step_q_w": 0.8201746987951808,
"calib/step_q_w_n": 1162.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2326.0,
"completions/max_terminated_length": 2326.0,
"completions/mean_length": 841.515625,
"completions/mean_terminated_length": 861.7120361328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 107.0,
"epoch": 0.1856,
"grad_norm": 0.0463547483086586,
"kl": 0.04796600341796875,
"learning_rate": 7.222222222222222e-07,
"loss": -0.1683,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.01758679747581482,
"mask/share_reasoning": 0.8539252281188965,
"mask/share_step_conf": 0.1050504595041275,
"num_tokens": 54721896.0,
"reward": 0.7721678018569946,
"reward_std": 0.29204386472702026,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.590755820274353,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6621733903884888,
"step": 174
},
{
"adv/mean_abs_final_conf": 0.5728576183319092,
"adv/mean_abs_reasoning": 0.4325658082962036,
"adv/mean_abs_step_conf": 0.7267290353775024,
"adv/ratio_final_to_reasoning": 1.3243247786696062,
"adv/ratio_step_to_reasoning": 1.6800427159047848,
"adv/std_final_conf": 0.7944455742835999,
"adv/std_reasoning": 0.7205823063850403,
"adv/std_step_conf": 0.9327433109283447,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6657325338894682,
"calib/avg_num_step_conf": 8.484375,
"calib/ece": 0.34783132530120486,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.6184738955823293,
"calib/gap": 0.19052593847758081,
"calib/mean_conf": 0.7836546184738957,
"calib/mu_c": 0.8884821428571429,
"calib/mu_w": 0.6979562043795621,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.340843373493976,
"calib/std_conf": 0.294405495144094,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8670323928944619,
"calib/step_q_c_n": 957.0,
"calib/step_q_gap": 0.04722169330598447,
"calib/step_q_w": 0.8198106995884774,
"calib/step_q_w_n": 1215.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1987.0,
"completions/max_terminated_length": 1987.0,
"completions/mean_length": 828.51953125,
"completions/mean_terminated_length": 841.670654296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 400.0,
"epoch": 0.18666666666666668,
"grad_norm": 0.036516882479190826,
"kl": 0.052318572998046875,
"learning_rate": 6.944444444444446e-07,
"loss": -0.0844,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01774880848824978,
"mask/share_reasoning": 0.856615424156189,
"mask/share_step_conf": 0.11001075804233551,
"num_tokens": 55039821.0,
"reward": 0.7852064967155457,
"reward_std": 0.21699534356594086,
"rewards/accuracy_reward_step": 0.4375,
"rewards/final_brier_reward_step": 0.6309511661529541,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.6574305295944214,
"step": 175
},
{
"adv/mean_abs_final_conf": 0.5725212097167969,
"adv/mean_abs_reasoning": 0.4553416967391968,
"adv/mean_abs_step_conf": 0.7647500038146973,
"adv/ratio_final_to_reasoning": 1.2573441303898778,
"adv/ratio_step_to_reasoning": 1.6795079591683393,
"adv/std_final_conf": 0.8109849095344543,
"adv/std_reasoning": 0.7206153869628906,
"adv/std_step_conf": 0.9289859533309937,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6485102109139605,
"calib/avg_num_step_conf": 8.4296875,
"calib/ece": 0.23387096774193544,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.6169354838709677,
"calib/gap": 0.16042986273853388,
"calib/mean_conf": 0.8108870967741936,
"calib/mu_c": 0.8775172413793104,
"calib/mu_w": 0.7170873786407765,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.23004032258064513,
"calib/std_conf": 0.25536821011381305,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8757729468599034,
"calib/step_q_c_n": 1242.0,
"calib/step_q_gap": 0.06768342720924836,
"calib/step_q_w": 0.808089519650655,
"calib/step_q_w_n": 916.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2603.0,
"completions/max_terminated_length": 2603.0,
"completions/mean_length": 777.57421875,
"completions/mean_terminated_length": 789.9166870117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 94.0,
"epoch": 0.18773333333333334,
"grad_norm": 0.03444397449493408,
"kl": 0.052021026611328125,
"learning_rate": 6.666666666666667e-07,
"loss": -0.1408,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018836980685591698,
"mask/share_reasoning": 0.8486654162406921,
"mask/share_step_conf": 0.11687257885932922,
"num_tokens": 55342944.0,
"reward": 0.8689711689949036,
"reward_std": 0.2312285155057907,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.6962413787841797,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.73466956615448,
"step": 176
},
{
"adv/mean_abs_final_conf": 0.5899331569671631,
"adv/mean_abs_reasoning": 0.37707242369651794,
"adv/mean_abs_step_conf": 0.7464255690574646,
"adv/ratio_final_to_reasoning": 1.5645088844841208,
"adv/ratio_step_to_reasoning": 1.9795283933524026,
"adv/std_final_conf": 0.8274363875389099,
"adv/std_reasoning": 0.6814361810684204,
"adv/std_step_conf": 0.9328826069831848,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.774943167466532,
"calib/avg_num_step_conf": 8.5546875,
"calib/ece": 0.2322352941176469,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.6235294117647059,
"calib/gap": 0.26361707501894405,
"calib/mean_conf": 0.8126274509803922,
"calib/mu_c": 0.9232432432432431,
"calib/mu_w": 0.6596261682242991,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2322352941176469,
"calib/std_conf": 0.24333741286811122,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8775059382422803,
"calib/step_q_c_n": 1263.0,
"calib/step_q_gap": 0.05978209789708078,
"calib/step_q_w": 0.8177238403451995,
"calib/step_q_w_n": 927.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1907.0,
"completions/max_terminated_length": 1907.0,
"completions/mean_length": 781.00390625,
"completions/mean_terminated_length": 784.0667114257812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 336.0,
"epoch": 0.1888,
"grad_norm": 0.05831204727292061,
"kl": 0.05739593505859375,
"learning_rate": 6.388888888888889e-07,
"loss": -0.0203,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.019267741590738297,
"mask/share_reasoning": 0.857624888420105,
"mask/share_step_conf": 0.11920113116502762,
"num_tokens": 55646713.0,
