zhaohq/PureRL-1.5B-v7-s2-l2-kl-w3-b0
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
main
PureRL-1.5B-v7-s2-l2-kl-w3-b0/trainer_state.json
ModelHub XC 66cd3c3000 初始化项目，由ModelHub XC社区提供模型 
Model: zhaohq/PureRL-1.5B-v7-s2-l2-kl-w3-b0
Source: Original Platform
2026-06-04 14:57:53 +08:00

12043 lines
496 KiB
JSON
Raw Permalink 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": 0.773959219455719,
"adv/mean_abs_reasoning": 0.47714588046073914,
"adv/mean_abs_step_conf": 0.7502421140670776,
"adv/ratio_final_to_reasoning": 1.622059942565935,
"adv/ratio_step_to_reasoning": 1.5723537492194897,
"adv/std_final_conf": 0.9294352531433105,
"adv/std_reasoning": 0.7393431663513184,
"adv/std_step_conf": 0.9357826709747314,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.38076182006817844,
"calib/avg_num_step_conf": 5.23046875,
"calib/ece": 0.2003187250996017,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.2948207171314741,
"calib/gap": -0.026059730250481805,
"calib/mean_conf": 0.8737051792828686,
"calib/mu_c": 0.865606936416185,
"calib/mu_w": 0.8916666666666668,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.19239043824701207,
"calib/std_conf": 0.09027744273295583,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7959393232205367,
"calib/step_q_c_n": 857.0,
"calib/step_q_gap": -0.006446568895645877,
"calib/step_q_w": 0.8023858921161826,
"calib/step_q_w_n": 482.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2492.0,
"completions/max_terminated_length": 2492.0,
"completions/mean_length": 474.94921875,
"completions/mean_terminated_length": 478.68896484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 138.0,
"epoch": 0.0010666666666666667,
"grad_norm": 0.04299849271774292,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.0136,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03466901555657387,
"mask/share_reasoning": 0.8340686559677124,
"mask/share_step_conf": 0.12344987690448761,
"num_tokens": 229171.0,
"reward": 1.264374852180481,
"reward_std": 0.26098379492759705,
"rewards/accuracy_reward_step": 0.67578125,
"rewards/final_brier_reward_step": 0.7142800688743591,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7420004606246948,
"step": 1
},
{
"adv/mean_abs_final_conf": 0.7672724723815918,
"adv/mean_abs_reasoning": 0.5104547739028931,
"adv/mean_abs_step_conf": 0.770934522151947,
"adv/ratio_final_to_reasoning": 1.503115479781084,
"adv/ratio_step_to_reasoning": 1.5102895722914849,
"adv/std_final_conf": 0.9330522418022156,
"adv/std_reasoning": 0.7575037479400635,
"adv/std_step_conf": 0.9358851313591003,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.44343065693430656,
"calib/avg_num_step_conf": 5.05859375,
"calib/ece": 0.3349411764705883,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.2823529411764706,
"calib/gap": 0.002352468143016151,
"calib/mean_conf": 0.8721960784313726,
"calib/mu_c": 0.8732846715328467,
"calib/mu_w": 0.8709322033898306,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3349411764705883,
"calib/std_conf": 0.07627016470309335,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7954391371340525,
"calib/step_q_c_n": 649.0,
"calib/step_q_gap": 0.011011892552009073,
"calib/step_q_w": 0.7844272445820434,
"calib/step_q_w_n": 646.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1966.0,
"completions/max_terminated_length": 1966.0,
"completions/mean_length": 492.9765625,
"completions/mean_terminated_length": 494.9098205566406,
"completions/min_length": 0.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.0021333333333333334,
"grad_norm": 0.04045659676194191,
"learning_rate": 5.000000000000001e-07,
"loss": -0.0158,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03364308178424835,
"mask/share_reasoning": 0.8523939251899719,
"mask/share_step_conf": 0.11005672812461853,
"num_tokens": 458661.0,
"reward": 1.198354721069336,
"reward_std": 0.24474793672561646,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.6320762038230896,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7291916012763977,
"step": 2
},
{
"adv/mean_abs_final_conf": 0.7855177521705627,
"adv/mean_abs_reasoning": 0.4552287757396698,
"adv/mean_abs_step_conf": 0.7503337860107422,
"adv/ratio_final_to_reasoning": 1.7255450314937346,
"adv/ratio_step_to_reasoning": 1.6482564943122864,
"adv/std_final_conf": 0.9304200410842896,
"adv/std_reasoning": 0.7206604480743408,
"adv/std_step_conf": 0.9357056617736816,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.4619900301718483,
"calib/avg_num_step_conf": 5.15625,
"calib/ece": 0.2755335968379446,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.3241106719367589,
"calib/gap": -0.007893217893217952,
"calib/mean_conf": 0.8825691699604744,
"calib/mu_c": 0.8794805194805194,
"calib/mu_w": 0.8873737373737374,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2747035573122529,
"calib/std_conf": 0.044117613178770984,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.804808972503618,
"calib/step_q_c_n": 691.0,
"calib/step_q_gap": 0.04946716010298213,
"calib/step_q_w": 0.7553418124006359,
"calib/step_q_w_n": 629.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2973.0,
"completions/max_terminated_length": 2973.0,
"completions/mean_length": 504.58203125,
"completions/mean_terminated_length": 508.55511474609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 183.0,
"epoch": 0.0032,
"grad_norm": 0.04432070627808571,
"learning_rate": 7.5e-07,
"loss": 0.0416,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03285394608974457,
"mask/share_reasoning": 0.8470344543457031,
"mask/share_step_conf": 0.11229914426803589,
"num_tokens": 693090.0,
"reward": 1.2373484373092651,
"reward_std": 0.2528243064880371,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.671227753162384,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7435314059257507,
"step": 3
},
{
"adv/mean_abs_final_conf": 0.744471549987793,
"adv/mean_abs_reasoning": 0.43164411187171936,
"adv/mean_abs_step_conf": 0.7570846676826477,
"adv/ratio_final_to_reasoning": 1.7247346355764561,
"adv/ratio_step_to_reasoning": 1.7539557400649688,
"adv/std_final_conf": 0.9300883412361145,
"adv/std_reasoning": 0.7013282775878906,
"adv/std_step_conf": 0.9355937838554382,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.4211795373085696,
"calib/avg_num_step_conf": 5.05078125,
"calib/ece": 0.27570866141732275,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.3110236220472441,
"calib/gap": -0.012273053111762744,
"calib/mean_conf": 0.885944881889764,
"calib/mu_c": 0.8811612903225807,
"calib/mu_w": 0.8934343434343435,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.27570866141732275,
"calib/std_conf": 0.043322778116439024,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7986074270557029,
"calib/step_q_c_n": 754.0,
"calib/step_q_gap": 0.00246642520041529,
"calib/step_q_w": 0.7961410018552876,
"calib/step_q_w_n": 539.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2598.0,
"completions/max_terminated_length": 2598.0,
"completions/mean_length": 508.6171875,
"completions/mean_terminated_length": 508.6171875,
"completions/min_length": 111.0,
"completions/min_terminated_length": 111.0,
"epoch": 0.004266666666666667,
"grad_norm": 0.03628219664096832,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.0335,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03336188197135925,
"mask/share_reasoning": 0.8539849519729614,
"mask/share_step_conf": 0.11265319585800171,
"num_tokens": 929464.0,
"reward": 1.2149934768676758,
"reward_std": 0.23332801461219788,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.6731215119361877,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.7186670899391174,
"step": 4
},
{
"adv/mean_abs_final_conf": 0.78107088804245,
"adv/mean_abs_reasoning": 0.4505809545516968,
"adv/mean_abs_step_conf": 0.7702205181121826,
"adv/ratio_final_to_reasoning": 1.7334751505854757,
"adv/ratio_step_to_reasoning": 1.7093943060210115,
"adv/std_final_conf": 0.9305281043052673,
"adv/std_reasoning": 0.7014207243919373,
"adv/std_step_conf": 0.9359554052352905,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.4787919017590442,
"calib/avg_num_step_conf": 4.55078125,
"calib/ece": 0.3518292682926829,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.32113821138211385,
"calib/gap": -0.0009764354463989156,
"calib/mean_conf": 0.8843495934959348,
"calib/mu_c": 0.8838931297709923,
"calib/mu_w": 0.8848695652173912,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.97265625,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.3518292682926829,
"calib/std_conf": 0.051436516044164424,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.8001442307692307,
"calib/step_q_c_n": 624.0,
"calib/step_q_gap": 0.026244045926347215,
"calib/step_q_w": 0.7739001848428835,
"calib/step_q_w_n": 541.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2823.0,
"completions/max_terminated_length": 2823.0,
"completions/mean_length": 518.09375,
"completions/mean_terminated_length": 520.1255493164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 164.0,
"epoch": 0.005333333333333333,
"grad_norm": 0.03847824037075043,
"learning_rate": 1.25e-06,
"loss": -0.0223,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03429059311747551,
"mask/share_reasoning": 0.8538260459899902,
"mask/share_step_conf": 0.10797706246376038,
"num_tokens": 1168784.0,
"reward": 1.1169718503952026,
"reward_std": 0.2400004267692566,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.5990898609161377,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6705517768859863,
"step": 5
},
{
"adv/mean_abs_final_conf": 0.7595282793045044,
"adv/mean_abs_reasoning": 0.3384864032268524,
"adv/mean_abs_step_conf": 0.775572657585144,
"adv/ratio_final_to_reasoning": 2.243895979465595,
"adv/ratio_step_to_reasoning": 2.2912963421616612,
"adv/std_final_conf": 0.9284359216690063,
"adv/std_reasoning": 0.6184952855110168,
"adv/std_step_conf": 0.9360383749008179,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.5289813938820561,
"calib/avg_num_step_conf": 5.06640625,
"calib/ece": 0.29503906250000006,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.3125,
"calib/gap": 0.002615578681804065,
"calib/mean_conf": 0.8823046875,
"calib/mu_c": 0.8833774834437087,
"calib/mu_w": 0.8807619047619046,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.29375000000000007,
"calib/std_conf": 0.04716743092990484,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.792674094707521,
"calib/step_q_c_n": 718.0,
"calib/step_q_gap": -0.009294817209577366,
"calib/step_q_w": 0.8019689119170984,
"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": 0.00390625,
"completions/max_length": 1232.0,
"completions/max_terminated_length": 1232.0,
"completions/mean_length": 435.56640625,
"completions/mean_terminated_length": 437.2745361328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 159.0,
"epoch": 0.0064,
"grad_norm": 0.039497025310993195,
"learning_rate": 1.5e-06,
"loss": -0.0103,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.037563733756542206,
"mask/share_reasoning": 0.8333431482315063,
"mask/share_step_conf": 0.12518689036369324,
"num_tokens": 1386241.0,
"reward": 1.2087037563323975,
"reward_std": 0.20760923624038696,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.6677120923995972,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7162538766860962,
"step": 6
},
{
"adv/mean_abs_final_conf": 0.7652530670166016,
"adv/mean_abs_reasoning": 0.4704771637916565,
"adv/mean_abs_step_conf": 0.7372301816940308,
"adv/ratio_final_to_reasoning": 1.6265466762494811,
"adv/ratio_step_to_reasoning": 1.566983986539465,
"adv/std_final_conf": 0.9297286868095398,
"adv/std_reasoning": 0.7392281293869019,
"adv/std_step_conf": 0.9354885816574097,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.46026745531696034,
"calib/avg_num_step_conf": 5.26953125,
"calib/ece": 0.27733333333333327,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.30980392156862746,
"calib/gap": -0.005463546354635529,
"calib/mean_conf": 0.8812549019607842,
"calib/mu_c": 0.879090909090909,
"calib/mu_w": 0.8845544554455446,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.27733333333333327,
"calib/std_conf": 0.045770031781768666,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7868014705882352,
"calib/step_q_c_n": 816.0,
"calib/step_q_gap": -0.001997778942721573,
"calib/step_q_w": 0.7887992495309568,
"calib/step_q_w_n": 533.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1202.0,
"completions/max_terminated_length": 1202.0,
"completions/mean_length": 524.5234375,
"completions/mean_terminated_length": 526.5804443359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 169.0,
"epoch": 0.007466666666666667,
"grad_norm": 0.034977804869413376,
"learning_rate": 1.75e-06,
"loss": 0.0616,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03061247244477272,
"mask/share_reasoning": 0.8583376407623291,
"mask/share_step_conf": 0.10714363306760788,
"num_tokens": 1627943.0,
"reward": 1.2274894714355469,
"reward_std": 0.25535452365875244,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6677761673927307,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7350075840950012,
"step": 7
},
{
"adv/mean_abs_final_conf": 0.7622976899147034,
"adv/mean_abs_reasoning": 0.4128996729850769,
"adv/mean_abs_step_conf": 0.7528561949729919,
"adv/ratio_final_to_reasoning": 1.8462056034184713,
"adv/ratio_step_to_reasoning": 1.823339286103532,
"adv/std_final_conf": 0.9313320517539978,
"adv/std_reasoning": 0.7013623118400574,
"adv/std_step_conf": 0.9357892274856567,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5052173913043478,
"calib/avg_num_step_conf": 5.0546875,
"calib/ece": 0.3403200000000002,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.312,
"calib/gap": 0.0033301127214171444,
"calib/mean_conf": 0.88032,
"calib/mu_c": 0.8818518518518519,
"calib/mu_w": 0.8785217391304347,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.3403200000000002,
"calib/std_conf": 0.0450144154688251,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7924,
"calib/step_q_c_n": 625.0,
"calib/step_q_gap": 0.05108460388639757,
"calib/step_q_w": 0.7413153961136024,
"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": 0.00390625,
"completions/max_length": 2684.0,
"completions/max_terminated_length": 2684.0,
"completions/mean_length": 535.90234375,
"completions/mean_terminated_length": 538.0039672851562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 163.0,
"epoch": 0.008533333333333334,
"grad_norm": 0.03353934735059738,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.0109,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03260628134012222,
"mask/share_reasoning": 0.856112003326416,
"mask/share_step_conf": 0.10737546533346176,
"num_tokens": 1871646.0,
"reward": 1.1858603954315186,
"reward_std": 0.24610640108585358,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.6166367530822754,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.727541983127594,
"step": 8
},
{
"adv/mean_abs_final_conf": 0.8015788793563843,
"adv/mean_abs_reasoning": 0.42125481367111206,
"adv/mean_abs_step_conf": 0.7736630439758301,
"adv/ratio_final_to_reasoning": 1.902836129920652,
"adv/ratio_step_to_reasoning": 1.836567841762053,
"adv/std_final_conf": 0.9309937357902527,
"adv/std_reasoning": 0.68181312084198,
"adv/std_step_conf": 0.9360260963439941,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.4510899558498896,
"calib/avg_num_step_conf": 4.765625,
"calib/ece": 0.26352226720647776,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.2550607287449393,
"calib/gap": -0.007388245033112639,
"calib/mean_conf": 0.8748582995951416,
"calib/mu_c": 0.8719867549668875,
"calib/mu_w": 0.8793750000000001,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.26352226720647776,
"calib/std_conf": 0.04740086781034078,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.7605013550135501,
"calib/step_q_c_n": 738.0,
"calib/step_q_gap": 0.0124930562583635,
"calib/step_q_w": 0.7480082987551866,
"calib/step_q_w_n": 482.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2593.0,
"completions/max_terminated_length": 2593.0,
"completions/mean_length": 491.46875,
"completions/mean_terminated_length": 499.2698669433594,
"completions/min_length": 0.0,
"completions/min_terminated_length": 194.0,
"epoch": 0.0096,
"grad_norm": 0.05028747767210007,
"learning_rate": 2.25e-06,
"loss": -0.0408,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.033226363360881805,
"mask/share_reasoning": 0.8455919027328491,
"mask/share_step_conf": 0.10555671155452728,
"num_tokens": 2104998.0,
"reward": 1.1680667400360107,
"reward_std": 0.28994813561439514,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.6573429703712463,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.6847077012062073,
"step": 9
},
{
"adv/mean_abs_final_conf": 0.7939379215240479,
"adv/mean_abs_reasoning": 0.4176982045173645,
"adv/mean_abs_step_conf": 0.7548781633377075,
"adv/ratio_final_to_reasoning": 1.9007453537929737,
"adv/ratio_step_to_reasoning": 1.8072334407325081,
"adv/std_final_conf": 0.9302391409873962,
"adv/std_reasoning": 0.6817051768302917,
"adv/std_step_conf": 0.9354656934738159,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.48880249867091974,
"calib/avg_num_step_conf": 4.8828125,
"calib/ece": 0.2816733067729084,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.35856573705179284,
"calib/gap": 0.0021325093035619025,
"calib/mean_conf": 0.8872509960159362,
"calib/mu_c": 0.888092105263158,
"calib/mu_w": 0.8859595959595961,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.2816733067729084,
"calib/std_conf": 0.04676819105147297,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.7871965317919075,
"calib/step_q_c_n": 692.0,
"calib/step_q_gap": 0.0032180371682516418,
"calib/step_q_w": 0.7839784946236559,
"calib/step_q_w_n": 558.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1937.0,
"completions/max_terminated_length": 1937.0,
"completions/mean_length": 511.09375,
"completions/mean_terminated_length": 513.0980834960938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 185.0,
"epoch": 0.010666666666666666,
"grad_norm": 0.03779426962137222,
"learning_rate": 2.5e-06,
"loss": 0.0688,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03259393572807312,
"mask/share_reasoning": 0.8553390502929688,
"mask/share_step_conf": 0.10816076397895813,
"num_tokens": 2342638.0,
"reward": 1.2232654094696045,
"reward_std": 0.24356569349765778,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6673445105552673,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7321712970733643,
"step": 10
},
{
"adv/mean_abs_final_conf": 0.7801154851913452,
"adv/mean_abs_reasoning": 0.3890037536621094,
"adv/mean_abs_step_conf": 0.7646522521972656,
"adv/ratio_final_to_reasoning": 2.0054189139494976,
"adv/ratio_step_to_reasoning": 1.9656680558960529,
"adv/std_final_conf": 0.9274177551269531,
"adv/std_reasoning": 0.6612036824226379,
"adv/std_step_conf": 0.9355158805847168,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.43217879256965946,
"calib/avg_num_step_conf": 5.6171875,
"calib/ece": 0.29846456692913387,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.35826771653543305,
"calib/gap": -0.02004127966976288,
"calib/mean_conf": 0.8767322834645669,
"calib/mu_c": 0.8686842105263157,
"calib/mu_w": 0.8887254901960786,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.28838582677165353,
"calib/std_conf": 0.09503807349936566,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7773716381418093,
"calib/step_q_c_n": 818.0,
"calib/step_q_gap": 0.01146841233535767,
"calib/step_q_w": 0.7659032258064516,
"calib/step_q_w_n": 620.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2896.0,
"completions/max_terminated_length": 2896.0,
"completions/mean_length": 535.8984375,
"completions/mean_terminated_length": 535.8984375,
"completions/min_length": 178.0,
"completions/min_terminated_length": 178.0,
"epoch": 0.011733333333333333,
"grad_norm": 0.039491284638643265,
"learning_rate": 2.7500000000000004e-06,
"loss": 0.054,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03238622844219208,
"mask/share_reasoning": 0.8498245477676392,
"mask/share_step_conf": 0.11778921633958817,
"num_tokens": 2584308.0,
"reward": 1.2056663036346436,
"reward_std": 0.2091943919658661,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6545777320861816,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.720174252986908,
"step": 11
},
{
"adv/mean_abs_final_conf": 0.7940953969955444,
"adv/mean_abs_reasoning": 0.44334566593170166,
"adv/mean_abs_step_conf": 0.7410632371902466,
"adv/ratio_final_to_reasoning": 1.7911427989867312,
"adv/ratio_step_to_reasoning": 1.6715247134149473,
"adv/std_final_conf": 0.9285423159599304,
"adv/std_reasoning": 0.7014912366867065,
"adv/std_step_conf": 0.9357432126998901,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.4787249031349067,
"calib/avg_num_step_conf": 5.58984375,
"calib/ece": 0.2275793650793651,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.39285714285714285,
"calib/gap": -0.005296935540683334,
"calib/mean_conf": 0.8849603174603174,
"calib/mu_c": 0.8831736526946108,
"calib/mu_w": 0.8884705882352941,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.22492063492063497,
"calib/std_conf": 0.05906344460652552,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.7653591790193843,
"calib/step_q_c_n": 877.0,
"calib/step_q_gap": -0.004514467190001925,
"calib/step_q_w": 0.7698736462093863,
"calib/step_q_w_n": 554.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2063.0,
"completions/max_terminated_length": 2063.0,
"completions/mean_length": 480.0546875,
"completions/mean_terminated_length": 481.9372863769531,
"completions/min_length": 0.0,
"completions/min_terminated_length": 142.0,
"epoch": 0.0128,
"grad_norm": 0.0577828474342823,
"learning_rate": 3e-06,
"loss": 0.0138,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.035471174865961075,
"mask/share_reasoning": 0.831091046333313,
"mask/share_step_conf": 0.12953157722949982,
"num_tokens": 2811378.0,
"reward": 1.2840325832366943,
"reward_std": 0.2465120553970337,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.7054234743118286,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7684301137924194,
"step": 12
},
{
"adv/mean_abs_final_conf": 0.7640044689178467,
"adv/mean_abs_reasoning": 0.4305124580860138,
"adv/mean_abs_step_conf": 0.7631770968437195,
"adv/ratio_final_to_reasoning": 1.7746396290469326,
"adv/ratio_step_to_reasoning": 1.7727177983110567,
"adv/std_final_conf": 0.930860698223114,
"adv/std_reasoning": 0.7013241648674011,
"adv/std_step_conf": 0.9360091090202332,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.5543396226415095,
"calib/avg_num_step_conf": 4.91015625,
"calib/ece": 0.2992578125,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.4140625,
"calib/gap": 0.011764779874213649,
"calib/mean_conf": 0.8851953125,
"calib/mu_c": 0.8900666666666666,
"calib/mu_w": 0.8783018867924529,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2992578125,
"calib/std_conf": 0.05742489097096609,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7743633762517882,
"calib/step_q_c_n": 699.0,
"calib/step_q_gap": 0.018467318904118013,
"calib/step_q_w": 0.7558960573476702,
"calib/step_q_w_n": 558.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1411.0,
"completions/max_terminated_length": 1411.0,
"completions/mean_length": 468.6484375,
"completions/mean_terminated_length": 470.4862976074219,
"completions/min_length": 0.0,
"completions/min_terminated_length": 85.0,
"epoch": 0.013866666666666666,
"grad_norm": 0.04164907708764076,
"learning_rate": 3.2500000000000002e-06,
"loss": 0.0422,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.0350169837474823,
"mask/share_reasoning": 0.8462280035018921,
"mask/share_step_conf": 0.11484874039888382,
"num_tokens": 3035944.0,
"reward": 1.262359619140625,
"reward_std": 0.22623397409915924,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.6652238368988037,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.771935224533081,
"step": 13
},
{
"adv/mean_abs_final_conf": 0.7712956070899963,
"adv/mean_abs_reasoning": 0.47973084449768066,
"adv/mean_abs_step_conf": 0.7793824672698975,
"adv/ratio_final_to_reasoning": 1.6077673886022672,
"adv/ratio_step_to_reasoning": 1.624624466425497,
"adv/std_final_conf": 0.9276650547981262,
"adv/std_reasoning": 0.7205986380577087,
"adv/std_step_conf": 0.9358716011047363,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.4078828828828829,
"calib/avg_num_step_conf": 5.36328125,
"calib/ece": 0.3528286852589641,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.6414342629482072,
"calib/gap": -0.010880952380952658,
"calib/mean_conf": 0.910597609561753,
"calib/mu_c": 0.9057857142857141,
"calib/mu_w": 0.9166666666666667,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3528286852589641,
"calib/std_conf": 0.045085591843685116,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7475923392612859,
"calib/step_q_c_n": 731.0,
"calib/step_q_gap": 0.03533377228309276,
"calib/step_q_w": 0.7122585669781931,
"calib/step_q_w_n": 642.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2698.0,
"completions/max_terminated_length": 2698.0,
"completions/mean_length": 536.953125,
"completions/mean_terminated_length": 541.1810913085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 136.0,
"epoch": 0.014933333333333333,
"grad_norm": 0.03964659944176674,
"learning_rate": 3.5e-06,
"loss": 0.0449,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.0325857549905777,
"mask/share_reasoning": 0.8444947004318237,
"mask/share_step_conf": 0.11510701477527618,
"num_tokens": 3278804.0,
"reward": 1.217205286026001,
"reward_std": 0.23579458892345428,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.6093109250068665,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7598154544830322,
"step": 14
},
{
"adv/mean_abs_final_conf": 0.7839799523353577,
"adv/mean_abs_reasoning": 0.40933963656425476,
"adv/mean_abs_step_conf": 0.7739418745040894,
"adv/ratio_final_to_reasoning": 1.9152309776683327,
"adv/ratio_step_to_reasoning": 1.8907083638420203,
"adv/std_final_conf": 0.9218094348907471,
"adv/std_reasoning": 0.6613172888755798,
"adv/std_step_conf": 0.9354211091995239,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.487977369165488,
"calib/avg_num_step_conf": 5.10546875,
"calib/ece": 0.3059215686274509,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.6549019607843137,
"calib/gap": 0.009904847627619895,
"calib/mean_conf": 0.909843137254902,
"calib/mu_c": 0.9137662337662338,
"calib/mu_w": 0.9038613861386139,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3059215686274509,
"calib/std_conf": 0.06864497255316152,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7073884514435695,
"calib/step_q_c_n": 762.0,
"calib/step_q_gap": -0.014597603602302112,
"calib/step_q_w": 0.7219860550458717,
"calib/step_q_w_n": 545.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1383.0,
"completions/max_terminated_length": 1383.0,
"completions/mean_length": 468.61328125,
"completions/mean_terminated_length": 470.4510192871094,
"completions/min_length": 0.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.016,
"grad_norm": 0.042239848524332047,
"learning_rate": 3.7500000000000005e-06,
"loss": 0.0464,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.033957935869693756,
"mask/share_reasoning": 0.84504234790802,
"mask/share_step_conf": 0.11709348857402802,
"num_tokens": 3506649.0,
"reward": 1.2645666599273682,
"reward_std": 0.22863566875457764,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6646316647529602,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7724852561950684,
"step": 15
},
{
"adv/mean_abs_final_conf": 0.7644206285476685,
"adv/mean_abs_reasoning": 0.44867604970932007,
"adv/mean_abs_step_conf": 0.7611935138702393,
"adv/ratio_final_to_reasoning": 1.7037250574059104,
"adv/ratio_step_to_reasoning": 1.696532530237322,
"adv/std_final_conf": 0.9242152571678162,
"adv/std_reasoning": 0.7205649018287659,
"adv/std_step_conf": 0.9357426166534424,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5098833218943033,
"calib/avg_num_step_conf": 6.55859375,
"calib/ece": 0.30337349397590363,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.7991967871485943,
"calib/gap": 0.01745161290322539,
"calib/mean_conf": 0.925863453815261,
"calib/mu_c": 0.9324516129032256,
"calib/mu_w": 0.9150000000000003,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.30337349397590363,
"calib/std_conf": 0.07098388745929748,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6563173359451517,
"calib/step_q_c_n": 1021.0,
"calib/step_q_gap": 0.024645603422355444,
"calib/step_q_w": 0.6316717325227963,
"calib/step_q_w_n": 658.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2619.0,
"completions/max_terminated_length": 2619.0,
"completions/mean_length": 627.3984375,
"completions/mean_terminated_length": 634.8379516601562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 202.0,
"epoch": 0.017066666666666667,
"grad_norm": 0.037402741611003876,
"learning_rate": 4.000000000000001e-06,
"loss": 0.0108,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.02552606537938118,
"mask/share_reasoning": 0.8506210446357727,
"mask/share_step_conf": 0.11213415116071701,
"num_tokens": 3776111.0,
"reward": 1.262770414352417,
"reward_std": 0.23914945125579834,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.6576437950134277,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7761359214782715,
"step": 16
},
{
"adv/mean_abs_final_conf": 0.7625212073326111,
"adv/mean_abs_reasoning": 0.3886851668357849,
"adv/mean_abs_step_conf": 0.7757367491722107,
"adv/ratio_final_to_reasoning": 1.9617965191215225,
"adv/ratio_step_to_reasoning": 1.9957971524546259,
"adv/std_final_conf": 0.9166886210441589,
"adv/std_reasoning": 0.6612927913665771,
"adv/std_step_conf": 0.9357386827468872,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5342518178339073,
"calib/avg_num_step_conf": 5.74609375,
"calib/ece": 0.1399604743083005,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.8616600790513834,
"calib/gap": 0.0019737849215459846,
"calib/mean_conf": 0.9325296442687746,
"calib/mu_c": 0.9329353233830845,
"calib/mu_w": 0.9309615384615385,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.13901185770750998,
"calib/std_conf": 0.03935151345673892,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6475511945392493,
"calib/step_q_c_n": 1172.0,
"calib/step_q_gap": 0.02213313433858033,
"calib/step_q_w": 0.6254180602006689,
"calib/step_q_w_n": 299.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2625.0,
"completions/max_terminated_length": 2625.0,
"completions/mean_length": 512.58203125,
"completions/mean_terminated_length": 516.6181030273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 153.0,
"epoch": 0.018133333333333335,
"grad_norm": 0.06586568057537079,
"learning_rate": 4.25e-06,
"loss": -0.001,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03208363801240921,
"mask/share_reasoning": 0.8363739848136902,
"mask/share_step_conf": 0.1237298846244812,
"num_tokens": 4010860.0,
"reward": 1.4073078632354736,
"reward_std": 0.2265596091747284,
"rewards/accuracy_reward_step": 0.78515625,
"rewards/final_brier_reward_step": 0.7993745803833008,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8310579061508179,
"step": 17
},
{
"adv/mean_abs_final_conf": 0.7089370489120483,
"adv/mean_abs_reasoning": 0.3773331940174103,
"adv/mean_abs_step_conf": 0.7654873728752136,
"adv/ratio_final_to_reasoning": 1.8788091271910146,
"adv/ratio_step_to_reasoning": 2.028677532249903,
"adv/std_final_conf": 0.9150155782699585,
"adv/std_reasoning": 0.6815622448921204,
"adv/std_step_conf": 0.9358413219451904,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5111811546118116,
"calib/avg_num_step_conf": 5.93359375,
"calib/ece": 0.38906882591093117,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.8866396761133604,
"calib/gap": 0.004445919044459012,
"calib/mean_conf": 0.9406477732793521,
"calib/mu_c": 0.9426277372262772,
"calib/mu_w": 0.9381818181818182,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.3875303643724696,
"calib/std_conf": 0.04027719521709381,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.621609947643979,
"calib/step_q_c_n": 764.0,
"calib/step_q_gap": 0.017954318504906253,
"calib/step_q_w": 0.6036556291390728,
"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": 0.00390625,
"completions/max_length": 2898.0,
"completions/max_terminated_length": 2898.0,
"completions/mean_length": 557.921875,
"completions/mean_terminated_length": 560.10986328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 178.0,
"epoch": 0.0192,
"grad_norm": 0.03878249600529671,
"learning_rate": 4.5e-06,
"loss": -0.0734,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.03142430633306503,
"mask/share_reasoning": 0.8538853526115417,
"mask/share_step_conf": 0.11078407615423203,
"num_tokens": 4264408.0,
"reward": 1.2024595737457275,
"reward_std": 0.22675225138664246,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.583086371421814,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7613070607185364,
"step": 18
},
{
"adv/mean_abs_final_conf": 0.6971423625946045,
"adv/mean_abs_reasoning": 0.3849453032016754,
"adv/mean_abs_step_conf": 0.760978102684021,
"adv/ratio_final_to_reasoning": 1.8110166737879823,
"adv/ratio_step_to_reasoning": 1.9768473504022452,
"adv/std_final_conf": 0.9060702919960022,
"adv/std_reasoning": 0.7011492848396301,
"adv/std_step_conf": 0.9357239603996277,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5756445701793866,
"calib/avg_num_step_conf": 5.19140625,
"calib/ece": 0.3559448818897637,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.952755905511811,
"calib/gap": 0.011833729827043005,
"calib/mean_conf": 0.9504330708661417,
"calib/mu_c": 0.9552317880794701,
"calib/mu_w": 0.9433980582524271,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3559448818897637,
"calib/std_conf": 0.03301469263736767,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6374509803921569,
"calib/step_q_c_n": 765.0,
"calib/step_q_gap": 0.023266583229036297,
"calib/step_q_w": 0.6141843971631206,
"calib/step_q_w_n": 564.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2597.0,
"completions/max_terminated_length": 2597.0,
"completions/mean_length": 518.5390625,
"completions/mean_terminated_length": 518.5390625,
"completions/min_length": 212.0,
"completions/min_terminated_length": 212.0,
"epoch": 0.020266666666666665,
"grad_norm": 0.03828023374080658,
"learning_rate": 4.75e-06,
"loss": 0.0769,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.030845241621136665,
"mask/share_reasoning": 0.857416033744812,
"mask/share_step_conf": 0.11173869669437408,
"num_tokens": 4501914.0,
"reward": 1.2848021984100342,
"reward_std": 0.20347543060779572,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.6318714618682861,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8106632232666016,
"step": 19
},
{
"adv/mean_abs_final_conf": 0.6998113393783569,
"adv/mean_abs_reasoning": 0.44451165199279785,
"adv/mean_abs_step_conf": 0.772807240486145,
"adv/ratio_final_to_reasoning": 1.5743374470410858,
"adv/ratio_step_to_reasoning": 1.738553392293677,
"adv/std_final_conf": 0.8987686634063721,
"adv/std_reasoning": 0.739271342754364,
"adv/std_step_conf": 0.9359571933746338,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.45597714133829487,
"calib/avg_num_step_conf": 5.5234375,
"calib/ece": 0.3706400000000001,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.996,
"calib/gap": -0.003477971958269599,
"calib/mean_conf": 0.96664,
"calib/mu_c": 0.965234899328859,
"calib/mu_w": 0.9687128712871286,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3706400000000001,
"calib/std_conf": 0.018652356419498316,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5813636363636364,
"calib/step_q_c_n": 814.0,
"calib/step_q_gap": -0.010903030303030303,
"calib/step_q_w": 0.5922666666666667,
"calib/step_q_w_n": 600.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2115.0,
"completions/max_terminated_length": 2115.0,
"completions/mean_length": 487.625,
"completions/mean_terminated_length": 487.625,
"completions/min_length": 162.0,
"completions/min_terminated_length": 162.0,
"epoch": 0.021333333333333333,
"grad_norm": 0.03156009316444397,
"learning_rate": 5e-06,
"loss": 0.0208,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.035062696784734726,
"mask/share_reasoning": 0.8308588266372681,
"mask/share_step_conf": 0.13407844305038452,
"num_tokens": 4731618.0,
"reward": 1.2386574745178223,
"reward_std": 0.26106804609298706,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.6013144850730896,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7825314998626709,
"step": 20
},
{
"adv/mean_abs_final_conf": 0.7398608922958374,
"adv/mean_abs_reasoning": 0.48953911662101746,
"adv/mean_abs_step_conf": 0.7502537965774536,
"adv/ratio_final_to_reasoning": 1.5113417236249367,
"adv/ratio_step_to_reasoning": 1.5325717008193067,
"adv/std_final_conf": 0.8951833844184875,
"adv/std_reasoning": 0.7393571138381958,
"adv/std_step_conf": 0.935997486114502,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5368434472208058,
"calib/avg_num_step_conf": 5.93359375,
"calib/ece": 0.39059055118110253,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.9921259842519685,
"calib/gap": 0.0013985211626722505,
"calib/mean_conf": 0.9732677165354332,
"calib/mu_c": 0.9738513513513514,
"calib/mu_w": 0.9724528301886791,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.39059055118110253,
"calib/std_conf": 0.016288225213710716,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5789975845410629,
"calib/step_q_c_n": 828.0,
"calib/step_q_gap": -0.0029416339828155502,
"calib/step_q_w": 0.5819392185238784,
"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": 0.0,
"completions/max_length": 2458.0,
"completions/max_terminated_length": 2458.0,
"completions/mean_length": 516.46484375,
"completions/mean_terminated_length": 516.46484375,
"completions/min_length": 151.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.0224,
"grad_norm": 0.024165457114577293,
"learning_rate": 4.9722222222222224e-06,
"loss": -0.0086,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03220517560839653,
"mask/share_reasoning": 0.8394327163696289,
"mask/share_step_conf": 0.12836214900016785,
"num_tokens": 4966793.0,
"reward": 1.256303310394287,
"reward_std": 0.25476592779159546,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.5999652147293091,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.7992894053459167,
"step": 21
},
{
"adv/mean_abs_final_conf": 0.6778824925422668,
"adv/mean_abs_reasoning": 0.4465583562850952,
"adv/mean_abs_step_conf": 0.7399794459342957,
"adv/ratio_final_to_reasoning": 1.5180154687542962,
"adv/ratio_step_to_reasoning": 1.6570722180414699,
"adv/std_final_conf": 0.8620956540107727,
"adv/std_reasoning": 0.7391456365585327,
"adv/std_step_conf": 0.9359897375106812,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.463719512195122,
"calib/avg_num_step_conf": 5.83203125,
"calib/ece": 0.3327952755905512,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0016897018970188915,
"calib/mean_conf": 0.9784645669291339,
"calib/mu_c": 0.9778658536585366,
"calib/mu_w": 0.9795555555555555,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.3327952755905512,
"calib/std_conf": 0.012813191575706534,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5943560209424084,
"calib/step_q_c_n": 955.0,
"calib/step_q_gap": 0.011902489343895373,
"calib/step_q_w": 0.582453531598513,
"calib/step_q_w_n": 538.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2354.0,
"completions/max_terminated_length": 2354.0,
"completions/mean_length": 504.71875,
"completions/mean_terminated_length": 504.71875,
"completions/min_length": 164.0,
"completions/min_terminated_length": 164.0,
"epoch": 0.023466666666666667,
"grad_norm": 0.03978949412703514,
"learning_rate": 4.944444444444445e-06,
"loss": 0.0607,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03203662857413292,
"mask/share_reasoning": 0.8401437997817993,
"mask/share_step_conf": 0.12781958281993866,
"num_tokens": 5197817.0,
"reward": 1.3022887706756592,
"reward_std": 0.2456250935792923,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6543769836425781,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8118190765380859,
"step": 22
},
{
"adv/mean_abs_final_conf": 0.7471860647201538,
"adv/mean_abs_reasoning": 0.5141867995262146,
"adv/mean_abs_step_conf": 0.7579343318939209,
"adv/ratio_final_to_reasoning": 1.4531412813565632,
"adv/ratio_step_to_reasoning": 1.4740447101954033,
"adv/std_final_conf": 0.8914094567298889,
"adv/std_reasoning": 0.7576435804367065,
"adv/std_step_conf": 0.9362046122550964,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6178689148073022,
"calib/avg_num_step_conf": 5.9375,
"calib/ece": 0.4417460317460319,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006001521298174661,
"calib/mean_conf": 0.9814285714285715,
"calib/mu_c": 0.9841911764705884,
"calib/mu_w": 0.9781896551724137,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.4417460317460319,
"calib/std_conf": 0.01289236797207704,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5900246913580247,
"calib/step_q_c_n": 810.0,
"calib/step_q_gap": -0.005792210050425961,
"calib/step_q_w": 0.5958169014084507,
"calib/step_q_w_n": 710.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2592.0,
"completions/max_terminated_length": 2592.0,
"completions/mean_length": 545.265625,
"completions/mean_terminated_length": 551.7312622070312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 164.0,
"epoch": 0.024533333333333334,
"grad_norm": 0.03615272417664528,
"learning_rate": 4.9166666666666665e-06,
"loss": -0.0227,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.032736070454120636,
"mask/share_reasoning": 0.8280990719795227,
"mask/share_step_conf": 0.12744614481925964,
"num_tokens": 5441341.0,
"reward": 1.1947392225265503,