"reward": 0.8887894153594971,
"reward_std": 0.18345263600349426,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.7657734155654907,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.6977428197860718,
"step": 177
},
{
"adv/mean_abs_final_conf": 0.6337563991546631,
"adv/mean_abs_reasoning": 0.4424035847187042,
"adv/mean_abs_step_conf": 0.7219555377960205,
"adv/ratio_final_to_reasoning": 1.4325299817758659,
"adv/ratio_step_to_reasoning": 1.6318935079494559,
"adv/std_final_conf": 0.8438934683799744,
"adv/std_reasoning": 0.7206499576568604,
"adv/std_step_conf": 0.92996746301651,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.749377141077546,
"calib/avg_num_step_conf": 8.71875,
"calib/ece": 0.13142857142857134,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.636734693877551,
"calib/gap": 0.27502257863593904,
"calib/mean_conf": 0.8103673469387754,
"calib/mu_c": 0.8956804733727811,
"calib/mu_w": 0.6206578947368421,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.12599999999999992,
"calib/std_conf": 0.2704060962570209,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8815164220824597,
"calib/step_q_c_n": 1431.0,
"calib/step_q_gap": 0.06092965554063701,
"calib/step_q_w": 0.8205867665418227,
"calib/step_q_w_n": 801.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2906.0,
"completions/max_terminated_length": 2906.0,
"completions/mean_length": 802.984375,
"completions/mean_terminated_length": 822.2560424804688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 400.0,
"epoch": 0.18986666666666666,
"grad_norm": 0.11171025782823563,
"kl": 0.0521392822265625,
"learning_rate": 6.111111111111112e-07,
"loss": -0.0644,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.018252698704600334,
"mask/share_reasoning": 0.8460112810134888,
"mask/share_step_conf": 0.11229851096868515,
"num_tokens": 55958349.0,
"reward": 0.9219865798950195,
"reward_std": 0.21582040190696716,
"rewards/accuracy_reward_step": 0.66015625,
"rewards/final_brier_reward_step": 0.780997633934021,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7395378947257996,
"step": 178
},
{
"adv/mean_abs_final_conf": 0.6373348236083984,
"adv/mean_abs_reasoning": 0.5426758527755737,
"adv/mean_abs_step_conf": 0.7440859079360962,
"adv/ratio_final_to_reasoning": 1.1744300402324541,
"adv/ratio_step_to_reasoning": 1.3711424677003576,
"adv/std_final_conf": 0.8385090231895447,
"adv/std_reasoning": 0.7928795218467712,
"adv/std_step_conf": 0.9288709163665771,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.7027412280701755,
"calib/avg_num_step_conf": 8.47265625,
"calib/ece": 0.255481171548117,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.698744769874477,
"calib/gap": 0.18927339181286595,
"calib/mean_conf": 0.8504602510460251,
"calib/mu_c": 0.9256944444444445,
"calib/mu_w": 0.7364210526315785,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.25171548117154796,
"calib/std_conf": 0.22995587508686588,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8779867986798678,
"calib/step_q_c_n": 1212.0,
"calib/step_q_gap": 0.07686872135489387,
"calib/step_q_w": 0.801118077324974,
"calib/step_q_w_n": 957.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3064.0,
"completions/max_terminated_length": 3064.0,
"completions/mean_length": 798.48046875,
"completions/mean_terminated_length": 834.33056640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 408.0,
"epoch": 0.19093333333333334,
"grad_norm": 0.042912401258945465,
"kl": 0.0484619140625,
"learning_rate": 5.833333333333334e-07,
"loss": -0.1355,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.018003320321440697,
"mask/share_reasoning": 0.829845130443573,
"mask/share_step_conf": 0.10918280482292175,
"num_tokens": 56269024.0,
"reward": 0.8372169733047485,
"reward_std": 0.2432376891374588,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.6839832067489624,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.6920132040977478,
"step": 179
},
{
"adv/mean_abs_final_conf": 0.5640825033187866,
"adv/mean_abs_reasoning": 0.4570828080177307,
"adv/mean_abs_step_conf": 0.7516995668411255,
"adv/ratio_final_to_reasoning": 1.2340925832784884,
"adv/ratio_step_to_reasoning": 1.6445588275373646,
"adv/std_final_conf": 0.8109744787216187,
"adv/std_reasoning": 0.757621169090271,
"adv/std_step_conf": 0.9324401617050171,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6544264874141876,
"calib/avg_num_step_conf": 9.0859375,
"calib/ece": 0.207827868852459,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.6229508196721312,
"calib/gap": 0.16543192219679614,
"calib/mean_conf": 0.8147950819672132,
"calib/mu_c": 0.8771710526315788,
"calib/mu_w": 0.7117391304347827,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.19983606557377048,
"calib/std_conf": 0.2515435118154883,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8635598455598456,
"calib/step_q_c_n": 1295.0,
"calib/step_q_gap": 0.08801183392066025,
"calib/step_q_w": 0.7755480116391853,
"calib/step_q_w_n": 1031.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2856.0,
"completions/max_terminated_length": 2856.0,
"completions/mean_length": 850.9765625,
"completions/mean_terminated_length": 867.9282836914062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 390.0,
"epoch": 0.192,
"grad_norm": 0.027483295649290085,
"kl": 0.0508270263671875,
"learning_rate": 5.555555555555555e-07,
"loss": -0.0867,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017159832641482353,
"mask/share_reasoning": 0.8568102121353149,
"mask/share_step_conf": 0.10649865865707397,
"num_tokens": 56590730.0,
"reward": 0.8719885349273682,
"reward_std": 0.22523637115955353,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.7079362869262695,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7266656756401062,
"step": 180
},
{
"adv/mean_abs_final_conf": 0.6659194827079773,
"adv/mean_abs_reasoning": 0.5319613218307495,
"adv/mean_abs_step_conf": 0.7389317750930786,
"adv/ratio_final_to_reasoning": 1.251819362385614,
"adv/ratio_step_to_reasoning": 1.3890704920990091,
"adv/std_final_conf": 0.8759973049163818,
"adv/std_reasoning": 0.7927730679512024,
"adv/std_step_conf": 0.9336047768592834,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6733273462056579,