"reward_std": 0.2796531021595001,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.5466066598892212,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7702640295028687,
"step": 23
},
{
"adv/mean_abs_final_conf": 0.6879265904426575,
"adv/mean_abs_reasoning": 0.5578153729438782,
"adv/mean_abs_step_conf": 0.770302414894104,
"adv/ratio_final_to_reasoning": 1.23325140146661,
"adv/ratio_step_to_reasoning": 1.3809271889170471,
"adv/std_final_conf": 0.8792401552200317,
"adv/std_reasoning": 0.8098664283752441,
"adv/std_step_conf": 0.9363515377044678,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.4586881868131868,
"calib/avg_num_step_conf": 6.328125,
"calib/ece": 0.4464049586776861,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.9958677685950413,
"calib/gap": 4.3956043956017155e-05,
"calib/mean_conf": 0.9835950413223141,
"calib/mu_c": 0.9836153846153847,
"calib/mu_w": 0.9835714285714287,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.4464049586776861,
"calib/std_conf": 0.01568969684161436,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5907673267326732,
"calib/step_q_c_n": 808.0,
"calib/step_q_gap": 0.013021021313953995,
"calib/step_q_w": 0.5777463054187192,
"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": 0.015625,
"completions/max_length": 2792.0,
"completions/max_terminated_length": 2792.0,
"completions/mean_length": 605.82421875,
"completions/mean_terminated_length": 615.4404907226562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 158.0,
"epoch": 0.0256,
"grad_norm": 0.023386143147945404,
"learning_rate": 4.888888888888889e-06,
"loss": -0.0244,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.029682423919439316,
"mask/share_reasoning": 0.8312495946884155,
"mask/share_step_conf": 0.12344293296337128,
"num_tokens": 5700944.0,
"reward": 1.157957911491394,
"reward_std": 0.3184521794319153,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.521700382232666,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.75179523229599,
"step": 24
},
{
"adv/mean_abs_final_conf": 0.6561360359191895,
"adv/mean_abs_reasoning": 0.39023175835609436,
"adv/mean_abs_step_conf": 0.7360990047454834,
"adv/ratio_final_to_reasoning": 1.681400915915337,
"adv/ratio_step_to_reasoning": 1.8863124002167404,
"adv/std_final_conf": 0.8390585780143738,
"adv/std_reasoning": 0.6613587737083435,
"adv/std_step_conf": 0.9361077547073364,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5486625845955526,
"calib/avg_num_step_conf": 6.02734375,
"calib/ece": 0.41007936507936527,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.9920634920634921,
"calib/gap": 0.005144698678698156,
"calib/mean_conf": 0.9775396825396826,
"calib/mu_c": 0.9797241379310345,
"calib/mu_w": 0.9745794392523364,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4061111111111113,
"calib/std_conf": 0.08808983967443393,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.597260606060606,
"calib/step_q_c_n": 825.0,
"calib/step_q_gap": 0.03227453363720778,
"calib/step_q_w": 0.5649860724233983,
"calib/step_q_w_n": 718.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2682.0,
"completions/max_terminated_length": 2682.0,
"completions/mean_length": 532.88671875,
"completions/mean_terminated_length": 532.88671875,
"completions/min_length": 150.0,
"completions/min_terminated_length": 150.0,
"epoch": 0.02666666666666667,
"grad_norm": 0.02353591099381447,
"learning_rate": 4.861111111111111e-06,
"loss": 0.061,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.030907901003956795,
"mask/share_reasoning": 0.8425581455230713,
"mask/share_step_conf": 0.12653397023677826,
"num_tokens": 5940587.0,
"reward": 1.2288556098937988,
"reward_std": 0.25345277786254883,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.5795210599899292,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7840168476104736,
"step": 25
},
{
"adv/mean_abs_final_conf": 0.6244549751281738,
"adv/mean_abs_reasoning": 0.38389983773231506,
"adv/mean_abs_step_conf": 0.7507979869842529,
"adv/ratio_final_to_reasoning": 1.6266091145461556,
"adv/ratio_step_to_reasoning": 1.9557132178518082,
"adv/std_final_conf": 0.8316320776939392,
"adv/std_reasoning": 0.6815283298492432,
"adv/std_step_conf": 0.935912013053894,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.4784319045035269,
"calib/avg_num_step_conf": 5.43359375,
"calib/ece": 0.37377510040160655,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0006918068366793539,
"calib/mean_conf": 0.9842168674698797,
"calib/mu_c": 0.9839473684210527,
"calib/mu_w": 0.984639175257732,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.37377510040160655,
"calib/std_conf": 0.010805377427554943,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.6245371577574967,
"calib/step_q_c_n": 767.0,
"calib/step_q_gap": 0.040594850065189036,
"calib/step_q_w": 0.5839423076923077,
"calib/step_q_w_n": 624.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2556.0,
"completions/max_terminated_length": 2556.0,
"completions/mean_length": 546.08984375,
"completions/mean_terminated_length": 550.3897705078125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 263.0,
"epoch": 0.027733333333333332,
"grad_norm": 0.04103153944015503,
"learning_rate": 4.833333333333333e-06,
"loss": 0.0347,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.02897307462990284,
"mask/share_reasoning": 0.853653073310852,
"mask/share_step_conf": 0.10956136882305145,
"num_tokens": 6185626.0,
"reward": 1.2445653676986694,
"reward_std": 0.2300690859556198,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6050347089767456,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.785407304763794,
"step": 26
},
{
"adv/mean_abs_final_conf": 0.5905928611755371,
"adv/mean_abs_reasoning": 0.44203656911849976,
"adv/mean_abs_step_conf": 0.7786446809768677,
"adv/ratio_final_to_reasoning": 1.3360724031346214,
"adv/ratio_step_to_reasoning": 1.7614938115405812,
"adv/std_final_conf": 0.8065117597579956,
"adv/std_reasoning": 0.7014051079750061,
"adv/std_step_conf": 0.9358586668968201,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.4908809794269848,
"calib/avg_num_step_conf": 6.02734375,
"calib/ece": 0.4384584980237154,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00021267196768914864,
"calib/mean_conf": 0.9878656126482214,
"calib/mu_c": 0.9877697841726619,
"calib/mu_w": 0.9879824561403511,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4384584980237154,
"calib/std_conf": 0.007290534995045996,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5710660486674393,
"calib/step_q_c_n": 863.0,
"calib/step_q_gap": 0.003007225138027536,
"calib/step_q_w": 0.5680588235294117,
"calib/step_q_w_n": 680.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2408.0,
"completions/max_terminated_length": 2408.0,
"completions/mean_length": 497.96875,
"completions/mean_terminated_length": 499.9216003417969,
"completions/min_length": 0.0,
"completions/min_terminated_length": 197.0,
"epoch": 0.0288,
"grad_norm": 0.02568461187183857,
"learning_rate": 4.805555555555556e-06,
"loss": -0.0224,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.031828589737415314,
"mask/share_reasoning": 0.8362923860549927,
"mask/share_step_conf": 0.12797272205352783,
"num_tokens": 6418322.0,
"reward": 1.209911823272705,
"reward_std": 0.24842794239521027,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.5534738302230835,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7800499200820923,
"step": 27
},
{
"adv/mean_abs_final_conf": 0.5335990786552429,
"adv/mean_abs_reasoning": 0.4032040238380432,
"adv/mean_abs_step_conf": 0.7723286151885986,
"adv/ratio_final_to_reasoning": 1.3233972061488555,
"adv/ratio_step_to_reasoning": 1.9154784415019215,
"adv/std_final_conf": 0.7860013842582703,
"adv/std_reasoning": 0.701285183429718,
"adv/std_step_conf": 0.9357082843780518,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5305719921104537,
"calib/avg_num_step_conf": 5.26171875,
"calib/ece": 0.35595141700404864,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.001831501831501714,
"calib/mean_conf": 0.9875303643724697,
"calib/mu_c": 0.988205128205128,
"calib/mu_w": 0.9863736263736262,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.35595141700404864,
"calib/std_conf": 0.007739044043112878,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6041807228915663,
"calib/step_q_c_n": 830.0,
"calib/step_q_gap": 0.026114958868355465,
"calib/step_q_w": 0.5780657640232109,
"calib/step_q_w_n": 517.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2583.0,
"completions/max_terminated_length": 2583.0,
"completions/mean_length": 580.2421875,
"completions/mean_terminated_length": 584.81103515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 201.0,
"epoch": 0.029866666666666666,
"grad_norm": 0.03119593672454357,
"learning_rate": 4.777777777777778e-06,
"loss": -0.0502,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.028895732015371323,
"mask/share_reasoning": 0.8566277027130127,
"mask/share_step_conf": 0.10666404664516449,
"num_tokens": 6673808.0,
"reward": 1.2516157627105713,
"reward_std": 0.24505168199539185,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.6188547015190125,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7847665548324585,
"step": 28
},
{
"adv/mean_abs_final_conf": 0.5490765571594238,
"adv/mean_abs_reasoning": 0.49996668100357056,
"adv/mean_abs_step_conf": 0.747068464756012,
"adv/ratio_final_to_reasoning": 1.0982262979150457,
"adv/ratio_step_to_reasoning": 1.4942365024334026,
"adv/std_final_conf": 0.7831160426139832,
"adv/std_reasoning": 0.7393505573272705,
"adv/std_step_conf": 0.935847818851471,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5112689393939395,
"calib/avg_num_step_conf": 5.92578125,
"calib/ece": 0.46464285714285714,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0010227272727271197,
"calib/mean_conf": 0.988452380952381,
"calib/mu_c": 0.9889393939393939,
"calib/mu_w": 0.9879166666666668,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.46464285714285714,
"calib/std_conf": 0.007318282328239851,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5883853459972863,
"calib/step_q_c_n": 737.0,
"calib/step_q_gap": 0.08215457676651705,
"calib/step_q_w": 0.5062307692307693,
"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": 0.00390625,
"completions/max_length": 2202.0,
"completions/max_terminated_length": 2202.0,
"completions/mean_length": 596.0078125,
"completions/mean_terminated_length": 598.3451538085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 186.0,
"epoch": 0.030933333333333334,
"grad_norm": 0.027221517637372017,
"learning_rate": 4.75e-06,
"loss": -0.067,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.02765384130179882,
"mask/share_reasoning": 0.8585842847824097,
"mask/share_step_conf": 0.10985566675662994,
"num_tokens": 6933514.0,
"reward": 1.2175838947296143,
"reward_std": 0.25506746768951416,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.5267691016197205,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8041993379592896,
"step": 29
},
{
"adv/mean_abs_final_conf": 0.6431605815887451,
"adv/mean_abs_reasoning": 0.5578954219818115,
"adv/mean_abs_step_conf": 0.7405064702033997,
"adv/ratio_final_to_reasoning": 1.1528335889619712,
"adv/ratio_step_to_reasoning": 1.3273212882315812,
"adv/std_final_conf": 0.8330176472663879,
"adv/std_reasoning": 0.775515079498291,
"adv/std_step_conf": 0.9361404776573181,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5194967345370727,
"calib/avg_num_step_conf": 6.26171875,
"calib/ece": 0.44262948207171326,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0014925086438725144,
"calib/mean_conf": 0.9884462151394423,
"calib/mu_c": 0.9891240875912408,
"calib/mu_w": 0.9876315789473683,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.44262948207171326,
"calib/std_conf": 0.006404064914397626,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5581136638452237,
"calib/step_q_c_n": 827.0,
"calib/step_q_gap": 0.01882242673182155,
"calib/step_q_w": 0.5392912371134021,
"calib/step_q_w_n": 776.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2570.0,
"completions/max_terminated_length": 2570.0,
"completions/mean_length": 637.87109375,
"completions/mean_terminated_length": 640.37255859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 198.0,
"epoch": 0.032,
"grad_norm": 0.03011404164135456,
"learning_rate": 4.722222222222222e-06,
"loss": 0.0648,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.02600308507680893,
"mask/share_reasoning": 0.8624289035797119,
"mask/share_step_conf": 0.10766175389289856,
"num_tokens": 7203793.0,
"reward": 1.2001268863677979,
"reward_std": 0.3211213946342468,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.5420949459075928,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7775168418884277,
"step": 30
},
{
"adv/mean_abs_final_conf": 0.5998179912567139,
"adv/mean_abs_reasoning": 0.46080482006073,
"adv/mean_abs_step_conf": 0.7407785654067993,
"adv/ratio_final_to_reasoning": 1.3016747333017549,
"adv/ratio_step_to_reasoning": 1.607575557280784,
"adv/std_final_conf": 0.8259286284446716,
"adv/std_reasoning": 0.7392586469650269,
"adv/std_step_conf": 0.9360550045967102,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5241233766233766,
"calib/avg_num_step_conf": 6.3515625,
"calib/ece": 0.5478800000000001,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0011883116883115719,
"calib/mean_conf": 0.9878800000000001,
"calib/mu_c": 0.9885454545454544,
"calib/mu_w": 0.9873571428571428,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.5478800000000001,
"calib/std_conf": 0.007893389639438817,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5275705329153606,
"calib/step_q_c_n": 638.0,
"calib/step_q_gap": 0.02370413615422695,
"calib/step_q_w": 0.5038663967611337,
"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": 0.0,
"completions/max_length": 2476.0,
"completions/max_terminated_length": 2476.0,
"completions/mean_length": 652.5,
"completions/mean_terminated_length": 652.5,
"completions/min_length": 213.0,
"completions/min_terminated_length": 213.0,
"epoch": 0.03306666666666667,
"grad_norm": 0.028559528291225433,
"learning_rate": 4.694444444444445e-06,
"loss": -0.0636,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.02662399224936962,
"mask/share_reasoning": 0.862493634223938,
"mask/share_step_conf": 0.11088240146636963,
"num_tokens": 7476745.0,
"reward": 1.1457479000091553,
"reward_std": 0.25158950686454773,
"rewards/accuracy_reward_step": 0.4296875,
"rewards/final_brier_reward_step": 0.44331127405166626,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7834672927856445,
"step": 31
},
{
"adv/mean_abs_final_conf": 0.5406360626220703,
"adv/mean_abs_reasoning": 0.5045670866966248,
"adv/mean_abs_step_conf": 0.7292282581329346,
"adv/ratio_final_to_reasoning": 1.0714849955069152,
"adv/ratio_step_to_reasoning": 1.4452553037241394,
"adv/std_final_conf": 0.8152623176574707,
"adv/std_reasoning": 0.7926533222198486,
"adv/std_step_conf": 0.936184287071228,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5180899608865711,
"calib/avg_num_step_conf": 6.00390625,
"calib/ece": 0.46423387096774205,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0010417209908737401,
"calib/mean_conf": 0.9884274193548388,
"calib/mu_c": 0.9889230769230769,
"calib/mu_w": 0.9878813559322032,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.46423387096774205,
"calib/std_conf": 0.005494386265730213,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5454992764109985,
"calib/step_q_c_n": 691.0,
"calib/step_q_gap": 0.07904537570177861,
"calib/step_q_w": 0.4664539007092199,
"calib/step_q_w_n": 846.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2560.0,
"completions/max_terminated_length": 2560.0,
"completions/mean_length": 589.296875,
"completions/mean_terminated_length": 596.2846069335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 200.0,
"epoch": 0.034133333333333335,
"grad_norm": 0.026078954339027405,
"learning_rate": 4.666666666666667e-06,
"loss": -0.0358,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.027657821774482727,
"mask/share_reasoning": 0.8504782915115356,
"mask/share_step_conf": 0.11014510691165924,
"num_tokens": 7734309.0,
"reward": 1.2053842544555664,
"reward_std": 0.2780751585960388,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.5188254117965698,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7983152866363525,
"step": 32
},
{
"adv/mean_abs_final_conf": 0.5705288052558899,
"adv/mean_abs_reasoning": 0.4791930019855499,
"adv/mean_abs_step_conf": 0.7540302276611328,
"adv/ratio_final_to_reasoning": 1.1906033746150038,
"adv/ratio_step_to_reasoning": 1.5735418183003236,
"adv/std_final_conf": 0.80317622423172,
"adv/std_reasoning": 0.739323616027832,
"adv/std_step_conf": 0.9357584714889526,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5451496388028895,
"calib/avg_num_step_conf": 5.84765625,
"calib/ece": 0.53152,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0016563467492261008,
"calib/mean_conf": 0.98752,
"calib/mu_c": 0.988421052631579,
"calib/mu_w": 0.9867647058823529,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.53152,
"calib/std_conf": 0.008015584819587408,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5033039647577092,
"calib/step_q_c_n": 681.0,
"calib/step_q_gap": 0.01259318044398372,
"calib/step_q_w": 0.4907107843137255,
"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": 2599.0,
"completions/max_terminated_length": 2599.0,
"completions/mean_length": 589.0625,
"completions/mean_terminated_length": 593.7008056640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.0352,
"grad_norm": 0.036052510142326355,
"learning_rate": 4.638888888888889e-06,
"loss": -0.0546,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.027194002643227577,
"mask/share_reasoning": 0.8590854406356812,
"mask/share_step_conf": 0.10590802878141403,
"num_tokens": 7991981.0,
"reward": 1.1576623916625977,
"reward_std": 0.2627679109573364,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/final_brier_reward_step": 0.4591601490974426,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7858948707580566,
"step": 33
},
{
"adv/mean_abs_final_conf": 0.5254691243171692,
"adv/mean_abs_reasoning": 0.44810813665390015,
"adv/mean_abs_step_conf": 0.7442312240600586,
"adv/ratio_final_to_reasoning": 1.172639105018125,
"adv/ratio_step_to_reasoning": 1.6608295256974355,
"adv/std_final_conf": 0.7536927461624146,
"adv/std_reasoning": 0.7013894319534302,
"adv/std_step_conf": 0.935812771320343,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.47804033156006437,
"calib/avg_num_step_conf": 6.46875,
"calib/ece": 0.449843137254902,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.996078431372549,
"calib/gap": 0.00027712482989006304,
"calib/mean_conf": 0.9870980392156862,
"calib/mu_c": 0.9872262773722628,
"calib/mu_w": 0.9869491525423727,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.449843137254902,
"calib/std_conf": 0.011384983060706762,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.44419434194341945,
"calib/step_q_c_n": 813.0,
"calib/step_q_gap": 0.033755433283870206,
"calib/step_q_w": 0.41043890865954924,
"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": 0.00390625,
"completions/max_length": 2204.0,
"completions/max_terminated_length": 2204.0,
"completions/mean_length": 520.625,
"completions/mean_terminated_length": 522.6666870117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 190.0,
"epoch": 0.03626666666666667,
"grad_norm": 0.02863895334303379,
"learning_rate": 4.611111111111112e-06,
"loss": 0.0052,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.030690640211105347,
"mask/share_reasoning": 0.8315275311470032,
"mask/share_step_conf": 0.13387557864189148,
"num_tokens": 8230373.0,
"reward": 1.257469654083252,
"reward_std": 0.1998538076877594,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.5468925833702087,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.8308983445167542,
"step": 34
},
{
"adv/mean_abs_final_conf": 0.7124243974685669,
"adv/mean_abs_reasoning": 0.6191345453262329,
"adv/mean_abs_step_conf": 0.7590646743774414,
"adv/ratio_final_to_reasoning": 1.1506778338352577,
"adv/ratio_step_to_reasoning": 1.2260092416220725,
"adv/std_final_conf": 0.886207103729248,
"adv/std_reasoning": 0.8266004323959351,
"adv/std_step_conf": 0.9362123012542725,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5425914249684741,
"calib/avg_num_step_conf": 5.33984375,
"calib/ece": 0.4697619047619047,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0026607818411098494,
"calib/mean_conf": 0.9856349206349206,
"calib/mu_c": 0.986923076923077,
"calib/mu_w": 0.9842622950819672,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4697619047619047,
"calib/std_conf": 0.01076414033061597,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4666979655712051,
"calib/step_q_c_n": 639.0,
"calib/step_q_gap": 0.03360730623054581,
"calib/step_q_w": 0.43309065934065927,
"calib/step_q_w_n": 728.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2906.0,
"completions/max_terminated_length": 2906.0,
"completions/mean_length": 607.5390625,
"completions/mean_terminated_length": 609.921630859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 226.0,
"epoch": 0.037333333333333336,
"grad_norm": 0.02912158891558647,
"learning_rate": 4.583333333333333e-06,
"loss": 0.0197,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.027022257447242737,
"mask/share_reasoning": 0.8741253018379211,
"mask/share_step_conf": 0.09494614601135254,
"num_tokens": 8495159.0,
"reward": 1.1994110345840454,
"reward_std": 0.2870052456855774,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.5224952697753906,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7889447212219238,
"step": 35
},
{
"adv/mean_abs_final_conf": 0.5671288967132568,
"adv/mean_abs_reasoning": 0.31426718831062317,
"adv/mean_abs_step_conf": 0.7686116695404053,
"adv/ratio_final_to_reasoning": 1.8046074098983058,
"adv/ratio_step_to_reasoning": 2.445726751405896,
"adv/std_final_conf": 0.759645402431488,
"adv/std_reasoning": 0.5961850881576538,
"adv/std_step_conf": 0.9357486963272095,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5037545787545787,
"calib/avg_num_step_conf": 5.73046875,
"calib/ece": 0.20880478087649412,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0011272893772892312,
"calib/mean_conf": 0.9856972111553786,
"calib/mu_c": 0.9859487179487179,
"calib/mu_w": 0.9848214285714286,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.20880478087649412,
"calib/std_conf": 0.01005026967345645,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4283938294010889,
"calib/step_q_c_n": 1102.0,
"calib/step_q_gap": 0.044585610223006666,
"calib/step_q_w": 0.3838082191780822,
"calib/step_q_w_n": 365.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2356.0,
"completions/max_terminated_length": 2356.0,
"completions/mean_length": 548.8671875,
"completions/mean_terminated_length": 548.8671875,
"completions/min_length": 179.0,
"completions/min_terminated_length": 179.0,
"epoch": 0.0384,
"grad_norm": 0.07427214831113815,
"learning_rate": 4.555555555555556e-06,
"loss": 0.0355,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03173261880874634,
"mask/share_reasoning": 0.8426610827445984,
"mask/share_step_conf": 0.12560628354549408,
"num_tokens": 8738381.0,
"reward": 1.3560553789138794,
"reward_std": 0.18213136494159698,
"rewards/accuracy_reward_step": 0.76171875,
"rewards/final_brier_reward_step": 0.7680597305297852,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7978066205978394,
"step": 36
},
{
"adv/mean_abs_final_conf": 0.5322921276092529,
"adv/mean_abs_reasoning": 0.4058647155761719,
"adv/mean_abs_step_conf": 0.7570114135742188,
"adv/ratio_final_to_reasoning": 1.3115013628459984,
"adv/ratio_step_to_reasoning": 1.8651816344753043,
"adv/std_final_conf": 0.7714137434959412,
"adv/std_reasoning": 0.6817934513092041,
"adv/std_step_conf": 0.9359712600708008,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5140953528860318,
"calib/avg_num_step_conf": 6.01953125,
"calib/ece": 0.5228163265306123,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9959183673469387,
"calib/gap": -0.0005852417302797397,
"calib/mean_conf": 0.9869795918367347,
"calib/mu_c": 0.9866666666666667,
"calib/mu_w": 0.9872519083969464,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.5222448979591837,
"calib/std_conf": 0.011059191017242254,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4620941558441558,
"calib/step_q_c_n": 616.0,
"calib/step_q_gap": 0.06104550719550711,
"calib/step_q_w": 0.4010486486486487,
"calib/step_q_w_n": 925.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2955.0,
"completions/max_terminated_length": 2955.0,
"completions/mean_length": 612.32421875,
"completions/mean_terminated_length": 614.7255249023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 208.0,
"epoch": 0.039466666666666664,
"grad_norm": 0.03133146092295647,
"learning_rate": 4.527777777777778e-06,
"loss": 0.0536,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.028429534286260605,
"mask/share_reasoning": 0.854699969291687,
"mask/share_step_conf": 0.11296417564153671,
"num_tokens": 9002232.0,
"reward": 1.170437216758728,
"reward_std": 0.24616439640522003,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/final_brier_reward_step": 0.4580800533294678,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8011627793312073,
"step": 37
},
{
"adv/mean_abs_final_conf": 0.5607985854148865,
"adv/mean_abs_reasoning": 0.4902248978614807,
"adv/mean_abs_step_conf": 0.7547855377197266,
"adv/ratio_final_to_reasoning": 1.1439618588555396,
"adv/ratio_step_to_reasoning": 1.5396719771116172,
"adv/std_final_conf": 0.8122161030769348,
"adv/std_reasoning": 0.7575083374977112,
"adv/std_step_conf": 0.935595691204071,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5287289983326919,
"calib/avg_num_step_conf": 5.52734375,
"calib/ece": 0.4382071713147411,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9960159362549801,
"calib/gap": 0.0012055918943183475,
"calib/mean_conf": 0.98800796812749,
"calib/mu_c": 0.9885507246376812,
"calib/mu_w": 0.9873451327433629,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4382071713147411,
"calib/std_conf": 0.009060219970361357,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5019422572178478,
"calib/step_q_c_n": 762.0,
"calib/step_q_gap": 0.017577785548628877,
"calib/step_q_w": 0.4843644716692189,
"calib/step_q_w_n": 653.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3062.0,
"completions/max_terminated_length": 3062.0,
"completions/mean_length": 557.86328125,
"completions/mean_terminated_length": 557.86328125,
"completions/min_length": 169.0,
"completions/min_terminated_length": 169.0,
"epoch": 0.04053333333333333,
"grad_norm": 0.03880874812602997,
"learning_rate": 4.5e-06,
"loss": 0.0294,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.030267592519521713,
"mask/share_reasoning": 0.8548325300216675,
"mask/share_step_conf": 0.11489984393119812,
"num_tokens": 9251933.0,
"reward": 1.242781400680542,
"reward_std": 0.22920340299606323,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.5500128269195557,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8158218264579773,
"step": 38
},
{
"adv/mean_abs_final_conf": 0.5456191301345825,
"adv/mean_abs_reasoning": 0.45056378841400146,
"adv/mean_abs_step_conf": 0.7524350881576538,
"adv/ratio_final_to_reasoning": 1.210969776455358,
"adv/ratio_step_to_reasoning": 1.6699857101393982,
"adv/std_final_conf": 0.7808905839920044,
"adv/std_reasoning": 0.720582127571106,
"adv/std_step_conf": 0.9357057213783264,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5267491010030914,
"calib/avg_num_step_conf": 5.6796875,
"calib/ece": 0.4690873015873016,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0012775219229070878,
"calib/mean_conf": 0.9889285714285715,
"calib/mu_c": 0.9895419847328244,
"calib/mu_w": 0.9882644628099173,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4690873015873016,
"calib/std_conf": 0.005274494742113267,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5766857142857144,
"calib/step_q_c_n": 700.0,
"calib/step_q_gap": 0.08383425539977268,
"calib/step_q_w": 0.4928514588859417,
"calib/step_q_w_n": 754.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2464.0,
"completions/max_terminated_length": 2464.0,
"completions/mean_length": 536.8359375,
"completions/mean_terminated_length": 541.06298828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 209.0,
"epoch": 0.0416,
"grad_norm": 0.026082059368491173,
"learning_rate": 4.472222222222223e-06,
"loss": -0.0718,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.029997561126947403,
"mask/share_reasoning": 0.8474767208099365,
"mask/share_step_conf": 0.1147131398320198,
"num_tokens": 9495451.0,
"reward": 1.2392897605895996,
"reward_std": 0.24177560210227966,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.5226644277572632,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8283482193946838,
"step": 39
},
{
"adv/mean_abs_final_conf": 0.5052137970924377,
"adv/mean_abs_reasoning": 0.43444058299064636,
"adv/mean_abs_step_conf": 0.7465513348579407,
"adv/ratio_final_to_reasoning": 1.1629065443531899,
"adv/ratio_step_to_reasoning": 1.7184198808471218,
"adv/std_final_conf": 0.7265942096710205,
"adv/std_reasoning": 0.6817202568054199,
"adv/std_step_conf": 0.9355455040931702,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5258540517081034,
"calib/avg_num_step_conf": 5.140625,
"calib/ece": 0.4883070866141732,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0018110236220473253,
"calib/mean_conf": 0.9883070866141732,
"calib/mu_c": 0.9892125984251967,
"calib/mu_w": 0.9874015748031494,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4883070866141732,
"calib/std_conf": 0.006265782354986256,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.607751572327044,
"calib/step_q_c_n": 636.0,
"calib/step_q_gap": 0.029222160562338062,
"calib/step_q_w": 0.578529411764706,
"calib/step_q_w_n": 680.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2583.0,
"completions/max_terminated_length": 2583.0,
"completions/mean_length": 553.20703125,
"completions/mean_terminated_length": 553.20703125,
"completions/min_length": 119.0,
"completions/min_terminated_length": 119.0,
"epoch": 0.042666666666666665,
"grad_norm": 0.039314769208431244,
"learning_rate": 4.444444444444444e-06,
"loss": 0.0366,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03152313828468323,
"mask/share_reasoning": 0.8555469512939453,
"mask/share_step_conf": 0.11292991787195206,
"num_tokens": 9743832.0,
"reward": 1.178615689277649,
"reward_std": 0.24467621743679047,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.5084191560745239,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.775577962398529,
"step": 40
},
{
"adv/mean_abs_final_conf": 0.5695716142654419,
"adv/mean_abs_reasoning": 0.4441681206226349,
"adv/mean_abs_step_conf": 0.7738885879516602,
"adv/ratio_final_to_reasoning": 1.2823333954427354,
"adv/ratio_step_to_reasoning": 1.742332580885866,
"adv/std_final_conf": 0.7791482210159302,
"adv/std_reasoning": 0.7014320492744446,
"adv/std_step_conf": 0.9351321458816528,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5342717717717718,
"calib/avg_num_step_conf": 5.21484375,
"calib/ece": 0.2792519685039371,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.9960629921259843,
"calib/gap": 0.002585585585585526,
"calib/mean_conf": 0.9879133858267717,
"calib/mu_c": 0.9886666666666666,
"calib/mu_w": 0.9860810810810811,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2792519685039371,
"calib/std_conf": 0.008463468157947225,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6133402922755742,
"calib/step_q_c_n": 958.0,
"calib/step_q_gap": 0.011828355936051693,
"calib/step_q_w": 0.6015119363395225,
"calib/step_q_w_n": 377.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2931.0,
"completions/max_terminated_length": 2931.0,
"completions/mean_length": 507.20703125,
"completions/mean_terminated_length": 509.19610595703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.04373333333333333,
"grad_norm": 0.032360244542360306,
"learning_rate": 4.416666666666667e-06,
"loss": 0.0269,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03386985510587692,
"mask/share_reasoning": 0.8410245180130005,
"mask/share_step_conf": 0.1211993470788002,
"num_tokens": 9980925.0,
"reward": 1.3508212566375732,
"reward_std": 0.23536597192287445,
"rewards/accuracy_reward_step": 0.703125,
"rewards/final_brier_reward_step": 0.7109558582305908,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8258119821548462,
"step": 41
},
{
"adv/mean_abs_final_conf": 0.3804103136062622,
"adv/mean_abs_reasoning": 0.2688877284526825,
"adv/mean_abs_step_conf": 0.7513008117675781,
"adv/ratio_final_to_reasoning": 1.4147552058077835,
"adv/ratio_step_to_reasoning": 2.7941059865057705,
"adv/std_final_conf": 0.6370423436164856,
"adv/std_reasoning": 0.5482489466667175,
"adv/std_step_conf": 0.9348362684249878,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.4952256131810522,
"calib/avg_num_step_conf": 5.73046875,
"calib/ece": 0.4145703125000001,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 1.0,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -3.182924545952481e-05,
"calib/mean_conf": 0.9887890625000001,
"calib/mu_c": 0.9887755102040816,
"calib/mu_w": 0.9888073394495411,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4145703125000001,
"calib/std_conf": 0.006099631576668688,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6229291716686675,
"calib/step_q_c_n": 833.0,
"calib/step_q_gap": 0.009916553372137482,
"calib/step_q_w": 0.61301261829653,
"calib/step_q_w_n": 634.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1134.0,
"completions/max_terminated_length": 1134.0,
"completions/mean_length": 441.42578125,
"completions/mean_terminated_length": 443.1568908691406,
"completions/min_length": 0.0,
"completions/min_terminated_length": 176.0,
"epoch": 0.0448,
"grad_norm": 0.02571403607726097,
"learning_rate": 4.388888888888889e-06,
"loss": 0.0051,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.034706950187683105,
"mask/share_reasoning": 0.8270288705825806,
"mask/share_step_conf": 0.13435789942741394,
"num_tokens": 10198298.0,
"reward": 1.2550263404846191,
"reward_std": 0.15356135368347168,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.5835870504379272,
"rewards/format_reward_step": 1.0,
"rewards/step_l2_reward": 0.805810809135437,
"step": 42
},
{
"adv/mean_abs_final_conf": 0.5298599600791931,
"adv/mean_abs_reasoning": 0.49790188670158386,
"adv/mean_abs_step_conf": 0.7524062395095825,
"adv/ratio_final_to_reasoning": 1.0641854835885836,
"adv/ratio_step_to_reasoning": 1.5111536220398682,
"adv/std_final_conf": 0.7560086846351624,
"adv/std_reasoning": 0.7393122911453247,
"adv/std_step_conf": 0.935416579246521,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.4993055555555555,
"calib/avg_num_step_conf": 5.375,
"calib/ece": 0.42125984251968507,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0008143939393943089,
"calib/mean_conf": 0.9881889763779528,
"calib/mu_c": 0.9885416666666669,
"calib/mu_w": 0.9877272727272726,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42125984251968507,
"calib/std_conf": 0.007144991399701208,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6314912280701753,
"calib/step_q_c_n": 798.0,
"calib/step_q_gap": 0.029743823225884647,
"calib/step_q_w": 0.6017474048442907,
"calib/step_q_w_n": 578.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2407.0,
"completions/max_terminated_length": 2407.0,
"completions/mean_length": 523.609375,
"completions/mean_terminated_length": 523.609375,
"completions/min_length": 143.0,
"completions/min_terminated_length": 143.0,
"epoch": 0.04586666666666667,
"grad_norm": 0.033340878784656525,
"learning_rate": 4.361111111111112e-06,
"loss": 0.0667,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03283454850316048,
"mask/share_reasoning": 0.84957355260849,
"mask/share_step_conf": 0.11759185045957565,
"num_tokens": 10437566.0,
"reward": 1.2455248832702637,
"reward_std": 0.24735292792320251,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.5728577971458435,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8036272525787354,
"step": 43
},
{
"adv/mean_abs_final_conf": 0.510025143623352,
"adv/mean_abs_reasoning": 0.33438119292259216,
"adv/mean_abs_step_conf": 0.7807687520980835,
"adv/ratio_final_to_reasoning": 1.5252805911886937,
"adv/ratio_step_to_reasoning": 2.3349661064186353,
"adv/std_final_conf": 0.7340036034584045,
"adv/std_reasoning": 0.5961487889289856,
"adv/std_step_conf": 0.9347970485687256,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.5302207968422351,
"calib/avg_num_step_conf": 5.71484375,
"calib/ece": 0.5129803921568628,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0017879610213397124,
"calib/mean_conf": 0.9874901960784314,
"calib/mu_c": 0.9884297520661157,
"calib/mu_w": 0.986641791044776,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.5129803921568628,
"calib/std_conf": 0.007869954154436087,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6270942812982999,
"calib/step_q_c_n": 647.0,
"calib/step_q_gap": 0.06045212443555481,
"calib/step_q_w": 0.5666421568627451,
"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": 0.0,
"completions/max_length": 2108.0,
"completions/max_terminated_length": 2108.0,
"completions/mean_length": 542.76171875,
"completions/mean_terminated_length": 542.76171875,
"completions/min_length": 200.0,
"completions/min_terminated_length": 200.0,
"epoch": 0.046933333333333334,
"grad_norm": 0.04081565886735916,
"learning_rate": 4.333333333333334e-06,
"loss": 0.0692,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.029616590589284897,
"mask/share_reasoning": 0.8555626273155212,
"mask/share_step_conf": 0.11482080817222595,
"num_tokens": 10682833.0,
"reward": 1.1988677978515625,
"reward_std": 0.18884721398353577,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/final_brier_reward_step": 0.4864230155944824,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.8087812662124634,
"step": 44
},
{
"adv/mean_abs_final_conf": 0.6177642345428467,
"adv/mean_abs_reasoning": 0.5329394340515137,
"adv/mean_abs_step_conf": 0.7456905245780945,
"adv/ratio_final_to_reasoning": 1.1591640533080423,
"adv/ratio_step_to_reasoning": 1.3992031306619663,
"adv/std_final_conf": 0.8467596173286438,
"adv/std_reasoning": 0.7926696538925171,
"adv/std_step_conf": 0.9357033371925354,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5511443102352193,
"calib/avg_num_step_conf": 5.52734375,
"calib/ece": 0.4658964143426296,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9960159362549801,
"calib/gap": 0.010575969485060344,
"calib/mean_conf": 0.9838247011952193,
"calib/mu_c": 0.9889230769230769,
"calib/mu_w": 0.9783471074380166,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.4658964143426296,
"calib/std_conf": 0.06266741043070152,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5720665742024966,
"calib/step_q_c_n": 721.0,
"calib/step_q_gap": 0.03484755402958595,
"calib/step_q_w": 0.5372190201729107,
"calib/step_q_w_n": 694.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2385.0,
"completions/max_terminated_length": 2385.0,
"completions/mean_length": 547.453125,
"completions/mean_terminated_length": 547.453125,
"completions/min_length": 158.0,
"completions/min_terminated_length": 158.0,
"epoch": 0.048,
"grad_norm": 0.02756119892001152,
"learning_rate": 4.305555555555556e-06,
"loss": 0.0009,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03192518651485443,
"mask/share_reasoning": 0.8486030101776123,
"mask/share_step_conf": 0.11947186291217804,
"num_tokens": 10928029.0,
"reward": 1.2208482027053833,
"reward_std": 0.25765717029571533,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.5241742134094238,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8099329471588135,
"step": 45
},
{
"adv/mean_abs_final_conf": 0.5813736319541931,
"adv/mean_abs_reasoning": 0.4622410535812378,
"adv/mean_abs_step_conf": 0.7636514902114868,
"adv/ratio_final_to_reasoning": 1.2577282512013357,
"adv/ratio_step_to_reasoning": 1.6520633212802178,
"adv/std_final_conf": 0.7970679402351379,
"adv/std_reasoning": 0.7206328511238098,
"adv/std_step_conf": 0.9353775978088379,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5165275569687335,
"calib/avg_num_step_conf": 6.234375,
"calib/ece": 0.435425101214575,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0006478537360893322,
"calib/mean_conf": 0.9860323886639677,
"calib/mu_c": 0.9863235294117648,
"calib/mu_w": 0.9856756756756755,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.435425101214575,
"calib/std_conf": 0.010706186657994667,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5459429280397022,
"calib/step_q_c_n": 806.0,
"calib/step_q_gap": 0.10791761158400609,
"calib/step_q_w": 0.43802531645569615,
"calib/step_q_w_n": 790.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2980.0,
"completions/max_terminated_length": 2980.0,
"completions/mean_length": 597.453125,
"completions/mean_terminated_length": 602.157470703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 209.0,
"epoch": 0.04906666666666667,
"grad_norm": 0.025706231594085693,
"learning_rate": 4.277777777777778e-06,
"loss": -0.0665,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.031699288636446,
"mask/share_reasoning": 0.8346951007843018,
"mask/share_step_conf": 0.12579315900802612,
"num_tokens": 11185745.0,
"reward": 1.2327275276184082,
"reward_std": 0.2366017997264862,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.5433730483055115,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8114316463470459,
"step": 46
},
{
"adv/mean_abs_final_conf": 0.5593074560165405,
"adv/mean_abs_reasoning": 0.4556795358657837,
"adv/mean_abs_step_conf": 0.7513096332550049,
"adv/ratio_final_to_reasoning": 1.227414031121379,
"adv/ratio_step_to_reasoning": 1.6487675528977372,
"adv/std_final_conf": 0.7847610712051392,
"adv/std_reasoning": 0.7206171154975891,
"adv/std_step_conf": 0.9353805184364319,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.535280778978258,
"calib/avg_num_step_conf": 5.48046875,
"calib/ece": 0.37621513944223106,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.002446311858076533,
"calib/mean_conf": 0.9857768924302789,
"calib/mu_c": 0.9867320261437909,
"calib/mu_w": 0.9842857142857143,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.37621513944223106,
"calib/std_conf": 0.010547465467741988,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.49233173076923076,
"calib/step_q_c_n": 832.0,
"calib/step_q_gap": 0.05213908628586822,
"calib/step_q_w": 0.44019264448336254,
"calib/step_q_w_n": 571.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2700.0,
"completions/max_terminated_length": 2700.0,
"completions/mean_length": 576.8515625,
"completions/mean_terminated_length": 576.8515625,
"completions/min_length": 185.0,
"completions/min_terminated_length": 185.0,
"epoch": 0.050133333333333335,
"grad_norm": 0.0265414509922266,
"learning_rate": 4.25e-06,
"loss": -0.0368,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03034675121307373,
"mask/share_reasoning": 0.8593940734863281,
"mask/share_step_conf": 0.11025922000408173,
"num_tokens": 11439395.0,
"reward": 1.3010945320129395,
"reward_std": 0.2340029627084732,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.6093801259994507,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8385920524597168,
"step": 47
},
{
"adv/mean_abs_final_conf": 0.6615752577781677,
"adv/mean_abs_reasoning": 0.5312079191207886,
"adv/mean_abs_step_conf": 0.7640124559402466,
"adv/ratio_final_to_reasoning": 1.245416783080253,
"adv/ratio_step_to_reasoning": 1.4382550192489163,
"adv/std_final_conf": 0.8361097574234009,
"adv/std_reasoning": 0.7576334476470947,
"adv/std_step_conf": 0.935871958732605,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5431520863041726,
"calib/avg_num_step_conf": 5.5546875,
"calib/ece": 0.4841732283464569,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.9960629921259843,
"calib/gap": 0.0037795275590550848,
"calib/mean_conf": 0.9841732283464568,
"calib/mu_c": 0.9860629921259841,
"calib/mu_w": 0.982283464566929,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4841732283464569,
"calib/std_conf": 0.013599498883883185,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5214637681159421,
"calib/step_q_c_n": 690.0,
"calib/step_q_gap": 0.06504300308861966,
"calib/step_q_w": 0.45642076502732243,
"calib/step_q_w_n": 732.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2896.0,
"completions/max_terminated_length": 2896.0,
"completions/mean_length": 526.75,
"completions/mean_terminated_length": 528.8157348632812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 141.0,
"epoch": 0.0512,
"grad_norm": 0.02948327362537384,
"learning_rate": 4.222222222222223e-06,
"loss": 0.0157,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03321732580661774,
"mask/share_reasoning": 0.8371387720108032,
"mask/share_step_conf": 0.12573771178722382,
"num_tokens": 11677931.0,
"reward": 1.2243374586105347,
"reward_std": 0.2565346360206604,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.5132398009300232,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8188894391059875,
"step": 48
},
{
"adv/mean_abs_final_conf": 0.6058684587478638,
"adv/mean_abs_reasoning": 0.4402289390563965,
"adv/mean_abs_step_conf": 0.782744824886322,
"adv/ratio_final_to_reasoning": 1.3762576809387075,
"adv/ratio_step_to_reasoning": 1.7780403681868056,
"adv/std_final_conf": 0.7842022776603699,
"adv/std_reasoning": 0.6817723512649536,
"adv/std_step_conf": 0.9354808926582336,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5498268698060941,
"calib/avg_num_step_conf": 5.78125,
"calib/ece": 0.3696356275303644,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0026842105263159555,
"calib/mean_conf": 0.9850202429149798,
"calib/mu_c": 0.9860526315789475,
"calib/mu_w": 0.9833684210526316,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.3696356275303644,
"calib/std_conf": 0.009810945466831327,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5215863689776734,
"calib/step_q_c_n": 851.0,
"calib/step_q_gap": 0.08287412732425531,
"calib/step_q_w": 0.4387122416534181,
"calib/step_q_w_n": 629.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2991.0,
"completions/max_terminated_length": 2991.0,
"completions/mean_length": 528.73828125,
"completions/mean_terminated_length": 532.9015502929688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 207.0,
"epoch": 0.05226666666666667,
"grad_norm": 0.02996394783258438,
"learning_rate": 4.194444444444445e-06,
"loss": -0.0307,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.031121041625738144,
"mask/share_reasoning": 0.8388256430625916,
"mask/share_step_conf": 0.1222408264875412,
"num_tokens": 11917824.0,
"reward": 1.2675228118896484,
"reward_std": 0.2394319772720337,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6055535078048706,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8092772364616394,
"step": 49
},
{
"adv/mean_abs_final_conf": 0.6307613849639893,
"adv/mean_abs_reasoning": 0.38215214014053345,
"adv/mean_abs_step_conf": 0.742942214012146,
"adv/ratio_final_to_reasoning": 1.6505504449930117,
"adv/ratio_step_to_reasoning": 1.9441006237435563,
"adv/std_final_conf": 0.798173189163208,
"adv/std_reasoning": 0.6613245010375977,
"adv/std_step_conf": 0.9352914094924927,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5886194730125367,
"calib/avg_num_step_conf": 5.75,
"calib/ece": 0.29016000000000003,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.992,
"calib/gap": 0.0062382704001202605,
"calib/mean_conf": 0.9812000000000002,
"calib/mu_c": 0.983121387283237,
"calib/mu_w": 0.9768831168831167,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.28968000000000005,
"calib/std_conf": 0.015929846201391904,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5423921765295887,
"calib/step_q_c_n": 997.0,
"calib/step_q_gap": 0.07986586074011504,
"calib/step_q_w": 0.4625263157894737,
"calib/step_q_w_n": 475.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2902.0,
"completions/max_terminated_length": 2902.0,
"completions/mean_length": 541.9765625,
"completions/mean_terminated_length": 548.4031982421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 170.0,
"epoch": 0.05333333333333334,
"grad_norm": 0.031243592500686646,
"learning_rate": 4.166666666666667e-06,
"loss": -0.0862,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03204977139830589,
"mask/share_reasoning": 0.8309839963912964,
"mask/share_step_conf": 0.12524744868278503,
"num_tokens": 12161930.0,
"reward": 1.345628261566162,
"reward_std": 0.23302677273750305,
"rewards/accuracy_reward_step": 0.67578125,
"rewards/final_brier_reward_step": 0.6890945434570312,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8358466625213623,
"step": 50
},
{
"adv/mean_abs_final_conf": 0.6398917436599731,
"adv/mean_abs_reasoning": 0.3863760232925415,
"adv/mean_abs_step_conf": 0.7543948888778687,
"adv/ratio_final_to_reasoning": 1.6561372991188024,
"adv/ratio_step_to_reasoning": 1.9524888797426352,
"adv/std_final_conf": 0.8142772912979126,
"adv/std_reasoning": 0.6612588167190552,
"adv/std_step_conf": 0.9354052543640137,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.586183176100629,
"calib/avg_num_step_conf": 5.84765625,
"calib/ece": 0.36023529411764726,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.9882352941176471,
"calib/gap": 0.001495676100628951,
"calib/mean_conf": 0.9787450980392157,
"calib/mu_c": 0.979308176100629,
"calib/mu_w": 0.9778125000000001,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35772549019607863,
"calib/std_conf": 0.03563422189536946,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5560224719101123,
"calib/step_q_c_n": 890.0,
"calib/step_q_gap": 0.1265167058475093,
"calib/step_q_w": 0.42950576606260304,
"calib/step_q_w_n": 607.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2332.0,
"completions/max_terminated_length": 2332.0,
"completions/mean_length": 536.2578125,
"completions/mean_terminated_length": 536.2578125,
"completions/min_length": 175.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.0544,
"grad_norm": 0.027206402271986008,
"learning_rate": 4.138888888888889e-06,
"loss": 0.0269,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.030329890549182892,
"mask/share_reasoning": 0.8490919470787048,
"mask/share_step_conf": 0.12057814747095108,
"num_tokens": 12408508.0,
"reward": 1.3251616954803467,
"reward_std": 0.21376600861549377,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.635941743850708,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8458625674247742,
"step": 51
},
{
"adv/mean_abs_final_conf": 0.6179524064064026,
"adv/mean_abs_reasoning": 0.42776966094970703,
"adv/mean_abs_step_conf": 0.7270907163619995,
"adv/ratio_final_to_reasoning": 1.4445914771853243,
"adv/ratio_step_to_reasoning": 1.699724834967863,
"adv/std_final_conf": 0.8091193437576294,
"adv/std_reasoning": 0.7014753222465515,
"adv/std_step_conf": 0.9353761672973633,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5595611285266457,
"calib/avg_num_step_conf": 5.78125,
"calib/ece": 0.29079681274900404,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9960159362549801,
"calib/gap": 0.0013942379459623844,
"calib/mean_conf": 0.9808366533864543,
"calib/mu_c": 0.981264367816092,
"calib/mu_w": 0.9798701298701297,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2892031872509961,
"calib/std_conf": 0.027938770758738494,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6002297702297702,
"calib/step_q_c_n": 1001.0,
"calib/step_q_gap": 0.10457215018801652,
"calib/step_q_w": 0.4956576200417537,
"calib/step_q_w_n": 479.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3004.0,
"completions/max_terminated_length": 3004.0,
"completions/mean_length": 539.76953125,
"completions/mean_terminated_length": 541.8862915039062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 214.0,
"epoch": 0.055466666666666664,
"grad_norm": 0.030449653044342995,
"learning_rate": 4.111111111111111e-06,
"loss": 0.0344,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03135004639625549,
"mask/share_reasoning": 0.842849612236023,
"mask/share_step_conf": 0.12189410626888275,
"num_tokens": 12654641.0,
"reward": 1.3621165752410889,
"reward_std": 0.23896150290966034,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/final_brier_reward_step": 0.690671443939209,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8507653474807739,
"step": 52
},
{
"adv/mean_abs_final_conf": 0.6005265116691589,
"adv/mean_abs_reasoning": 0.35591036081314087,
"adv/mean_abs_step_conf": 0.7810007929801941,
"adv/ratio_final_to_reasoning": 1.6872970775482588,
"adv/ratio_step_to_reasoning": 2.194374985869639,
"adv/std_final_conf": 0.7910946011543274,
"adv/std_reasoning": 0.64017653465271,
"adv/std_step_conf": 0.9354143142700195,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.5703883495145631,
"calib/avg_num_step_conf": 5.72265625,
"calib/ece": 0.3865490196078431,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.996078431372549,
"calib/gap": 0.004896525293817233,
"calib/mean_conf": 0.9826274509803922,
"calib/mu_c": 0.9846052631578949,
"calib/mu_w": 0.9797087378640776,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3865490196078431,
"calib/std_conf": 0.015958230753959433,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5978512396694214,
"calib/step_q_c_n": 847.0,
"calib/step_q_gap": 0.06527842413544083,
"calib/step_q_w": 0.5325728155339806,
"calib/step_q_w_n": 618.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2317.0,
"completions/max_terminated_length": 2317.0,
"completions/mean_length": 543.703125,
"completions/mean_terminated_length": 543.703125,
"completions/min_length": 171.0,
"completions/min_terminated_length": 171.0,
"epoch": 0.05653333333333333,
"grad_norm": 0.02810330130159855,
"learning_rate": 4.083333333333334e-06,
"loss": -0.0089,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.02985529601573944,
"mask/share_reasoning": 0.8550183773040771,
"mask/share_step_conf": 0.11512631177902222,
"num_tokens": 12899653.0,
"reward": 1.2754499912261963,
"reward_std": 0.20352619886398315,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6095238327980042,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.811703622341156,
"step": 53
},
{
"adv/mean_abs_final_conf": 0.48447397351264954,
"adv/mean_abs_reasoning": 0.3503273129463196,
"adv/mean_abs_step_conf": 0.7499665021896362,
"adv/ratio_final_to_reasoning": 1.382918075779279,
"adv/ratio_step_to_reasoning": 2.140759439743007,
"adv/std_final_conf": 0.7099431157112122,
"adv/std_reasoning": 0.6401990056037903,
"adv/std_step_conf": 0.9351658225059509,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.6277173913043479,
"calib/avg_num_step_conf": 5.8828125,
"calib/ece": 0.251372549019608,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.9764705882352941,
"calib/gap": 0.04779929577464814,
"calib/mean_conf": 0.9729411764705883,
"calib/mu_c": 0.9862500000000002,
"calib/mu_w": 0.938450704225352,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.251372549019608,
"calib/std_conf": 0.09634688386236717,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5997388781431334,
"calib/step_q_c_n": 1034.0,
"calib/step_q_gap": 0.13666684424482833,
"calib/step_q_w": 0.4630720338983051,
"calib/step_q_w_n": 472.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2310.0,
"completions/max_terminated_length": 2310.0,
"completions/mean_length": 483.65234375,
"completions/mean_terminated_length": 483.65234375,
"completions/min_length": 182.0,
"completions/min_terminated_length": 182.0,
"epoch": 0.0576,
"grad_norm": 0.030784349888563156,
"learning_rate": 4.055555555555556e-06,
"loss": 0.0105,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03423851728439331,
"mask/share_reasoning": 0.8335533142089844,
"mask/share_step_conf": 0.13220810890197754,
"num_tokens": 13129700.0,
"reward": 1.4109441041946411,
"reward_std": 0.198542058467865,
"rewards/accuracy_reward_step": 0.71875,
"rewards/final_brier_reward_step": 0.7428370714187622,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8684317469596863,
"step": 54
},
{
"adv/mean_abs_final_conf": 0.5835695266723633,
"adv/mean_abs_reasoning": 0.4531455636024475,
"adv/mean_abs_step_conf": 0.7731038928031921,
"adv/ratio_final_to_reasoning": 1.2878191326271904,
"adv/ratio_step_to_reasoning": 1.7060828901360483,
"adv/std_final_conf": 0.8207448124885559,
"adv/std_reasoning": 0.7205977439880371,
"adv/std_step_conf": 0.9359610676765442,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.6237734165923283,
"calib/avg_num_step_conf": 5.60546875,
"calib/ece": 0.44613545816733063,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9800796812749004,
"calib/gap": 0.02398814833694407,
"calib/mean_conf": 0.9760159362549802,
"calib/mu_c": 0.9872932330827068,
"calib/mu_w": 0.9633050847457627,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44613545816733063,
"calib/std_conf": 0.07534293274727284,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5862,
"calib/step_q_c_n": 700.0,
"calib/step_q_gap": 0.12093469387755101,
"calib/step_q_w": 0.46526530612244904,
"calib/step_q_w_n": 735.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2808.0,
"completions/max_terminated_length": 2808.0,
"completions/mean_length": 525.796875,
"completions/mean_terminated_length": 529.93701171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 178.0,
"epoch": 0.058666666666666666,
"grad_norm": 0.03521295264363289,
"learning_rate": 4.027777777777779e-06,
"loss": -0.0709,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03193287178874016,
"mask/share_reasoning": 0.8432319760322571,
"mask/share_step_conf": 0.11702266335487366,
"num_tokens": 13372128.0,
"reward": 1.231523036956787,
"reward_std": 0.2821798324584961,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5472296476364136,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8079081177711487,
"step": 55
},
{
"adv/mean_abs_final_conf": 0.5679474472999573,
"adv/mean_abs_reasoning": 0.5074265599250793,
"adv/mean_abs_step_conf": 0.7627795338630676,
"adv/ratio_final_to_reasoning": 1.1192702395866188,
"adv/ratio_step_to_reasoning": 1.503231391702655,
"adv/std_final_conf": 0.786614716053009,
"adv/std_reasoning": 0.7575352191925049,
"adv/std_step_conf": 0.9355356097221375,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6151002506265665,
"calib/avg_num_step_conf": 5.94140625,
"calib/ece": 0.4554940711462451,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.9920948616600791,
"calib/gap": 0.01192543859649109,
"calib/mean_conf": 0.9811857707509883,
"calib/mu_c": 0.9868421052631579,
"calib/mu_w": 0.9749166666666668,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4554940711462451,
"calib/std_conf": 0.033753598477446724,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.551680981595092,
"calib/step_q_c_n": 815.0,
"calib/step_q_gap": 0.05923055666591365,
"calib/step_q_w": 0.4924504249291784,
"calib/step_q_w_n": 706.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2259.0,
"completions/max_terminated_length": 2259.0,
"completions/mean_length": 535.55859375,
"completions/mean_terminated_length": 537.6588745117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 204.0,
"epoch": 0.05973333333333333,
"grad_norm": 0.027493732050061226,
"learning_rate": 4.000000000000001e-06,
"loss": -0.0323,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03047093003988266,
"mask/share_reasoning": 0.8427596688270569,
"mask/share_step_conf": 0.12286314368247986,
"num_tokens": 13616071.0,
"reward": 1.2343281507492065,
"reward_std": 0.2484326958656311,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5414933562278748,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8131908178329468,
"step": 56
},
{
"adv/mean_abs_final_conf": 0.5659802556037903,
"adv/mean_abs_reasoning": 0.4423058032989502,
"adv/mean_abs_step_conf": 0.7457073926925659,
"adv/ratio_final_to_reasoning": 1.2796129993828043,
"adv/ratio_step_to_reasoning": 1.6859543490740716,
"adv/std_final_conf": 0.8015991449356079,
"adv/std_reasoning": 0.7013233304023743,
"adv/std_step_conf": 0.9359928965568542,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.6317563102947006,
"calib/avg_num_step_conf": 5.8203125,
"calib/ece": 0.347843137254902,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.984313725490196,
"calib/gap": 0.019145632350997976,
"calib/mean_conf": 0.9792156862745097,
"calib/mu_c": 0.986273291925466,
"calib/mu_w": 0.967127659574468,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.347843137254902,
"calib/std_conf": 0.05214020078839857,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5801402373247033,
"calib/step_q_c_n": 927.0,
"calib/step_q_gap": 0.085859597893087,
"calib/step_q_w": 0.4942806394316163,
"calib/step_q_w_n": 563.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2627.0,
"completions/max_terminated_length": 2627.0,
"completions/mean_length": 521.7265625,
"completions/mean_terminated_length": 521.7265625,
"completions/min_length": 147.0,
"completions/min_terminated_length": 147.0,
"epoch": 0.0608,
"grad_norm": 0.02348768338561058,
"learning_rate": 3.972222222222223e-06,
"loss": 0.0309,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03096698224544525,
"mask/share_reasoning": 0.8459857702255249,
"mask/share_step_conf": 0.12304721772670746,
"num_tokens": 13856425.0,
"reward": 1.303645372390747,
"reward_std": 0.2623915672302246,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.6460027098655701,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8185346126556396,
"step": 57
},
{
"adv/mean_abs_final_conf": 0.6994427442550659,
"adv/mean_abs_reasoning": 0.5924781560897827,
"adv/mean_abs_step_conf": 0.7661586999893188,
"adv/ratio_final_to_reasoning": 1.180537606434682,
"adv/ratio_step_to_reasoning": 1.2931425270524521,
"adv/std_final_conf": 0.8900883197784424,
"adv/std_reasoning": 0.8265689611434937,
"adv/std_step_conf": 0.9355772137641907,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5210778061224489,
"calib/avg_num_step_conf": 7.28125,
"calib/ece": 0.4255555555555556,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9880952380952381,
"calib/gap": 0.000571428571428223,
"calib/mean_conf": 0.9788888888888889,
"calib/mu_c": 0.979142857142857,
"calib/mu_w": 0.9785714285714288,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42444444444444446,
"calib/std_conf": 0.023978238223214852,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5137659574468085,
"calib/step_q_c_n": 940.0,
"calib/step_q_gap": 0.02425297043382152,
"calib/step_q_w": 0.489512987012987,
"calib/step_q_w_n": 924.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2888.0,
"completions/max_terminated_length": 2888.0,
"completions/mean_length": 599.9921875,
"completions/mean_terminated_length": 602.3451538085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 186.0,
"epoch": 0.06186666666666667,
"grad_norm": 0.02133341133594513,
"learning_rate": 3.944444444444445e-06,
"loss": 0.0349,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.027864959090948105,
"mask/share_reasoning": 0.8431490063667297,
"mask/share_step_conf": 0.12507976591587067,
"num_tokens": 14116343.0,
"reward": 1.2402286529541016,
"reward_std": 0.30855613946914673,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.5645425319671631,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8052229881286621,
"step": 58
},
{
"adv/mean_abs_final_conf": 0.5506974458694458,
"adv/mean_abs_reasoning": 0.37096497416496277,
"adv/mean_abs_step_conf": 0.7607747912406921,
"adv/ratio_final_to_reasoning": 1.4844998429004206,
"adv/ratio_step_to_reasoning": 2.050799520771971,
"adv/std_final_conf": 0.7599132061004639,
"adv/std_reasoning": 0.6403107047080994,
"adv/std_step_conf": 0.9347589015960693,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5710547504025765,
"calib/avg_num_step_conf": 6.234375,
"calib/ece": 0.34059055118110243,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.984251968503937,
"calib/gap": 0.015193236714975522,
"calib/mean_conf": 0.9783858267716535,
"calib/mu_c": 0.9838888888888888,
"calib/mu_w": 0.9686956521739133,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34059055118110243,
"calib/std_conf": 0.06270912066731307,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6074947589098533,
"calib/step_q_c_n": 954.0,
"calib/step_q_gap": 0.10428603616842025,
"calib/step_q_w": 0.503208722741433,
"calib/step_q_w_n": 642.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2898.0,
"completions/max_terminated_length": 2898.0,
"completions/mean_length": 562.83984375,
"completions/mean_terminated_length": 562.83984375,
"completions/min_length": 135.0,
"completions/min_terminated_length": 135.0,
"epoch": 0.06293333333333333,
"grad_norm": 0.02957838773727417,
"learning_rate": 3.916666666666667e-06,
"loss": 0.0696,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.031655244529247284,
"mask/share_reasoning": 0.8429268002510071,
"mask/share_step_conf": 0.12541794776916504,
"num_tokens": 14366678.0,
"reward": 1.3069705963134766,
"reward_std": 0.21562360227108002,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/final_brier_reward_step": 0.6509472727775574,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8189969062805176,
"step": 59
},
{
"adv/mean_abs_final_conf": 0.6753536462783813,
"adv/mean_abs_reasoning": 0.5778477191925049,
"adv/mean_abs_step_conf": 0.7475611567497253,
"adv/ratio_final_to_reasoning": 1.1687398320480231,
"adv/ratio_step_to_reasoning": 1.2936992427596343,
"adv/std_final_conf": 0.8445501327514648,
"adv/std_reasoning": 0.7929325699806213,
"adv/std_step_conf": 0.9358199834823608,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.6227464140018423,
"calib/avg_num_step_conf": 6.0,
"calib/ece": 0.3924302788844621,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9840637450199203,
"calib/gap": 0.005486248190551213,
"calib/mean_conf": 0.9701195219123507,
"calib/mu_c": 0.9723489932885904,
"calib/mu_w": 0.9668627450980392,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.38446215139442225,
"calib/std_conf": 0.10780121173883471,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.574626168224299,
"calib/step_q_c_n": 856.0,
"calib/step_q_gap": 0.08472910940076961,
"calib/step_q_w": 0.4898970588235294,
"calib/step_q_w_n": 680.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2702.0,
"completions/max_terminated_length": 2702.0,
"completions/mean_length": 532.43359375,
"completions/mean_terminated_length": 536.6259765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 209.0,
"epoch": 0.064,
"grad_norm": 0.031429000198841095,
"learning_rate": 3.88888888888889e-06,
"loss": -0.049,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03032679483294487,
"mask/share_reasoning": 0.838133692741394,
"mask/share_step_conf": 0.12372701615095139,
"num_tokens": 14611837.0,
"reward": 1.285160779953003,
"reward_std": 0.31759822368621826,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.5961679220199585,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8308267593383789,
"step": 60
},
{
"adv/mean_abs_final_conf": 0.5964518785476685,
"adv/mean_abs_reasoning": 0.4378543496131897,
"adv/mean_abs_step_conf": 0.7456066012382507,
"adv/ratio_final_to_reasoning": 1.362215264218769,
"adv/ratio_step_to_reasoning": 1.702864438589993,
"adv/std_final_conf": 0.7979384660720825,
"adv/std_reasoning": 0.7013479471206665,
"adv/std_step_conf": 0.9356203675270081,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.6158138268916713,
"calib/avg_num_step_conf": 5.67578125,
"calib/ece": 0.3293725490196078,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.996078431372549,
"calib/gap": 0.007047495917256175,
"calib/mean_conf": 0.9842745098039215,
"calib/mu_c": 0.9867065868263472,
"calib/mu_w": 0.979659090909091,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3293725490196078,
"calib/std_conf": 0.013021970019589218,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6622426868905742,
"calib/step_q_c_n": 923.0,
"calib/step_q_gap": 0.11929929066415912,
"calib/step_q_w": 0.5429433962264151,
"calib/step_q_w_n": 530.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1737.0,
"completions/max_terminated_length": 1737.0,
"completions/mean_length": 459.84765625,
"completions/mean_terminated_length": 459.84765625,
"completions/min_length": 119.0,
"completions/min_terminated_length": 119.0,
"epoch": 0.06506666666666666,
"grad_norm": 0.026003647595643997,
"learning_rate": 3.861111111111112e-06,
"loss": -0.0255,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.036012012511491776,
"mask/share_reasoning": 0.827039361000061,
"mask/share_step_conf": 0.1369486153125763,
"num_tokens": 14833622.0,
"reward": 1.3126163482666016,
"reward_std": 0.2654881477355957,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.6656820178031921,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8153221607208252,
"step": 61
},
{
"adv/mean_abs_final_conf": 0.6631834506988525,
"adv/mean_abs_reasoning": 0.5495285987854004,
"adv/mean_abs_step_conf": 0.7788981199264526,
"adv/ratio_final_to_reasoning": 1.2068224513968129,
"adv/ratio_step_to_reasoning": 1.4173932378551688,
"adv/std_final_conf": 0.8523333072662354,
"adv/std_reasoning": 0.7753791213035583,
"adv/std_step_conf": 0.9356539249420166,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.517351069982649,
"calib/avg_num_step_conf": 6.375,
"calib/ece": 0.44724,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.98,
"calib/gap": 0.004370541738962785,
"calib/mean_conf": 0.9741200000000001,
"calib/mu_c": 0.9761654135338345,
"calib/mu_w": 0.9717948717948717,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.44468,
"calib/std_conf": 0.07066700503063647,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5814967741935484,
"calib/step_q_c_n": 775.0,
"calib/step_q_gap": 0.0856858056987993,
"calib/step_q_w": 0.4958109684947491,
"calib/step_q_w_n": 857.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3007.0,
"completions/max_terminated_length": 3007.0,
"completions/mean_length": 560.5859375,
"completions/mean_terminated_length": 567.2332153320312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 197.0,
"epoch": 0.06613333333333334,
"grad_norm": 0.03154468163847923,
"learning_rate": 3.833333333333334e-06,
"loss": -0.0568,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.029570816084742546,
"mask/share_reasoning": 0.8387907147407532,
"mask/share_step_conf": 0.11991971731185913,
"num_tokens": 15084212.0,
"reward": 1.1733200550079346,
"reward_std": 0.3146999180316925,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5397816300392151,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7538197636604309,
"step": 62
},
{
"adv/mean_abs_final_conf": 0.6640254259109497,
"adv/mean_abs_reasoning": 0.475774347782135,
"adv/mean_abs_step_conf": 0.7512675523757935,
"adv/ratio_final_to_reasoning": 1.3956730307263603,
"adv/ratio_step_to_reasoning": 1.5790417366507774,
"adv/std_final_conf": 0.8391026258468628,
"adv/std_reasoning": 0.7393614053726196,
"adv/std_step_conf": 0.9358152747154236,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6436034658511723,
"calib/avg_num_step_conf": 6.12890625,
"calib/ece": 0.4007905138339921,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.9525691699604744,
"calib/gap": 0.02507581549439386,
"calib/mean_conf": 0.9699604743083005,
"calib/mu_c": 0.980763888888889,
"calib/mu_w": 0.9556880733944951,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4007905138339921,
"calib/std_conf": 0.06619230591708436,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5627252252252253,
"calib/step_q_c_n": 888.0,
"calib/step_q_gap": 0.03624945430011517,
"calib/step_q_w": 0.5264757709251101,
"calib/step_q_w_n": 681.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2497.0,
"completions/max_terminated_length": 2497.0,
"completions/mean_length": 585.31640625,
"completions/mean_terminated_length": 587.61181640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 184.0,
"epoch": 0.0672,
"grad_norm": 0.03112921118736267,
"learning_rate": 3.8055555555555556e-06,
"loss": -0.0232,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.02907847985625267,
"mask/share_reasoning": 0.8485139608383179,
"mask/share_step_conf": 0.11850129067897797,
"num_tokens": 15342693.0,
"reward": 1.2638373374938965,
"reward_std": 0.2874062657356262,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.5950124859809875,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8112529516220093,
"step": 63
},
{
"adv/mean_abs_final_conf": 0.6524174213409424,
"adv/mean_abs_reasoning": 0.5324134826660156,
"adv/mean_abs_step_conf": 0.7383209466934204,
"adv/ratio_final_to_reasoning": 1.2253961302294922,
"adv/ratio_step_to_reasoning": 1.3867435193345228,
"adv/std_final_conf": 0.8418984413146973,
"adv/std_reasoning": 0.7927736043930054,
"adv/std_step_conf": 0.9360528588294983,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5773400119617225,
"calib/avg_num_step_conf": 5.52734375,
"calib/ece": 0.2762301587301588,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.9722222222222222,
"calib/gap": 0.028032296650717825,
"calib/mean_conf": 0.97234126984127,
"calib/mu_c": 0.9807954545454546,
"calib/mu_w": 0.9527631578947368,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2750793650793652,
"calib/std_conf": 0.07486810229493363,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6347746781115882,
"calib/step_q_c_n": 932.0,
"calib/step_q_gap": 0.08777674850496286,
"calib/step_q_w": 0.5469979296066253,
"calib/step_q_w_n": 483.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2466.0,
"completions/max_terminated_length": 2466.0,
"completions/mean_length": 513.578125,
"completions/mean_terminated_length": 517.6220703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 182.0,
"epoch": 0.06826666666666667,
"grad_norm": 0.036455199122428894,
"learning_rate": 3.777777777777778e-06,
"loss": 0.0045,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.031546011567115784,
"mask/share_reasoning": 0.845130205154419,
"mask/share_step_conf": 0.11551132053136826,
"num_tokens": 15577945.0,
"reward": 1.3307418823242188,
"reward_std": 0.30955109000205994,
"rewards/accuracy_reward_step": 0.6875,
"rewards/final_brier_reward_step": 0.7089675664901733,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8098517656326294,
"step": 64
},
{
"adv/mean_abs_final_conf": 0.4953068494796753,
"adv/mean_abs_reasoning": 0.295492023229599,
"adv/mean_abs_step_conf": 0.7505620718002319,
"adv/ratio_final_to_reasoning": 1.6762105591419605,
"adv/ratio_step_to_reasoning": 2.5400417364804495,
"adv/std_final_conf": 0.7653894424438477,
"adv/std_reasoning": 0.5959096550941467,
"adv/std_step_conf": 0.9354201555252075,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.5859900373599003,
"calib/avg_num_step_conf": 5.49609375,
"calib/ece": 0.4091796875000001,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 1.0,
"calib/frac_conf_gt_0.9": 0.96875,
"calib/gap": 0.010745952677459591,
"calib/mean_conf": 0.9794921875000001,
"calib/mu_c": 0.9841095890410959,
"calib/mu_w": 0.9733636363636363,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4091796875000001,
"calib/std_conf": 0.028670677555035978,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.672340425531915,
"calib/step_q_c_n": 799.0,
"calib/step_q_gap": 0.03360687290033604,
"calib/step_q_w": 0.6387335526315789,
"calib/step_q_w_n": 608.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1161.0,
"completions/max_terminated_length": 1161.0,
"completions/mean_length": 444.41796875,
"completions/mean_terminated_length": 446.1607971191406,
"completions/min_length": 0.0,
"completions/min_terminated_length": 173.0,
"epoch": 0.06933333333333333,
"grad_norm": 0.036643534898757935,
"learning_rate": 3.7500000000000005e-06,
"loss": -0.0162,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.036343321204185486,
"mask/share_reasoning": 0.828540563583374,
"mask/share_step_conf": 0.13120990991592407,
"num_tokens": 15796740.0,
"reward": 1.2438958883285522,
"reward_std": 0.20089185237884521,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.5919605493545532,
"rewards/format_reward_step": 1.0,
"rewards/step_l2_reward": 0.7908843159675598,
"step": 65
},
{
"adv/mean_abs_final_conf": 0.6261056661605835,
"adv/mean_abs_reasoning": 0.36768579483032227,
"adv/mean_abs_step_conf": 0.7685141563415527,
"adv/ratio_final_to_reasoning": 1.7028279986979522,
"adv/ratio_step_to_reasoning": 2.090138284227713,
"adv/std_final_conf": 0.8404014706611633,
"adv/std_reasoning": 0.6612588167190552,
"adv/std_step_conf": 0.9333556294441223,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6741760000000001,
"calib/avg_num_step_conf": 6.80078125,
"calib/ece": 0.4415200000000001,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.88,
"calib/gap": 0.06288000000000016,
"calib/mean_conf": 0.9415200000000001,
"calib/mu_c": 0.9729599999999999,
"calib/mu_w": 0.9100799999999998,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4415200000000001,
"calib/std_conf": 0.13628532422825282,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6151608751608751,
"calib/step_q_c_n": 777.0,
"calib/step_q_gap": 0.18636419466294984,
"calib/step_q_w": 0.4287966804979253,
"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": 0.00390625,
"completions/max_length": 3064.0,
"completions/max_terminated_length": 3064.0,
"completions/mean_length": 634.59375,
"completions/mean_terminated_length": 637.0823974609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 176.0,
"epoch": 0.0704,
"grad_norm": 0.033459797501564026,
"learning_rate": 3.7222222222222225e-06,
"loss": 0.0443,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.0286815594881773,
"mask/share_reasoning": 0.8461880683898926,
"mask/share_step_conf": 0.12122409045696259,
"num_tokens": 16065548.0,
"reward": 1.1990708112716675,
"reward_std": 0.2781738042831421,
"rewards/accuracy_reward_step": 0.48828125,
"rewards/final_brier_reward_step": 0.5546156167984009,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7752785682678223,
"step": 66
},
{
"adv/mean_abs_final_conf": 0.5760310888290405,
"adv/mean_abs_reasoning": 0.3161531090736389,
"adv/mean_abs_step_conf": 0.7715531587600708,
"adv/ratio_final_to_reasoning": 1.8220003925214299,
"adv/ratio_step_to_reasoning": 2.4404414716046943,
"adv/std_final_conf": 0.7942258715629578,
"adv/std_reasoning": 0.6186121702194214,
"adv/std_step_conf": 0.9347555637359619,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7356792144026186,
"calib/avg_num_step_conf": 6.1328125,
"calib/ece": 0.33548,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.908,
"calib/gap": 0.053210583742498985,
"calib/mean_conf": 0.95948,
"calib/mu_c": 0.9794871794871796,
"calib/mu_w": 0.9262765957446806,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.33548,
"calib/std_conf": 0.07957970595572718,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6555046826222686,
"calib/step_q_c_n": 961.0,
"calib/step_q_gap": 0.1265555857421372,
"calib/step_q_w": 0.5289490968801314,
"calib/step_q_w_n": 609.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2415.0,
"completions/max_terminated_length": 2415.0,
"completions/mean_length": 553.71484375,
"completions/mean_terminated_length": 560.2806396484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 207.0,
"epoch": 0.07146666666666666,
"grad_norm": 0.02705869823694229,
"learning_rate": 3.694444444444445e-06,
"loss": -0.078,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.029447246342897415,
"mask/share_reasoning": 0.8373123407363892,
"mask/share_step_conf": 0.12152165174484253,
"num_tokens": 16312307.0,
"reward": 1.3032801151275635,
"reward_std": 0.25523093342781067,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.655727744102478,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8164318799972534,
"step": 67
},
{
"adv/mean_abs_final_conf": 0.6699117422103882,
"adv/mean_abs_reasoning": 0.3923490047454834,
"adv/mean_abs_step_conf": 0.7669743299484253,
"adv/ratio_final_to_reasoning": 1.707438362549077,
"adv/ratio_step_to_reasoning": 1.9548267503468275,
"adv/std_final_conf": 0.8446117639541626,
"adv/std_reasoning": 0.6613078713417053,
"adv/std_step_conf": 0.9356321692466736,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7924491392801252,
"calib/avg_num_step_conf": 5.9375,
"calib/ece": 0.3712000000000002,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.86,
"calib/gap": 0.08661971830985882,
"calib/mean_conf": 0.9392,
"calib/mu_c": 0.9766197183098588,
"calib/mu_w": 0.89,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.3712000000000002,
"calib/std_conf": 0.11468984261912647,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.6703259452411995,
"calib/step_q_c_n": 767.0,
"calib/step_q_gap": 0.19024626396629907,
"calib/step_q_w": 0.4800796812749004,
"calib/step_q_w_n": 753.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2676.0,
"completions/max_terminated_length": 2676.0,
"completions/mean_length": 539.84765625,
"completions/mean_terminated_length": 541.9647216796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 163.0,
"epoch": 0.07253333333333334,
"grad_norm": 0.035893820226192474,
"learning_rate": 3.6666666666666666e-06,
"loss": 0.0214,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03266291320323944,
"mask/share_reasoning": 0.8366715908050537,
"mask/share_step_conf": 0.12675921618938446,
"num_tokens": 16554596.0,
"reward": 1.2877839803695679,
"reward_std": 0.25560203194618225,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.6310445070266724,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8191367387771606,
"step": 68
},
{
"adv/mean_abs_final_conf": 0.7307136058807373,
"adv/mean_abs_reasoning": 0.4282193183898926,
"adv/mean_abs_step_conf": 0.7660757303237915,
"adv/ratio_final_to_reasoning": 1.706400375929384,
"adv/ratio_step_to_reasoning": 1.7889798461317468,
"adv/std_final_conf": 0.8997459411621094,
"adv/std_reasoning": 0.7013484835624695,
"adv/std_step_conf": 0.9359602928161621,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7487881107284092,
"calib/avg_num_step_conf": 6.60546875,
"calib/ece": 0.38139442231075693,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.601593625498008,
"calib/gap": 0.13963324403622912,
"calib/mean_conf": 0.8384462151394423,
"calib/mu_c": 0.912991452991453,
"calib/mu_w": 0.7733582089552239,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3768525896414342,
"calib/std_conf": 0.20519637881818759,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6215023474178404,
"calib/step_q_c_n": 639.0,
"calib/step_q_gap": 0.19641679608704188,
"calib/step_q_w": 0.4250855513307985,
"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": 0.00390625,
"completions/max_length": 2853.0,
"completions/max_terminated_length": 2853.0,
"completions/mean_length": 630.01171875,
"completions/mean_terminated_length": 632.4823608398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 235.0,
"epoch": 0.0736,
"grad_norm": 0.03961062431335449,
"learning_rate": 3.638888888888889e-06,
"loss": 0.0818,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.027123235166072845,
"mask/share_reasoning": 0.8559575080871582,
"mask/share_step_conf": 0.11301306635141373,
"num_tokens": 16820375.0,
"reward": 1.26645028591156,
"reward_std": 0.23711402714252472,
"rewards/accuracy_reward_step": 0.45703125,
"rewards/final_brier_reward_step": 0.6274238228797913,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8089882731437683,
"step": 69
},
{
"adv/mean_abs_final_conf": 0.7056593894958496,
"adv/mean_abs_reasoning": 0.41198545694351196,
"adv/mean_abs_step_conf": 0.7569766044616699,
"adv/ratio_final_to_reasoning": 1.7128259689822105,
"adv/ratio_step_to_reasoning": 1.837386712816565,
"adv/std_final_conf": 0.8766942024230957,
"adv/std_reasoning": 0.6816877722740173,
"adv/std_step_conf": 0.9351736307144165,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8069104349547884,