"calib/avg_num_step_conf": 8.6640625,
"calib/ece": 0.2990799999999999,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.628,
"calib/gap": 0.17728718968503476,
"calib/mean_conf": 0.81668,
"calib/mu_c": 0.9010687022900765,
"calib/mu_w": 0.7237815126050418,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2958799999999999,
"calib/std_conf": 0.2502538263443738,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8699122036874452,
"calib/step_q_c_n": 1139.0,
"calib/step_q_gap": 0.04246085985055925,
"calib/step_q_w": 0.8274513438368859,
"calib/step_q_w_n": 1079.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2591.0,
"completions/max_terminated_length": 2591.0,
"completions/mean_length": 791.1953125,
"completions/mean_terminated_length": 803.7540283203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 422.0,
"epoch": 0.19306666666666666,
"grad_norm": 0.05116986483335495,
"kl": 0.05887603759765625,
"learning_rate": 5.277777777777779e-07,
"loss": -0.0833,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.018377797678112984,
"mask/share_reasoning": 0.852816104888916,
"mask/share_step_conf": 0.11318107694387436,
"num_tokens": 56899540.0,
"reward": 0.8266412019729614,
"reward_std": 0.24478717148303986,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.6745378971099854,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.6810882091522217,
"step": 181
},
{
"adv/mean_abs_final_conf": 0.5257282257080078,
"adv/mean_abs_reasoning": 0.3534095287322998,
"adv/mean_abs_step_conf": 0.7751988172531128,
"adv/ratio_final_to_reasoning": 1.4875892780645306,
"adv/ratio_step_to_reasoning": 2.1934858973208655,
"adv/std_final_conf": 0.776384711265564,
"adv/std_reasoning": 0.6401922702789307,
"adv/std_step_conf": 0.9291584491729736,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.709341755319149,
"calib/avg_num_step_conf": 8.46484375,
"calib/ece": 0.23803149606299207,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.7047244094488189,
"calib/gap": 0.19531781914893642,
"calib/mean_conf": 0.8537795275590552,
"calib/mu_c": 0.9260625000000001,
"calib/mu_w": 0.7307446808510637,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2309448818897637,
"calib/std_conf": 0.23002714366998875,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8782753515914139,
"calib/step_q_c_n": 1351.0,
"calib/step_q_gap": 0.06087339080710019,
"calib/step_q_w": 0.8174019607843137,
"calib/step_q_w_n": 816.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2306.0,
"completions/max_terminated_length": 2306.0,
"completions/mean_length": 754.671875,
"completions/mean_terminated_length": 760.6141967773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 461.0,
"epoch": 0.19413333333333332,
"grad_norm": 0.0377509668469429,
"kl": 0.0584564208984375,
"learning_rate": 5.000000000000001e-07,
"loss": -0.0351,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.019371334463357925,
"mask/share_reasoning": 0.8548128604888916,
"mask/share_step_conf": 0.11800329387187958,
"num_tokens": 57198896.0,
"reward": 0.8838473558425903,
"reward_std": 0.1849229335784912,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.7490218877792358,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.6952352523803711,
"step": 182
},
{
"adv/mean_abs_final_conf": 0.6666151881217957,
"adv/mean_abs_reasoning": 0.5118650197982788,
"adv/mean_abs_step_conf": 0.7944930791854858,
"adv/ratio_final_to_reasoning": 1.3023261257127952,
"adv/ratio_step_to_reasoning": 1.5521534944868631,
"adv/std_final_conf": 0.8759217858314514,
"adv/std_reasoning": 0.7394871711730957,
"adv/std_step_conf": 0.9344072341918945,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7245745943806885,
"calib/avg_num_step_conf": 8.2578125,
"calib/ece": 0.2508906882591092,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.6072874493927125,
"calib/gap": 0.2473934837092734,
"calib/mean_conf": 0.7893522267206476,
"calib/mu_c": 0.9035338345864663,
"calib/mu_w": 0.6561403508771929,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2508906882591092,
"calib/std_conf": 0.27891124614075113,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8831876138433515,
"calib/step_q_c_n": 1098.0,
"calib/step_q_gap": 0.10113052722917837,
"calib/step_q_w": 0.7820570866141732,
"calib/step_q_w_n": 1016.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2034.0,
"completions/max_terminated_length": 2034.0,
"completions/mean_length": 830.46875,
"completions/mean_terminated_length": 850.4000244140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 351.0,
"epoch": 0.1952,
"grad_norm": 0.07475608587265015,
"kl": 0.05438232421875,
"learning_rate": 4.7222222222222226e-07,
"loss": -0.1073,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.0174814835190773,
"mask/share_reasoning": 0.8535329103469849,
"mask/share_step_conf": 0.10554810613393784,
"num_tokens": 57518176.0,
"reward": 0.8417513370513916,
"reward_std": 0.26017308235168457,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.7079120874404907,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.6787155866622925,
"step": 183
},
{
"adv/mean_abs_final_conf": 0.6304896473884583,
"adv/mean_abs_reasoning": 0.4839908480644226,
"adv/mean_abs_step_conf": 0.7729167938232422,
"adv/ratio_final_to_reasoning": 1.302689193214942,
"adv/ratio_step_to_reasoning": 1.5969657213856272,
"adv/std_final_conf": 0.8440559506416321,
"adv/std_reasoning": 0.7576839923858643,
"adv/std_step_conf": 0.9316712617874146,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6616161616161615,
"calib/avg_num_step_conf": 9.109375,
"calib/ece": 0.194320987654321,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.7037037037037037,
"calib/gap": 0.17838461538461503,
"calib/mean_conf": 0.8474074074074074,
"calib/mu_c": 0.9046666666666665,
"calib/mu_w": 0.7262820512820515,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.18135802469135803,
"calib/std_conf": 0.24131725840558851,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8664727540500737,
"calib/step_q_c_n": 1358.0,
"calib/step_q_gap": 0.07795119347512514,
"calib/step_q_w": 0.7885215605749486,
"calib/step_q_w_n": 974.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2701.0,