"calib/avg_num_step_conf": 6.0234375,
"calib/ece": 0.20653225806451614,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.5201612903225806,
"calib/gap": 0.3008685895320442,
"calib/mean_conf": 0.7592741935483871,
"calib/mu_c": 0.8915107913669066,
"calib/mu_w": 0.5906422018348624,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.20266129032258062,
"calib/std_conf": 0.28556727185267905,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5986434108527132,
"calib/step_q_c_n": 774.0,
"calib/step_q_gap": 0.20463299418604652,
"calib/step_q_w": 0.3940104166666667,
"calib/step_q_w_n": 768.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2552.0,
"completions/max_terminated_length": 2552.0,
"completions/mean_length": 609.78125,
"completions/mean_terminated_length": 612.172607421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 142.0,
"epoch": 0.07466666666666667,
"grad_norm": 0.03800712525844574,
"learning_rate": 3.6111111111111115e-06,
"loss": -0.0097,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.030685387551784515,
"mask/share_reasoning": 0.8435661792755127,
"mask/share_step_conf": 0.12184222042560577,
"num_tokens": 17083471.0,
"reward": 1.3623967170715332,
"reward_std": 0.23922546207904816,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.7564241886138916,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8330126404762268,
"step": 70
},
{
"adv/mean_abs_final_conf": 0.7434986233711243,
"adv/mean_abs_reasoning": 0.46663838624954224,
"adv/mean_abs_step_conf": 0.7635763883590698,
"adv/ratio_final_to_reasoning": 1.5933078916776613,
"adv/ratio_step_to_reasoning": 1.6363342812323531,
"adv/std_final_conf": 0.9111484885215759,
"adv/std_reasoning": 0.7392783164978027,
"adv/std_step_conf": 0.9355947971343994,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.6251235788433019,
"calib/avg_num_step_conf": 6.484375,
"calib/ece": 0.2691372549019608,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.4666666666666667,
"calib/gap": 0.0944432773109245,
"calib/mean_conf": 0.7530588235294118,
"calib/mu_c": 0.7971323529411766,
"calib/mu_w": 0.7026890756302521,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.24443137254901962,
"calib/std_conf": 0.2693813836795812,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.515,
"calib/step_q_c_n": 890.0,
"calib/step_q_gap": 0.05715584415584413,
"calib/step_q_w": 0.4578441558441559,
"calib/step_q_w_n": 770.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2472.0,
"completions/max_terminated_length": 2472.0,
"completions/mean_length": 563.7109375,
"completions/mean_terminated_length": 563.7109375,
"completions/min_length": 225.0,
"completions/min_terminated_length": 225.0,
"epoch": 0.07573333333333333,
"grad_norm": 0.0358133465051651,
"learning_rate": 3.5833333333333335e-06,
"loss": -0.0344,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03034607507288456,
"mask/share_reasoning": 0.8403059244155884,
"mask/share_step_conf": 0.1293479949235916,
"num_tokens": 17332189.0,
"reward": 1.295250415802002,
"reward_std": 0.2442229986190796,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.6744769811630249,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8056682348251343,
"step": 71
},
{
"adv/mean_abs_final_conf": 0.7388667464256287,
"adv/mean_abs_reasoning": 0.3799426555633545,
"adv/mean_abs_step_conf": 0.7495532035827637,
"adv/ratio_final_to_reasoning": 1.9446796394315997,
"adv/ratio_step_to_reasoning": 1.972806139577496,
"adv/std_final_conf": 0.9300165176391602,
"adv/std_reasoning": 0.681516170501709,
"adv/std_step_conf": 0.9357799887657166,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.798740157480315,
"calib/avg_num_step_conf": 5.90625,
"calib/ece": 0.18142857142857138,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.36904761904761907,
"calib/gap": 0.29151874015748,
"calib/mean_conf": 0.6754761904761906,
"calib/mu_c": 0.8200787401574801,
"calib/mu_w": 0.52856,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.1764682539682539,
"calib/std_conf": 0.2909224713798188,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5878830083565459,
"calib/step_q_c_n": 718.0,
"calib/step_q_gap": 0.16847494790314543,
"calib/step_q_w": 0.4194080604534005,
"calib/step_q_w_n": 794.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2354.0,
"completions/max_terminated_length": 2354.0,
"completions/mean_length": 524.52734375,
"completions/mean_terminated_length": 528.657470703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 229.0,
"epoch": 0.0768,
"grad_norm": 0.08097515255212784,
"learning_rate": 3.555555555555556e-06,
"loss": -0.0603,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.029277801513671875,
"mask/share_reasoning": 0.8439998626708984,
"mask/share_step_conf": 0.11890986561775208,
"num_tokens": 17570876.0,
"reward": 1.3871753215789795,
"reward_std": 0.23918916285037994,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.7695007920265198,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8539873957633972,
"step": 72
},
{
"adv/mean_abs_final_conf": 0.740119457244873,
"adv/mean_abs_reasoning": 0.476161926984787,
"adv/mean_abs_step_conf": 0.7462303638458252,
"adv/ratio_final_to_reasoning": 1.5543440483189226,
"adv/ratio_step_to_reasoning": 1.5671777216023965,
"adv/std_final_conf": 0.9278834462165833,
"adv/std_reasoning": 0.7207294702529907,
"adv/std_step_conf": 0.9352030158042908,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.788538984617416,
"calib/avg_num_step_conf": 6.16796875,
"calib/ece": 0.08587301587301585,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.30952380952380953,
"calib/gap": 0.3112537136066548,
"calib/mean_conf": 0.6396825396825397,
"calib/mu_c": 0.7619607843137255,
"calib/mu_w": 0.45070707070707067,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.059206349206349165,
"calib/std_conf": 0.3082166742189103,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5835434007134364,
"calib/step_q_c_n": 841.0,
"calib/step_q_gap": 0.22873310261045532,
"calib/step_q_w": 0.35481029810298104,
"calib/step_q_w_n": 738.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2976.0,
"completions/max_terminated_length": 2976.0,
"completions/mean_length": 553.109375,
"completions/mean_terminated_length": 553.109375,
"completions/min_length": 235.0,
"completions/min_terminated_length": 235.0,
"epoch": 0.07786666666666667,
"grad_norm": 0.0581560842692852,
"learning_rate": 3.5277777777777784e-06,
"loss": -0.0039,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.029643334448337555,
"mask/share_reasoning": 0.8481807708740234,
"mask/share_step_conf": 0.1221759170293808,
"num_tokens": 17819504.0,
"reward": 1.4292724132537842,
"reward_std": 0.21470966935157776,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.7984195351600647,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8722500801086426,
"step": 73
},
{
"adv/mean_abs_final_conf": 0.7195020914077759,
"adv/mean_abs_reasoning": 0.5106623768806458,
"adv/mean_abs_step_conf": 0.7628088593482971,
"adv/ratio_final_to_reasoning": 1.4089584899573306,
"adv/ratio_step_to_reasoning": 1.4937635782136034,
"adv/std_final_conf": 0.9145419001579285,
"adv/std_reasoning": 0.7752376198768616,
"adv/std_step_conf": 0.935356080532074,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7208216619981327,
"calib/avg_num_step_conf": 5.70703125,
"calib/ece": 0.16078740157480317,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.2204724409448819,
"calib/gap": 0.2381313414254589,
"calib/mean_conf": 0.5514173228346457,
"calib/mu_c": 0.6779831932773107,
"calib/mu_w": 0.4398518518518518,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.12185039370078743,
"calib/std_conf": 0.3114955653921489,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.49009921259842515,
"calib/step_q_c_n": 635.0,
"calib/step_q_gap": 0.09882802615774716,
"calib/step_q_w": 0.391271186440678,
"calib/step_q_w_n": 826.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1243.0,
"completions/max_terminated_length": 1243.0,
"completions/mean_length": 511.04296875,
"completions/mean_terminated_length": 513.047119140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 172.0,
"epoch": 0.07893333333333333,
"grad_norm": 0.04147012531757355,
"learning_rate": 3.5e-06,
"loss": -0.047,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03193596377968788,
"mask/share_reasoning": 0.8396960496902466,
"mask/share_step_conf": 0.12446170300245285,
"num_tokens": 18054259.0,
"reward": 1.3750741481781006,
"reward_std": 0.20677241683006287,
"rewards/accuracy_reward_step": 0.46484375,
"rewards/final_brier_reward_step": 0.7596992254257202,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8495213985443115,
"step": 74
},
{
"adv/mean_abs_final_conf": 0.7395988702774048,
"adv/mean_abs_reasoning": 0.3775707483291626,
"adv/mean_abs_step_conf": 0.7631238698959351,
"adv/ratio_final_to_reasoning": 1.958835194596774,
"adv/ratio_step_to_reasoning": 2.0211413974015033,
"adv/std_final_conf": 0.909938633441925,
"adv/std_reasoning": 0.6612691283226013,
"adv/std_step_conf": 0.9357552528381348,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7381803589668758,
"calib/avg_num_step_conf": 5.8515625,
"calib/ece": 0.15396078431372548,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.4117647058823529,
"calib/gap": 0.3014767255216695,
"calib/mean_conf": 0.656156862745098,
"calib/mu_c": 0.7471910112359551,
"calib/mu_w": 0.4457142857142856,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.056039215686274534,
"calib/std_conf": 0.3397532768341895,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5519630709426628,
"calib/step_q_c_n": 1029.0,
"calib/step_q_gap": 0.19177117328807858,
"calib/step_q_w": 0.36019189765458426,
"calib/step_q_w_n": 469.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2300.0,
"completions/max_terminated_length": 2300.0,
"completions/mean_length": 493.79296875,
"completions/mean_terminated_length": 493.79296875,
"completions/min_length": 212.0,
"completions/min_terminated_length": 212.0,
"epoch": 0.08,
"grad_norm": 0.05103519186377525,
"learning_rate": 3.4722222222222224e-06,
"loss": 0.0104,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03264331445097923,
"mask/share_reasoning": 0.8353633284568787,
"mask/share_step_conf": 0.1319933533668518,
"num_tokens": 18285422.0,
"reward": 1.4125370979309082,
"reward_std": 0.2209436148405075,
"rewards/accuracy_reward_step": 0.6953125,
"rewards/final_brier_reward_step": 0.7917730212211609,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.848291277885437,
"step": 75
},
{
"adv/mean_abs_final_conf": 0.7181636691093445,
"adv/mean_abs_reasoning": 0.3028116524219513,
"adv/mean_abs_step_conf": 0.7950438261032104,
"adv/ratio_final_to_reasoning": 2.3716513660069563,
"adv/ratio_step_to_reasoning": 2.625539076004119,
"adv/std_final_conf": 0.8976917266845703,
"adv/std_reasoning": 0.6183298826217651,
"adv/std_step_conf": 0.9354302883148193,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.7375237122428133,
"calib/avg_num_step_conf": 5.7578125,
"calib/ece": 0.1915294117647059,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.34901960784313724,
"calib/gap": 0.29070698963957414,
"calib/mean_conf": 0.5864313725490197,
"calib/mu_c": 0.6742134831460675,
"calib/mu_w": 0.3835064935064934,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.03996078431372553,
"calib/std_conf": 0.35051588058913247,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4840788177339902,
"calib/step_q_c_n": 1015.0,
"calib/step_q_gap": 0.09244483080588561,
"calib/step_q_w": 0.3916339869281046,
"calib/step_q_w_n": 459.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2446.0,
"completions/max_terminated_length": 2446.0,
"completions/mean_length": 522.375,
"completions/mean_terminated_length": 522.375,
"completions/min_length": 185.0,
"completions/min_terminated_length": 185.0,
"epoch": 0.08106666666666666,
"grad_norm": 0.048353470861911774,
"learning_rate": 3.444444444444445e-06,
"loss": 0.0597,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03172909840941429,
"mask/share_reasoning": 0.8478416204452515,
"mask/share_step_conf": 0.12042926251888275,
"num_tokens": 18522206.0,
"reward": 1.414903163909912,
"reward_std": 0.17302893102169037,
"rewards/accuracy_reward_step": 0.6953125,
"rewards/final_brier_reward_step": 0.7734194993972778,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.8590527176856995,
"step": 76
},
{
"adv/mean_abs_final_conf": 0.744149923324585,
"adv/mean_abs_reasoning": 0.46821609139442444,
"adv/mean_abs_step_conf": 0.7633011341094971,
"adv/ratio_final_to_reasoning": 1.5893300913866164,
"adv/ratio_step_to_reasoning": 1.6302325958859314,
"adv/std_final_conf": 0.909368097782135,
"adv/std_reasoning": 0.7392141819000244,
"adv/std_step_conf": 0.9352315664291382,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7154060027090612,
"calib/avg_num_step_conf": 6.23046875,
"calib/ece": 0.16301587301587295,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.36507936507936506,
"calib/gap": 0.2542767519783275,
"calib/mean_conf": 0.6429365079365079,
"calib/mu_c": 0.7266863905325444,
"calib/mu_w": 0.47240963855421686,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.06765873015873013,
"calib/std_conf": 0.3292308676639026,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4918036072144289,
"calib/step_q_c_n": 998.0,
"calib/step_q_gap": 0.11434967086602021,
"calib/step_q_w": 0.3774539363484087,
"calib/step_q_w_n": 597.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2986.0,
"completions/max_terminated_length": 2986.0,
"completions/mean_length": 556.34765625,
"completions/mean_terminated_length": 558.5294189453125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.08213333333333334,
"grad_norm": 0.045243460685014725,
"learning_rate": 3.416666666666667e-06,
"loss": 0.022,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03196696564555168,
"mask/share_reasoning": 0.8324825763702393,
"mask/share_step_conf": 0.13164415955543518,
"num_tokens": 18769295.0,
"reward": 1.3950183391571045,
"reward_std": 0.21794402599334717,
"rewards/accuracy_reward_step": 0.66015625,
"rewards/final_brier_reward_step": 0.7689140439033508,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8464987874031067,
"step": 77
},
{
"adv/mean_abs_final_conf": 0.697986364364624,
"adv/mean_abs_reasoning": 0.39741477370262146,
"adv/mean_abs_step_conf": 0.7532141804695129,
"adv/ratio_final_to_reasoning": 1.7563171038198873,
"adv/ratio_step_to_reasoning": 1.895284801448096,
"adv/std_final_conf": 0.8897503614425659,
"adv/std_reasoning": 0.6815274953842163,
"adv/std_step_conf": 0.9348727464675903,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.66003300330033,
"calib/avg_num_step_conf": 6.09765625,
"calib/ece": 0.1886055776892431,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.42231075697211157,
"calib/gap": 0.22551155115511556,
"calib/mean_conf": 0.652589641434263,
"calib/mu_c": 0.7433333333333333,
"calib/mu_w": 0.5178217821782177,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.12179282868525904,
"calib/std_conf": 0.34061706795804225,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5183182844243792,
"calib/step_q_c_n": 886.0,
"calib/step_q_gap": 0.13551828442437924,
"calib/step_q_w": 0.3828,
"calib/step_q_w_n": 675.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2880.0,
"completions/max_terminated_length": 2880.0,
"completions/mean_length": 587.96484375,
"completions/mean_terminated_length": 590.2706298828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 218.0,
"epoch": 0.0832,
"grad_norm": 0.04956922307610512,
"learning_rate": 3.3888888888888893e-06,
"loss": 0.1075,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.028507526963949203,
"mask/share_reasoning": 0.8540570735931396,
"mask/share_step_conf": 0.11352914571762085,
"num_tokens": 19027838.0,
"reward": 1.3692377805709839,
"reward_std": 0.23170146346092224,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.7343155741691589,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8454394340515137,
"step": 78
},
{
"adv/mean_abs_final_conf": 0.6280073523521423,
"adv/mean_abs_reasoning": 0.3942117691040039,
"adv/mean_abs_step_conf": 0.7584797143936157,
"adv/ratio_final_to_reasoning": 1.5930710383901723,
"adv/ratio_step_to_reasoning": 1.9240412738502182,
"adv/std_final_conf": 0.8607851266860962,
"adv/std_reasoning": 0.7012507915496826,
"adv/std_step_conf": 0.9304596781730652,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6052868625277162,
"calib/avg_num_step_conf": 6.18359375,
"calib/ece": 0.20690476190476187,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.5277777777777778,
"calib/gap": 0.131879157427938,
"calib/mean_conf": 0.7615079365079365,
"calib/mu_c": 0.8075609756097563,
"calib/mu_w": 0.6756818181818183,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.15880952380952376,
"calib/std_conf": 0.2860016301993651,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5340417457305503,
"calib/step_q_c_n": 1054.0,
"calib/step_q_gap": 0.029372558584992547,
"calib/step_q_w": 0.5046691871455578,
"calib/step_q_w_n": 529.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1238.0,
"completions/max_terminated_length": 1238.0,
"completions/mean_length": 532.50390625,
"completions/mean_terminated_length": 538.8181762695312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 174.0,
"epoch": 0.08426666666666667,
"grad_norm": 0.04731028527021408,
"learning_rate": 3.3611111111111117e-06,
"loss": -0.1041,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.028967462480068207,
"mask/share_reasoning": 0.8404107689857483,
"mask/share_step_conf": 0.11890304088592529,
"num_tokens": 19270535.0,
"reward": 1.3517359495162964,
"reward_std": 0.2281036078929901,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.7270851731300354,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8256934285163879,
"step": 79
},
{
"adv/mean_abs_final_conf": 0.5573588013648987,
"adv/mean_abs_reasoning": 0.41972774267196655,
"adv/mean_abs_step_conf": 0.7844960689544678,
"adv/ratio_final_to_reasoning": 1.3279055556746844,
"adv/ratio_step_to_reasoning": 1.8690593668181275,
"adv/std_final_conf": 0.8060113191604614,
"adv/std_reasoning": 0.7012417316436768,
"adv/std_step_conf": 0.9352529048919678,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.5498077942735948,
"calib/avg_num_step_conf": 5.92578125,
"calib/ece": 0.32324218749999994,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 1.0,
"calib/frac_conf_gt_0.9": 0.86328125,
"calib/gap": 0.03450424178154843,
"calib/mean_conf": 0.9175390625000002,
"calib/mu_c": 0.9299390243902439,
"calib/mu_w": 0.8954347826086955,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3000781249999999,
"calib/std_conf": 0.19404966351844335,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6749524815205913,
"calib/step_q_c_n": 947.0,
"calib/step_q_gap": 0.03404020081883685,
"calib/step_q_w": 0.6409122807017544,
"calib/step_q_w_n": 570.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1760.0,
"completions/max_terminated_length": 1760.0,
"completions/mean_length": 478.0703125,
"completions/mean_terminated_length": 479.94512939453125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 218.0,
"epoch": 0.08533333333333333,
"grad_norm": 0.04117859527468681,
"learning_rate": 3.3333333333333333e-06,
"loss": -0.0116,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03346114605665207,
"mask/share_reasoning": 0.8312935829162598,
"mask/share_step_conf": 0.13133902847766876,
"num_tokens": 19495081.0,
"reward": 1.2891900539398193,
"reward_std": 0.23486578464508057,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6713261604309082,
"rewards/format_reward_step": 1.0,
"rewards/step_l2_reward": 0.7894644141197205,
"step": 80
},
{
"adv/mean_abs_final_conf": 0.5828555822372437,
"adv/mean_abs_reasoning": 0.4114669859409332,
"adv/mean_abs_step_conf": 0.7615097761154175,
"adv/ratio_final_to_reasoning": 1.4165306140038987,
"adv/ratio_step_to_reasoning": 1.8507190178915922,
"adv/std_final_conf": 0.8089916110038757,
"adv/std_reasoning": 0.7014133930206299,
"adv/std_step_conf": 0.934392511844635,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7270913685330496,
"calib/avg_num_step_conf": 6.40234375,
"calib/ece": 0.2655823293172691,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.7670682730923695,
"calib/gap": 0.2501363213193246,
"calib/mean_conf": 0.8478313253012049,
"calib/mu_c": 0.9513013698630136,
"calib/mu_w": 0.701165048543689,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2635341365461848,
"calib/std_conf": 0.28008195097094235,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7073521850899742,
"calib/step_q_c_n": 778.0,
"calib/step_q_gap": 0.2678283755661647,
"calib/step_q_w": 0.43952380952380954,
"calib/step_q_w_n": 861.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2411.0,
"completions/max_terminated_length": 2411.0,
"completions/mean_length": 552.83984375,
"completions/mean_terminated_length": 559.395263671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 218.0,
"epoch": 0.0864,
"grad_norm": 0.03954046592116356,
"learning_rate": 3.3055555555555558e-06,
"loss": -0.0206,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.030756745487451553,
"mask/share_reasoning": 0.8393161296844482,
"mask/share_step_conf": 0.11820834130048752,
"num_tokens": 19742856.0,
"reward": 1.2981681823730469,
"reward_std": 0.3054763376712799,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.7119581699371338,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7875015139579773,
"step": 81
},
{
"adv/mean_abs_final_conf": 0.4838182330131531,
"adv/mean_abs_reasoning": 0.4532151222229004,
"adv/mean_abs_step_conf": 0.7782233953475952,
"adv/ratio_final_to_reasoning": 1.0675244697047013,
"adv/ratio_step_to_reasoning": 1.717117009535373,
"adv/std_final_conf": 0.740546464920044,
"adv/std_reasoning": 0.7205851078033447,
"adv/std_step_conf": 0.9346796274185181,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6594101123595506,
"calib/avg_num_step_conf": 6.1953125,
"calib/ece": 0.29353413654618477,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.8795180722891566,
"calib/gap": 0.13055688202247184,
"calib/mean_conf": 0.9193975903614459,
"calib/mu_c": 0.9660624999999999,
"calib/mu_w": 0.8355056179775281,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2851807228915663,
"calib/std_conf": 0.2104213561946316,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7590744920993229,
"calib/step_q_c_n": 886.0,
"calib/step_q_gap": 0.2732459206707515,
"calib/step_q_w": 0.4858285714285714,
"calib/step_q_w_n": 700.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3005.0,
"completions/max_terminated_length": 3005.0,
"completions/mean_length": 501.48046875,
"completions/mean_terminated_length": 511.4701232910156,
"completions/min_length": 0.0,
"completions/min_terminated_length": 220.0,
"epoch": 0.08746666666666666,
"grad_norm": 0.04062870889902115,
"learning_rate": 3.277777777777778e-06,
"loss": -0.0729,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03170660138130188,
"mask/share_reasoning": 0.8233975768089294,
"mask/share_step_conf": 0.12536457180976868,
"num_tokens": 19976787.0,
"reward": 1.303025245666504,
"reward_std": 0.2826133072376251,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.689989447593689,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7982649207115173,
"step": 82
},
{
"adv/mean_abs_final_conf": 0.4387087821960449,
"adv/mean_abs_reasoning": 0.36228880286216736,
"adv/mean_abs_step_conf": 0.755591630935669,
"adv/ratio_final_to_reasoning": 1.2109366304730966,
"adv/ratio_step_to_reasoning": 2.085605806655674,
"adv/std_final_conf": 0.7206283807754517,
"adv/std_reasoning": 0.6611847877502441,
"adv/std_step_conf": 0.9289259910583496,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.576388888888889,
"calib/avg_num_step_conf": 5.6015625,
"calib/ece": 0.37199203187251006,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.8884462151394422,
"calib/gap": 0.07596962616822422,
"calib/mean_conf": 0.9238645418326695,
"calib/mu_c": 0.9562499999999999,
"calib/mu_w": 0.8802803738317757,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.36107569721115546,
"calib/std_conf": 0.20589084829598686,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7503230337078652,
"calib/step_q_c_n": 712.0,
"calib/step_q_gap": 0.16447815836160484,
"calib/step_q_w": 0.5858448753462604,
"calib/step_q_w_n": 722.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2452.0,
"completions/max_terminated_length": 2452.0,
"completions/mean_length": 542.39453125,
"completions/mean_terminated_length": 551.0040283203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.08853333333333334,
"grad_norm": 0.06323331594467163,
"learning_rate": 3.2500000000000002e-06,
"loss": -0.0869,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.031102724373340607,
"mask/share_reasoning": 0.8401299715042114,
"mask/share_step_conf": 0.11314228922128677,
"num_tokens": 20222904.0,
"reward": 1.235900640487671,
"reward_std": 0.23764778673648834,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.6153316497802734,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.7739379405975342,
"step": 83
},
{
"adv/mean_abs_final_conf": 0.43967533111572266,
"adv/mean_abs_reasoning": 0.36439767479896545,
"adv/mean_abs_step_conf": 0.7603832483291626,
"adv/ratio_final_to_reasoning": 1.2065810555961618,
"adv/ratio_step_to_reasoning": 2.0866852368052524,
"adv/std_final_conf": 0.7023812532424927,
"adv/std_reasoning": 0.6402844190597534,
"adv/std_step_conf": 0.9353541731834412,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5685534989858012,
"calib/avg_num_step_conf": 5.16015625,
"calib/ece": 0.42261904761904767,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.9246031746031746,
"calib/gap": 0.060053245436105684,
"calib/mean_conf": 0.9490476190476191,
"calib/mu_c": 0.9766911764705883,
"calib/mu_w": 0.9166379310344827,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4159920634920636,
"calib/std_conf": 0.16705634716840811,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7947251114413076,
"calib/step_q_c_n": 673.0,
"calib/step_q_gap": 0.1365152348980978,
"calib/step_q_w": 0.6582098765432098,
"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": 0.0,
"completions/max_length": 2789.0,
"completions/max_terminated_length": 2789.0,
"completions/mean_length": 497.1640625,
"completions/mean_terminated_length": 497.1640625,
"completions/min_length": 131.0,
"completions/min_terminated_length": 131.0,
"epoch": 0.0896,
"grad_norm": 0.03995829448103905,
"learning_rate": 3.2222222222222227e-06,
"loss": 0.0282,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.0344880037009716,
"mask/share_reasoning": 0.8458722829818726,
"mask/share_step_conf": 0.11963975429534912,
"num_tokens": 20456098.0,
"reward": 1.1661443710327148,
"reward_std": 0.2615615427494049,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.5767694711685181,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7261971235275269,
"step": 84
},
{
"adv/mean_abs_final_conf": 0.5147285461425781,
"adv/mean_abs_reasoning": 0.47191280126571655,
"adv/mean_abs_step_conf": 0.7632074952125549,
"adv/ratio_final_to_reasoning": 1.0907280852776733,
"adv/ratio_step_to_reasoning": 1.6172638105293125,
"adv/std_final_conf": 0.7584975957870483,
"adv/std_reasoning": 0.7393365502357483,
"adv/std_step_conf": 0.9357002377510071,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6181605351170568,
"calib/avg_num_step_conf": 6.11328125,
"calib/ece": 0.4077551020408164,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9020408163265307,
"calib/gap": 0.09645819397993294,
"calib/mean_conf": 0.9338775510204083,
"calib/mu_c": 0.9791538461538462,
"calib/mu_w": 0.8826956521739132,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40551020408163274,
"calib/std_conf": 0.1852920070143077,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7397910863509749,
"calib/step_q_c_n": 718.0,
"calib/step_q_gap": 0.1751984063037494,
"calib/step_q_w": 0.5645926800472255,
"calib/step_q_w_n": 847.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2646.0,
"completions/max_terminated_length": 2646.0,
"completions/mean_length": 568.9453125,
"completions/mean_terminated_length": 577.9761962890625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 169.0,
"epoch": 0.09066666666666667,
"grad_norm": 0.04487134516239166,
"learning_rate": 3.1944444444444443e-06,
"loss": -0.0152,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.030954696238040924,
"mask/share_reasoning": 0.8301589488983154,
"mask/share_step_conf": 0.12326133251190186,
"num_tokens": 20709572.0,
"reward": 1.1941413879394531,
"reward_std": 0.30876463651657104,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.5761609077453613,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7595765590667725,
"step": 85
},
{
"adv/mean_abs_final_conf": 0.3884989321231842,
"adv/mean_abs_reasoning": 0.24427273869514465,
"adv/mean_abs_step_conf": 0.7575057148933411,
"adv/ratio_final_to_reasoning": 1.5904309838194248,
"adv/ratio_step_to_reasoning": 3.1010653048710335,
"adv/std_final_conf": 0.6616693139076233,
"adv/std_reasoning": 0.5481932163238525,
"adv/std_step_conf": 0.9352168440818787,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6341478696741853,
"calib/avg_num_step_conf": 5.7734375,
"calib/ece": 0.39470355731225304,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.8853754940711462,
"calib/gap": 0.12376441102756863,
"calib/mean_conf": 0.9203952569169962,
"calib/mu_c": 0.979097744360902,
"calib/mu_w": 0.8553333333333334,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.39470355731225304,
"calib/std_conf": 0.21292393633701728,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7278800000000001,
"calib/step_q_c_n": 750.0,
"calib/step_q_gap": 0.155640989010989,
"calib/step_q_w": 0.5722390109890111,
"calib/step_q_w_n": 728.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2666.0,
"completions/max_terminated_length": 2666.0,
"completions/mean_length": 534.9140625,
"completions/mean_terminated_length": 537.0117797851562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 158.0,
"epoch": 0.09173333333333333,
"grad_norm": 0.04348112642765045,
"learning_rate": 3.1666666666666667e-06,
"loss": 0.0043,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03243178874254227,
"mask/share_reasoning": 0.8390860557556152,
"mask/share_step_conf": 0.1245758980512619,
"num_tokens": 20952022.0,
"reward": 1.2344986200332642,
"reward_std": 0.21900911629199982,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.6001824140548706,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7840167880058289,
"step": 86
},
{
"adv/mean_abs_final_conf": 0.4687391519546509,
"adv/mean_abs_reasoning": 0.41405802965164185,
"adv/mean_abs_step_conf": 0.775758683681488,
"adv/ratio_final_to_reasoning": 1.132061494735445,
"adv/ratio_step_to_reasoning": 1.8735506333113614,
"adv/std_final_conf": 0.7078585028648376,
"adv/std_reasoning": 0.681672990322113,
"adv/std_step_conf": 0.9350662231445312,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5451519093900666,
"calib/avg_num_step_conf": 5.2265625,
"calib/ece": 0.269484126984127,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9603174603174603,
"calib/gap": 0.037394199127573136,
"calib/mean_conf": 0.9661507936507937,
"calib/mu_c": 0.9769832402234636,
"calib/mu_w": 0.9395890410958905,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.26265873015873015,
"calib/std_conf": 0.1309302229930924,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7454516129032258,
"calib/step_q_c_n": 930.0,
"calib/step_q_gap": 0.06878494623655917,
"calib/step_q_w": 0.6766666666666666,
"calib/step_q_w_n": 408.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2914.0,
"completions/max_terminated_length": 2914.0,
"completions/mean_length": 489.05859375,
"completions/mean_terminated_length": 489.05859375,
"completions/min_length": 126.0,
"completions/min_terminated_length": 126.0,
"epoch": 0.0928,
"grad_norm": 0.06873677670955658,
"learning_rate": 3.138888888888889e-06,
"loss": 0.0182,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03524698317050934,
"mask/share_reasoning": 0.8414642214775085,
"mask/share_step_conf": 0.12328878790140152,
"num_tokens": 21182717.0,
"reward": 1.326061487197876,
"reward_std": 0.2892857789993286,
"rewards/accuracy_reward_step": 0.69921875,
"rewards/final_brier_reward_step": 0.7155144810676575,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8003355264663696,
"step": 87
},
{
"adv/mean_abs_final_conf": 0.4815911054611206,
"adv/mean_abs_reasoning": 0.32384055852890015,
"adv/mean_abs_step_conf": 0.7832763195037842,
"adv/ratio_final_to_reasoning": 1.487124119501364,
"adv/ratio_step_to_reasoning": 2.4187097597099876,
"adv/std_final_conf": 0.737276554107666,
"adv/std_reasoning": 0.6186534762382507,
"adv/std_step_conf": 0.9355406761169434,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.6192062367115521,
"calib/avg_num_step_conf": 5.83984375,
"calib/ece": 0.28964143426294814,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9322709163346613,
"calib/gap": 0.11223033309709418,
"calib/mean_conf": 0.9392828685258965,
"calib/mu_c": 0.9772891566265058,
"calib/mu_w": 0.8650588235294117,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.28378486055776886,
"calib/std_conf": 0.19620372082373552,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6254573804573805,
"calib/step_q_c_n": 962.0,
"calib/step_q_gap": 0.09386263374068249,
"calib/step_q_w": 0.531594746716698,
"calib/step_q_w_n": 533.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2839.0,
"completions/max_terminated_length": 2839.0,
"completions/mean_length": 555.05078125,
"completions/mean_terminated_length": 557.2274780273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 204.0,
"epoch": 0.09386666666666667,
"grad_norm": 0.06078055128455162,
"learning_rate": 3.1111111111111116e-06,
"loss": 0.0042,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.029890595003962517,
"mask/share_reasoning": 0.8496705889701843,
"mask/share_step_conf": 0.1165325790643692,
"num_tokens": 21434658.0,
"reward": 1.3150005340576172,
"reward_std": 0.25842803716659546,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/final_brier_reward_step": 0.6966882944107056,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8037657737731934,
"step": 88
},
{
"adv/mean_abs_final_conf": 0.5294369459152222,
"adv/mean_abs_reasoning": 0.3573133647441864,
"adv/mean_abs_step_conf": 0.7569603323936462,
"adv/ratio_final_to_reasoning": 1.4817160457859315,
"adv/ratio_step_to_reasoning": 2.118477524442954,
"adv/std_final_conf": 0.7610372304916382,
"adv/std_reasoning": 0.6402999758720398,
"adv/std_step_conf": 0.9327251315116882,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.6629285077099528,
"calib/avg_num_step_conf": 6.40625,
"calib/ece": 0.34573705179282876,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.8207171314741036,
"calib/gap": 0.19132662163884306,
"calib/mean_conf": 0.8756175298804783,
"calib/mu_c": 0.9655639097744362,
"calib/mu_w": 0.7742372881355931,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.34573705179282876,
"calib/std_conf": 0.27125633106510616,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.6052157829839705,
"calib/step_q_c_n": 811.0,
"calib/step_q_gap": 0.14473327393692587,
"calib/step_q_w": 0.46048250904704463,
"calib/step_q_w_n": 829.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2479.0,
"completions/max_terminated_length": 2479.0,
"completions/mean_length": 568.96484375,
"completions/mean_terminated_length": 577.99609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 187.0,
"epoch": 0.09493333333333333,
"grad_norm": 0.04653610289096832,
"learning_rate": 3.0833333333333336e-06,
"loss": -0.0857,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.029766708612442017,
"mask/share_reasoning": 0.8333158493041992,
"mask/share_step_conf": 0.12129238992929459,
"num_tokens": 21689201.0,
"reward": 1.269324541091919,
"reward_std": 0.27564674615859985,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.6335089802742004,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8033512830734253,
"step": 89
},
{
"adv/mean_abs_final_conf": 0.4718265235424042,
"adv/mean_abs_reasoning": 0.4269935190677643,
"adv/mean_abs_step_conf": 0.7298752665519714,
"adv/ratio_final_to_reasoning": 1.1049969202636185,
"adv/ratio_step_to_reasoning": 1.7093357017349473,
"adv/std_final_conf": 0.7413638830184937,
"adv/std_reasoning": 0.7205032110214233,
"adv/std_step_conf": 0.9347279667854309,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.616210241049779,
"calib/avg_num_step_conf": 6.6328125,
"calib/ece": 0.26588932806324106,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.8616600790513834,
"calib/gap": 0.15416773641420622,
"calib/mean_conf": 0.888102766798419,
"calib/mu_c": 0.9380701754385965,
"calib/mu_w": 0.7839024390243903,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.23905138339920942,
"calib/std_conf": 0.27309591857684906,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5750236518448438,
"calib/step_q_c_n": 1057.0,
"calib/step_q_gap": 0.13070227898992964,
"calib/step_q_w": 0.4443213728549142,
"calib/step_q_w_n": 641.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2454.0,
"completions/max_terminated_length": 2454.0,
"completions/mean_length": 536.16015625,
"completions/mean_terminated_length": 540.3818969726562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 194.0,
"epoch": 0.096,
"grad_norm": 0.053866349160671234,
"learning_rate": 3.055555555555556e-06,
"loss": -0.0463,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.030385639518499374,
"mask/share_reasoning": 0.8322639465332031,
"mask/share_step_conf": 0.12953796982765198,
"num_tokens": 21929778.0,
"reward": 1.369779109954834,
"reward_std": 0.23515094816684723,
"rewards/accuracy_reward_step": 0.66796875,
"rewards/final_brier_reward_step": 0.7164187431335449,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8463352918624878,
"step": 90
},
{
"adv/mean_abs_final_conf": 0.5278657674789429,
"adv/mean_abs_reasoning": 0.41073352098464966,
"adv/mean_abs_step_conf": 0.7537205219268799,
"adv/ratio_final_to_reasoning": 1.2851782007309571,
"adv/ratio_step_to_reasoning": 1.835059675967009,
"adv/std_final_conf": 0.7727110385894775,
"adv/std_reasoning": 0.7014096975326538,
"adv/std_step_conf": 0.9352238178253174,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6680456171735242,
"calib/avg_num_step_conf": 6.7734375,
"calib/ece": 0.24536000000000005,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.836,
"calib/gap": 0.20369409660107352,
"calib/mean_conf": 0.87168,
"calib/mu_c": 0.935232558139535,
"calib/mu_w": 0.7315384615384615,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.21452000000000007,
"calib/std_conf": 0.29014544214927795,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5232943469785575,
"calib/step_q_c_n": 1026.0,
"calib/step_q_gap": 0.1492971718373145,
"calib/step_q_w": 0.373997175141243,
"calib/step_q_w_n": 708.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2402.0,
"completions/max_terminated_length": 2402.0,
"completions/mean_length": 541.12109375,
"completions/mean_terminated_length": 547.53759765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 225.0,
"epoch": 0.09706666666666666,
"grad_norm": 0.07357340306043625,
"learning_rate": 3.0277777777777776e-06,
"loss": -0.0439,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.028301572427153587,
"mask/share_reasoning": 0.8381121158599854,
"mask/share_step_conf": 0.1218675896525383,
"num_tokens": 22176017.0,
"reward": 1.365419626235962,
"reward_std": 0.2590673863887787,
"rewards/accuracy_reward_step": 0.671875,
"rewards/final_brier_reward_step": 0.7370996475219727,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8324166536331177,
"step": 91
},
{
"adv/mean_abs_final_conf": 0.5962716341018677,
"adv/mean_abs_reasoning": 0.4923080503940582,
"adv/mean_abs_step_conf": 0.7713586091995239,
"adv/ratio_final_to_reasoning": 1.211175875804984,
"adv/ratio_step_to_reasoning": 1.5668210352889926,
"adv/std_final_conf": 0.7960085272789001,
"adv/std_reasoning": 0.7395338416099548,
"adv/std_step_conf": 0.9347488880157471,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6264086232239099,
"calib/avg_num_step_conf": 6.03515625,
"calib/ece": 0.28157258064516133,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.7903225806451613,
"calib/gap": 0.20725484706376407,
"calib/mean_conf": 0.8207661290322581,
"calib/mu_c": 0.8968152866242037,
"calib/mu_w": 0.6895604395604397,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.2346370967741936,
"calib/std_conf": 0.3455614554575189,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5519620253164557,
"calib/step_q_c_n": 948.0,
"calib/step_q_gap": 0.14090674893454608,
"calib/step_q_w": 0.4110552763819096,
"calib/step_q_w_n": 597.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2518.0,
"completions/max_terminated_length": 2518.0,
"completions/mean_length": 513.265625,
"completions/mean_terminated_length": 523.4900512695312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 167.0,
"epoch": 0.09813333333333334,
"grad_norm": 0.053371768444776535,
"learning_rate": 3e-06,
"loss": -0.0806,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.03073771856725216,
"mask/share_reasoning": 0.8260927200317383,
"mask/share_step_conf": 0.12363831698894501,
"num_tokens": 22414133.0,
"reward": 1.3340181112289429,
"reward_std": 0.2972875237464905,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6785469055175781,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.837713360786438,
"step": 92
},
{
"adv/mean_abs_final_conf": 0.5813256502151489,
"adv/mean_abs_reasoning": 0.4211743474006653,
"adv/mean_abs_step_conf": 0.7620203495025635,
"adv/ratio_final_to_reasoning": 1.3802494235531655,
"adv/ratio_step_to_reasoning": 1.8092753136687352,
"adv/std_final_conf": 0.7933601140975952,
"adv/std_reasoning": 0.6817123889923096,
"adv/std_step_conf": 0.9354146718978882,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6722418995146269,
"calib/avg_num_step_conf": 7.109375,
"calib/ece": 0.29296442687747026,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.5849802371541502,
"calib/gap": 0.3127922077922078,
"calib/mean_conf": 0.6094861660079053,
"calib/mu_c": 0.7318831168831169,
"calib/mu_w": 0.4190909090909091,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.14687747035573115,
"calib/std_conf": 0.4599906907529452,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.46842459173871276,
"calib/step_q_c_n": 1041.0,
"calib/step_q_gap": 0.11452215271432242,
"calib/step_q_w": 0.35390243902439034,
"calib/step_q_w_n": 779.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2351.0,
"completions/max_terminated_length": 2351.0,
"completions/mean_length": 594.62109375,
"completions/mean_terminated_length": 596.9530029296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 150.0,
"epoch": 0.0992,
"grad_norm": 0.07864338159561157,
"learning_rate": 2.9722222222222225e-06,
"loss": -0.0145,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.028944937512278557,
"mask/share_reasoning": 0.8290570974349976,
"mask/share_step_conf": 0.13809169828891754,
"num_tokens": 22672132.0,
"reward": 1.3323159217834473,
"reward_std": 0.2134857177734375,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6910336017608643,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8278146386146545,
"step": 93
},
{
"adv/mean_abs_final_conf": 0.706995964050293,
"adv/mean_abs_reasoning": 0.42493805289268494,
"adv/mean_abs_step_conf": 0.7471293807029724,
"adv/ratio_final_to_reasoning": 1.6637624219284493,
"adv/ratio_step_to_reasoning": 1.7582077566766057,
"adv/std_final_conf": 0.8768224716186523,
"adv/std_reasoning": 0.7206136584281921,
"adv/std_step_conf": 0.9347711801528931,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.68680105420614,
"calib/avg_num_step_conf": 7.77734375,
"calib/ece": 0.28625000000000006,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.5125,
"calib/gap": 0.33865659947289695,
"calib/mean_conf": 0.5400000000000001,
"calib/mu_c": 0.6825179856115108,
"calib/mu_w": 0.34386138613861383,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.12354166666666673,
"calib/std_conf": 0.4673025429704685,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.45993839835728956,
"calib/step_q_c_n": 974.0,
"calib/step_q_gap": 0.15902394408000342,
"calib/step_q_w": 0.30091445427728614,
"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": 0.03125,
"completions/max_length": 2980.0,
"completions/max_terminated_length": 2980.0,
"completions/mean_length": 567.58203125,
"completions/mean_terminated_length": 585.89111328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 173.0,
"epoch": 0.10026666666666667,
"grad_norm": 0.0639803558588028,
"learning_rate": 2.944444444444445e-06,
"loss": -0.1048,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.02839193120598793,
"mask/share_reasoning": 0.8084524869918823,
"mask/share_step_conf": 0.13190564513206482,
"num_tokens": 22926113.0,
"reward": 1.272699236869812,
"reward_std": 0.3097056448459625,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.6576046943664551,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.7954593896865845,
"step": 94
},
{
"adv/mean_abs_final_conf": 0.6146141886711121,
"adv/mean_abs_reasoning": 0.4706389009952545,
"adv/mean_abs_step_conf": 0.7723755836486816,
"adv/ratio_final_to_reasoning": 1.3059145501389595,
"adv/ratio_step_to_reasoning": 1.6411214245472445,
"adv/std_final_conf": 0.8057188987731934,
"adv/std_reasoning": 0.7392786145210266,
"adv/std_step_conf": 0.9354429841041565,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7381290584415585,
"calib/avg_num_step_conf": 7.19140625,
"calib/ece": 0.26336000000000004,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.576,
"calib/gap": 0.3906114718614719,
"calib/mean_conf": 0.5952000000000001,
"calib/mu_c": 0.7451948051948052,
"calib/mu_w": 0.3545833333333333,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.12128000000000003,
"calib/std_conf": 0.46847855874095246,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4801834862385321,
"calib/step_q_c_n": 1090.0,
"calib/step_q_gap": 0.1356428737218876,
"calib/step_q_w": 0.3445406125166445,
"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": 0.01171875,
"completions/max_length": 2733.0,
"completions/max_terminated_length": 2733.0,
"completions/mean_length": 578.84765625,
"completions/mean_terminated_length": 585.7114868164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.10133333333333333,
"grad_norm": 0.06708774715662003,
"learning_rate": 2.916666666666667e-06,
"loss": -0.1013,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.028501030057668686,
"mask/share_reasoning": 0.8217149972915649,
"mask/share_step_conf": 0.13806524872779846,
"num_tokens": 23180426.0,
"reward": 1.346002459526062,
"reward_std": 0.26025962829589844,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6996574401855469,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8395332098007202,
"step": 95
},
{
"adv/mean_abs_final_conf": 0.5579100251197815,
"adv/mean_abs_reasoning": 0.30279800295829773,
"adv/mean_abs_step_conf": 0.7457443475723267,
"adv/ratio_final_to_reasoning": 1.8425155373188462,
"adv/ratio_step_to_reasoning": 2.462844339416046,
"adv/std_final_conf": 0.7927126884460449,
"adv/std_reasoning": 0.5959977507591248,
"adv/std_step_conf": 0.9352127909660339,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7185032894736842,
"calib/avg_num_step_conf": 7.33203125,
"calib/ece": 0.2919685039370078,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.5787401574803149,
"calib/gap": 0.33855921052631566,
"calib/mean_conf": 0.6063779527559054,
"calib/mu_c": 0.6916842105263157,
"calib/mu_w": 0.353125,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.07515748031496058,
"calib/std_conf": 0.46109877209135486,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4926863468634687,
"calib/step_q_c_n": 1355.0,
"calib/step_q_gap": 0.15048328172936903,
"calib/step_q_w": 0.34220306513409965,
"calib/step_q_w_n": 522.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2968.0,
"completions/max_terminated_length": 2968.0,
"completions/mean_length": 533.7578125,
"completions/mean_terminated_length": 535.8510131835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 216.0,
"epoch": 0.1024,
"grad_norm": 0.06763053685426712,
"learning_rate": 2.888888888888889e-06,
"loss": 0.0039,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.029817262664437294,
"mask/share_reasoning": 0.8207234144210815,
"mask/share_step_conf": 0.1455530822277069,
"num_tokens": 23422884.0,
"reward": 1.3436824083328247,
"reward_std": 0.21160925924777985,
"rewards/accuracy_reward_step": 0.7421875,
"rewards/final_brier_reward_step": 0.7009460926055908,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8197718262672424,
"step": 96
},
{
"adv/mean_abs_final_conf": 0.6728768348693848,
"adv/mean_abs_reasoning": 0.43920740485191345,
"adv/mean_abs_step_conf": 0.7557384967803955,
"adv/ratio_final_to_reasoning": 1.5320252514783013,
"adv/ratio_step_to_reasoning": 1.7206870568022552,
"adv/std_final_conf": 0.8519013524055481,
"adv/std_reasoning": 0.7014325857162476,
"adv/std_step_conf": 0.9350149035453796,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.682139522475657,
"calib/avg_num_step_conf": 7.40625,
"calib/ece": 0.30369477911646575,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.5461847389558233,
"calib/gap": 0.3056162464985993,
"calib/mean_conf": 0.5771887550200803,
"calib/mu_c": 0.7023809523809523,
"calib/mu_w": 0.396764705882353,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.1452610441767067,
"calib/std_conf": 0.4634480173365173,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5188142292490118,
"calib/step_q_c_n": 1012.0,
"calib/step_q_gap": 0.14113323377389864,
"calib/step_q_w": 0.37768099547511313,
"calib/step_q_w_n": 884.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2542.0,
"completions/max_terminated_length": 2542.0,
"completions/mean_length": 551.26171875,
"completions/mean_terminated_length": 555.6023559570312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 169.0,
"epoch": 0.10346666666666667,
"grad_norm": 0.07822365313768387,
"learning_rate": 2.861111111111111e-06,
"loss": -0.0378,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.029504206031560898,
"mask/share_reasoning": 0.8202545642852783,
"mask/share_step_conf": 0.14242875576019287,
"num_tokens": 23669079.0,
"reward": 1.3232464790344238,
"reward_std": 0.2652709484100342,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.6721296906471252,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8324941992759705,
"step": 97
},
{
"adv/mean_abs_final_conf": 0.6271836161613464,
"adv/mean_abs_reasoning": 0.420247346162796,
"adv/mean_abs_step_conf": 0.7645853757858276,
"adv/ratio_final_to_reasoning": 1.4924154117522663,
"adv/ratio_step_to_reasoning": 1.8193699086195811,
"adv/std_final_conf": 0.8403021097183228,
"adv/std_reasoning": 0.7013960480690002,
"adv/std_step_conf": 0.9347314834594727,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6450785612258148,
"calib/avg_num_step_conf": 7.28125,
"calib/ece": 0.3208695652173913,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.5573122529644269,
"calib/gap": 0.27845799246851055,
"calib/mean_conf": 0.5872727272727272,
"calib/mu_c": 0.6995364238410595,
"calib/mu_w": 0.42107843137254897,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.15565217391304345,
"calib/std_conf": 0.46653807923338414,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5182462686567164,
"calib/step_q_c_n": 1072.0,
"calib/step_q_gap": 0.10553162219206996,
"calib/step_q_w": 0.41271464646464645,
"calib/step_q_w_n": 792.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2442.0,
"completions/max_terminated_length": 2442.0,
"completions/mean_length": 572.4453125,
"completions/mean_terminated_length": 576.9527587890625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.10453333333333334,
"grad_norm": 0.07259117811918259,
"learning_rate": 2.8333333333333335e-06,
"loss": -0.0795,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.029141124337911606,
"mask/share_reasoning": 0.8251111507415771,
"mask/share_step_conf": 0.13793519139289856,
"num_tokens": 23921809.0,
"reward": 1.3164103031158447,
"reward_std": 0.23620383441448212,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.6614941358566284,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8286319375038147,
"step": 98
},
{
"adv/mean_abs_final_conf": 0.6719821691513062,
"adv/mean_abs_reasoning": 0.5113678574562073,
"adv/mean_abs_step_conf": 0.7439428567886353,
"adv/ratio_final_to_reasoning": 1.314087616875399,
"adv/ratio_step_to_reasoning": 1.4548095777653474,
"adv/std_final_conf": 0.8572866320610046,
"adv/std_reasoning": 0.7576091289520264,
"adv/std_step_conf": 0.935447096824646,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6780647270349789,
"calib/avg_num_step_conf": 8.0859375,
"calib/ece": 0.2859677419354839,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.3709677419354839,
"calib/gap": 0.33985616214449177,
"calib/mean_conf": 0.4055645161290322,
"calib/mu_c": 0.5878260869565218,
"calib/mu_w": 0.24796992481203006,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.11391129032258067,
"calib/std_conf": 0.4630958279208193,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5245192307692308,
"calib/step_q_c_n": 832.0,
"calib/step_q_gap": 0.1616597800422518,
"calib/step_q_w": 0.362859450726979,
"calib/step_q_w_n": 1238.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2284.0,
"completions/max_terminated_length": 2284.0,
"completions/mean_length": 658.0078125,
"completions/mean_terminated_length": 671.1155395507812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.1056,
"grad_norm": 0.06929034739732742,
"learning_rate": 2.805555555555556e-06,
"loss": -0.1218,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.025643540546298027,
"mask/share_reasoning": 0.8259391784667969,
"mask/share_step_conf": 0.12888604402542114,
"num_tokens": 24196059.0,
"reward": 1.3131260871887207,
"reward_std": 0.26187536120414734,
"rewards/accuracy_reward_step": 0.44921875,
"rewards/final_brier_reward_step": 0.6805578470230103,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8310502767562866,
"step": 99
},
{
"adv/mean_abs_final_conf": 0.5885224342346191,
"adv/mean_abs_reasoning": 0.3973303437232971,
"adv/mean_abs_step_conf": 0.7406108975410461,
"adv/ratio_final_to_reasoning": 1.4811917678365616,
"adv/ratio_step_to_reasoning": 1.8639676260336446,
"adv/std_final_conf": 0.8200313448905945,
"adv/std_reasoning": 0.7013905644416809,
"adv/std_step_conf": 0.9343773126602173,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7538900067069081,
"calib/avg_num_step_conf": 7.73046875,
"calib/ece": 0.2663967611336032,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.5182186234817814,
"calib/gap": 0.42807444668008054,
"calib/mean_conf": 0.5378137651821862,
"calib/mu_c": 0.7197887323943662,
"calib/mu_w": 0.2917142857142857,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.11465587044534414,
"calib/std_conf": 0.4797149095727484,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5245932539682541,
"calib/step_q_c_n": 1008.0,
"calib/step_q_gap": 0.14648820762427872,
"calib/step_q_w": 0.37810504634397535,
"calib/step_q_w_n": 971.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2402.0,
"completions/max_terminated_length": 2402.0,
"completions/mean_length": 620.04296875,
"completions/mean_terminated_length": 634.9240112304688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 169.0,
"epoch": 0.10666666666666667,
"grad_norm": 0.061593785881996155,
"learning_rate": 2.7777777777777783e-06,
"loss": -0.1101,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.026141732931137085,
"mask/share_reasoning": 0.8237094283103943,
"mask/share_step_conf": 0.12671136856079102,
"num_tokens": 24462198.0,
"reward": 1.3486356735229492,
"reward_std": 0.2595185339450836,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.7075608968734741,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8425114154815674,
"step": 100
},
{
"adv/mean_abs_final_conf": 0.6654077172279358,
"adv/mean_abs_reasoning": 0.5122069120407104,
"adv/mean_abs_step_conf": 0.7619589567184448,
"adv/ratio_final_to_reasoning": 1.299099449042673,
"adv/ratio_step_to_reasoning": 1.4875999109084337,
"adv/std_final_conf": 0.8433419466018677,
"adv/std_reasoning": 0.7394994497299194,
"adv/std_step_conf": 0.9344830513000488,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5954233409610984,
"calib/avg_num_step_conf": 8.53125,
"calib/ece": 0.3698387096774194,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.40725806451612906,
"calib/gap": 0.21982085648904887,
"calib/mean_conf": 0.43032258064516127,
"calib/mu_c": 0.5322556390977444,
"calib/mu_w": 0.31243478260869556,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.13193548387096776,
"calib/std_conf": 0.4750431822202961,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5072187776793623,
"calib/step_q_c_n": 1129.0,
"calib/step_q_gap": 0.07490598147083144,
"calib/step_q_w": 0.43231279620853086,
"calib/step_q_w_n": 1055.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2630.0,
"completions/max_terminated_length": 2630.0,
"completions/mean_length": 677.53515625,
"completions/mean_terminated_length": 682.8700561523438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 200.0,
"epoch": 0.10773333333333333,
"grad_norm": 0.06848599761724472,
"learning_rate": 2.7500000000000004e-06,
"loss": -0.0228,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.024894852191209793,
"mask/share_reasoning": 0.8335383534431458,
"mask/share_step_conf": 0.13375428318977356,
"num_tokens": 24742639.0,
"reward": 1.250705361366272,
"reward_std": 0.2775850296020508,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.6042609214782715,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7997466921806335,
"step": 101
},
{
"adv/mean_abs_final_conf": 0.5570687055587769,
"adv/mean_abs_reasoning": 0.36170411109924316,
"adv/mean_abs_step_conf": 0.7299972772598267,
"adv/ratio_final_to_reasoning": 1.5401226816742766,
"adv/ratio_step_to_reasoning": 2.0182166993936446,
"adv/std_final_conf": 0.7902530431747437,
"adv/std_reasoning": 0.681429922580719,
"adv/std_step_conf": 0.9347419738769531,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.7456252095206168,
"calib/avg_num_step_conf": 7.15234375,
"calib/ece": 0.2704761904761904,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.5079365079365079,
"calib/gap": 0.4434837412001341,
"calib/mean_conf": 0.5265079365079364,
"calib/mu_c": 0.6936942675159236,
"calib/mu_w": 0.25021052631578944,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.08698412698412697,
"calib/std_conf": 0.4850132013590634,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5594901610017889,
"calib/step_q_c_n": 1118.0,
"calib/step_q_gap": 0.11871877250249024,
"calib/step_q_w": 0.4407713884992987,
"calib/step_q_w_n": 713.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2903.0,
"completions/max_terminated_length": 2903.0,
"completions/mean_length": 553.609375,
"completions/mean_terminated_length": 555.7803955078125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 137.0,
"epoch": 0.1088,
"grad_norm": 0.060042575001716614,
"learning_rate": 2.7222222222222224e-06,
"loss": -0.0549,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03207315504550934,
"mask/share_reasoning": 0.8154199719429016,
"mask/share_step_conf": 0.14860066771507263,
"num_tokens": 24991059.0,
"reward": 1.3538480997085571,
"reward_std": 0.22371500730514526,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.7175117135047913,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8353265523910522,
"step": 102
},
{
"adv/mean_abs_final_conf": 0.504610002040863,
"adv/mean_abs_reasoning": 0.46583327651023865,
"adv/mean_abs_step_conf": 0.7451016306877136,
"adv/ratio_final_to_reasoning": 1.0832416392000113,
"adv/ratio_step_to_reasoning": 1.5995028012373798,
"adv/std_final_conf": 0.747572124004364,
"adv/std_reasoning": 0.7575631141662598,
"adv/std_step_conf": 0.9354161620140076,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.703758865248227,
"calib/avg_num_step_conf": 7.3671875,
"calib/ece": 0.2804098360655738,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.6270491803278688,
"calib/gap": 0.36996028368794326,
"calib/mean_conf": 0.6375409836065574,
"calib/mu_c": 0.7800666666666666,
"calib/mu_w": 0.4101063829787233,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.15159836065573773,
"calib/std_conf": 0.4707289629044141,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5817972831765935,
"calib/step_q_c_n": 957.0,
"calib/step_q_gap": 0.12517726164806825,
"calib/step_q_w": 0.45662002152852527,
"calib/step_q_w_n": 929.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2726.0,
"completions/max_terminated_length": 2726.0,
"completions/mean_length": 634.44921875,
"completions/mean_terminated_length": 647.087646484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 166.0,
"epoch": 0.10986666666666667,
"grad_norm": 0.05146903917193413,
"learning_rate": 2.6944444444444444e-06,
"loss": -0.1129,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.028226887807250023,
"mask/share_reasoning": 0.8312579393386841,
"mask/share_step_conf": 0.12098385393619537,
"num_tokens": 25258030.0,
"reward": 1.3042078018188477,
"reward_std": 0.28497809171676636,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.6827234029769897,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8089398145675659,
"step": 103
},
{
"adv/mean_abs_final_conf": 0.6268337965011597,
"adv/mean_abs_reasoning": 0.4484860301017761,
"adv/mean_abs_step_conf": 0.7349182367324829,
"adv/ratio_final_to_reasoning": 1.39766626924569,
"adv/ratio_step_to_reasoning": 1.638664723995318,
"adv/std_final_conf": 0.8485493659973145,
"adv/std_reasoning": 0.72063809633255,
"adv/std_step_conf": 0.935498833656311,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7556918795851252,
"calib/avg_num_step_conf": 7.4140625,
"calib/ece": 0.22392857142857148,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.43253968253968256,
"calib/gap": 0.4920882873766758,
"calib/mean_conf": 0.4551984126984126,
"calib/mu_c": 0.7168644067796609,
"calib/mu_w": 0.2247761194029851,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.10543650793650798,
"calib/std_conf": 0.4773425531139598,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5774069319640566,
"calib/step_q_c_n": 779.0,
"calib/step_q_gap": 0.13528896949756863,
"calib/step_q_w": 0.44211796246648793,
"calib/step_q_w_n": 1119.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2481.0,
"completions/max_terminated_length": 2481.0,
"completions/mean_length": 598.61328125,
"completions/mean_terminated_length": 605.7114868164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 165.0,
"epoch": 0.11093333333333333,
"grad_norm": 0.08386190980672836,
"learning_rate": 2.666666666666667e-06,
"loss": -0.0847,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.02749706618487835,
"mask/share_reasoning": 0.830479621887207,
"mask/share_step_conf": 0.1303045153617859,
"num_tokens": 25517955.0,
"reward": 1.3406147956848145,
"reward_std": 0.26186612248420715,
"rewards/accuracy_reward_step": 0.4609375,
"rewards/final_brier_reward_step": 0.7465121150016785,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8243899345397949,
"step": 104
},
{
"adv/mean_abs_final_conf": 0.7227871417999268,
"adv/mean_abs_reasoning": 0.6233465671539307,
"adv/mean_abs_step_conf": 0.7583866715431213,
"adv/ratio_final_to_reasoning": 1.1595269467834255,
"adv/ratio_step_to_reasoning": 1.2166372793320341,
"adv/std_final_conf": 0.8908979892730713,
"adv/std_reasoning": 0.842984139919281,
"adv/std_step_conf": 0.9357901811599731,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5856815203145479,
"calib/avg_num_step_conf": 6.73046875,
"calib/ece": 0.4117269076305221,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.5100401606425703,
"calib/gap": 0.14942857142857147,
"calib/mean_conf": 0.5240160642570281,
"calib/mu_c": 0.5894285714285714,
"calib/mu_w": 0.43999999999999995,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.95703125,
"calib/pce": 0.18674698795180725,
"calib/std_conf": 0.4868940325560858,
"calib/step_conf_rate": 0.95703125,
"calib/step_q_c": 0.5554122621564482,
"calib/step_q_c_n": 946.0,
"calib/step_q_gap": 0.08958214632633232,
"calib/step_q_w": 0.4658301158301159,
"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": 0.01171875,
"completions/max_length": 3006.0,
"completions/max_terminated_length": 3006.0,
"completions/mean_length": 615.9921875,
"completions/mean_terminated_length": 623.2964477539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 232.0,
"epoch": 0.112,
"grad_norm": 0.06290662288665771,
"learning_rate": 2.6388888888888893e-06,
"loss": -0.0662,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.027417458593845367,
"mask/share_reasoning": 0.8350158333778381,
"mask/share_step_conf": 0.12584799528121948,
"num_tokens": 25781409.0,
"reward": 1.20177161693573,
"reward_std": 0.345975786447525,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/final_brier_reward_step": 0.5574495792388916,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.7734373807907104,
"step": 105
},
{
"adv/mean_abs_final_conf": 0.5670008659362793,
"adv/mean_abs_reasoning": 0.4488065838813782,
"adv/mean_abs_step_conf": 0.768217921257019,
"adv/ratio_final_to_reasoning": 1.263352380067001,
"adv/ratio_step_to_reasoning": 1.711690400379828,
"adv/std_final_conf": 0.776646077632904,
"adv/std_reasoning": 0.7206236720085144,
"adv/std_step_conf": 0.9351850748062134,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.766190103146625,
"calib/avg_num_step_conf": 7.46484375,
"calib/ece": 0.23493975903614456,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.5100401606425703,
"calib/gap": 0.4839992166079122,
"calib/mean_conf": 0.5307630522088354,
"calib/mu_c": 0.7465217391304347,
"calib/mu_w": 0.26252252252252256,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.10574297188755019,
"calib/std_conf": 0.48095435171547984,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6151315789473685,
"calib/step_q_c_n": 988.0,
"calib/step_q_gap": 0.19577079888236304,
"calib/step_q_w": 0.41936078006500543,
"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": 0.01171875,
"completions/max_length": 2453.0,
"completions/max_terminated_length": 2453.0,
"completions/mean_length": 600.07421875,
"completions/mean_terminated_length": 607.1897583007812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 259.0,
"epoch": 0.11306666666666666,
"grad_norm": 0.0558047778904438,
"learning_rate": 2.6111111111111113e-06,
"loss": -0.0252,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.02651432901620865,
"mask/share_reasoning": 0.8316283226013184,
"mask/share_step_conf": 0.13013863563537598,
"num_tokens": 26039612.0,
"reward": 1.343740701675415,
"reward_std": 0.2567639648914337,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.7355890274047852,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8255555629730225,
"step": 106
},
{
"adv/mean_abs_final_conf": 0.5723678469657898,
"adv/mean_abs_reasoning": 0.4326947331428528,
"adv/mean_abs_step_conf": 0.7519802451133728,
"adv/ratio_final_to_reasoning": 1.3227982758385561,
"adv/ratio_step_to_reasoning": 1.7379001580430815,
"adv/std_final_conf": 0.8108200430870056,
"adv/std_reasoning": 0.7205343246459961,
"adv/std_step_conf": 0.9351160526275635,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6481891610213653,
"calib/avg_num_step_conf": 7.31640625,
"calib/ece": 0.30758893280632416,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.6679841897233202,
"calib/gap": 0.28726810838978634,
"calib/mean_conf": 0.6830830039525692,
"calib/mu_c": 0.7977631578947367,
"calib/mu_w": 0.5104950495049504,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.19494071146245062,
"calib/std_conf": 0.4509594306920598,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5780399619410086,
"calib/step_q_c_n": 1051.0,
"calib/step_q_gap": 0.07975529040816182,
"calib/step_q_w": 0.4982846715328468,
"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": 0.00390625,
"completions/max_length": 2507.0,
"completions/max_terminated_length": 2507.0,
"completions/mean_length": 547.1875,
"completions/mean_terminated_length": 549.3333740234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 159.0,
"epoch": 0.11413333333333334,
"grad_norm": 0.04952119290828705,
"learning_rate": 2.5833333333333337e-06,
"loss": 0.0278,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.029747068881988525,
"mask/share_reasoning": 0.8243297338485718,
"mask/share_step_conf": 0.1420169472694397,
"num_tokens": 26284308.0,
"reward": 1.3272364139556885,
"reward_std": 0.2504537105560303,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6797593832015991,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8291535377502441,
"step": 107
},
{
"adv/mean_abs_final_conf": 0.548678457736969,
"adv/mean_abs_reasoning": 0.40821319818496704,
"adv/mean_abs_step_conf": 0.7383649349212646,
"adv/ratio_final_to_reasoning": 1.3440977905088585,
"adv/ratio_step_to_reasoning": 1.8087728133344216,
"adv/std_final_conf": 0.792629063129425,
"adv/std_reasoning": 0.7014238834381104,
"adv/std_step_conf": 0.9351604580879211,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7042369395310573,
"calib/avg_num_step_conf": 7.74609375,
"calib/ece": 0.2576583333333333,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.6547619047619048,
"calib/gap": 0.3955978033730976,
"calib/mean_conf": 0.6669448412698411,
"calib/mu_c": 0.7689839572192514,
"calib/mu_w": 0.3733861538461538,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.09126984126984128,
"calib/std_conf": 0.4603743121794018,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6037661351556569,
"calib/step_q_c_n": 1317.0,
"calib/step_q_gap": 0.11863100002052174,
"calib/step_q_w": 0.4851351351351352,
"calib/step_q_w_n": 666.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2240.0,
"completions/max_terminated_length": 2240.0,
"completions/mean_length": 591.82421875,
"completions/mean_terminated_length": 596.4842529296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 199.0,
"epoch": 0.1152,
"grad_norm": 0.04725135862827301,
"learning_rate": 2.5555555555555557e-06,
"loss": -0.04,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.028656262904405594,
"mask/share_reasoning": 0.8208492994308472,
"mask/share_step_conf": 0.14268191158771515,
"num_tokens": 26539047.0,
"reward": 1.3706966638565063,
"reward_std": 0.2445434033870697,
"rewards/accuracy_reward_step": 0.73046875,
"rewards/final_brier_reward_step": 0.7252894043922424,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8373488187789917,
"step": 108
},
{
"adv/mean_abs_final_conf": 0.45445898175239563,
"adv/mean_abs_reasoning": 0.32057422399520874,
"adv/mean_abs_step_conf": 0.750169038772583,
"adv/ratio_final_to_reasoning": 1.417640433122246,
"adv/ratio_step_to_reasoning": 2.3400790912740224,
"adv/std_final_conf": 0.7065565586090088,
"adv/std_reasoning": 0.6186202168464661,
"adv/std_step_conf": 0.9343803524971008,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7798996655518395,
"calib/avg_num_step_conf": 9.0859375,
"calib/ece": 0.22836734693877553,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.5061224489795918,
"calib/gap": 0.501391304347826,
"calib/mean_conf": 0.5226530612244898,
"calib/mu_c": 0.7579999999999999,
"calib/mu_w": 0.25660869565217387,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.11020408163265306,
"calib/std_conf": 0.4844225491903719,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5867817307692307,
"calib/step_q_c_n": 1040.0,
"calib/step_q_gap": 0.20645824709893518,
"calib/step_q_w": 0.3803234836702955,
"calib/step_q_w_n": 1286.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3071.0,
"completions/max_terminated_length": 3071.0,
"completions/mean_length": 640.05078125,
"completions/mean_terminated_length": 650.2103271484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 174.0,
"epoch": 0.11626666666666667,
"grad_norm": 0.04770355299115181,
"learning_rate": 2.5277777777777778e-06,
"loss": 0.019,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.02566307783126831,
"mask/share_reasoning": 0.8131533861160278,
"mask/share_step_conf": 0.14555855095386505,
"num_tokens": 26807500.0,
"reward": 1.3314653635025024,
"reward_std": 0.21549201011657715,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.7329742312431335,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.818884551525116,
"step": 109
},
{
"adv/mean_abs_final_conf": 0.5815128087997437,
"adv/mean_abs_reasoning": 0.4468216001987457,
"adv/mean_abs_step_conf": 0.7596118450164795,
"adv/ratio_final_to_reasoning": 1.3014429216069399,
"adv/ratio_step_to_reasoning": 1.700033849479535,
"adv/std_final_conf": 0.8087716698646545,
"adv/std_reasoning": 0.7014685273170471,
"adv/std_step_conf": 0.9353334307670593,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.6179768203577727,
"calib/avg_num_step_conf": 7.109375,
"calib/ece": 0.38184313725490193,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.5098039215686274,
"calib/gap": 0.1932275132275133,
"calib/mean_conf": 0.5376862745098039,
"calib/mu_c": 0.6195238095238095,
"calib/mu_w": 0.4262962962962962,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1715294117647059,
"calib/std_conf": 0.47674276741467103,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6170449897750512,
"calib/step_q_c_n": 978.0,
"calib/step_q_gap": 0.10406399215034812,
"calib/step_q_w": 0.5129809976247031,
"calib/step_q_w_n": 842.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1489.0,
"completions/max_terminated_length": 1489.0,
"completions/mean_length": 542.19921875,
"completions/mean_terminated_length": 544.3255004882812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 149.0,
"epoch": 0.11733333333333333,
"grad_norm": 0.04419422522187233,
"learning_rate": 2.5e-06,
"loss": -0.0314,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.030844062566757202,
"mask/share_reasoning": 0.8253611922264099,
"mask/share_step_conf": 0.1398884356021881,
"num_tokens": 27051223.0,
"reward": 1.2631927728652954,
"reward_std": 0.21499332785606384,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.6189863085746765,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.7970589399337769,
"step": 110
},
{
"adv/mean_abs_final_conf": 0.5841657519340515,
"adv/mean_abs_reasoning": 0.43651944398880005,
"adv/mean_abs_step_conf": 0.7689560651779175,
"adv/ratio_final_to_reasoning": 1.3382353523501687,
"adv/ratio_step_to_reasoning": 1.7615620008845398,
"adv/std_final_conf": 0.7941992282867432,
"adv/std_reasoning": 0.7013795375823975,
"adv/std_step_conf": 0.9354787468910217,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6561019782669266,
"calib/avg_num_step_conf": 7.54296875,
"calib/ece": 0.33289795918367343,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.6244897959183674,
"calib/gap": 0.2811744218445249,
"calib/mean_conf": 0.6323265306122449,
"calib/mu_c": 0.7436486486486485,
"calib/mu_w": 0.46247422680412364,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.18057142857142855,
"calib/std_conf": 0.4751634471853076,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5799950347567031,
"calib/step_q_c_n": 1007.0,
"calib/step_q_gap": 0.18059027285194124,
"calib/step_q_w": 0.3994047619047619,
"calib/step_q_w_n": 924.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3038.0,
"completions/max_terminated_length": 3038.0,
"completions/mean_length": 632.640625,
"completions/mean_terminated_length": 635.12158203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 179.0,
"epoch": 0.1184,
"grad_norm": 0.05241648480296135,
"learning_rate": 2.4722222222222226e-06,
"loss": 0.0322,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.029879005625844002,
"mask/share_reasoning": 0.8296252489089966,
"mask/share_step_conf": 0.13658946752548218,
"num_tokens": 27320587.0,
"reward": 1.2603142261505127,
"reward_std": 0.28884023427963257,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.6361085772514343,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7887442708015442,
"step": 111
},
{
"adv/mean_abs_final_conf": 0.5549716353416443,
"adv/mean_abs_reasoning": 0.4541664123535156,
"adv/mean_abs_step_conf": 0.7850562334060669,
"adv/ratio_final_to_reasoning": 1.2219565785716087,
"adv/ratio_step_to_reasoning": 1.7285651515660565,
"adv/std_final_conf": 0.7948229312896729,
"adv/std_reasoning": 0.7392616271972656,
"adv/std_step_conf": 0.9357273578643799,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7439711757990867,
"calib/avg_num_step_conf": 7.51953125,
"calib/ece": 0.28107438016528924,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.4462809917355372,
"calib/gap": 0.45555793378995435,
"calib/mean_conf": 0.4566115702479338,
"calib/mu_c": 0.6373287671232877,
"calib/mu_w": 0.18177083333333333,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.0671900826446281,
"calib/std_conf": 0.48851952720558744,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5791213270142179,
"calib/step_q_c_n": 1055.0,
"calib/step_q_gap": 0.19313282126709147,
"calib/step_q_w": 0.38598850574712645,
"calib/step_q_w_n": 870.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3037.0,
"completions/max_terminated_length": 3037.0,
"completions/mean_length": 631.64453125,
"completions/mean_terminated_length": 644.2271118164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 239.0,
"epoch": 0.11946666666666667,
"grad_norm": 0.11762329190969467,
"learning_rate": 2.4444444444444447e-06,
"loss": -0.0457,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.02612355723977089,
"mask/share_reasoning": 0.8241636753082275,
"mask/share_step_conf": 0.13018152117729187,
"num_tokens": 27590208.0,
"reward": 1.282677412033081,
"reward_std": 0.27923113107681274,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.6753547191619873,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.7934376001358032,
"step": 112
},
{
"adv/mean_abs_final_conf": 0.5706948041915894,
"adv/mean_abs_reasoning": 0.32284876704216003,
"adv/mean_abs_step_conf": 0.7501890063285828,
"adv/ratio_final_to_reasoning": 1.7676846327155515,
"adv/ratio_step_to_reasoning": 2.32365454947089,
"adv/std_final_conf": 0.7929322123527527,
"adv/std_reasoning": 0.5961199998855591,
"adv/std_step_conf": 0.9351218938827515,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.7311046511627907,
"calib/avg_num_step_conf": 7.78125,
"calib/ece": 0.2763779527559055,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.5196850393700787,
"calib/gap": 0.4366860465116278,
"calib/mean_conf": 0.5346456692913386,
"calib/mu_c": 0.6824999999999999,
"calib/mu_w": 0.2458139534883721,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.0748031496062992,
"calib/std_conf": 0.48556614456627994,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6204979591836733,
"calib/step_q_c_n": 1225.0,
"calib/step_q_gap": 0.1989334220259158,
"calib/step_q_w": 0.42156453715775755,
"calib/step_q_w_n": 767.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2499.0,
"completions/max_terminated_length": 2499.0,
"completions/mean_length": 558.078125,
"completions/mean_terminated_length": 560.2667236328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 222.0,
"epoch": 0.12053333333333334,
"grad_norm": 0.057893529534339905,
"learning_rate": 2.4166666666666667e-06,
"loss": -0.0251,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.0291130430996418,
"mask/share_reasoning": 0.8191037178039551,
"mask/share_step_conf": 0.14787697792053223,
"num_tokens": 27838276.0,
"reward": 1.386540174484253,
"reward_std": 0.22876328229904175,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.710267186164856,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8669533729553223,
"step": 113
},
{
"adv/mean_abs_final_conf": 0.4646427035331726,
"adv/mean_abs_reasoning": 0.3929080367088318,
"adv/mean_abs_step_conf": 0.7613837718963623,
"adv/ratio_final_to_reasoning": 1.182573681681906,
"adv/ratio_step_to_reasoning": 1.9378167427524342,
"adv/std_final_conf": 0.7412422895431519,
"adv/std_reasoning": 0.6816239953041077,
"adv/std_step_conf": 0.9348096251487732,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.798599012775842,
"calib/avg_num_step_conf": 7.63671875,
"calib/ece": 0.16599959999999997,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.696,
"calib/gap": 0.5592514663182346,
"calib/mean_conf": 0.7072803999999999,
"calib/mu_c": 0.8907148809523809,
"calib/mu_w": 0.33146341463414636,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.10064,
"calib/std_conf": 0.44446557101291884,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.6734562607204116,
"calib/step_q_c_n": 1166.0,
"calib/step_q_gap": 0.24828515805881463,
"calib/step_q_w": 0.425171102661597,
"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": 0.0234375,
"completions/max_length": 2005.0,
"completions/max_terminated_length": 2005.0,
"completions/mean_length": 554.14453125,
"completions/mean_terminated_length": 567.4440307617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 165.0,
"epoch": 0.1216,
"grad_norm": 0.04622293636202812,
"learning_rate": 2.388888888888889e-06,
"loss": -0.1354,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.030705967918038368,
"mask/share_reasoning": 0.8009713888168335,
"mask/share_step_conf": 0.14488518238067627,
"num_tokens": 28085161.0,
"reward": 1.4095648527145386,
"reward_std": 0.2490803748369217,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.8027671575546265,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8460718989372253,
"step": 114
},
{
"adv/mean_abs_final_conf": 0.48520588874816895,
"adv/mean_abs_reasoning": 0.36930257081985474,
"adv/mean_abs_step_conf": 0.802715539932251,
"adv/ratio_final_to_reasoning": 1.3138437884984333,
"adv/ratio_step_to_reasoning": 2.1735985702731933,
"adv/std_final_conf": 0.7171012163162231,
"adv/std_reasoning": 0.6187039017677307,
"adv/std_step_conf": 0.935295581817627,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6699300699300699,
"calib/avg_num_step_conf": 7.79296875,
"calib/ece": 0.2988134387351779,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.7351778656126482,
"calib/gap": 0.33630048951048963,
"calib/mean_conf": 0.743083794466403,
"calib/mu_c": 0.8893013986013987,
"calib/mu_w": 0.553000909090909,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23833992094861664,
"calib/std_conf": 0.4289959877253122,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.651243830207305,
"calib/step_q_c_n": 1013.0,
"calib/step_q_gap": 0.1571603271523151,
"calib/step_q_w": 0.49408350305498994,
"calib/step_q_w_n": 982.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 1356.0,
"completions/max_terminated_length": 1356.0,
"completions/mean_length": 528.60546875,
"completions/mean_terminated_length": 534.87353515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 204.0,
"epoch": 0.12266666666666666,
"grad_norm": 0.03641931712627411,
"learning_rate": 2.361111111111111e-06,
"loss": -0.0446,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.029507558792829514,
"mask/share_reasoning": 0.8150744438171387,
"mask/share_step_conf": 0.14369924366474152,
"num_tokens": 28325748.0,
"reward": 1.2941392660140991,
"reward_std": 0.24585412442684174,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.6956202983856201,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7916415929794312,
"step": 115
},
{
"adv/mean_abs_final_conf": 0.543901264667511,
"adv/mean_abs_reasoning": 0.38514238595962524,
"adv/mean_abs_step_conf": 0.7905224561691284,
"adv/ratio_final_to_reasoning": 1.4122082754208427,
"adv/ratio_step_to_reasoning": 2.0525459803637385,
"adv/std_final_conf": 0.7767652869224548,
"adv/std_reasoning": 0.6614387631416321,
"adv/std_step_conf": 0.9356350898742676,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6349728629579376,