"completions/max_terminated_length": 2701.0,
"completions/mean_length": 785.9140625,
"completions/mean_terminated_length": 817.8617553710938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 444.0,
"epoch": 0.19626666666666667,
"grad_norm": 0.054167453199625015,
"kl": 0.053737640380859375,
"learning_rate": 4.444444444444445e-07,
"loss": -0.1561,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.018033988773822784,
"mask/share_reasoning": 0.8329838514328003,
"mask/share_step_conf": 0.10991968959569931,
"num_tokens": 57824650.0,
"reward": 0.894181489944458,
"reward_std": 0.2465510070323944,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.7339491844177246,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.735663652420044,
"step": 184
},
{
"adv/mean_abs_final_conf": 0.5086137056350708,
"adv/mean_abs_reasoning": 0.39162346720695496,
"adv/mean_abs_step_conf": 0.7781357169151306,
"adv/ratio_final_to_reasoning": 1.2987314301221176,
"adv/ratio_step_to_reasoning": 1.9869486434629868,
"adv/std_final_conf": 0.7587011456489563,
"adv/std_reasoning": 0.681826114654541,
"adv/std_step_conf": 0.9302533864974976,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.7149665670792431,
"calib/avg_num_step_conf": 9.171875,
"calib/ece": 0.26950207468879667,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.7053941908713693,
"calib/gap": 0.21292360221937678,
"calib/mean_conf": 0.843941908713693,
"calib/mu_c": 0.9314084507042251,
"calib/mu_w": 0.7184848484848483,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2621161825726141,
"calib/std_conf": 0.24912923196358036,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8771261487050961,
"calib/step_q_c_n": 1197.0,
"calib/step_q_gap": 0.14506706964341065,
"calib/step_q_w": 0.7320590790616854,
"calib/step_q_w_n": 1151.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2567.0,
"completions/max_terminated_length": 2567.0,
"completions/mean_length": 786.78125,
"completions/mean_terminated_length": 828.8723754882812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 464.0,
"epoch": 0.19733333333333333,
"grad_norm": 0.045686863362789154,
"kl": 0.04729461669921875,
"learning_rate": 4.1666666666666667e-07,
"loss": -0.2002,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01783943921327591,
"mask/share_reasoning": 0.8219054937362671,
"mask/share_step_conf": 0.10947383940219879,
"num_tokens": 58132986.0,
"reward": 0.833606481552124,
"reward_std": 0.2180936634540558,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.6881363391876221,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.6806391477584839,
"step": 185
},
{
"adv/mean_abs_final_conf": 0.6431267857551575,
"adv/mean_abs_reasoning": 0.5003708600997925,
"adv/mean_abs_step_conf": 0.787868857383728,
"adv/ratio_final_to_reasoning": 1.2853002383609913,
"adv/ratio_step_to_reasoning": 1.5745698245229505,
"adv/std_final_conf": 0.8440446853637695,
"adv/std_reasoning": 0.7578926682472229,
"adv/std_step_conf": 0.9339975118637085,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6725707782045811,
"calib/avg_num_step_conf": 8.6796875,
"calib/ece": 0.2716182572614107,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.6597510373443983,
"calib/gap": 0.17574619433774386,
"calib/mean_conf": 0.8106224066390042,
"calib/mu_c": 0.8828169014084508,
"calib/mu_w": 0.707070707070707,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.24651452282157665,
"calib/std_conf": 0.2773631518500008,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8730307813733229,
"calib/step_q_c_n": 1267.0,
"calib/step_q_gap": 0.09363811121625487,
"calib/step_q_w": 0.779392670157068,
"calib/step_q_w_n": 955.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 3024.0,
"completions/max_terminated_length": 3024.0,
"completions/mean_length": 814.22265625,
"completions/mean_terminated_length": 840.4878540039062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 357.0,
"epoch": 0.1984,
"grad_norm": 0.05286160111427307,
"kl": 0.05419921875,
"learning_rate": 3.8888888888888895e-07,
"loss": -0.1288,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01790950819849968,
"mask/share_reasoning": 0.8364917039871216,
"mask/share_step_conf": 0.11434879899024963,
"num_tokens": 58446467.0,
"reward": 0.8335369825363159,
"reward_std": 0.25954774022102356,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.6725648641586304,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.6960716247558594,
"step": 186
},
{
"adv/mean_abs_final_conf": 0.6640216708183289,
"adv/mean_abs_reasoning": 0.5353648662567139,
"adv/mean_abs_step_conf": 0.779157280921936,
"adv/ratio_final_to_reasoning": 1.240316114617657,
"adv/ratio_step_to_reasoning": 1.4553761930061373,
"adv/std_final_conf": 0.8601834774017334,
"adv/std_reasoning": 0.7929431796073914,
"adv/std_step_conf": 0.9341410398483276,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7007133152173914,
"calib/avg_num_step_conf": 9.73828125,
"calib/ece": 0.28650205761316866,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.6090534979423868,
"calib/gap": 0.19911209239130456,
"calib/mean_conf": 0.8013168724279836,
"calib/mu_c": 0.8955468750000001,
"calib/mu_w": 0.6964347826086955,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.28053497942386824,
"calib/std_conf": 0.25901664786592343,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8671660649819495,
"calib/step_q_c_n": 1108.0,
"calib/step_q_gap": 0.08116606498194934,
"calib/step_q_w": 0.7860000000000001,
"calib/step_q_w_n": 1385.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2983.0,
"completions/max_terminated_length": 2983.0,
"completions/mean_length": 866.58984375,
"completions/mean_terminated_length": 887.3880615234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 468.0,
"epoch": 0.19946666666666665,
"grad_norm": 0.08439423888921738,
"kl": 0.05120849609375,
"learning_rate": 3.611111111111111e-07,
"loss": -0.0256,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01723765954375267,
"mask/share_reasoning": 0.8504993915557861,
"mask/share_step_conf": 0.1088254451751709,
"num_tokens": 58769858.0,
"reward": 0.8159067034721375,
"reward_std": 0.25626736879348755,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.6715812683105469,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6703883409500122,
"step": 187
},
{
"adv/mean_abs_final_conf": 0.6788389086723328,
"adv/mean_abs_reasoning": 0.39350107312202454,
"adv/mean_abs_step_conf": 0.772308349609375,
"adv/ratio_final_to_reasoning": 1.7251259400297114,
"adv/ratio_step_to_reasoning": 1.9626588143252215,
"adv/std_final_conf": 0.8910698890686035,
"adv/std_reasoning": 0.6816974878311157,
"adv/std_step_conf": 0.9345043897628784,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.7079327778964609,
"calib/avg_num_step_conf": 8.54296875,
"calib/ece": 0.19024590163934435,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.6147540983606558,
"calib/gap": 0.23190201523890885,
"calib/mean_conf": 0.7911475409836065,
"calib/mu_c": 0.8795364238410595,
"calib/mu_w": 0.6476344086021506,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.18127049180327878,
"calib/std_conf": 0.2817820827626987,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8599916387959866,
"calib/step_q_c_n": 1196.0,
"calib/step_q_gap": 0.05549113425511876,
"calib/step_q_w": 0.8045005045408679,
"calib/step_q_w_n": 991.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2889.0,
"completions/max_terminated_length": 2889.0,
"completions/mean_length": 838.92578125,
"completions/mean_terminated_length": 862.510009765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 374.0,
"epoch": 0.20053333333333334,
"grad_norm": 0.058961011469364166,
"kl": 0.05727386474609375,
"learning_rate": 3.3333333333333335e-07,
"loss": -0.1537,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.017490971833467484,
"mask/share_reasoning": 0.8490737080574036,
"mask/share_step_conf": 0.10609160363674164,
"num_tokens": 59088695.0,
"reward": 0.8640185594558716,
"reward_std": 0.23061273992061615,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.7286055088043213,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6908378601074219,
"step": 188
},
{
"adv/mean_abs_final_conf": 0.6422284245491028,
"adv/mean_abs_reasoning": 0.4257098436355591,
"adv/mean_abs_step_conf": 0.7781589031219482,
"adv/ratio_final_to_reasoning": 1.5086060004261976,
"adv/ratio_step_to_reasoning": 1.8279091140493173,
"adv/std_final_conf": 0.8458375334739685,
"adv/std_reasoning": 0.6818082332611084,
"adv/std_step_conf": 0.9307503700256348,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.7350993377483444,
"calib/avg_num_step_conf": 8.56640625,
"calib/ece": 0.18626016260162592,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.5447154471544715,
"calib/gap": 0.25046357615894055,
"calib/mean_conf": 0.753739837398374,
"calib/mu_c": 0.8504635761589405,
"calib/mu_w": 0.6,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.16308943089430886,
"calib/std_conf": 0.2916853142674879,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8517623918174665,
"calib/step_q_c_n": 1271.0,
"calib/step_q_gap": 0.07708777142701095,
"calib/step_q_w": 0.7746746203904555,
"calib/step_q_w_n": 922.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3007.0,
"completions/max_terminated_length": 3007.0,
"completions/mean_length": 779.9453125,
"completions/mean_terminated_length": 805.1047973632812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 376.0,
"epoch": 0.2016,
"grad_norm": 0.067589171230793,
"kl": 0.059906005859375,
"learning_rate": 3.055555555555556e-07,
"loss": -0.162,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.018148701637983322,
"mask/share_reasoning": 0.8403629064559937,
"mask/share_step_conf": 0.11023841053247452,
"num_tokens": 59396129.0,
"reward": 0.8732390403747559,
"reward_std": 0.2376517951488495,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.742780864238739,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6943222284317017,
"step": 189
},
{
"adv/mean_abs_final_conf": 0.6645102500915527,
"adv/mean_abs_reasoning": 0.48422133922576904,
"adv/mean_abs_step_conf": 0.7489802837371826,
"adv/ratio_final_to_reasoning": 1.3723274797307594,
"adv/ratio_step_to_reasoning": 1.5467725667248409,
"adv/std_final_conf": 0.8758800625801086,
"adv/std_reasoning": 0.7576428055763245,
"adv/std_step_conf": 0.9346054196357727,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.652066647597254,
"calib/avg_num_step_conf": 8.93359375,
"calib/ece": 0.20122950819672128,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.5901639344262295,
"calib/gap": 0.17271453089244837,
"calib/mean_conf": 0.7770491803278687,
"calib/mu_c": 0.8421710526315789,
"calib/mu_w": 0.6694565217391305,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1776639344262295,
"calib/std_conf": 0.29038185296625374,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8654778554778556,
"calib/step_q_c_n": 1287.0,
"calib/step_q_gap": 0.06381785547785568,
"calib/step_q_w": 0.8016599999999999,
"calib/step_q_w_n": 1000.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2796.0,
"completions/max_terminated_length": 2796.0,
"completions/mean_length": 830.8671875,
"completions/mean_terminated_length": 844.0556030273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 428.0,
"epoch": 0.20266666666666666,
"grad_norm": 0.06410624831914902,
"kl": 0.05812835693359375,
"learning_rate": 2.7777777777777776e-07,
"loss": -0.0981,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01752610132098198,
"mask/share_reasoning": 0.8577367663383484,
"mask/share_step_conf": 0.10911212116479874,
"num_tokens": 59714439.0,
"reward": 0.8567715883255005,
"reward_std": 0.2456488162279129,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.7035819888114929,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6998047828674316,
"step": 190
},
{
"adv/mean_abs_final_conf": 0.6169793605804443,
"adv/mean_abs_reasoning": 0.37427815794944763,
"adv/mean_abs_step_conf": 0.7451186180114746,
"adv/ratio_final_to_reasoning": 1.6484514190212978,
"adv/ratio_step_to_reasoning": 1.9908151255572735,
"adv/std_final_conf": 0.8437768816947937,
"adv/std_reasoning": 0.6816237568855286,
"adv/std_step_conf": 0.933275043964386,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.7011305241521069,
"calib/avg_num_step_conf": 9.58203125,
"calib/ece": 0.27098360655737697,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.6352459016393442,