"calib/avg_num_step_conf": 8.4765625,
"calib/ece": 0.33713114754098356,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.7704918032786885,
"calib/gap": 0.25052645861601086,
"calib/mean_conf": 0.781311475409836,
"calib/mu_c": 0.8942537313432836,
"calib/mu_w": 0.6437272727272727,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.28463114754098356,
"calib/std_conf": 0.40257969543904026,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6374273305084746,
"calib/step_q_c_n": 944.0,
"calib/step_q_gap": 0.1480227628086377,
"calib/step_q_w": 0.48940456769983687,
"calib/step_q_w_n": 1226.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2961.0,
"completions/max_terminated_length": 2961.0,
"completions/mean_length": 607.95703125,
"completions/mean_terminated_length": 622.5480346679688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 152.0,
"epoch": 0.12373333333333333,
"grad_norm": 0.030581878498196602,
"learning_rate": 2.3333333333333336e-06,
"loss": -0.0749,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.02716742642223835,
"mask/share_reasoning": 0.8129376173019409,
"mask/share_step_conf": 0.1364574283361435,
"num_tokens": 28585905.0,
"reward": 1.233880877494812,
"reward_std": 0.3017020523548126,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/final_brier_reward_step": 0.6295531392097473,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.7722293138504028,
"step": 116
},
{
"adv/mean_abs_final_conf": 0.5185819864273071,
"adv/mean_abs_reasoning": 0.3688926100730896,
"adv/mean_abs_step_conf": 0.7403764724731445,
"adv/ratio_final_to_reasoning": 1.4057803606436008,
"adv/ratio_step_to_reasoning": 2.007024408340552,
"adv/std_final_conf": 0.7588999271392822,
"adv/std_reasoning": 0.6612929105758667,
"adv/std_step_conf": 0.9353594779968262,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6061151079136691,
"calib/avg_num_step_conf": 7.84765625,
"calib/ece": 0.361501581027668,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.7905138339920948,
"calib/gap": 0.19482906727249794,
"calib/mean_conf": 0.803794861660079,
"calib/mu_c": 0.8915834532374102,
"calib/mu_w": 0.6967543859649122,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3079446640316206,
"calib/std_conf": 0.38466862985657124,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6479385530227949,
"calib/step_q_c_n": 1009.0,
"calib/step_q_gap": 0.12708855302279487,
"calib/step_q_w": 0.52085,
"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": 0.00390625,
"completions/max_length": 2040.0,
"completions/max_terminated_length": 2040.0,
"completions/mean_length": 579.16015625,
"completions/mean_terminated_length": 581.431396484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 216.0,
"epoch": 0.1248,
"grad_norm": 0.030159028246998787,
"learning_rate": 2.305555555555556e-06,
"loss": -0.0532,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.02752252295613289,
"mask/share_reasoning": 0.8242526650428772,
"mask/share_step_conf": 0.14431855082511902,
"num_tokens": 28840770.0,
"reward": 1.268562912940979,
"reward_std": 0.274804949760437,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.6287655830383301,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.801445722579956,
"step": 117
},
{
"adv/mean_abs_final_conf": 0.43278244137763977,
"adv/mean_abs_reasoning": 0.2965746819972992,
"adv/mean_abs_step_conf": 0.7554448246955872,
"adv/ratio_final_to_reasoning": 1.459269680281006,
"adv/ratio_step_to_reasoning": 2.547233026123473,
"adv/std_final_conf": 0.7201955318450928,
"adv/std_reasoning": 0.5961441397666931,
"adv/std_step_conf": 0.9351212382316589,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5915766596417281,
"calib/avg_num_step_conf": 8.55859375,
"calib/ece": 0.3528,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.868,
"calib/gap": 0.15894625922023187,
"calib/mean_conf": 0.88744,
"calib/mu_c": 0.9535616438356165,
"calib/mu_w": 0.7946153846153846,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.32812,
"calib/std_conf": 0.303349050435303,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6735756056808688,
"calib/step_q_c_n": 1197.0,
"calib/step_q_gap": 0.1388673561838869,
"calib/step_q_w": 0.5347082494969819,
"calib/step_q_w_n": 994.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2845.0,
"completions/max_terminated_length": 2845.0,
"completions/mean_length": 653.11328125,
"completions/mean_terminated_length": 653.11328125,
"completions/min_length": 213.0,
"completions/min_terminated_length": 213.0,
"epoch": 0.12586666666666665,
"grad_norm": 0.05803324282169342,
"learning_rate": 2.277777777777778e-06,
"loss": -0.0084,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.02645675651729107,
"mask/share_reasoning": 0.828969419002533,
"mask/share_step_conf": 0.14457379281520844,
"num_tokens": 29111975.0,
"reward": 1.26542329788208,
"reward_std": 0.23771917819976807,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.6349507570266724,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.793260395526886,
"step": 118
},
{
"adv/mean_abs_final_conf": 0.5820865631103516,
"adv/mean_abs_reasoning": 0.4483203887939453,
"adv/mean_abs_step_conf": 0.789027214050293,
"adv/ratio_final_to_reasoning": 1.2983718288527073,
"adv/ratio_step_to_reasoning": 1.7599628162638428,
"adv/std_final_conf": 0.7938616275787354,
"adv/std_reasoning": 0.7014293074607849,
"adv/std_step_conf": 0.9352133274078369,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6387527233115468,
"calib/avg_num_step_conf": 8.3984375,
"calib/ece": 0.3209756097560975,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7357723577235772,
"calib/gap": 0.250061274509804,
"calib/mean_conf": 0.755691056910569,
"calib/mu_c": 0.859375,
"calib/mu_w": 0.609313725490196,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.245650406504065,
"calib/std_conf": 0.41067083984684005,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.6519414893617023,
"calib/step_q_c_n": 1128.0,
"calib/step_q_gap": 0.16325254611316997,
"calib/step_q_w": 0.4886889432485323,
"calib/step_q_w_n": 1022.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2806.0,
"completions/max_terminated_length": 2806.0,
"completions/mean_length": 629.90625,
"completions/mean_terminated_length": 645.0240478515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 138.0,
"epoch": 0.12693333333333334,
"grad_norm": 0.03724614530801773,
"learning_rate": 2.25e-06,
"loss": -0.1336,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.02622862160205841,
"mask/share_reasoning": 0.8168337345123291,
"mask/share_step_conf": 0.13350018858909607,
"num_tokens": 29378295.0,
"reward": 1.276491403579712,
"reward_std": 0.3191832900047302,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.6544101238250732,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7969425916671753,
"step": 119
},
{
"adv/mean_abs_final_conf": 0.5171021223068237,
"adv/mean_abs_reasoning": 0.42247337102890015,
"adv/mean_abs_step_conf": 0.7612936496734619,
"adv/ratio_final_to_reasoning": 1.2239874930991812,
"adv/ratio_step_to_reasoning": 1.80199203518885,
"adv/std_final_conf": 0.7414100170135498,
"adv/std_reasoning": 0.7014121413230896,
"adv/std_step_conf": 0.9345976710319519,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6686746987951807,
"calib/avg_num_step_conf": 7.640625,
"calib/ece": 0.24758064516129036,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.7862903225806451,
"calib/gap": 0.3137685286600951,
"calib/mean_conf": 0.8005645161290323,
"calib/mu_c": 0.9055757575757577,
"calib/mu_w": 0.5918072289156626,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.19141129032258067,
"calib/std_conf": 0.3854789698028282,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6643690165361184,
"calib/step_q_c_n": 1149.0,
"calib/step_q_gap": 0.18397248617676265,
"calib/step_q_w": 0.4803965303593557,
"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": 0.01953125,
"completions/max_length": 2875.0,
"completions/max_terminated_length": 2875.0,
"completions/mean_length": 547.33984375,
"completions/mean_terminated_length": 558.2430419921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 238.0,
"epoch": 0.128,
"grad_norm": 0.03176447004079819,
"learning_rate": 2.222222222222222e-06,
"loss": -0.0537,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.028437182307243347,
"mask/share_reasoning": 0.8139419555664062,
"mask/share_step_conf": 0.13808965682983398,
"num_tokens": 29625102.0,
"reward": 1.3524608612060547,
"reward_std": 0.2827880382537842,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.7267366647720337,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8277642726898193,
"step": 120
},
{
"adv/mean_abs_final_conf": 0.641068160533905,
"adv/mean_abs_reasoning": 0.4862363636493683,
"adv/mean_abs_step_conf": 0.7601180076599121,
"adv/ratio_final_to_reasoning": 1.3184290778305263,
"adv/ratio_step_to_reasoning": 1.5632685345764137,
"adv/std_final_conf": 0.858704686164856,
"adv/std_reasoning": 0.7575821876525879,
"adv/std_step_conf": 0.9351223111152649,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6479188166494668,
"calib/avg_num_step_conf": 8.3203125,
"calib/ece": 0.2900403225806451,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.75,
"calib/gap": 0.2653340213278297,
"calib/mean_conf": 0.7735887096774193,
"calib/mu_c": 0.8752287581699348,
"calib/mu_w": 0.6098947368421052,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.22334677419354837,
"calib/std_conf": 0.39741049940058065,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6422859744990893,
"calib/step_q_c_n": 1098.0,
"calib/step_q_gap": 0.15888481170839164,
"calib/step_q_w": 0.48340116279069767,
"calib/step_q_w_n": 1032.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2712.0,
"completions/max_terminated_length": 2712.0,
"completions/mean_length": 626.875,
"completions/mean_terminated_length": 639.362548828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 219.0,
"epoch": 0.12906666666666666,
"grad_norm": 0.030280577018857002,
"learning_rate": 2.1944444444444445e-06,
"loss": -0.0602,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.026005305349826813,
"mask/share_reasoning": 0.8234038352966309,
"mask/share_step_conf": 0.13105961680412292,
"num_tokens": 29890638.0,
"reward": 1.3218499422073364,
"reward_std": 0.308574378490448,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6845276951789856,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8225547671318054,
"step": 121
},
{
"adv/mean_abs_final_conf": 0.5707475543022156,
"adv/mean_abs_reasoning": 0.43730437755584717,
"adv/mean_abs_step_conf": 0.7497875094413757,
"adv/ratio_final_to_reasoning": 1.3051494190206836,
"adv/ratio_step_to_reasoning": 1.714566667802501,
"adv/std_final_conf": 0.7942202687263489,
"adv/std_reasoning": 0.720486581325531,
"adv/std_step_conf": 0.9346208572387695,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7199184463335407,
"calib/avg_num_step_conf": 8.484375,
"calib/ece": 0.2172,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.744,
"calib/gap": 0.40112101734743244,
"calib/mean_conf": 0.7680799999999999,
"calib/mu_c": 0.9140880503144654,
"calib/mu_w": 0.5129670329670329,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.17464000000000002,
"calib/std_conf": 0.4014527538826954,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6792504258943782,
"calib/step_q_c_n": 1174.0,
"calib/step_q_gap": 0.24258709924107158,
"calib/step_q_w": 0.43666332665330665,
"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": 0.0078125,
"completions/max_length": 2677.0,
"completions/max_terminated_length": 2677.0,
"completions/mean_length": 617.8203125,
"completions/mean_terminated_length": 622.68505859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 244.0,
"epoch": 0.13013333333333332,
"grad_norm": 0.03379069268703461,
"learning_rate": 2.166666666666667e-06,
"loss": 0.0004,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.026336126029491425,
"mask/share_reasoning": 0.8244964480400085,
"mask/share_step_conf": 0.14135491847991943,
"num_tokens": 30156144.0,
"reward": 1.3796439170837402,
"reward_std": 0.3015453815460205,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.7574304342269897,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8411630392074585,
"step": 122
},
{
"adv/mean_abs_final_conf": 0.6780246496200562,
"adv/mean_abs_reasoning": 0.49683240056037903,
"adv/mean_abs_step_conf": 0.7450475692749023,
"adv/ratio_final_to_reasoning": 1.364694912922969,
"adv/ratio_step_to_reasoning": 1.4995953734791865,
"adv/std_final_conf": 0.8590086698532104,
"adv/std_reasoning": 0.7394485473632812,
"adv/std_step_conf": 0.9357582330703735,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6953206465067779,
"calib/avg_num_step_conf": 8.56640625,
"calib/ece": 0.27651405622489966,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.5742971887550201,
"calib/gap": 0.3350961939520333,
"calib/mean_conf": 0.6227630522088354,
"calib/mu_c": 0.7734890510948904,
"calib/mu_w": 0.4383928571428571,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.17453815261044187,
"calib/std_conf": 0.4485328292063186,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6023872180451129,
"calib/step_q_c_n": 1064.0,
"calib/step_q_gap": 0.15375125701765496,
"calib/step_q_w": 0.4486359610274579,
"calib/step_q_w_n": 1129.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2769.0,
"completions/max_terminated_length": 2769.0,
"completions/mean_length": 726.8359375,
"completions/mean_terminated_length": 726.8359375,
"completions/min_length": 274.0,
"completions/min_terminated_length": 274.0,
"epoch": 0.1312,
"grad_norm": 0.035880621522665024,
"learning_rate": 2.138888888888889e-06,
"loss": 0.0256,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.023776985704898834,
"mask/share_reasoning": 0.8434326648712158,
"mask/share_step_conf": 0.13279032707214355,
"num_tokens": 30447502.0,
"reward": 1.3135013580322266,
"reward_std": 0.29589012265205383,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.6924653649330139,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8164873719215393,
"step": 123
},
{
"adv/mean_abs_final_conf": 0.5248005390167236,
"adv/mean_abs_reasoning": 0.33113977313041687,
"adv/mean_abs_step_conf": 0.7580268383026123,
"adv/ratio_final_to_reasoning": 1.5848308829094806,
"adv/ratio_step_to_reasoning": 2.2891446446817163,
"adv/std_final_conf": 0.7578999400138855,
"adv/std_reasoning": 0.6185828447341919,
"adv/std_step_conf": 0.935204803943634,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6887034659820281,
"calib/avg_num_step_conf": 8.40234375,
"calib/ece": 0.29213438735177866,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.5810276679841897,
"calib/gap": 0.3372108115818001,
"calib/mean_conf": 0.6046245059288538,
"calib/mu_c": 0.7139181286549708,
"calib/mu_w": 0.37670731707317073,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.11043478260869564,
"calib/std_conf": 0.4690166709833766,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5899855177407676,
"calib/step_q_c_n": 1381.0,
"calib/step_q_gap": 0.18094655670180648,
"calib/step_q_w": 0.4090389610389611,
"calib/step_q_w_n": 770.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2552.0,
"completions/max_terminated_length": 2552.0,
"completions/mean_length": 630.87890625,
"completions/mean_terminated_length": 633.3529663085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 142.0,
"epoch": 0.13226666666666667,
"grad_norm": 0.03069070540368557,
"learning_rate": 2.1111111111111114e-06,
"loss": -0.0143,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.026175033301115036,
"mask/share_reasoning": 0.8293299674987793,
"mask/share_step_conf": 0.14058873057365417,
"num_tokens": 30715823.0,
"reward": 1.3607532978057861,
"reward_std": 0.23503340780735016,
"rewards/accuracy_reward_step": 0.66796875,
"rewards/final_brier_reward_step": 0.6918461322784424,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8495957851409912,
"step": 124
},
{
"adv/mean_abs_final_conf": 0.6436437368392944,
"adv/mean_abs_reasoning": 0.5224178433418274,
"adv/mean_abs_step_conf": 0.7601853013038635,
"adv/ratio_final_to_reasoning": 1.2320477660602163,
"adv/ratio_step_to_reasoning": 1.455128899198913,
"adv/std_final_conf": 0.8280212879180908,
"adv/std_reasoning": 0.7755022048950195,
"adv/std_step_conf": 0.9357369542121887,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6191798941798942,
"calib/avg_num_step_conf": 8.4375,
"calib/ece": 0.35554655870445345,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.582995951417004,
"calib/gap": 0.20311375661375675,
"calib/mean_conf": 0.617085020242915,
"calib/mu_c": 0.7091851851851853,
"calib/mu_w": 0.5060714285714285,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.21303643724696356,
"calib/std_conf": 0.45976539174194386,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6052729044834307,
"calib/step_q_c_n": 1026.0,
"calib/step_q_gap": 0.17805950060336018,
"calib/step_q_w": 0.42721340388007056,
"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": 0.0078125,
"completions/max_length": 3008.0,
"completions/max_terminated_length": 3008.0,
"completions/mean_length": 684.1484375,
"completions/mean_terminated_length": 689.535400390625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 203.0,
"epoch": 0.13333333333333333,
"grad_norm": 0.03402207791805267,
"learning_rate": 2.0833333333333334e-06,
"loss": 0.0411,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.02608790621161461,
"mask/share_reasoning": 0.8333162069320679,
"mask/share_step_conf": 0.13278338313102722,
"num_tokens": 30995773.0,
"reward": 1.2464786767959595,
"reward_std": 0.30777016282081604,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.6141093969345093,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7902053594589233,
"step": 125
},
{
"adv/mean_abs_final_conf": 0.5593317151069641,
"adv/mean_abs_reasoning": 0.42206138372421265,
"adv/mean_abs_step_conf": 0.759347677230835,
"adv/ratio_final_to_reasoning": 1.325237836666071,
"adv/ratio_step_to_reasoning": 1.7991403774741332,
"adv/std_final_conf": 0.7963278889656067,
"adv/std_reasoning": 0.7015090584754944,
"adv/std_step_conf": 0.9353612661361694,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.7362362714728246,
"calib/avg_num_step_conf": 9.1640625,
"calib/ece": 0.23920833333333336,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.5291666666666667,
"calib/gap": 0.4673176569980287,
"calib/mean_conf": 0.5505416666666666,
"calib/mu_c": 0.7569402985074626,
"calib/mu_w": 0.2896226415094339,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.1157083333333333,
"calib/std_conf": 0.47981867053838395,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5592987512007686,
"calib/step_q_c_n": 1041.0,
"calib/step_q_gap": 0.24736695043448503,
"calib/step_q_w": 0.3119318007662835,
"calib/step_q_w_n": 1305.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2628.0,
"completions/max_terminated_length": 2628.0,
"completions/mean_length": 612.140625,
"completions/mean_terminated_length": 637.0243530273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 217.0,
"epoch": 0.1344,
"grad_norm": 0.024109316989779472,
"learning_rate": 2.0555555555555555e-06,
"loss": -0.1785,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.0260985866189003,
"mask/share_reasoning": 0.7924785017967224,
"mask/share_step_conf": 0.1423604041337967,
"num_tokens": 31257945.0,
"reward": 1.302736759185791,
"reward_std": 0.285793662071228,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/final_brier_reward_step": 0.706494927406311,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.8033955097198486,
"step": 126
},
{
"adv/mean_abs_final_conf": 0.7272650599479675,
"adv/mean_abs_reasoning": 0.491754412651062,
"adv/mean_abs_step_conf": 0.7553401589393616,
"adv/ratio_final_to_reasoning": 1.4789192353704788,
"adv/ratio_step_to_reasoning": 1.5360109426721793,
"adv/std_final_conf": 0.8917009234428406,
"adv/std_reasoning": 0.7577455043792725,
"adv/std_step_conf": 0.9358600974082947,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.7085182534001431,
"calib/avg_num_step_conf": 9.0,
"calib/ece": 0.2907172995780591,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 0.45569620253164556,
"calib/gap": 0.3459642090193272,
"calib/mean_conf": 0.4984810126582278,
"calib/mu_c": 0.6590551181102362,
"calib/mu_w": 0.31309090909090903,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.12666666666666668,
"calib/std_conf": 0.46878446445254,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5359112825458052,
"calib/step_q_c_n": 1037.0,
"calib/step_q_gap": 0.22067055642110123,
"calib/step_q_w": 0.315240726124704,
"calib/step_q_w_n": 1267.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2900.0,
"completions/max_terminated_length": 2900.0,
"completions/mean_length": 640.9140625,
"completions/mean_terminated_length": 661.5886840820312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 146.0,
"epoch": 0.13546666666666668,
"grad_norm": 0.032043445855379105,
"learning_rate": 2.027777777777778e-06,
"loss": -0.0016,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.026310283690690994,
"mask/share_reasoning": 0.8025298118591309,
"mask/share_step_conf": 0.13990992307662964,
"num_tokens": 31525691.0,
"reward": 1.245164394378662,
"reward_std": 0.34896039962768555,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.6501039266586304,
"rewards/format_reward_step": 0.921875,
"rewards/step_l2_reward": 0.7783154845237732,
"step": 127
},
{
"adv/mean_abs_final_conf": 0.6428596377372742,
"adv/mean_abs_reasoning": 0.5563979148864746,
"adv/mean_abs_step_conf": 0.7521604299545288,
"adv/ratio_final_to_reasoning": 1.1553954832279356,
"adv/ratio_step_to_reasoning": 1.3518390522868815,
"adv/std_final_conf": 0.8436455130577087,
"adv/std_reasoning": 0.7928293347358704,
"adv/std_step_conf": 0.9357814192771912,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6912990116917511,
"calib/avg_num_step_conf": 6.9765625,
"calib/ece": 0.283064,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.492,
"calib/gap": 0.3543325366578386,
"calib/mean_conf": 0.526776,
"calib/mu_c": 0.6869343065693431,
"calib/mu_w": 0.3326017699115045,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.13092000000000004,
"calib/std_conf": 0.47018616507081534,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5741783029001075,
"calib/step_q_c_n": 931.0,
"calib/step_q_gap": 0.1583537414965987,
"calib/step_q_w": 0.4158245614035088,
"calib/step_q_w_n": 855.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2654.0,
"completions/max_terminated_length": 2654.0,
"completions/mean_length": 579.63671875,
"completions/mean_terminated_length": 588.8373413085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 211.0,
"epoch": 0.13653333333333334,
"grad_norm": 0.024526173248887062,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.0954,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.02894548699259758,
"mask/share_reasoning": 0.8233139514923096,
"mask/share_step_conf": 0.13211554288864136,
"num_tokens": 31780742.0,
"reward": 1.3008251190185547,
"reward_std": 0.3150237202644348,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.6897578239440918,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8051649332046509,
"step": 128
},
{
"adv/mean_abs_final_conf": 0.6891508102416992,
"adv/mean_abs_reasoning": 0.4242488145828247,
"adv/mean_abs_step_conf": 0.7488288879394531,
"adv/ratio_final_to_reasoning": 1.6244024415704255,
"adv/ratio_step_to_reasoning": 1.7650700772748105,
"adv/std_final_conf": 0.8731085658073425,
"adv/std_reasoning": 0.7013865113258362,
"adv/std_step_conf": 0.9358651638031006,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6691920885015086,
"calib/avg_num_step_conf": 8.26171875,
"calib/ece": 0.37857142857142856,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.4365079365079365,
"calib/gap": 0.21326315789473688,
"calib/mean_conf": 0.4896031746031746,
"calib/mu_c": 0.57,
"calib/mu_w": 0.3567368421052631,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1225793650793651,
"calib/std_conf": 0.4602706256962427,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.50537734375,
"calib/step_q_c_n": 1280.0,
"calib/step_q_gap": 0.0858204575224551,
"calib/step_q_w": 0.4195568862275449,
"calib/step_q_w_n": 835.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2562.0,
"completions/max_terminated_length": 2562.0,
"completions/mean_length": 601.7578125,
"completions/mean_terminated_length": 604.11767578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 192.0,
"epoch": 0.1376,
"grad_norm": 0.048718493431806564,
"learning_rate": 1.9722222222222224e-06,
"loss": 0.0562,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.027454271912574768,
"mask/share_reasoning": 0.822243332862854,
"mask/share_step_conf": 0.1463961899280548,
"num_tokens": 32037176.0,
"reward": 1.2862615585327148,
"reward_std": 0.27578479051589966,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6257296800613403,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8136311769485474,
"step": 129
},
{
"adv/mean_abs_final_conf": 0.5660302639007568,
"adv/mean_abs_reasoning": 0.32305261492729187,
"adv/mean_abs_step_conf": 0.7698035836219788,
"adv/ratio_final_to_reasoning": 1.7521302653072501,
"adv/ratio_step_to_reasoning": 2.3829046664588533,
"adv/std_final_conf": 0.7834324836730957,
"adv/std_reasoning": 0.5962008833885193,
"adv/std_step_conf": 0.9351350665092468,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7171706105755186,
"calib/avg_num_step_conf": 7.0234375,
"calib/ece": 0.2849392712550607,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.5506072874493927,
"calib/gap": 0.37632632193981885,
"calib/mean_conf": 0.5885829959514169,
"calib/mu_c": 0.7165644171779141,
"calib/mu_w": 0.34023809523809523,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.10680161943319838,
"calib/std_conf": 0.4692858043757504,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6083996212121212,
"calib/step_q_c_n": 1056.0,
"calib/step_q_gap": 0.25159369129298376,
"calib/step_q_w": 0.35680592991913745,
"calib/step_q_w_n": 742.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 3031.0,
"completions/max_terminated_length": 3031.0,
"completions/mean_length": 562.48046875,
"completions/mean_terminated_length": 569.1502075195312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 205.0,
"epoch": 0.13866666666666666,
"grad_norm": 0.026100866496562958,
"learning_rate": 1.944444444444445e-06,
"loss": -0.0812,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.029097583144903183,
"mask/share_reasoning": 0.8250260949134827,
"mask/share_step_conf": 0.13415755331516266,
"num_tokens": 32286459.0,
"reward": 1.3368961811065674,
"reward_std": 0.2628602981567383,
"rewards/accuracy_reward_step": 0.63671875,
"rewards/final_brier_reward_step": 0.69388747215271,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8297960758209229,
"step": 130
},
{
"adv/mean_abs_final_conf": 0.6387416124343872,
"adv/mean_abs_reasoning": 0.360165536403656,
"adv/mean_abs_step_conf": 0.7710878849029541,
"adv/ratio_final_to_reasoning": 1.7734667753399833,
"adv/ratio_step_to_reasoning": 2.140926343487669,
"adv/std_final_conf": 0.8598631620407104,
"adv/std_reasoning": 0.6611831188201904,
"adv/std_step_conf": 0.9360483884811401,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.7330898268398269,
"calib/avg_num_step_conf": 7.390625,
"calib/ece": 0.25274590163934435,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.4139344262295082,
"calib/gap": 0.38307359307359296,
"calib/mean_conf": 0.46454918032786885,
"calib/mu_c": 0.6717857142857142,
"calib/mu_w": 0.28871212121212125,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.12913934426229515,
"calib/std_conf": 0.4619455046005314,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5429871794871794,
"calib/step_q_c_n": 780.0,
"calib/step_q_gap": 0.1899745895591219,
"calib/step_q_w": 0.35301258992805756,
"calib/step_q_w_n": 1112.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2298.0,
"completions/max_terminated_length": 2298.0,
"completions/mean_length": 585.35546875,
"completions/mean_terminated_length": 601.8112182617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 211.0,
"epoch": 0.13973333333333332,
"grad_norm": 0.043049655854701996,
"learning_rate": 1.916666666666667e-06,
"loss": -0.1359,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.026013091206550598,
"mask/share_reasoning": 0.8193711042404175,
"mask/share_step_conf": 0.1272720843553543,
"num_tokens": 32542518.0,
"reward": 1.2662631273269653,
"reward_std": 0.33393311500549316,
"rewards/accuracy_reward_step": 0.4375,
"rewards/final_brier_reward_step": 0.686969518661499,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.7844971418380737,
"step": 131
},
{
"adv/mean_abs_final_conf": 0.5512483716011047,
"adv/mean_abs_reasoning": 0.46704182028770447,
"adv/mean_abs_step_conf": 0.7561357021331787,
"adv/ratio_final_to_reasoning": 1.1802976685503834,
"adv/ratio_step_to_reasoning": 1.6189892838020121,
"adv/std_final_conf": 0.7938889265060425,
"adv/std_reasoning": 0.7393816709518433,
"adv/std_step_conf": 0.9344714879989624,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8069264069264069,
"calib/avg_num_step_conf": 7.61328125,
"calib/ece": 0.1663855421686748,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.5662650602409639,
"calib/gap": 0.5592402597402596,
"calib/mean_conf": 0.6087951807228916,
"calib/mu_c": 0.7974545454545453,
"calib/mu_w": 0.23821428571428574,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.056265060240963956,
"calib/std_conf": 0.4545761897261351,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6008884859474161,
"calib/step_q_c_n": 1103.0,
"calib/step_q_gap": 0.28548659469446097,
"calib/step_q_w": 0.3154018912529551,
"calib/step_q_w_n": 846.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2979.0,
"completions/max_terminated_length": 2979.0,
"completions/mean_length": 631.93359375,
"completions/mean_terminated_length": 639.4268798828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 172.0,
"epoch": 0.1408,
"grad_norm": 0.03704225271940231,
"learning_rate": 1.888888888888889e-06,
"loss": -0.0146,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.0284785944968462,
"mask/share_reasoning": 0.8273619413375854,
"mask/share_step_conf": 0.1324407309293747,
"num_tokens": 32809885.0,
"reward": 1.4165544509887695,
"reward_std": 0.250224232673645,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.794607400894165,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8579226732254028,
"step": 132
},
{
"adv/mean_abs_final_conf": 0.7426465749740601,
"adv/mean_abs_reasoning": 0.522463858127594,
"adv/mean_abs_step_conf": 0.7531304359436035,
"adv/ratio_final_to_reasoning": 1.421431479749732,
"adv/ratio_step_to_reasoning": 1.4414976734327087,
"adv/std_final_conf": 0.8922584056854248,
"adv/std_reasoning": 0.7756352424621582,
"adv/std_step_conf": 0.9363987445831299,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.727751263335205,
"calib/avg_num_step_conf": 9.1328125,
"calib/ece": 0.28414937759336106,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.4605809128630705,
"calib/gap": 0.36234348680516554,
"calib/mean_conf": 0.5073858921161826,
"calib/mu_c": 0.7133653846153846,
"calib/mu_w": 0.35102189781021903,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.18000000000000008,
"calib/std_conf": 0.4655976087191073,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5147852760736197,
"calib/step_q_c_n": 815.0,
"calib/step_q_gap": 0.1762888873671194,
"calib/step_q_w": 0.3384963887065003,
"calib/step_q_w_n": 1523.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2813.0,
"completions/max_terminated_length": 2813.0,
"completions/mean_length": 719.05078125,
"completions/mean_terminated_length": 736.3080444335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 190.0,
"epoch": 0.14186666666666667,
"grad_norm": 0.03149598091840744,
"learning_rate": 1.8611111111111113e-06,
"loss": -0.0652,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.0236146692186594,
"mask/share_reasoning": 0.825226366519928,
"mask/share_step_conf": 0.12772145867347717,
"num_tokens": 33100306.0,
"reward": 1.2592096328735352,
"reward_std": 0.399488627910614,
"rewards/accuracy_reward_step": 0.41015625,
"rewards/final_brier_reward_step": 0.6683304309844971,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.7898881435394287,
"step": 133
},
{
"adv/mean_abs_final_conf": 0.7197145223617554,
"adv/mean_abs_reasoning": 0.5542685985565186,
"adv/mean_abs_step_conf": 0.7648236751556396,
"adv/ratio_final_to_reasoning": 1.2984941312499167,
"adv/ratio_step_to_reasoning": 1.3798791364826901,
"adv/std_final_conf": 0.8917521834373474,
"adv/std_reasoning": 0.8099950551986694,
"adv/std_step_conf": 0.9360350370407104,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.7137681159420289,
"calib/avg_num_step_conf": 7.9375,
"calib/ece": 0.25962083333333336,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.6208333333333333,
"calib/gap": 0.3615971867007671,
"calib/mean_conf": 0.6703791666666667,
"calib/mu_c": 0.8240579710144926,
"calib/mu_w": 0.4624607843137255,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.17750000000000005,
"calib/std_conf": 0.43895969108712657,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5716602687140115,
"calib/step_q_c_n": 1042.0,
"calib/step_q_gap": 0.2143956222493651,
"calib/step_q_w": 0.35726464646464645,
"calib/step_q_w_n": 990.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 3024.0,
"completions/max_terminated_length": 3024.0,
"completions/mean_length": 712.8828125,
"completions/mean_terminated_length": 724.1984252929688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 229.0,
"epoch": 0.14293333333333333,
"grad_norm": 0.022994376718997955,
"learning_rate": 1.8333333333333333e-06,
"loss": -0.1002,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.02312406152486801,
"mask/share_reasoning": 0.8467497229576111,
"mask/share_step_conf": 0.1145012304186821,
"num_tokens": 33391756.0,
"reward": 1.2713760137557983,
"reward_std": 0.4110109806060791,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.6849120259284973,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.7812637090682983,
"step": 134
},
{
"adv/mean_abs_final_conf": 0.565937340259552,
"adv/mean_abs_reasoning": 0.5127121210098267,
"adv/mean_abs_step_conf": 0.7459043264389038,
"adv/ratio_final_to_reasoning": 1.1038111194736222,
"adv/ratio_step_to_reasoning": 1.4548209333724875,
"adv/std_final_conf": 0.7928846478462219,
"adv/std_reasoning": 0.7754972577095032,
"adv/std_step_conf": 0.9352908730506897,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6286195286195286,
"calib/avg_num_step_conf": 7.7421875,
"calib/ece": 0.3159385542168675,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.6746987951807228,
"calib/gap": 0.23028989898989904,
"calib/mean_conf": 0.7117722891566265,
"calib/mu_c": 0.8033333333333333,
"calib/mu_w": 0.5730434343434343,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.2126506024096386,
"calib/std_conf": 0.42404818803797983,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.527002700270027,
"calib/step_q_c_n": 1111.0,
"calib/step_q_gap": 0.08112439946635308,
"calib/step_q_w": 0.44587830080367397,
"calib/step_q_w_n": 871.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2268.0,
"completions/max_terminated_length": 2268.0,
"completions/mean_length": 622.4453125,
"completions/mean_terminated_length": 629.8261108398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 186.0,
"epoch": 0.144,
"grad_norm": 0.023047512397170067,
"learning_rate": 1.8055555555555557e-06,
"loss": -0.0883,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.027006974443793297,
"mask/share_reasoning": 0.8312538266181946,
"mask/share_step_conf": 0.13002046942710876,
"num_tokens": 33656982.0,
"reward": 1.2960524559020996,
"reward_std": 0.3031335771083832,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.6604580879211426,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8099638819694519,
"step": 135
},
{
"adv/mean_abs_final_conf": 0.6360206007957458,
"adv/mean_abs_reasoning": 0.503818929195404,
"adv/mean_abs_step_conf": 0.7464261054992676,
"adv/ratio_final_to_reasoning": 1.2623991754566806,
"adv/ratio_step_to_reasoning": 1.4815364454275384,
"adv/std_final_conf": 0.8439712524414062,
"adv/std_reasoning": 0.7929543852806091,
"adv/std_step_conf": 0.9357492327690125,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7431534985622347,
"calib/avg_num_step_conf": 8.578125,
"calib/ece": 0.2198353909465022,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.6090534979423868,
"calib/gap": 0.41825961933452,
"calib/mean_conf": 0.6647736625514404,
"calib/mu_c": 0.8523880597014926,
"calib/mu_w": 0.43412844036697257,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.16658436213991784,
"calib/std_conf": 0.4341268259791531,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5619423306772907,
"calib/step_q_c_n": 1004.0,
"calib/step_q_gap": 0.2213466930933981,
"calib/step_q_w": 0.3405956375838926,
"calib/step_q_w_n": 1192.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2777.0,
"completions/max_terminated_length": 2777.0,
"completions/mean_length": 609.71875,
"completions/mean_terminated_length": 624.35205078125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 170.0,
"epoch": 0.14506666666666668,
"grad_norm": 0.02454623207449913,
"learning_rate": 1.777777777777778e-06,
"loss": -0.1694,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.027274496853351593,
"mask/share_reasoning": 0.8087484836578369,
"mask/share_step_conf": 0.14053946733474731,
"num_tokens": 33921558.0,
"reward": 1.3231323957443237,
"reward_std": 0.3496283292770386,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/final_brier_reward_step": 0.7197468876838684,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8159933686256409,
"step": 136
},
{
"adv/mean_abs_final_conf": 0.5607572793960571,
"adv/mean_abs_reasoning": 0.37462568283081055,
"adv/mean_abs_step_conf": 0.7560116052627563,
"adv/ratio_final_to_reasoning": 1.4968468663407357,
"adv/ratio_step_to_reasoning": 2.0180453180626925,
"adv/std_final_conf": 0.7927778959274292,
"adv/std_reasoning": 0.6614594459533691,
"adv/std_step_conf": 0.935052752494812,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6742097701149427,
"calib/avg_num_step_conf": 8.97265625,
"calib/ece": 0.29374999999999996,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6693548387096774,
"calib/gap": 0.2998563218390804,
"calib/mean_conf": 0.7189112903225806,
"calib/mu_c": 0.8591666666666666,
"calib/mu_w": 0.5593103448275862,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.24020161290322575,
"calib/std_conf": 0.4140116780357143,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5718988549618321,
"calib/step_q_c_n": 1048.0,
"calib/step_q_gap": 0.17152255392099947,
"calib/step_q_w": 0.40037630104083266,
"calib/step_q_w_n": 1249.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2734.0,
"completions/max_terminated_length": 2734.0,
"completions/mean_length": 649.9375,
"completions/mean_terminated_length": 660.2540283203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 240.0,
"epoch": 0.14613333333333334,
"grad_norm": 0.027212072163820267,
"learning_rate": 1.75e-06,
"loss": -0.0234,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.024864552542567253,
"mask/share_reasoning": 0.8198833465576172,
"mask/share_step_conf": 0.139627143740654,
"num_tokens": 34194926.0,
"reward": 1.2909659147262573,
"reward_std": 0.2898910343647003,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.6685027480125427,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8090583086013794,
"step": 137
},
{
"adv/mean_abs_final_conf": 0.5885981321334839,
"adv/mean_abs_reasoning": 0.5579100251197815,
"adv/mean_abs_step_conf": 0.7548394203186035,
"adv/ratio_final_to_reasoning": 1.0550054769263444,
"adv/ratio_step_to_reasoning": 1.3529769789609747,
"adv/std_final_conf": 0.7946194410324097,
"adv/std_reasoning": 0.7931153178215027,
"adv/std_step_conf": 0.935364305973053,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.7048546691403834,
"calib/avg_num_step_conf": 8.69921875,
"calib/ece": 0.22147058823529409,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 0.7100840336134454,
"calib/gap": 0.3526623376623378,
"calib/mean_conf": 0.7524789915966387,
"calib/mu_c": 0.876948051948052,
"calib/mu_w": 0.5242857142857142,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.16344537815126048,
"calib/std_conf": 0.40030890065698216,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5912727272727274,
"calib/step_q_c_n": 1100.0,
"calib/step_q_gap": 0.2486462853916271,
"calib/step_q_w": 0.3426264418811003,
"calib/step_q_w_n": 1127.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2808.0,
"completions/max_terminated_length": 2808.0,
"completions/mean_length": 601.58203125,
"completions/mean_terminated_length": 626.0365600585938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 186.0,
"epoch": 0.1472,
"grad_norm": 0.024571401998400688,
"learning_rate": 1.7222222222222224e-06,
"loss": -0.1332,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.02681715041399002,
"mask/share_reasoning": 0.8012897968292236,
"mask/share_step_conf": 0.13283054530620575,
"num_tokens": 34453267.0,
"reward": 1.292980432510376,
"reward_std": 0.37091195583343506,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.7039051055908203,
"rewards/format_reward_step": 0.921875,
"rewards/step_l2_reward": 0.7886841297149658,
"step": 138
},
{
"adv/mean_abs_final_conf": 0.5235190391540527,
"adv/mean_abs_reasoning": 0.3951146602630615,
"adv/mean_abs_step_conf": 0.7521218657493591,
"adv/ratio_final_to_reasoning": 1.3249800420098345,
"adv/ratio_step_to_reasoning": 1.9035534273737336,
"adv/std_final_conf": 0.7528417110443115,
"adv/std_reasoning": 0.6815641522407532,
"adv/std_step_conf": 0.9350164532661438,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7040816326530612,
"calib/avg_num_step_conf": 7.5,