"calib/gap": 0.19729359369647137,
"calib/mean_conf": 0.8028688524590163,
"calib/mu_c": 0.8877697841726618,
"calib/mu_w": 0.6904761904761905,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2520901639344262,
"calib/std_conf": 0.2717203224272977,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8665474209650583,
"calib/step_q_c_n": 1202.0,
"calib/step_q_gap": 0.0922228805973524,
"calib/step_q_w": 0.7743245403677059,
"calib/step_q_w_n": 1251.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2811.0,
"completions/max_terminated_length": 2811.0,
"completions/mean_length": 791.0625,
"completions/mean_terminated_length": 819.8866577148438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 399.0,
"epoch": 0.20373333333333332,
"grad_norm": 0.03953538462519646,
"kl": 0.057159423828125,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.1979,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01807655580341816,
"mask/share_reasoning": 0.8292906284332275,
"mask/share_step_conf": 0.11747653782367706,
"num_tokens": 60021119.0,
"reward": 0.8270571231842041,
"reward_std": 0.20282432436943054,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.689464807510376,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.6654306650161743,
"step": 191
},
{
"adv/mean_abs_final_conf": 0.5836527347564697,
"adv/mean_abs_reasoning": 0.39761674404144287,
"adv/mean_abs_step_conf": 0.7857372164726257,
"adv/ratio_final_to_reasoning": 1.467877657324302,
"adv/ratio_step_to_reasoning": 1.97611702285538,
"adv/std_final_conf": 0.8277207612991333,
"adv/std_reasoning": 0.6614072322845459,
"adv/std_step_conf": 0.9336384534835815,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.756342062193126,
"calib/avg_num_step_conf": 9.07421875,
"calib/ece": 0.16856000000000007,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.58,
"calib/gap": 0.27840698308783396,
"calib/mean_conf": 0.7709600000000001,
"calib/mu_c": 0.8756410256410254,
"calib/mu_w": 0.5972340425531915,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.15776000000000007,
"calib/std_conf": 0.2851972622589495,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8703448275862069,
"calib/step_q_c_n": 1392.0,
"calib/step_q_gap": 0.11804622393421982,
"calib/step_q_w": 0.7522986036519871,
"calib/step_q_w_n": 931.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2954.0,
"completions/max_terminated_length": 2954.0,
"completions/mean_length": 818.40234375,
"completions/mean_terminated_length": 831.3928833007812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 356.0,
"epoch": 0.2048,
"grad_norm": 0.05431187525391579,
"kl": 0.05841064453125,
"learning_rate": 2.2222222222222224e-07,
"loss": -0.0675,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.018427275121212006,
"mask/share_reasoning": 0.8483537435531616,
"mask/share_step_conf": 0.11759395897388458,
"num_tokens": 60335606.0,
"reward": 0.9095621109008789,
"reward_std": 0.23026014864444733,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.7744953036308289,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7274413704872131,
"step": 192
},
{
"adv/mean_abs_final_conf": 0.7208178639411926,
"adv/mean_abs_reasoning": 0.5967491269111633,
"adv/mean_abs_step_conf": 0.7681723833084106,
"adv/ratio_final_to_reasoning": 1.2079076976152814,
"adv/ratio_step_to_reasoning": 1.2872618470085615,
"adv/std_final_conf": 0.9065069556236267,
"adv/std_reasoning": 0.8100512623786926,
"adv/std_step_conf": 0.9342743754386902,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.7010898109243697,
"calib/avg_num_step_conf": 8.7265625,
"calib/ece": 0.24729838709677424,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.5887096774193549,
"calib/gap": 0.23204306722689072,
"calib/mean_conf": 0.7710887096774194,
"calib/mu_c": 0.8758823529411766,
"calib/mu_w": 0.6438392857142858,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23500000000000004,
"calib/std_conf": 0.29827870470246487,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8694429708222812,
"calib/step_q_c_n": 1131.0,
"calib/step_q_gap": 0.10144659729553596,
"calib/step_q_w": 0.7679963735267452,
"calib/step_q_w_n": 1103.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2984.0,
"completions/max_terminated_length": 2984.0,
"completions/mean_length": 845.15234375,
"completions/mean_terminated_length": 855.1739501953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 463.0,
"epoch": 0.20586666666666667,
"grad_norm": 0.04506753757596016,
"kl": 0.05387115478515625,
"learning_rate": 1.9444444444444447e-07,
"loss": -0.045,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.017770592123270035,
"mask/share_reasoning": 0.8608217239379883,
"mask/share_step_conf": 0.10968892276287079,
"num_tokens": 60657677.0,
"reward": 0.8366061449050903,
"reward_std": 0.26884931325912476,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.6995316743850708,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.6752431988716125,
"step": 193
},
{
"adv/mean_abs_final_conf": 0.6042905449867249,
"adv/mean_abs_reasoning": 0.4425891637802124,
"adv/mean_abs_step_conf": 0.7328290939331055,
"adv/ratio_final_to_reasoning": 1.36535323148312,
"adv/ratio_step_to_reasoning": 1.655777307500969,
"adv/std_final_conf": 0.8594117760658264,
"adv/std_reasoning": 0.720699667930603,
"adv/std_step_conf": 0.9311245679855347,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.699829293274155,
"calib/avg_num_step_conf": 9.1015625,
"calib/ece": 0.20357723577235765,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.5853658536585366,
"calib/gap": 0.23855786958006153,
"calib/mean_conf": 0.7653658536585366,
"calib/mu_c": 0.8633103448275863,
"calib/mu_w": 0.6247524752475248,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1897560975609755,
"calib/std_conf": 0.30406672535509977,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8639723801787165,
"calib/step_q_c_n": 1231.0,
"calib/step_q_gap": 0.1521434447829021,
"calib/step_q_w": 0.7118289353958144,
"calib/step_q_w_n": 1099.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2497.0,
"completions/max_terminated_length": 2497.0,
"completions/mean_length": 770.81640625,