"calib/ece": 0.2350396825396825,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.7857142857142857,
"calib/gap": 0.2928571428571429,
"calib/mean_conf": 0.8327380952380952,
"calib/mu_c": 0.9303571428571429,
"calib/mu_w": 0.6375,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.20055555555555551,
"calib/std_conf": 0.33762952079262504,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.632515404699739,
"calib/step_q_c_n": 1149.0,
"calib/step_q_gap": 0.15545963297470655,
"calib/step_q_w": 0.47705577172503244,
"calib/step_q_w_n": 771.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2996.0,
"completions/max_terminated_length": 2996.0,
"completions/mean_length": 580.10546875,
"completions/mean_terminated_length": 584.6732177734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 163.0,
"epoch": 0.14826666666666666,
"grad_norm": 0.023692641407251358,
"learning_rate": 1.6944444444444446e-06,
"loss": 0.0227,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.02832508645951748,
"mask/share_reasoning": 0.8266885280609131,
"mask/share_step_conf": 0.13717389106750488,
"num_tokens": 34704870.0,
"reward": 1.3846945762634277,
"reward_std": 0.28370344638824463,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.751888632774353,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8450783491134644,
"step": 139
},
{
"adv/mean_abs_final_conf": 0.5185952186584473,
"adv/mean_abs_reasoning": 0.4070984721183777,
"adv/mean_abs_step_conf": 0.7555028200149536,
"adv/ratio_final_to_reasoning": 1.2738815155947039,
"adv/ratio_step_to_reasoning": 1.8558232755913304,
"adv/std_final_conf": 0.7767248153686523,
"adv/std_reasoning": 0.6816426515579224,
"adv/std_step_conf": 0.934374213218689,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7118879178804273,
"calib/avg_num_step_conf": 7.6171875,
"calib/ece": 0.22605458167330678,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.8087649402390438,
"calib/gap": 0.3221040851713136,
"calib/mean_conf": 0.8498019920318725,
"calib/mu_c": 0.9640141975308641,
"calib/mu_w": 0.6419101123595505,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2152191235059761,
"calib/std_conf": 0.324334355103517,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6633154121863799,
"calib/step_q_c_n": 1116.0,
"calib/step_q_gap": 0.21230222273793858,
"calib/step_q_w": 0.4510131894484413,
"calib/step_q_w_n": 834.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2684.0,
"completions/max_terminated_length": 2684.0,
"completions/mean_length": 598.13671875,
"completions/mean_terminated_length": 605.229248046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 206.0,
"epoch": 0.14933333333333335,
"grad_norm": 0.030547764152288437,
"learning_rate": 1.6666666666666667e-06,
"loss": -0.0224,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.027081944048404694,
"mask/share_reasoning": 0.8269479274749756,
"mask/share_step_conf": 0.13425138592720032,
"num_tokens": 34963009.0,
"reward": 1.3870867490768433,
"reward_std": 0.26042604446411133,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/final_brier_reward_step": 0.7565398216247559,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8474887013435364,
"step": 140
},
{
"adv/mean_abs_final_conf": 0.47909656167030334,
"adv/mean_abs_reasoning": 0.4322686195373535,
"adv/mean_abs_step_conf": 0.7472922801971436,
"adv/ratio_final_to_reasoning": 1.1083306537103448,
"adv/ratio_step_to_reasoning": 1.7287682853244173,
"adv/std_final_conf": 0.7406020760536194,
"adv/std_reasoning": 0.7205878496170044,
"adv/std_step_conf": 0.9347190260887146,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7789922480620155,
"calib/avg_num_step_conf": 7.76171875,
"calib/ece": 0.1749797570850204,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.7732793522267206,
"calib/gap": 0.45480465116279084,
"calib/mean_conf": 0.8055060728744938,
"calib/mu_c": 0.9436046511627908,
"calib/mu_w": 0.48879999999999996,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.14206477732793538,
"calib/std_conf": 0.3698172857802966,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.614653937947494,
"calib/step_q_c_n": 1257.0,
"calib/step_q_gap": 0.22399640370091856,
"calib/step_q_w": 0.39065753424657546,
"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": 0.0078125,
"completions/max_length": 2514.0,
"completions/max_terminated_length": 2514.0,
"completions/mean_length": 641.890625,
"completions/mean_terminated_length": 646.9448852539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 62.0,
"epoch": 0.1504,
"grad_norm": 0.02207571640610695,
"learning_rate": 1.638888888888889e-06,
"loss": -0.1009,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.025763999670743942,
"mask/share_reasoning": 0.8380931615829468,
"mask/share_step_conf": 0.12833033502101898,
"num_tokens": 35234429.0,
"reward": 1.4104137420654297,
"reward_std": 0.2923741936683655,
"rewards/accuracy_reward_step": 0.671875,
"rewards/final_brier_reward_step": 0.8029515743255615,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8452661633491516,
"step": 141
},
{
"adv/mean_abs_final_conf": 0.5865595936775208,
"adv/mean_abs_reasoning": 0.3907625079154968,
"adv/mean_abs_step_conf": 0.746477484703064,
"adv/ratio_final_to_reasoning": 1.5010641548148869,
"adv/ratio_step_to_reasoning": 1.9103098930475981,
"adv/std_final_conf": 0.8279939889907837,
"adv/std_reasoning": 0.701130747795105,
"adv/std_step_conf": 0.935562014579773,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7004310344827586,
"calib/avg_num_step_conf": 7.96875,
"calib/ece": 0.29021811023622057,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.7362204724409449,
"calib/gap": 0.32836880309845107,
"calib/mean_conf": 0.7786007874015748,
"calib/mu_c": 0.9285644927536233,
"calib/mu_w": 0.6001956896551722,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2627559055118111,
"calib/std_conf": 0.3905670701666905,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6373262906309752,
"calib/step_q_c_n": 1046.0,
"calib/step_q_gap": 0.1907065723211161,
"calib/step_q_w": 0.44661971830985914,
"calib/step_q_w_n": 994.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2626.0,
"completions/max_terminated_length": 2626.0,
"completions/mean_length": 624.6328125,
"completions/mean_terminated_length": 627.0823974609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 167.0,
"epoch": 0.15146666666666667,
"grad_norm": 0.02521516941487789,
"learning_rate": 1.6111111111111113e-06,
"loss": 0.0182,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.02745303511619568,
"mask/share_reasoning": 0.8278377056121826,
"mask/share_step_conf": 0.14080297946929932,
"num_tokens": 35499495.0,
"reward": 1.3193681240081787,
"reward_std": 0.2825583815574646,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.701399564743042,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8155432939529419,
"step": 142
},
{
"adv/mean_abs_final_conf": 0.4948643743991852,
"adv/mean_abs_reasoning": 0.44098764657974243,
"adv/mean_abs_step_conf": 0.7755706310272217,
"adv/ratio_final_to_reasoning": 1.1221728731797942,
"adv/ratio_step_to_reasoning": 1.7587128279952342,
"adv/std_final_conf": 0.7587860226631165,
"adv/std_reasoning": 0.7205314040184021,
"adv/std_step_conf": 0.9349554181098938,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.7777046783625731,
"calib/avg_num_step_conf": 8.8984375,
"calib/ece": 0.2130962343096235,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.7196652719665272,
"calib/gap": 0.4375840643274854,
"calib/mean_conf": 0.7689121338912134,
"calib/mu_c": 0.9428472222222223,
"calib/mu_w": 0.5052631578947369,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.18974895397489547,
"calib/std_conf": 0.38111550185258836,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6166698564593303,
"calib/step_q_c_n": 1045.0,
"calib/step_q_gap": 0.2781864825744965,
"calib/step_q_w": 0.33848337388483374,
"calib/step_q_w_n": 1233.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2999.0,
"completions/max_terminated_length": 2999.0,
"completions/mean_length": 657.38671875,
"completions/mean_terminated_length": 678.5927124023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 210.0,
"epoch": 0.15253333333333333,
"grad_norm": 0.02659987099468708,
"learning_rate": 1.5833333333333333e-06,
"loss": -0.0362,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.02512693963944912,
"mask/share_reasoning": 0.8076825737953186,
"mask/share_step_conf": 0.13594046235084534,
"num_tokens": 35775122.0,
"reward": 1.3324463367462158,
"reward_std": 0.2990143597126007,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.7442284822463989,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.8107225298881531,
"step": 143
},
{
"adv/mean_abs_final_conf": 0.4784291386604309,
"adv/mean_abs_reasoning": 0.3893376588821411,
"adv/mean_abs_step_conf": 0.7565593719482422,
"adv/ratio_final_to_reasoning": 1.228828313279757,
"adv/ratio_step_to_reasoning": 1.9431959757513853,
"adv/std_final_conf": 0.758713960647583,
"adv/std_reasoning": 0.6614014506340027,
"adv/std_step_conf": 0.9349938035011292,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7353927203065134,
"calib/avg_num_step_conf": 7.359375,
"calib/ece": 0.18304878048780482,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.7439024390243902,
"calib/gap": 0.3871599616858238,
"calib/mean_conf": 0.7889837398373983,
"calib/mu_c": 0.9022988505747126,
"calib/mu_w": 0.5151388888888888,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.13235772357723571,
"calib/std_conf": 0.3750308710055105,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.5998940505297473,
"calib/step_q_c_n": 1227.0,
"calib/step_q_gap": 0.2120706106515129,
"calib/step_q_w": 0.3878234398782344,
"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": 0.01953125,
"completions/max_length": 2403.0,
"completions/max_terminated_length": 2403.0,
"completions/mean_length": 601.13671875,
"completions/mean_terminated_length": 613.111572265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.1536,
"grad_norm": 0.0394146703183651,
"learning_rate": 1.5555555555555558e-06,
"loss": -0.1006,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.028188010677695274,
"mask/share_reasoning": 0.8198911547660828,
"mask/share_step_conf": 0.1323896050453186,
"num_tokens": 36033141.0,
"reward": 1.3731136322021484,
"reward_std": 0.2826434373855591,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/final_brier_reward_step": 0.7666667699813843,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8268896341323853,
"step": 144
},
{
"adv/mean_abs_final_conf": 0.6184766292572021,
"adv/mean_abs_reasoning": 0.5618990659713745,
"adv/mean_abs_step_conf": 0.7520120143890381,
"adv/ratio_final_to_reasoning": 1.1006899044902665,
"adv/ratio_step_to_reasoning": 1.3383400328117803,
"adv/std_final_conf": 0.8280069828033447,
"adv/std_reasoning": 0.7929149866104126,
"adv/std_step_conf": 0.9355904459953308,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6229311846689896,
"calib/avg_num_step_conf": 8.15234375,
"calib/ece": 0.2644354838709678,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.8104838709677419,
"calib/gap": 0.20515098722415792,
"calib/mean_conf": 0.8543548387096774,
"calib/mu_c": 0.9238414634146341,
"calib/mu_w": 0.7186904761904762,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.22875000000000006,
"calib/std_conf": 0.3157919944629842,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6530969845150774,
"calib/step_q_c_n": 1227.0,
"calib/step_q_gap": 0.19349233335228666,
"calib/step_q_w": 0.4596046511627907,
"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": 0.01953125,
"completions/max_length": 2685.0,
"completions/max_terminated_length": 2685.0,
"completions/mean_length": 572.71875,
"completions/mean_terminated_length": 584.1275024414062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 207.0,
"epoch": 0.15466666666666667,
"grad_norm": 0.02873079851269722,
"learning_rate": 1.527777777777778e-06,
"loss": -0.1331,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.028295297175645828,
"mask/share_reasoning": 0.8084123134613037,
"mask/share_step_conf": 0.14376112818717957,
"num_tokens": 36282461.0,
"reward": 1.3241336345672607,
"reward_std": 0.3417121171951294,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.708076536655426,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8091577291488647,
"step": 145
},
{
"adv/mean_abs_final_conf": 0.6596519947052002,
"adv/mean_abs_reasoning": 0.5093032121658325,
"adv/mean_abs_step_conf": 0.7680574059486389,
"adv/ratio_final_to_reasoning": 1.2952048582218898,
"adv/ratio_step_to_reasoning": 1.508055295159918,
"adv/std_final_conf": 0.84257972240448,
"adv/std_reasoning": 0.7754387855529785,
"adv/std_step_conf": 0.9353065490722656,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7108323133414931,
"calib/avg_num_step_conf": 7.49609375,
"calib/ece": 0.30551440329218105,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.7160493827160493,
"calib/gap": 0.3522419420644636,
"calib/mean_conf": 0.7636213991769547,
"calib/mu_c": 0.9506140350877194,
"calib/mu_w": 0.5983720930232558,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.3,
"calib/std_conf": 0.38390034912799154,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.6353417721518988,
"calib/step_q_c_n": 790.0,
"calib/step_q_gap": 0.2182204258277181,
"calib/step_q_w": 0.4171213463241807,
"calib/step_q_w_n": 1129.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2622.0,
"completions/max_terminated_length": 2622.0,
"completions/mean_length": 607.8046875,
"completions/mean_terminated_length": 632.5121459960938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 188.0,
"epoch": 0.15573333333333333,
"grad_norm": 0.02385423704981804,
"learning_rate": 1.5e-06,
"loss": -0.1869,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.026042595505714417,
"mask/share_reasoning": 0.8098737001419067,
"mask/share_step_conf": 0.12502123415470123,
"num_tokens": 36545275.0,
"reward": 1.2390952110290527,
"reward_std": 0.3953307271003723,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/final_brier_reward_step": 0.6571453213691711,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.7710694074630737,
"step": 146
},
{
"adv/mean_abs_final_conf": 0.49504554271698,
"adv/mean_abs_reasoning": 0.36164918541908264,
"adv/mean_abs_step_conf": 0.7685758471488953,
"adv/ratio_final_to_reasoning": 1.3688556830103635,
"adv/ratio_step_to_reasoning": 2.1251972301783617,
"adv/std_final_conf": 0.7386893630027771,
"adv/std_reasoning": 0.6612387895584106,
"adv/std_step_conf": 0.9356215596199036,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6546900161030595,
"calib/avg_num_step_conf": 7.9609375,
"calib/ece": 0.36841463414634146,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7195121951219512,
"calib/gap": 0.27287439613526576,
"calib/mean_conf": 0.7658130081300812,
"calib/mu_c": 0.9188888888888889,
"calib/mu_w": 0.6460144927536231,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.34760162601626016,
"calib/std_conf": 0.38750288432377356,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6030394431554523,
"calib/step_q_c_n": 862.0,
"calib/step_q_gap": 0.14825202818946592,
"calib/step_q_w": 0.4547874149659864,
"calib/step_q_w_n": 1176.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2981.0,
"completions/max_terminated_length": 2981.0,
"completions/mean_length": 656.359375,
"completions/mean_terminated_length": 661.5275268554688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 259.0,
"epoch": 0.1568,
"grad_norm": 0.025043372064828873,
"learning_rate": 1.4722222222222225e-06,
"loss": -0.0195,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.024738822132349014,
"mask/share_reasoning": 0.836007833480835,
"mask/share_step_conf": 0.13144081830978394,
"num_tokens": 36816983.0,
"reward": 1.1915602684020996,
"reward_std": 0.2968427538871765,
"rewards/accuracy_reward_step": 0.421875,
"rewards/final_brier_reward_step": 0.606521487236023,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7500181198120117,
"step": 147
},
{
"adv/mean_abs_final_conf": 0.5144132971763611,
"adv/mean_abs_reasoning": 0.45546209812164307,
"adv/mean_abs_step_conf": 0.7453041076660156,
"adv/ratio_final_to_reasoning": 1.129431624053542,
"adv/ratio_step_to_reasoning": 1.6363691089548413,
"adv/std_final_conf": 0.7764376997947693,
"adv/std_reasoning": 0.7394083142280579,
"adv/std_step_conf": 0.9351122379302979,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.68747591522158,
"calib/avg_num_step_conf": 7.41015625,
"calib/ece": 0.20294354838709677,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.7943548387096774,
"calib/gap": 0.330892485549133,
"calib/mean_conf": 0.8280241935483871,
"calib/mu_c": 0.9280924855491329,
"calib/mu_w": 0.5972,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.1666935483870968,
"calib/std_conf": 0.34947040494268067,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6147808764940239,
"calib/step_q_c_n": 1255.0,
"calib/step_q_gap": 0.18976530017003634,
"calib/step_q_w": 0.42501557632398757,
"calib/step_q_w_n": 642.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2486.0,
"completions/max_terminated_length": 2486.0,
"completions/mean_length": 575.42578125,
"completions/mean_terminated_length": 589.2360229492188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 171.0,
"epoch": 0.15786666666666666,
"grad_norm": 0.024046506732702255,
"learning_rate": 1.4444444444444445e-06,
"loss": -0.1088,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.029452763497829437,
"mask/share_reasoning": 0.8054896593093872,
"mask/share_step_conf": 0.14162006974220276,
"num_tokens": 37069404.0,
"reward": 1.372555136680603,
"reward_std": 0.3066759705543518,
"rewards/accuracy_reward_step": 0.67578125,
"rewards/final_brier_reward_step": 0.7648323774337769,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8256858587265015,
"step": 148
},
{
"adv/mean_abs_final_conf": 0.5331004858016968,
"adv/mean_abs_reasoning": 0.5124787092208862,
"adv/mean_abs_step_conf": 0.7494461536407471,
"adv/ratio_final_to_reasoning": 1.0402392845005435,
"adv/ratio_step_to_reasoning": 1.46239471056294,
"adv/std_final_conf": 0.7765871286392212,
"adv/std_reasoning": 0.7575570940971375,
"adv/std_step_conf": 0.9348960518836975,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.7253521126760563,
"calib/avg_num_step_conf": 7.828125,
"calib/ece": 0.25684426229508195,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.8073770491803278,
"calib/gap": 0.3722328086164043,
"calib/mean_conf": 0.8304508196721311,
"calib/mu_c": 0.986056338028169,
"calib/mu_w": 0.6138235294117648,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.2526639344262295,
"calib/std_conf": 0.3521343028638739,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.6404268292682928,
"calib/step_q_c_n": 984.0,
"calib/step_q_gap": 0.25766212338593986,
"calib/step_q_w": 0.3827647058823529,
"calib/step_q_w_n": 1020.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2420.0,
"completions/max_terminated_length": 2420.0,
"completions/mean_length": 620.30078125,
"completions/mean_terminated_length": 635.1880493164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 197.0,
"epoch": 0.15893333333333334,
"grad_norm": 0.028948621824383736,
"learning_rate": 1.4166666666666667e-06,
"loss": -0.0542,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.026255900040268898,
"mask/share_reasoning": 0.8194175958633423,
"mask/share_step_conf": 0.13088896870613098,
"num_tokens": 37332657.0,
"reward": 1.3208997249603271,
"reward_std": 0.3231900632381439,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.7129691243171692,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8140244483947754,
"step": 149
},
{
"adv/mean_abs_final_conf": 0.5323728919029236,
"adv/mean_abs_reasoning": 0.4642510414123535,
"adv/mean_abs_step_conf": 0.7564442157745361,
"adv/ratio_final_to_reasoning": 1.1467349438427288,
"adv/ratio_step_to_reasoning": 1.6293861473595554,
"adv/std_final_conf": 0.7764654755592346,
"adv/std_reasoning": 0.7205824255943298,
"adv/std_step_conf": 0.9355114698410034,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6243245216883306,
"calib/avg_num_step_conf": 7.39453125,
"calib/ece": 0.26751004016064256,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.8313253012048193,
"calib/gap": 0.17992697531765733,
"calib/mean_conf": 0.8614056224899599,
"calib/mu_c": 0.9206586826347305,
"calib/mu_w": 0.7407317073170732,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2291164658634538,
"calib/std_conf": 0.3127042820309887,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5925986277873071,
"calib/step_q_c_n": 1166.0,
"calib/step_q_gap": 0.1243042674021626,
"calib/step_q_w": 0.4682943603851445,
"calib/step_q_w_n": 727.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2356.0,
"completions/max_terminated_length": 2356.0,
"completions/mean_length": 556.68359375,
"completions/mean_terminated_length": 565.5198974609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 171.0,
"epoch": 0.16,
"grad_norm": 0.031456056982278824,
"learning_rate": 1.3888888888888892e-06,
"loss": -0.017,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.030916960909962654,
"mask/share_reasoning": 0.8093185424804688,
"mask/share_step_conf": 0.14413949847221375,
"num_tokens": 37580128.0,
"reward": 1.3363224267959595,
"reward_std": 0.2611082196235657,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.7043378949165344,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8220440745353699,
"step": 150
},
{
"adv/mean_abs_final_conf": 0.5166769027709961,
"adv/mean_abs_reasoning": 0.4023810029029846,
"adv/mean_abs_step_conf": 0.7550654411315918,
"adv/ratio_final_to_reasoning": 1.2840489462559657,
"adv/ratio_step_to_reasoning": 1.8764937601033829,
"adv/std_final_conf": 0.7582067847251892,
"adv/std_reasoning": 0.6817334294319153,
"adv/std_step_conf": 0.9352085590362549,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6951489533011272,
"calib/avg_num_step_conf": 8.140625,
"calib/ece": 0.2702845528455285,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.7276422764227642,
"calib/gap": 0.3219202898550724,
"calib/mean_conf": 0.778089430894309,
"calib/mu_c": 0.9194202898550724,
"calib/mu_w": 0.5975,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.24369918699186996,
"calib/std_conf": 0.37643277524484753,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.6179471544715448,
"calib/step_q_c_n": 984.0,
"calib/step_q_gap": 0.24471988174427206,
"calib/step_q_w": 0.37322727272727274,
"calib/step_q_w_n": 1100.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2930.0,
"completions/max_terminated_length": 2930.0,
"completions/mean_length": 636.46484375,
"completions/mean_terminated_length": 649.1434326171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 216.0,
"epoch": 0.16106666666666666,
"grad_norm": 0.027021048590540886,
"learning_rate": 1.3611111111111112e-06,
"loss": -0.0869,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.02588428556919098,
"mask/share_reasoning": 0.8255480527877808,
"mask/share_step_conf": 0.12903639674186707,
"num_tokens": 37850087.0,
"reward": 1.2872600555419922,
"reward_std": 0.30778902769088745,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.6913784742355347,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.7919613122940063,
"step": 151
},
{
"adv/mean_abs_final_conf": 0.6841462850570679,
"adv/mean_abs_reasoning": 0.6128799915313721,
"adv/mean_abs_step_conf": 0.7404754161834717,
"adv/ratio_final_to_reasoning": 1.1162809922177852,
"adv/ratio_step_to_reasoning": 1.2081899008210129,
"adv/std_final_conf": 0.8762879371643066,
"adv/std_reasoning": 0.8267136812210083,
"adv/std_step_conf": 0.9357709884643555,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6138136288998359,
"calib/avg_num_step_conf": 8.19921875,
"calib/ece": 0.38019876033057853,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.756198347107438,
"calib/gap": 0.18923923097974815,
"calib/mean_conf": 0.7949252066115703,
"calib/mu_c": 0.8856349206349207,
"calib/mu_w": 0.6963956896551725,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 0.9609375,
"calib/nonempty_step_conf_rate": 0.94921875,
"calib/pce": 0.32723140495867775,
"calib/std_conf": 0.370573986635919,
"calib/step_conf_rate": 0.94921875,
"calib/step_q_c": 0.5821627408993576,
"calib/step_q_c_n": 934.0,
"calib/step_q_gap": 0.15284085248734047,
"calib/step_q_w": 0.4293218884120171,
"calib/step_q_w_n": 1165.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2792.0,
"completions/max_terminated_length": 2792.0,
"completions/mean_length": 629.27734375,
"completions/mean_terminated_length": 636.7391357421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 109.0,
"epoch": 0.16213333333333332,
"grad_norm": 0.027300406247377396,
"learning_rate": 1.3333333333333334e-06,
"loss": -0.1419,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.026354603469371796,
"mask/share_reasoning": 0.8259069919586182,
"mask/share_step_conf": 0.13601967692375183,
"num_tokens": 38116574.0,
"reward": 1.1926360130310059,
"reward_std": 0.40411001443862915,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/final_brier_reward_step": 0.5822392702102661,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.7593287229537964,
"step": 152
},
{
"adv/mean_abs_final_conf": 0.49810728430747986,
"adv/mean_abs_reasoning": 0.4407353401184082,
"adv/mean_abs_step_conf": 0.7415911555290222,
"adv/ratio_final_to_reasoning": 1.1301732331554308,
"adv/ratio_step_to_reasoning": 1.6826223994876062,
"adv/std_final_conf": 0.762622594833374,
"adv/std_reasoning": 0.7393104434013367,
"adv/std_step_conf": 0.9351751804351807,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.582449160035367,
"calib/avg_num_step_conf": 7.80859375,
"calib/ece": 0.32271604938271603,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.7860082304526749,
"calib/gap": 0.11059681697612733,
"calib/mean_conf": 0.8395061728395061,
"calib/mu_c": 0.8791025641025642,
"calib/mu_w": 0.7685057471264368,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.98046875,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.26012345679012344,
"calib/std_conf": 0.3278006315996392,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.5794348659003832,
"calib/step_q_c_n": 1044.0,
"calib/step_q_gap": 0.18048198631923135,
"calib/step_q_w": 0.39895287958115183,
"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": 0.02734375,
"completions/max_length": 2868.0,
"completions/max_terminated_length": 2868.0,
"completions/mean_length": 606.0078125,
"completions/mean_terminated_length": 623.0441284179688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 206.0,
"epoch": 0.1632,
"grad_norm": 0.035739362239837646,
"learning_rate": 1.3055555555555556e-06,
"loss": -0.1231,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.026105670258402824,
"mask/share_reasoning": 0.8216968774795532,
"mask/share_step_conf": 0.12485368549823761,
"num_tokens": 38379032.0,
"reward": 1.2565110921859741,
"reward_std": 0.2758803963661194,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.6402719020843506,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.780515730381012,
"step": 153
},
{
"adv/mean_abs_final_conf": 0.4910415709018707,
"adv/mean_abs_reasoning": 0.41510850191116333,
"adv/mean_abs_step_conf": 0.7597833871841431,
"adv/ratio_final_to_reasoning": 1.1829234251794671,
"adv/ratio_step_to_reasoning": 1.8303248034817245,
"adv/std_final_conf": 0.7406032085418701,
"adv/std_reasoning": 0.7013994455337524,
"adv/std_step_conf": 0.9353368878364563,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.609903381642512,
"calib/avg_num_step_conf": 7.10546875,
"calib/ece": 0.36212851405622504,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.8032128514056225,
"calib/gap": 0.17549353701527615,
"calib/mean_conf": 0.8347389558232933,
"calib/mu_c": 0.9129710144927536,
"calib/mu_w": 0.7374774774774775,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.3213253012048194,
"calib/std_conf": 0.3449696339126116,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5856728778467909,
"calib/step_q_c_n": 966.0,
"calib/step_q_gap": 0.1229765120789128,
"calib/step_q_w": 0.4626963657678781,
"calib/step_q_w_n": 853.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2499.0,
"completions/max_terminated_length": 2499.0,
"completions/mean_length": 578.46484375,
"completions/mean_terminated_length": 583.0196533203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 181.0,
"epoch": 0.16426666666666667,
"grad_norm": 0.026899857446551323,
"learning_rate": 1.2777777777777779e-06,
"loss": -0.0082,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.028808027505874634,
"mask/share_reasoning": 0.829174280166626,
"mask/share_step_conf": 0.134205162525177,
"num_tokens": 38631559.0,
"reward": 1.2400684356689453,
"reward_std": 0.2998583912849426,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.6244043111801147,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7766944169998169,
"step": 154
},
{
"adv/mean_abs_final_conf": 0.5528605580329895,
"adv/mean_abs_reasoning": 0.5002375841140747,
"adv/mean_abs_step_conf": 0.7550753355026245,
"adv/ratio_final_to_reasoning": 1.1051959620589296,
"adv/ratio_step_to_reasoning": 1.5094334361938633,
"adv/std_final_conf": 0.7846831679344177,
"adv/std_reasoning": 0.775325357913971,
"adv/std_step_conf": 0.9355731010437012,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6421815889029003,
"calib/avg_num_step_conf": 7.59765625,
"calib/ece": 0.31912698412698415,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.6468253968253969,
"calib/gap": 0.23936696090794463,
"calib/mean_conf": 0.6987301587301586,
"calib/mu_c": 0.8222131147540984,
"calib/mu_w": 0.5828461538461538,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.26686507936507936,
"calib/std_conf": 0.41400575896420855,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.539650110375276,
"calib/step_q_c_n": 906.0,
"calib/step_q_gap": 0.11493403722801904,
"calib/step_q_w": 0.42471607314725696,
"calib/step_q_w_n": 1039.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2685.0,
"completions/max_terminated_length": 2685.0,
"completions/mean_length": 585.1953125,
"completions/mean_terminated_length": 587.490234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 205.0,
"epoch": 0.16533333333333333,
"grad_norm": 0.027479618787765503,
"learning_rate": 1.25e-06,
"loss": -0.0124,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.027545053511857986,
"mask/share_reasoning": 0.8279811143875122,
"mask/share_step_conf": 0.14056754112243652,
"num_tokens": 38888585.0,
"reward": 1.2375892400741577,
"reward_std": 0.2888715863227844,
"rewards/accuracy_reward_step": 0.48046875,
"rewards/final_brier_reward_step": 0.6382601261138916,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7727559804916382,
"step": 155
},
{
"adv/mean_abs_final_conf": 0.5545140504837036,
"adv/mean_abs_reasoning": 0.4322436451911926,
"adv/mean_abs_step_conf": 0.7410604953765869,
"adv/ratio_final_to_reasoning": 1.282873806596804,
"adv/ratio_step_to_reasoning": 1.7144508742258004,
"adv/std_final_conf": 0.7995932698249817,
"adv/std_reasoning": 0.7205738425254822,
"adv/std_step_conf": 0.9348962306976318,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6980792724311116,
"calib/avg_num_step_conf": 7.79296875,
"calib/ece": 0.24755102040816335,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.6816326530612244,
"calib/gap": 0.3311680466947198,
"calib/mean_conf": 0.7417551020408162,
"calib/mu_c": 0.8850359712230216,
"calib/mu_w": 0.5538679245283018,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.21097959183673476,
"calib/std_conf": 0.3934833726026877,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.5458590733590734,
"calib/step_q_c_n": 1036.0,
"calib/step_q_gap": 0.168372107769918,
"calib/step_q_w": 0.37748696558915534,
"calib/step_q_w_n": 959.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3048.0,
"completions/max_terminated_length": 3048.0,
"completions/mean_length": 598.31640625,
"completions/mean_terminated_length": 607.8135375976562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 148.0,
"epoch": 0.1664,
"grad_norm": 0.03416966274380684,
"learning_rate": 1.2222222222222223e-06,
"loss": -0.0556,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.028070878237485886,
"mask/share_reasoning": 0.8195517659187317,
"mask/share_step_conf": 0.13675233721733093,
"num_tokens": 39146514.0,
"reward": 1.2828004360198975,
"reward_std": 0.26979345083236694,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.6931429505348206,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.787010133266449,
"step": 156
},
{
"adv/mean_abs_final_conf": 0.5185093879699707,
"adv/mean_abs_reasoning": 0.44146716594696045,
"adv/mean_abs_step_conf": 0.7368483543395996,
"adv/ratio_final_to_reasoning": 1.1745140476251554,
"adv/ratio_step_to_reasoning": 1.6690898240620853,
"adv/std_final_conf": 0.7898529171943665,
"adv/std_reasoning": 0.7574617862701416,
"adv/std_step_conf": 0.9352151155471802,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7419051204819278,
"calib/avg_num_step_conf": 8.0703125,
"calib/ece": 0.19674796747967488,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7479674796747967,
"calib/gap": 0.3956746987951808,
"calib/mean_conf": 0.79,
"calib/mu_c": 0.9186746987951807,
"calib/mu_w": 0.5229999999999999,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.15597560975609762,
"calib/std_conf": 0.3765472146730723,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5667429837518465,
"calib/step_q_c_n": 1354.0,
"calib/step_q_gap": 0.1697626466731948,
"calib/step_q_w": 0.3969803370786517,
"calib/step_q_w_n": 712.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2964.0,
"completions/max_terminated_length": 2964.0,
"completions/mean_length": 597.6796875,
"completions/mean_terminated_length": 607.1666870117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.16746666666666668,
"grad_norm": 0.041828371584415436,
"learning_rate": 1.1944444444444446e-06,
"loss": -0.017,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.02795276790857315,
"mask/share_reasoning": 0.8079496622085571,
"mask/share_step_conf": 0.1484726071357727,
"num_tokens": 39403248.0,
"reward": 1.3735963106155396,
"reward_std": 0.2610359191894531,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.7679358720779419,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8279096484184265,
"step": 157
},
{
"adv/mean_abs_final_conf": 0.5031821727752686,
"adv/mean_abs_reasoning": 0.42407286167144775,
"adv/mean_abs_step_conf": 0.7414059638977051,
"adv/ratio_final_to_reasoning": 1.1865465071073353,
"adv/ratio_step_to_reasoning": 1.7482985376039284,
"adv/std_final_conf": 0.7528063058853149,
"adv/std_reasoning": 0.7013627290725708,
"adv/std_step_conf": 0.9352811574935913,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6027310363247863,
"calib/avg_num_step_conf": 6.875,
"calib/ece": 0.3140873015873016,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.8134920634920635,
"calib/gap": 0.14328525641025647,
"calib/mean_conf": 0.8659920634920636,
"calib/mu_c": 0.9205769230769232,
"calib/mu_w": 0.7772916666666667,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.280515873015873,
"calib/std_conf": 0.29557619263279233,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.6128088803088804,
"calib/step_q_c_n": 1036.0,
"calib/step_q_gap": 0.14988070351330024,
"calib/step_q_w": 0.4629281767955801,
"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": 0.01171875,
"completions/max_length": 1946.0,
"completions/max_terminated_length": 1946.0,
"completions/mean_length": 561.27734375,
"completions/mean_terminated_length": 567.9328002929688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 167.0,
"epoch": 0.16853333333333334,
"grad_norm": 0.03425997868180275,
"learning_rate": 1.1666666666666668e-06,
"loss": -0.0805,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.030332986265420914,
"mask/share_reasoning": 0.8240467309951782,
"mask/share_step_conf": 0.13390159606933594,
"num_tokens": 39652175.0,
"reward": 1.3016924858093262,
"reward_std": 0.2724427580833435,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.6727285385131836,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8059532046318054,
"step": 158
},
{
"adv/mean_abs_final_conf": 0.4782908856868744,
"adv/mean_abs_reasoning": 0.4805372357368469,
"adv/mean_abs_step_conf": 0.739375650882721,
"adv/ratio_final_to_reasoning": 0.9953253361385658,
"adv/ratio_step_to_reasoning": 1.5386438259024318,
"adv/std_final_conf": 0.7405784726142883,
"adv/std_reasoning": 0.7393211722373962,
"adv/std_step_conf": 0.9332762360572815,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6278726544381193,
"calib/avg_num_step_conf": 7.671875,
"calib/ece": 0.29825203252032517,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7032520325203252,
"calib/gap": 0.18947712418300655,
"calib/mean_conf": 0.7711788617886178,
"calib/mu_c": 0.8428104575163398,
"calib/mu_w": 0.6533333333333332,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.22373983739837394,
"calib/std_conf": 0.3682288863246713,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.578344562078922,
"calib/step_q_c_n": 1039.0,
"calib/step_q_gap": 0.18837699451135453,
"calib/step_q_w": 0.3899675675675675,
"calib/step_q_w_n": 925.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1482.0,
"completions/max_terminated_length": 1482.0,
"completions/mean_length": 534.94921875,
"completions/mean_terminated_length": 552.2056274414062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 208.0,
"epoch": 0.1696,
"grad_norm": 0.029841555282473564,
"learning_rate": 1.138888888888889e-06,
"loss": -0.1624,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.02883203700184822,
"mask/share_reasoning": 0.8057365417480469,
"mask/share_step_conf": 0.134181410074234,
"num_tokens": 39893906.0,
"reward": 1.3323006629943848,
"reward_std": 0.2667738199234009,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.6689214706420898,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.841980516910553,
"step": 159
},
{
"adv/mean_abs_final_conf": 0.5557472705841064,
"adv/mean_abs_reasoning": 0.5128598213195801,
"adv/mean_abs_step_conf": 0.7253711223602295,
"adv/ratio_final_to_reasoning": 1.083624116145768,
"adv/ratio_step_to_reasoning": 1.4143652752790445,
"adv/std_final_conf": 0.7941585779190063,
"adv/std_reasoning": 0.7755072116851807,
"adv/std_step_conf": 0.9360411763191223,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.6706115107913669,
"calib/avg_num_step_conf": 8.74609375,
"calib/ece": 0.26849372384937237,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.6736401673640168,
"calib/gap": 0.2852575539568346,
"calib/mean_conf": 0.738702928870293,
"calib/mu_c": 0.8580575539568346,
"calib/mu_w": 0.5728,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.21280334728033473,
"calib/std_conf": 0.395331003243248,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.591060762100927,
"calib/step_q_c_n": 971.0,
"calib/step_q_gap": 0.26627369585487015,
"calib/step_q_w": 0.3247870662460568,
"calib/step_q_w_n": 1268.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2616.0,
"completions/max_terminated_length": 2616.0,
"completions/mean_length": 600.89453125,
"completions/mean_terminated_length": 630.4467163085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 202.0,
"epoch": 0.17066666666666666,
"grad_norm": 0.02899031713604927,
"learning_rate": 1.111111111111111e-06,
"loss": -0.2876,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.02542033977806568,
"mask/share_reasoning": 0.8041006922721863,
"mask/share_step_conf": 0.1236039474606514,
"num_tokens": 40152575.0,
"reward": 1.2626303434371948,
"reward_std": 0.3472943902015686,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.6632011532783508,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.783764123916626,
"step": 160
},
{
"adv/mean_abs_final_conf": 0.502620279788971,
"adv/mean_abs_reasoning": 0.34476661682128906,
"adv/mean_abs_step_conf": 0.7360956072807312,
"adv/ratio_final_to_reasoning": 1.4578565767854081,
"adv/ratio_step_to_reasoning": 2.1350547627477776,
"adv/std_final_conf": 0.775112509727478,
"adv/std_reasoning": 0.6613057851791382,
"adv/std_step_conf": 0.9351096153259277,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7037220843672455,
"calib/avg_num_step_conf": 7.28515625,
"calib/ece": 0.20486055776892437,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.6812749003984063,
"calib/gap": 0.3267311827956989,
"calib/mean_conf": 0.7501195219123507,
"calib/mu_c": 0.8347311827956989,
"calib/mu_w": 0.508,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.10697211155378496,
"calib/std_conf": 0.3778809463559852,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5683183856502243,
"calib/step_q_c_n": 1338.0,
"calib/step_q_gap": 0.14742654504301367,
"calib/step_q_w": 0.42089184060721063,
"calib/step_q_w_n": 527.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2242.0,
"completions/max_terminated_length": 2242.0,
"completions/mean_length": 548.92578125,
"completions/mean_terminated_length": 557.638916015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 190.0,
"epoch": 0.17173333333333332,
"grad_norm": 0.03485341742634773,
"learning_rate": 1.0833333333333335e-06,
"loss": -0.0284,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03024507313966751,
"mask/share_reasoning": 0.8108487129211426,
"mask/share_step_conf": 0.14328116178512573,
"num_tokens": 40397020.0,
"reward": 1.3916150331497192,
"reward_std": 0.22925958037376404,
"rewards/accuracy_reward_step": 0.7265625,
"rewards/final_brier_reward_step": 0.775180459022522,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8333216309547424,
"step": 161
},
{
"adv/mean_abs_final_conf": 0.5494496822357178,
"adv/mean_abs_reasoning": 0.3901814818382263,
"adv/mean_abs_step_conf": 0.7636798620223999,
"adv/ratio_final_to_reasoning": 1.4081900546564787,
"adv/ratio_step_to_reasoning": 1.9572427128641392,
"adv/std_final_conf": 0.7924970984458923,
"adv/std_reasoning": 0.6614232063293457,