"completions/mean_terminated_length": 795.6814575195312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 436.0,
"epoch": 0.20693333333333333,
"grad_norm": 0.04400273412466049,
"kl": 0.0572967529296875,
"learning_rate": 1.6666666666666668e-07,
"loss": -0.1635,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.018388589844107628,
"mask/share_reasoning": 0.8372006416320801,
"mask/share_step_conf": 0.11316081881523132,
"num_tokens": 60960950.0,
"reward": 0.8722167611122131,
"reward_std": 0.22317549586296082,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.7207523584365845,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.718212366104126,
"step": 194
},
{
"adv/mean_abs_final_conf": 0.6190444231033325,
"adv/mean_abs_reasoning": 0.3682692050933838,
"adv/mean_abs_step_conf": 0.7552182674407959,
"adv/ratio_final_to_reasoning": 1.6809562530387476,
"adv/ratio_step_to_reasoning": 2.0507233757144903,
"adv/std_final_conf": 0.8602829575538635,
"adv/std_reasoning": 0.6612259745597839,
"adv/std_step_conf": 0.9333444237709045,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.658011583011583,
"calib/avg_num_step_conf": 8.7578125,
"calib/ece": 0.22916996047430815,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.5889328063241107,
"calib/gap": 0.18003217503217495,
"calib/mean_conf": 0.7805533596837946,
"calib/mu_c": 0.8552702702702701,
"calib/mu_w": 0.6752380952380952,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.21237154150197615,
"calib/std_conf": 0.2846113953485178,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8665965732087227,
"calib/step_q_c_n": 1284.0,
"calib/step_q_gap": 0.03806838949264757,
"calib/step_q_w": 0.8285281837160752,
"calib/step_q_w_n": 958.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2330.0,
"completions/max_terminated_length": 2330.0,
"completions/mean_length": 810.71484375,
"completions/mean_terminated_length": 817.0984497070312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 402.0,
"epoch": 0.208,
"grad_norm": 0.049184273928403854,
"kl": 0.06110382080078125,
"learning_rate": 1.3888888888888888e-07,
"loss": -0.0271,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.01838977076113224,
"mask/share_reasoning": 0.858869194984436,
"mask/share_step_conf": 0.11492855846881866,
"num_tokens": 61274477.0,
"reward": 0.8680363893508911,
"reward_std": 0.18345674872398376,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.7168843746185303,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7059070467948914,
"step": 195
},
{
"adv/mean_abs_final_conf": 0.4948877692222595,
"adv/mean_abs_reasoning": 0.27376601099967957,
"adv/mean_abs_step_conf": 0.7344076633453369,
"adv/ratio_final_to_reasoning": 1.807703474274017,
"adv/ratio_step_to_reasoning": 2.682610820326401,
"adv/std_final_conf": 0.7586008906364441,
"adv/std_reasoning": 0.572577714920044,
"adv/std_step_conf": 0.9284154772758484,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6725761217948718,
"calib/avg_num_step_conf": 9.08203125,
"calib/ece": 0.24845238095238098,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.7698412698412699,
"calib/gap": 0.21816506410256375,
"calib/mean_conf": 0.8559920634920636,
"calib/mu_c": 0.9391025641025639,
"calib/mu_w": 0.7209375000000001,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.24269841269841275,
"calib/std_conf": 0.2712467685721276,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.873566384180791,
"calib/step_q_c_n": 1416.0,
"calib/step_q_gap": 0.08908893643601656,
"calib/step_q_w": 0.7844774477447745,
"calib/step_q_w_n": 909.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2989.0,
"completions/max_terminated_length": 2989.0,
"completions/mean_length": 749.15625,
"completions/mean_terminated_length": 752.0941772460938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 406.0,
"epoch": 0.20906666666666668,
"grad_norm": 0.049532778561115265,
"kl": 0.05275726318359375,
"learning_rate": 1.1111111111111112e-07,
"loss": 0.0266,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.01997404731810093,
"mask/share_reasoning": 0.8479194641113281,
"mask/share_step_conf": 0.12820029258728027,
"num_tokens": 61568805.0,
"reward": 0.8798357844352722,
"reward_std": 0.17858222126960754,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.7258324027061462,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7150891423225403,
"step": 196
},
{
"adv/mean_abs_final_conf": 0.6538981199264526,
"adv/mean_abs_reasoning": 0.5470722317695618,
"adv/mean_abs_step_conf": 0.755987823009491,
"adv/ratio_final_to_reasoning": 1.1952683429231117,
"adv/ratio_step_to_reasoning": 1.3818793554265587,
"adv/std_final_conf": 0.8599660396575928,
"adv/std_reasoning": 0.7929037809371948,
"adv/std_step_conf": 0.9324612617492676,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7064814814814814,
"calib/avg_num_step_conf": 9.3671875,
"calib/ece": 0.24326612903225808,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.5443548387096774,
"calib/gap": 0.21345238095238106,
"calib/mean_conf": 0.7538306451612904,
"calib/mu_c": 0.8467857142857144,
"calib/mu_w": 0.6333333333333333,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.21629032258064515,
"calib/std_conf": 0.28916635522888545,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8520353982300884,
"calib/step_q_c_n": 1243.0,
"calib/step_q_gap": 0.1004076926023828,
"calib/step_q_w": 0.7516277056277056,
"calib/step_q_w_n": 1155.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2801.0,
"completions/max_terminated_length": 2801.0,
"completions/mean_length": 814.2109375,
"completions/mean_terminated_length": 827.1349487304688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 475.0,
"epoch": 0.21013333333333334,
"grad_norm": 0.08905098587274551,
"kl": 0.06281280517578125,
"learning_rate": 8.333333333333334e-08,
"loss": 0.0467,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018287431448698044,
"mask/share_reasoning": 0.8483529686927795,
"mask/share_step_conf": 0.11773453652858734,
"num_tokens": 61882299.0,
"reward": 0.8597941398620605,
"reward_std": 0.20589497685432434,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.7164624929428101,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.700782060623169,
"step": 197
},
{
"adv/mean_abs_final_conf": 0.58735591173172,
"adv/mean_abs_reasoning": 0.4391744136810303,
"adv/mean_abs_step_conf": 0.7891101837158203,
"adv/ratio_final_to_reasoning": 1.337409223840424,
"adv/ratio_step_to_reasoning": 1.7968036368551887,
"adv/std_final_conf": 0.8107753396034241,
"adv/std_reasoning": 0.7014614343643188,
"adv/std_step_conf": 0.9324103593826294,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.666535770384121,
"calib/avg_num_step_conf": 9.44921875,
"calib/ece": 0.17724279835390927,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.588477366255144,
"calib/gap": 0.21069041839211777,
"calib/mean_conf": 0.7669547325102881,
"calib/mu_c": 0.8484563758389263,
"calib/mu_w": 0.6377659574468085,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.1655144032921809,
"calib/std_conf": 0.29970456304725857,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.870100154083205,
"calib/step_q_c_n": 1298.0,
"calib/step_q_gap": 0.11071567593869103,
"calib/step_q_w": 0.7593844781445139,
"calib/step_q_w_n": 1121.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2972.0,
"completions/max_terminated_length": 2972.0,
"completions/mean_length": 781.1953125,
"completions/mean_terminated_length": 812.951171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 392.0,
"epoch": 0.2112,
"grad_norm": 0.05872613564133644,
"kl": 0.05881500244140625,
"learning_rate": 5.555555555555556e-08,
"loss": -0.1555,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.018448233604431152,
"mask/share_reasoning": 0.8231604099273682,
"mask/share_step_conf": 0.11932889372110367,
"num_tokens": 62187669.0,
"reward": 0.8654526472091675,
"reward_std": 0.21482285857200623,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.7112331986427307,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.7134221196174622,
"step": 198
},
{
"adv/mean_abs_final_conf": 0.6843090057373047,
"adv/mean_abs_reasoning": 0.6271142959594727,
"adv/mean_abs_step_conf": 0.777929961681366,
"adv/ratio_final_to_reasoning": 1.091203007404456,
"adv/ratio_step_to_reasoning": 1.2404915127810128,
"adv/std_final_conf": 0.876064121723175,
"adv/std_reasoning": 0.8429972529411316,
"adv/std_step_conf": 0.9352033734321594,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6885495666305526,
"calib/avg_num_step_conf": 8.58984375,
"calib/ece": 0.24605691056910572,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.6056910569105691,
"calib/gap": 0.20585319609967523,
"calib/mean_conf": 0.7724796747967478,
"calib/mu_c": 0.8595070422535213,
"calib/mu_w": 0.653653846153846,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.22065040650406506,
"calib/std_conf": 0.2989819270061604,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8696669442131559,
"calib/step_q_c_n": 1201.0,
"calib/step_q_gap": 0.09017796625724406,
"calib/step_q_w": 0.7794889779559119,
"calib/step_q_w_n": 998.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2965.0,
"completions/max_terminated_length": 2965.0,
"completions/mean_length": 834.5078125,
"completions/mean_terminated_length": 847.7540283203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 399.0,
"epoch": 0.21226666666666666,
"grad_norm": 0.03624487668275833,
"kl": 0.05606842041015625,
"learning_rate": 2.777777777777778e-08,
"loss": -0.074,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.018301576375961304,
"mask/share_reasoning": 0.8528690934181213,
"mask/share_step_conf": 0.11320433020591736,
"num_tokens": 62505503.0,
"reward": 0.848616361618042,
"reward_std": 0.28091567754745483,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.7004519701004028,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.693655788898468,
"step": 199
},
{
"adv/mean_abs_final_conf": 0.5117772817611694,
"adv/mean_abs_reasoning": 0.33070600032806396,
"adv/mean_abs_step_conf": 0.7476848363876343,
"adv/ratio_final_to_reasoning": 1.5475294710512684,
"adv/ratio_step_to_reasoning": 2.2608747214925726,
"adv/std_final_conf": 0.7761644124984741,
"adv/std_reasoning": 0.6402080655097961,
"adv/std_step_conf": 0.925877034664154,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7651164421997755,
"calib/avg_num_step_conf": 8.72265625,
"calib/ece": 0.16052,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.624,
"calib/gap": 0.2816526374859708,
"calib/mean_conf": 0.8009200000000001,
"calib/mu_c": 0.9000617283950617,
"calib/mu_w": 0.6184090909090909,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.15672,
"calib/std_conf": 0.2658938765748471,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8784808259587019,
"calib/step_q_c_n": 1356.0,
"calib/step_q_gap": 0.06750021022324004,
"calib/step_q_w": 0.8109806157354619,
"calib/step_q_w_n": 877.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2239.0,
"completions/max_terminated_length": 2239.0,
"completions/mean_length": 788.03515625,
"completions/mean_terminated_length": 803.7330932617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 447.0,
"epoch": 0.21333333333333335,
"grad_norm": 0.04874600097537041,
"kl": 0.0551300048828125,
"learning_rate": 0.0,
"loss": -0.1078,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.018403755500912666,
"mask/share_reasoning": 0.850422739982605,
"mask/share_step_conf": 0.11164219677448273,
"num_tokens": 62815288.0,
"reward": 0.9284921288490295,
"reward_std": 0.15986087918281555,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/final_brier_reward_step": 0.7874097228050232,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.747699499130249,
"step": 200
},
{
"epoch": 0.21333333333333335,
"step": 200,
"total_flos": 0.0,
"train_loss": 0.2748976366652641,
"train_runtime": 22212.2891,
"train_samples_per_second": 2.305,
"train_steps_per_second": 0.009
}
],
"logging_steps": 1,
"max_steps": 200,
"num_input_tokens_seen": 62815288,
"num_train_epochs": 1,
"save_steps": 25,
"stateful_callbacks": {
"TrainerControl": {
"args": {
"should_epoch_stop": false,
"should_evaluate": false,
"should_log": false,
"should_save": true,
"should_training_stop": true
},
"attributes": {}
}
},
"total_flos": 0.0,
"train_batch_size": 8,
"trial_name": null,
"trial_params": null
}
Reference in New Issue View Git Blame Copy Permalink