"adv/std_step_conf": 0.9354234933853149,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7297174487772637,
"calib/avg_num_step_conf": 7.3046875,
"calib/ece": 0.1895528455284553,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.7154471544715447,
"calib/gap": 0.35837904824851274,
"calib/mean_conf": 0.7747560975609756,
"calib/mu_c": 0.873820224719101,
"calib/mu_w": 0.5154411764705883,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.12036585365853661,
"calib/std_conf": 0.3684549997666794,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.5647347266881029,
"calib/step_q_c_n": 1244.0,
"calib/step_q_gap": 0.21722673946765558,
"calib/step_q_w": 0.3475079872204473,
"calib/step_q_w_n": 626.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1621.0,
"completions/max_terminated_length": 1621.0,
"completions/mean_length": 526.6171875,
"completions/mean_terminated_length": 537.1076049804688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 209.0,
"epoch": 0.1728,
"grad_norm": 0.031547948718070984,
"learning_rate": 1.0555555555555557e-06,
"loss": -0.0952,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.030170854181051254,
"mask/share_reasoning": 0.8118693828582764,
"mask/share_step_conf": 0.13842850923538208,
"num_tokens": 40635978.0,
"reward": 1.3593121767044067,
"reward_std": 0.28900277614593506,
"rewards/accuracy_reward_step": 0.703125,
"rewards/final_brier_reward_step": 0.759925365447998,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.8145056962966919,
"step": 162
},
{
"adv/mean_abs_final_conf": 0.5519617199897766,
"adv/mean_abs_reasoning": 0.4598156213760376,
"adv/mean_abs_step_conf": 0.7201074361801147,
"adv/ratio_final_to_reasoning": 1.20039792980061,
"adv/ratio_step_to_reasoning": 1.566078668717543,
"adv/std_final_conf": 0.8111170530319214,
"adv/std_reasoning": 0.7574796676635742,
"adv/std_step_conf": 0.9355893731117249,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.7833166499833166,
"calib/avg_num_step_conf": 7.70703125,
"calib/ece": 0.19756097560975622,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.6544715447154471,
"calib/gap": 0.44075275275275283,
"calib/mean_conf": 0.7343089430894308,
"calib/mu_c": 0.9331851851851852,
"calib/mu_w": 0.4924324324324324,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.1915447154471546,
"calib/std_conf": 0.3867271076271738,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6137882352941176,
"calib/step_q_c_n": 850.0,
"calib/step_q_gap": 0.2832005238070295,
"calib/step_q_w": 0.33058771148708815,
"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": 2150.0,
"completions/max_terminated_length": 2150.0,
"completions/mean_length": 607.08984375,
"completions/mean_terminated_length": 619.1832885742188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.17386666666666667,
"grad_norm": 0.030279580503702164,
"learning_rate": 1.0277777777777777e-06,
"loss": -0.1582,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.028492219746112823,
"mask/share_reasoning": 0.8214021921157837,
"mask/share_step_conf": 0.1305743008852005,
"num_tokens": 40896225.0,
"reward": 1.3520934581756592,
"reward_std": 0.3107609152793884,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.7559499740600586,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.825290322303772,
"step": 163
},
{
"adv/mean_abs_final_conf": 0.6321254968643188,
"adv/mean_abs_reasoning": 0.48479893803596497,
"adv/mean_abs_step_conf": 0.7586383819580078,
"adv/ratio_final_to_reasoning": 1.3038920824067985,
"adv/ratio_step_to_reasoning": 1.5648515754416277,
"adv/std_final_conf": 0.8424050807952881,
"adv/std_reasoning": 0.739427924156189,
"adv/std_step_conf": 0.9356386661529541,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.7565453813104188,
"calib/avg_num_step_conf": 8.234375,
"calib/ece": 0.2139183673469388,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6285714285714286,
"calib/gap": 0.4061983082706766,
"calib/mean_conf": 0.7064897959183672,
"calib/mu_c": 0.8921804511278195,
"calib/mu_w": 0.4859821428571429,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.18877551020408165,
"calib/std_conf": 0.3947325688567956,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5480887011615628,
"calib/step_q_c_n": 947.0,
"calib/step_q_gap": 0.21559688376276864,
"calib/step_q_w": 0.3324918173987942,
"calib/step_q_w_n": 1161.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2880.0,
"completions/max_terminated_length": 2880.0,
"completions/mean_length": 679.82421875,
"completions/mean_terminated_length": 687.8853759765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 226.0,
"epoch": 0.17493333333333333,
"grad_norm": 0.026235386729240417,
"learning_rate": 1.0000000000000002e-06,
"loss": -0.0761,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.02434472180902958,
"mask/share_reasoning": 0.8386214375495911,
"mask/share_step_conf": 0.1253151148557663,
"num_tokens": 41176396.0,
"reward": 1.3394906520843506,
"reward_std": 0.29882270097732544,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.7377316355705261,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8229684829711914,
"step": 164
},
{
"adv/mean_abs_final_conf": 0.5562580823898315,
"adv/mean_abs_reasoning": 0.43912598490715027,
"adv/mean_abs_step_conf": 0.7661980390548706,
"adv/ratio_final_to_reasoning": 1.2667391625832571,
"adv/ratio_step_to_reasoning": 1.7448250966448209,
"adv/std_final_conf": 0.7766196727752686,
"adv/std_reasoning": 0.7014268040657043,
"adv/std_step_conf": 0.9359138011932373,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7032210553278688,
"calib/avg_num_step_conf": 7.41796875,
"calib/ece": 0.30076,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.704,
"calib/gap": 0.3004316086065574,
"calib/mean_conf": 0.76628,
"calib/mu_c": 0.912890625,
"calib/mu_w": 0.6124590163934426,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.27752,
"calib/std_conf": 0.37215502361247255,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5824116161616162,
"calib/step_q_c_n": 792.0,
"calib/step_q_gap": 0.18214061345158905,
"calib/step_q_w": 0.40027100271002714,
"calib/step_q_w_n": 1107.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2981.0,
"completions/max_terminated_length": 2981.0,
"completions/mean_length": 643.3359375,
"completions/mean_terminated_length": 648.4015502929688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 166.0,
"epoch": 0.176,
"grad_norm": 0.03587152808904648,
"learning_rate": 9.722222222222224e-07,
"loss": -0.0357,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.026652254164218903,
"mask/share_reasoning": 0.8428949117660522,
"mask/share_step_conf": 0.12264031916856766,
"num_tokens": 41446666.0,
"reward": 1.278690218925476,
"reward_std": 0.31332042813301086,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.6767804622650146,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7934250235557556,
"step": 165
},
{
"adv/mean_abs_final_conf": 0.49421077966690063,
"adv/mean_abs_reasoning": 0.3963538110256195,
"adv/mean_abs_step_conf": 0.7633610963821411,
"adv/ratio_final_to_reasoning": 1.2468929676443956,
"adv/ratio_step_to_reasoning": 1.9259587649904015,
"adv/std_final_conf": 0.7540462017059326,
"adv/std_reasoning": 0.6614139676094055,
"adv/std_step_conf": 0.9339258670806885,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.8318401937046005,
"calib/avg_num_step_conf": 7.5546875,
"calib/ece": 0.12331983805668015,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.6558704453441295,
"calib/gap": 0.5746246973365617,
"calib/mean_conf": 0.7174898785425101,
"calib/mu_c": 0.8803389830508475,
"calib/mu_w": 0.3057142857142857,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98046875,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.06210526315789472,
"calib/std_conf": 0.40888534035530694,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.526060606060606,
"calib/step_q_c_n": 1287.0,
"calib/step_q_gap": 0.24952892136199706,
"calib/step_q_w": 0.27653168469860895,
"calib/step_q_w_n": 647.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2197.0,
"completions/max_terminated_length": 2197.0,
"completions/mean_length": 613.37109375,
"completions/mean_terminated_length": 620.644287109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 233.0,
"epoch": 0.17706666666666668,
"grad_norm": 0.033427972346544266,
"learning_rate": 9.444444444444445e-07,
"loss": -0.0804,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.026531532406806946,
"mask/share_reasoning": 0.8271982669830322,
"mask/share_step_conf": 0.13455137610435486,
"num_tokens": 41709873.0,
"reward": 1.4548362493515015,
"reward_std": 0.2148941308259964,
"rewards/accuracy_reward_step": 0.69140625,
"rewards/final_brier_reward_step": 0.8294359445571899,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8748839497566223,
"step": 166
},
{
"adv/mean_abs_final_conf": 0.49704134464263916,
"adv/mean_abs_reasoning": 0.4293529689311981,
"adv/mean_abs_step_conf": 0.7540600895881653,
"adv/ratio_final_to_reasoning": 1.1576520499670466,
"adv/ratio_step_to_reasoning": 1.7562708171443902,
"adv/std_final_conf": 0.7590051293373108,
"adv/std_reasoning": 0.7013872265815735,
"adv/std_step_conf": 0.9354445934295654,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6131402482522471,
"calib/avg_num_step_conf": 7.32421875,
"calib/ece": 0.2698393574297187,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.7831325301204819,
"calib/gap": 0.18734626908260799,
"calib/mean_conf": 0.8412449799196787,
"calib/mu_c": 0.9059509202453987,
"calib/mu_w": 0.7186046511627907,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.22823293172690748,
"calib/std_conf": 0.31642908476884707,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5313021702838063,
"calib/step_q_c_n": 1198.0,
"calib/step_q_gap": 0.03786051592634693,
"calib/step_q_w": 0.4934416543574594,
"calib/step_q_w_n": 677.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 3066.0,
"completions/max_terminated_length": 3066.0,
"completions/mean_length": 578.46484375,
"completions/mean_terminated_length": 585.3241577148438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.17813333333333334,
"grad_norm": 0.041853420436382294,
"learning_rate": 9.166666666666666e-07,
"loss": -0.0108,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.028374582529067993,
"mask/share_reasoning": 0.8254715800285339,
"mask/share_step_conf": 0.13443510234355927,
"num_tokens": 41963568.0,
"reward": 1.3114798069000244,
"reward_std": 0.2623671293258667,
"rewards/accuracy_reward_step": 0.63671875,
"rewards/final_brier_reward_step": 0.7038777470588684,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7986034154891968,
"step": 167
},
{
"adv/mean_abs_final_conf": 0.5935930609703064,
"adv/mean_abs_reasoning": 0.4941191077232361,
"adv/mean_abs_step_conf": 0.7486543655395508,
"adv/ratio_final_to_reasoning": 1.2013157388416302,
"adv/ratio_step_to_reasoning": 1.5151293561367067,
"adv/std_final_conf": 0.8258097767829895,
"adv/std_reasoning": 0.7576125264167786,
"adv/std_step_conf": 0.9353750348091125,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7621713895527483,
"calib/avg_num_step_conf": 7.65625,
"calib/ece": 0.20207317073170739,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.6991869918699187,
"calib/gap": 0.36288221235916696,
"calib/mean_conf": 0.7828048780487804,
"calib/mu_c": 0.9317931034482759,
"calib/mu_w": 0.568910891089109,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.19772357723577244,
"calib/std_conf": 0.35477328518173923,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5597534833869239,
"calib/step_q_c_n": 933.0,
"calib/step_q_gap": 0.19168142885917316,
"calib/step_q_w": 0.3680720545277507,
"calib/step_q_w_n": 1027.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2855.0,
"completions/max_terminated_length": 2855.0,
"completions/mean_length": 629.69140625,
"completions/mean_terminated_length": 642.235107421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 184.0,
"epoch": 0.1792,
"grad_norm": 0.02765379659831524,
"learning_rate": 8.88888888888889e-07,
"loss": -0.0696,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.02691633626818657,
"mask/share_reasoning": 0.8299218416213989,
"mask/share_step_conf": 0.1236305758357048,
"num_tokens": 42229441.0,
"reward": 1.343550682067871,
"reward_std": 0.3048199713230133,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.7402839660644531,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8206743001937866,
"step": 168
},
{
"adv/mean_abs_final_conf": 0.5480157136917114,
"adv/mean_abs_reasoning": 0.3762734830379486,
"adv/mean_abs_step_conf": 0.7599032521247864,
"adv/ratio_final_to_reasoning": 1.4564292685924987,
"adv/ratio_step_to_reasoning": 2.0195503706227083,
"adv/std_final_conf": 0.7893987894058228,
"adv/std_reasoning": 0.6403399109840393,
"adv/std_step_conf": 0.9350810647010803,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.696174018289403,
"calib/avg_num_step_conf": 6.55078125,
"calib/ece": 0.28497975708502027,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.7368421052631579,
"calib/gap": 0.2515384615384615,
"calib/mean_conf": 0.821012145748988,
"calib/mu_c": 0.9269230769230768,
"calib/mu_w": 0.6753846153846154,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.26352226720647776,
"calib/std_conf": 0.31959518272987264,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5789989118607182,
"calib/step_q_c_n": 919.0,
"calib/step_q_gap": 0.1252390174016152,
"calib/step_q_w": 0.453759894459103,
"calib/step_q_w_n": 758.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2842.0,
"completions/max_terminated_length": 2842.0,
"completions/mean_length": 606.80078125,
"completions/mean_terminated_length": 613.99609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 178.0,
"epoch": 0.18026666666666666,
"grad_norm": 0.04610948637127876,
"learning_rate": 8.611111111111112e-07,
"loss": -0.0505,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.028334004804491997,
"mask/share_reasoning": 0.8365331888198853,
"mask/share_step_conf": 0.1234140545129776,
"num_tokens": 42488966.0,
"reward": 1.3015222549438477,
"reward_std": 0.2599875032901764,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.6834961175918579,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8086023330688477,
"step": 169
},
{
"adv/mean_abs_final_conf": 0.5640392303466797,
"adv/mean_abs_reasoning": 0.4219498932361603,
"adv/mean_abs_step_conf": 0.7920935153961182,
"adv/ratio_final_to_reasoning": 1.3367445741501673,
"adv/ratio_step_to_reasoning": 1.8772217462152383,
"adv/std_final_conf": 0.7761444449424744,
"adv/std_reasoning": 0.6817612648010254,
"adv/std_step_conf": 0.9347199201583862,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6794758938244854,
"calib/avg_num_step_conf": 6.97265625,
"calib/ece": 0.28219512195121943,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.7723577235772358,
"calib/gap": 0.26059452871072586,
"calib/mean_conf": 0.8260975609756097,
"calib/mu_c": 0.9362676056338027,
"calib/mu_w": 0.6756730769230769,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.96484375,
"calib/nonempty_step_conf_rate": 0.95703125,
"calib/pce": 0.26552845528455277,
"calib/std_conf": 0.32923077172549003,
"calib/step_conf_rate": 0.95703125,
"calib/step_q_c": 0.5710483234714004,
"calib/step_q_c_n": 1014.0,
"calib/step_q_gap": 0.09902497716789849,
"calib/step_q_w": 0.47202334630350196,
"calib/step_q_w_n": 771.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1783.0,
"completions/max_terminated_length": 1783.0,
"completions/mean_length": 602.64453125,
"completions/mean_terminated_length": 605.0078735351562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 204.0,
"epoch": 0.18133333333333335,
"grad_norm": 0.032945066690444946,
"learning_rate": 8.333333333333333e-07,
"loss": -0.0215,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.02677723951637745,
"mask/share_reasoning": 0.8398348689079285,
"mask/share_step_conf": 0.12948164343833923,
"num_tokens": 42747395.0,
"reward": 1.2866040468215942,
"reward_std": 0.32099831104278564,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.6776843667030334,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7969805002212524,
"step": 170
},
{
"adv/mean_abs_final_conf": 0.6232767105102539,
"adv/mean_abs_reasoning": 0.48395657539367676,
"adv/mean_abs_step_conf": 0.7597352266311646,
"adv/ratio_final_to_reasoning": 1.2878773472666356,
"adv/ratio_step_to_reasoning": 1.5698417280789176,
"adv/std_final_conf": 0.8432072401046753,
"adv/std_reasoning": 0.7394060492515564,
"adv/std_step_conf": 0.9357650876045227,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6812934540207267,
"calib/avg_num_step_conf": 7.4765625,
"calib/ece": 0.305748987854251,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6720647773279352,
"calib/gap": 0.24974419519874058,
"calib/mean_conf": 0.7538461538461538,
"calib/mu_c": 0.8761904761904762,
"calib/mu_w": 0.6264462809917356,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.2747368421052631,
"calib/std_conf": 0.37176048051651633,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5969871043376319,
"calib/step_q_c_n": 853.0,
"calib/step_q_gap": 0.21684572827731147,
"calib/step_q_w": 0.38014137606032045,
"calib/step_q_w_n": 1061.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3026.0,
"completions/max_terminated_length": 3026.0,
"completions/mean_length": 604.0,
"completions/mean_terminated_length": 611.162109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 177.0,
"epoch": 0.1824,
"grad_norm": 0.045893196016550064,
"learning_rate": 8.055555555555557e-07,
"loss": -0.0116,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.028350790962576866,
"mask/share_reasoning": 0.8282653093338013,
"mask/share_step_conf": 0.13166514039039612,
"num_tokens": 43008915.0,
"reward": 1.2718992233276367,
"reward_std": 0.3006881773471832,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/final_brier_reward_step": 0.6478304862976074,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8030619621276855,
"step": 171
},
{
"adv/mean_abs_final_conf": 0.4694840908050537,
"adv/mean_abs_reasoning": 0.4077759385108948,
"adv/mean_abs_step_conf": 0.7584755420684814,
"adv/ratio_final_to_reasoning": 1.1513285764714394,
"adv/ratio_step_to_reasoning": 1.8600301548891336,
"adv/std_final_conf": 0.7224934101104736,
"adv/std_reasoning": 0.6815813779830933,
"adv/std_step_conf": 0.9330750107765198,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7557434813248767,
"calib/avg_num_step_conf": 6.8984375,
"calib/ece": 0.20354581673306776,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.7768924302788844,
"calib/gap": 0.3175362931642002,
"calib/mean_conf": 0.857808764940239,
"calib/mu_c": 0.9666060606060607,
"calib/mu_w": 0.6490697674418605,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.20199203187250997,
"calib/std_conf": 0.2910955732261504,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5770408888888889,
"calib/step_q_c_n": 1125.0,
"calib/step_q_gap": 0.15041062367828045,
"calib/step_q_w": 0.4266302652106085,
"calib/step_q_w_n": 641.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3006.0,
"completions/max_terminated_length": 3006.0,
"completions/mean_length": 586.81640625,
"completions/mean_terminated_length": 586.81640625,
"completions/min_length": 246.0,
"completions/min_terminated_length": 246.0,
"epoch": 0.18346666666666667,
"grad_norm": 0.039151061326265335,
"learning_rate": 7.777777777777779e-07,
"loss": 0.0318,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.02954472042620182,
"mask/share_reasoning": 0.8346071243286133,
"mask/share_step_conf": 0.13584816455841064,
"num_tokens": 43262492.0,
"reward": 1.4082257747650146,
"reward_std": 0.24087859690189362,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.7774074077606201,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8570220470428467,
"step": 172
},
{
"adv/mean_abs_final_conf": 0.5621016025543213,
"adv/mean_abs_reasoning": 0.464821457862854,
"adv/mean_abs_step_conf": 0.7540398240089417,
"adv/ratio_final_to_reasoning": 1.2092849696284242,
"adv/ratio_step_to_reasoning": 1.6222138871898246,
"adv/std_final_conf": 0.8106486797332764,
"adv/std_reasoning": 0.7574560046195984,
"adv/std_step_conf": 0.9359198808670044,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6301006889242183,
"calib/avg_num_step_conf": 6.94140625,
"calib/ece": 0.31708,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.872,
"calib/gap": 0.16937334393216774,
"calib/mean_conf": 0.9077200000000001,
"calib/mu_c": 0.9768243243243244,
"calib/mu_w": 0.8074509803921567,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.3164,
"calib/std_conf": 0.2519984158680368,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5796259124087592,
"calib/step_q_c_n": 1096.0,
"calib/step_q_gap": 0.038377747944735585,
"calib/step_q_w": 0.5412481644640236,
"calib/step_q_w_n": 681.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 1650.0,
"completions/max_terminated_length": 1650.0,
"completions/mean_length": 565.44140625,
"completions/mean_terminated_length": 572.146240234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 138.0,
"epoch": 0.18453333333333333,
"grad_norm": 0.029890544712543488,
"learning_rate": 7.5e-07,
"loss": -0.1264,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.030869608744978905,
"mask/share_reasoning": 0.8199148774147034,
"mask/share_step_conf": 0.13749675452709198,
"num_tokens": 43510405.0,
"reward": 1.2570515871047974,
"reward_std": 0.3210733234882355,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.6597632169723511,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7720918655395508,
"step": 173
},
{
"adv/mean_abs_final_conf": 0.6808509826660156,
"adv/mean_abs_reasoning": 0.6322292685508728,
"adv/mean_abs_step_conf": 0.7378787994384766,
"adv/ratio_final_to_reasoning": 1.0769051933115785,
"adv/ratio_step_to_reasoning": 1.1671063586312005,
"adv/std_final_conf": 0.8761774301528931,
"adv/std_reasoning": 0.8268210291862488,
"adv/std_step_conf": 0.9360378980636597,
"calib/answer_extract_rate": 0.91015625,
"calib/auroc": 0.5756345926800472,
"calib/avg_num_step_conf": 7.52734375,
"calib/ece": 0.3387982832618026,
"calib/final_conf_rate": 0.91015625,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.6781115879828327,
"calib/gap": 0.12098878394332935,
"calib/mean_conf": 0.7917596566523606,
"calib/mu_c": 0.8499173553719008,
"calib/mu_w": 0.7289285714285715,
"calib/nonempty_final_conf_rate": 0.91015625,
"calib/nonempty_reasoning_rate": 0.97265625,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.30562231759656655,
"calib/std_conf": 0.3279849948204185,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.5179640718562875,
"calib/step_q_c_n": 835.0,
"calib/step_q_gap": 0.10083952973174531,
"calib/step_q_w": 0.41712454212454214,
"calib/step_q_w_n": 1092.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 3071.0,
"completions/max_terminated_length": 3071.0,
"completions/mean_length": 609.5390625,
"completions/mean_terminated_length": 639.516357421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 211.0,
"epoch": 0.1856,
"grad_norm": 0.0357220433652401,
"learning_rate": 7.222222222222222e-07,
"loss": -0.2293,
"mask/has_final_conf_rate": 0.91015625,
"mask/share_final_conf": 0.02538606896996498,
"mask/share_reasoning": 0.8105248808860779,
"mask/share_step_conf": 0.11721404641866684,
"num_tokens": 43770679.0,
"reward": 1.1440212726593018,
"reward_std": 0.38225239515304565,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/final_brier_reward_step": 0.5724663734436035,
"rewards/format_reward_step": 0.91015625,
"rewards/step_l2_reward": 0.7195067405700684,
"step": 174
},
{
"adv/mean_abs_final_conf": 0.6520330905914307,
"adv/mean_abs_reasoning": 0.5189327001571655,
"adv/mean_abs_step_conf": 0.7571752071380615,
"adv/ratio_final_to_reasoning": 1.2564887323422747,
"adv/ratio_step_to_reasoning": 1.459100971876972,
"adv/std_final_conf": 0.8270578384399414,
"adv/std_reasoning": 0.7576694488525391,
"adv/std_step_conf": 0.9357485175132751,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.75656194750442,
"calib/avg_num_step_conf": 7.09765625,
"calib/ece": 0.23403292181069957,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.5473251028806584,
"calib/gap": 0.3640922072623417,
"calib/mean_conf": 0.663909465020576,
"calib/mu_c": 0.8571929824561402,
"calib/mu_w": 0.49310077519379847,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.21440329218106996,
"calib/std_conf": 0.39867365496285584,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5756865284974093,
"calib/step_q_c_n": 772.0,
"calib/step_q_gap": 0.20009801175099784,
"calib/step_q_w": 0.37558851674641147,
"calib/step_q_w_n": 1045.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2466.0,
"completions/max_terminated_length": 2466.0,
"completions/mean_length": 621.71875,
"completions/mean_terminated_length": 636.6400146484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 190.0,
"epoch": 0.18666666666666668,
"grad_norm": 0.03226575627923012,
"learning_rate": 6.944444444444446e-07,
"loss": -0.1542,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.02714480832219124,
"mask/share_reasoning": 0.8222717046737671,
"mask/share_step_conf": 0.12714600563049316,
"num_tokens": 44035663.0,
"reward": 1.2900152206420898,
"reward_std": 0.3247193992137909,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/final_brier_reward_step": 0.6946007609367371,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.803652286529541,
"step": 175
},
{
"adv/mean_abs_final_conf": 0.5556491017341614,
"adv/mean_abs_reasoning": 0.460569828748703,
"adv/mean_abs_step_conf": 0.7447050213813782,
"adv/ratio_final_to_reasoning": 1.2064383445258975,
"adv/ratio_step_to_reasoning": 1.6169209854771134,
"adv/std_final_conf": 0.7920377850532532,
"adv/std_reasoning": 0.7206613421440125,
"adv/std_step_conf": 0.9342063069343567,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7433612858141161,
"calib/avg_num_step_conf": 6.44140625,
"calib/ece": 0.18899598393574296,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.6184738955823293,
"calib/gap": 0.3336415094339622,
"calib/mean_conf": 0.7320481927710843,
"calib/mu_c": 0.8526415094339622,
"calib/mu_w": 0.519,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.14124497991967871,
"calib/std_conf": 0.3658782437419884,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5316715257531585,
"calib/step_q_c_n": 1029.0,
"calib/step_q_gap": 0.15713926768864228,
"calib/step_q_w": 0.3745322580645162,
"calib/step_q_w_n": 620.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2412.0,
"completions/max_terminated_length": 2412.0,
"completions/mean_length": 562.90234375,
"completions/mean_terminated_length": 567.3346557617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 177.0,
"epoch": 0.18773333333333334,
"grad_norm": 0.03709743171930313,
"learning_rate": 6.666666666666667e-07,
"loss": -0.0819,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03193432092666626,
"mask/share_reasoning": 0.8266017436981201,
"mask/share_step_conf": 0.13365145027637482,
"num_tokens": 44283830.0,
"reward": 1.4114222526550293,
"reward_std": 0.22874104976654053,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.7592554688453674,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8724194765090942,
"step": 176
},
{
"adv/mean_abs_final_conf": 0.5837680101394653,
"adv/mean_abs_reasoning": 0.45486849546432495,
"adv/mean_abs_step_conf": 0.7547392845153809,
"adv/ratio_final_to_reasoning": 1.283377538696236,
"adv/ratio_step_to_reasoning": 1.6592472154945594,
"adv/std_final_conf": 0.7963255047798157,
"adv/std_reasoning": 0.720579981803894,
"adv/std_step_conf": 0.9347895979881287,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7630876068376069,
"calib/avg_num_step_conf": 7.2578125,
"calib/ece": 0.19181451612903228,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.6653225806451613,
"calib/gap": 0.37500534188034207,
"calib/mean_conf": 0.7494758064516129,
"calib/mu_c": 0.9067361111111113,
"calib/mu_w": 0.5317307692307692,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.18032258064516132,
"calib/std_conf": 0.3656701308227984,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5499669239250276,
"calib/step_q_c_n": 907.0,
"calib/step_q_gap": 0.1828691321269203,
"calib/step_q_w": 0.36709779179810725,
"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": 0.00390625,
"completions/max_length": 2871.0,
"completions/max_terminated_length": 2871.0,
"completions/mean_length": 602.71484375,
"completions/mean_terminated_length": 605.0784912109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 197.0,
"epoch": 0.1888,
"grad_norm": 0.04204968363046646,
"learning_rate": 6.388888888888889e-07,
"loss": -0.0151,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.0288112610578537,
"mask/share_reasoning": 0.8356721997261047,
"mask/share_step_conf": 0.13161027431488037,
"num_tokens": 44541957.0,
"reward": 1.3634679317474365,
"reward_std": 0.2572265863418579,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.752631664276123,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8340271711349487,
"step": 177
},
{
"adv/mean_abs_final_conf": 0.5842238664627075,
"adv/mean_abs_reasoning": 0.5355183482170105,
"adv/mean_abs_step_conf": 0.7687432169914246,
"adv/ratio_final_to_reasoning": 1.0909502324390197,
"adv/ratio_step_to_reasoning": 1.4355123770285894,
"adv/std_final_conf": 0.8104052543640137,
"adv/std_reasoning": 0.7754615545272827,
"adv/std_step_conf": 0.9334971904754639,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8196256381168463,
"calib/avg_num_step_conf": 6.71484375,
"calib/ece": 0.11960000000000004,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.632,
"calib/gap": 0.4809600113442996,
"calib/mean_conf": 0.7414400000000001,
"calib/mu_c": 0.9068902439024391,
"calib/mu_w": 0.4259302325581395,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.10252000000000006,
"calib/std_conf": 0.36475022467436535,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5849324324324324,
"calib/step_q_c_n": 1036.0,
"calib/step_q_gap": 0.2448592259902655,
"calib/step_q_w": 0.34007320644216693,
"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": 0.00390625,
"completions/max_length": 3070.0,
"completions/max_terminated_length": 3070.0,
"completions/mean_length": 577.5703125,
"completions/mean_terminated_length": 579.8353271484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 126.0,
"epoch": 0.18986666666666666,
"grad_norm": 0.03877848759293556,
"learning_rate": 6.111111111111112e-07,
"loss": 0.0605,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.030248302966356277,
"mask/share_reasoning": 0.8328169584274292,
"mask/share_step_conf": 0.13302844762802124,
"num_tokens": 44795887.0,
"reward": 1.4552483558654785,
"reward_std": 0.23271851241588593,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.8311171531677246,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8783615827560425,
"step": 178
},
{
"adv/mean_abs_final_conf": 0.5848767757415771,
"adv/mean_abs_reasoning": 0.4726482927799225,
"adv/mean_abs_step_conf": 0.7531768679618835,
"adv/ratio_final_to_reasoning": 1.2374460770852953,
"adv/ratio_step_to_reasoning": 1.5935249940965779,
"adv/std_final_conf": 0.828561544418335,
"adv/std_reasoning": 0.7392279505729675,
"adv/std_step_conf": 0.9353491067886353,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7433426144070822,
"calib/avg_num_step_conf": 6.70703125,
"calib/ece": 0.2012096774193548,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.6895161290322581,
"calib/gap": 0.3211594202898551,
"calib/mean_conf": 0.7751612903225806,
"calib/mu_c": 0.8878260869565218,
"calib/mu_w": 0.5666666666666667,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.16358870967741934,
"calib/std_conf": 0.3512752892307421,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5760626185958255,
"calib/step_q_c_n": 1054.0,
"calib/step_q_gap": 0.19030092930472442,
"calib/step_q_w": 0.3857616892911011,
"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": 0.0078125,
"completions/max_length": 2766.0,
"completions/max_terminated_length": 2766.0,
"completions/mean_length": 576.4140625,
"completions/mean_terminated_length": 580.9527587890625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 193.0,
"epoch": 0.19093333333333334,
"grad_norm": 0.03158321604132652,
"learning_rate": 5.833333333333334e-07,
"loss": -0.0242,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.029713882133364677,
"mask/share_reasoning": 0.8328595161437988,
"mask/share_step_conf": 0.12961412966251373,
"num_tokens": 45049713.0,
"reward": 1.3898929357528687,
"reward_std": 0.2539806663990021,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/final_brier_reward_step": 0.7549265623092651,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8522734045982361,
"step": 179
},
{
"adv/mean_abs_final_conf": 0.5628108978271484,
"adv/mean_abs_reasoning": 0.46406090259552,
"adv/mean_abs_step_conf": 0.7706776857376099,
"adv/ratio_final_to_reasoning": 1.2127953350073533,
"adv/ratio_step_to_reasoning": 1.6607253087410814,
"adv/std_final_conf": 0.7955601811408997,
"adv/std_reasoning": 0.720676839351654,
"adv/std_step_conf": 0.9343445897102356,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.7493001119820828,
"calib/avg_num_step_conf": 7.2734375,
"calib/ece": 0.17995528455284543,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6544715447154471,
"calib/gap": 0.3767978723404255,
"calib/mean_conf": 0.7510203252032519,
"calib/mu_c": 0.8949999999999999,
"calib/mu_w": 0.5182021276595744,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.15654471544715437,
"calib/std_conf": 0.3682110393586609,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.5186773905272565,
"calib/step_q_c_n": 1119.0,
"calib/step_q_gap": 0.15622786159051355,
"calib/step_q_w": 0.3624495289367429,
"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": 0.01953125,
"completions/max_length": 2983.0,
"completions/max_terminated_length": 2983.0,
"completions/mean_length": 630.296875,
"completions/mean_terminated_length": 642.8526000976562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 243.0,
"epoch": 0.192,
"grad_norm": 0.043362926691770554,
"learning_rate": 5.555555555555555e-07,
"loss": -0.189,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.025322385132312775,
"mask/share_reasoning": 0.8317726254463196,
"mask/share_step_conf": 0.12337373197078705,
"num_tokens": 45314925.0,
"reward": 1.3708226680755615,
"reward_std": 0.25812745094299316,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.7541399002075195,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8386746048927307,
"step": 180
},
{
"adv/mean_abs_final_conf": 0.5692725777626038,
"adv/mean_abs_reasoning": 0.49015751481056213,
"adv/mean_abs_step_conf": 0.7483924627304077,
"adv/ratio_final_to_reasoning": 1.1614074263099248,
"adv/ratio_step_to_reasoning": 1.5268407402050117,
"adv/std_final_conf": 0.8138942122459412,
"adv/std_reasoning": 0.7575979232788086,
"adv/std_step_conf": 0.9353100061416626,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7661928868120457,
"calib/avg_num_step_conf": 6.49609375,
"calib/ece": 0.21764940239043823,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.6533864541832669,
"calib/gap": 0.33786928868120447,
"calib/mean_conf": 0.7564541832669323,
"calib/mu_c": 0.900486111111111,
"calib/mu_w": 0.5626168224299065,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.20019920318725096,
"calib/std_conf": 0.35548824078313745,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5786396103896104,
"calib/step_q_c_n": 924.0,
"calib/step_q_gap": 0.1654731692529392,
"calib/step_q_w": 0.4131664411366712,
"calib/step_q_w_n": 739.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1782.0,
"completions/max_terminated_length": 1782.0,
"completions/mean_length": 518.62109375,
"completions/mean_terminated_length": 524.770751953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 206.0,
"epoch": 0.19306666666666666,
"grad_norm": 0.05061223730444908,
"learning_rate": 5.277777777777779e-07,
"loss": -0.1146,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.030669106170535088,
"mask/share_reasoning": 0.8251529335975647,
"mask/share_step_conf": 0.13245923817157745,
"num_tokens": 45553956.0,
"reward": 1.358069896697998,
"reward_std": 0.26986604928970337,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.7428753972053528,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8335071206092834,
"step": 181
},
{
"adv/mean_abs_final_conf": 0.5297540426254272,
"adv/mean_abs_reasoning": 0.39702683687210083,
"adv/mean_abs_step_conf": 0.7356059551239014,
"adv/ratio_final_to_reasoning": 1.3343028567010533,
"adv/ratio_step_to_reasoning": 1.8527864789172204,
"adv/std_final_conf": 0.7600104808807373,
"adv/std_reasoning": 0.681530237197876,
"adv/std_step_conf": 0.9350858330726624,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.800428746427113,
"calib/avg_num_step_conf": 7.05078125,
"calib/ece": 0.20266932270916346,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.701195219123506,
"calib/gap": 0.35521709541309365,
"calib/mean_conf": 0.7966135458167332,
"calib/mu_c": 0.9282278481012658,
"calib/mu_w": 0.5730107526881721,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.18490039840637462,
"calib/std_conf": 0.3345605520735549,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5788622754491017,
"calib/step_q_c_n": 1002.0,
"calib/step_q_gap": 0.21201296038060852,
"calib/step_q_w": 0.3668493150684932,
"calib/step_q_w_n": 803.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2697.0,
"completions/max_terminated_length": 2697.0,
"completions/mean_length": 556.67578125,
"completions/mean_terminated_length": 561.05908203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 214.0,
"epoch": 0.19413333333333332,
"grad_norm": 0.04269418865442276,
"learning_rate": 5.000000000000001e-07,
"loss": -0.0214,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.028990257531404495,
"mask/share_reasoning": 0.8292310237884521,
"mask/share_step_conf": 0.13396620750427246,
"num_tokens": 45802625.0,
"reward": 1.417762041091919,
"reward_std": 0.26264894008636475,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/final_brier_reward_step": 0.7771191596984863,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8694368600845337,
"step": 182
},
{
"adv/mean_abs_final_conf": 0.6165659427642822,
"adv/mean_abs_reasoning": 0.5332478284835815,
"adv/mean_abs_step_conf": 0.7585545182228088,
"adv/ratio_final_to_reasoning": 1.156246513966378,
"adv/ratio_step_to_reasoning": 1.42251778198505,
"adv/std_final_conf": 0.843794047832489,
"adv/std_reasoning": 0.7929291129112244,
"adv/std_step_conf": 0.9355603456497192,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.682703737947431,
"calib/avg_num_step_conf": 6.8203125,
"calib/ece": 0.2678137651821862,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.6072874493927125,
"calib/gap": 0.27145951657641,
"calib/mean_conf": 0.7070040485829959,
"calib/mu_c": 0.8311940298507462,
"calib/mu_w": 0.5597345132743362,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2161538461538461,
"calib/std_conf": 0.38714559604135523,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5481409477521264,
"calib/step_q_c_n": 823.0,
"calib/step_q_gap": 0.16593076357011127,
"calib/step_q_w": 0.38221018418201513,
"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": 0.01953125,
"completions/max_length": 2836.0,
"completions/max_terminated_length": 2836.0,
"completions/mean_length": 595.52734375,
"completions/mean_terminated_length": 607.3904418945312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 218.0,
"epoch": 0.1952,
"grad_norm": 0.03282076120376587,
"learning_rate": 4.7222222222222226e-07,
"loss": -0.1704,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.0280508603900671,
"mask/share_reasoning": 0.8383486270904541,
"mask/share_step_conf": 0.11406920850276947,
"num_tokens": 46061760.0,
"reward": 1.2791484594345093,
"reward_std": 0.3096787929534912,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/final_brier_reward_step": 0.6807718276977539,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.7903251051902771,
"step": 183
},
{
"adv/mean_abs_final_conf": 0.6084942817687988,
"adv/mean_abs_reasoning": 0.43991637229919434,
"adv/mean_abs_step_conf": 0.7658754587173462,
"adv/ratio_final_to_reasoning": 1.3832044454007089,
"adv/ratio_step_to_reasoning": 1.740956933961216,
"adv/std_final_conf": 0.814765214920044,
"adv/std_reasoning": 0.7014064788818359,
"adv/std_step_conf": 0.93467777967453,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6863379443700729,
"calib/avg_num_step_conf": 6.9765625,
"calib/ece": 0.21659919028340083,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7894736842105263,
"calib/gap": 0.2433556448014279,
"calib/mean_conf": 0.8442914979757083,
"calib/mu_c": 0.9240963855421687,
"calib/mu_w": 0.6807407407407408,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.19441295546558707,
"calib/std_conf": 0.30395137990264104,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.541272051996286,
"calib/step_q_c_n": 1077.0,
"calib/step_q_gap": 0.14540463309642704,
"calib/step_q_w": 0.395867418899859,
"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": 0.015625,
"completions/max_length": 2915.0,
"completions/max_terminated_length": 2915.0,
"completions/mean_length": 576.0234375,
"completions/mean_terminated_length": 585.1666870117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 214.0,
"epoch": 0.19626666666666667,
"grad_norm": 0.031147437170147896,
"learning_rate": 4.444444444444445e-07,
"loss": -0.0018,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.02811710722744465,
"mask/share_reasoning": 0.8329929113388062,
"mask/share_step_conf": 0.12326496839523315,
"num_tokens": 46314502.0,
"reward": 1.3443111181259155,
"reward_std": 0.27114933729171753,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.734431266784668,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8157672882080078,
"step": 184
},
{
"adv/mean_abs_final_conf": 0.5487991571426392,
"adv/mean_abs_reasoning": 0.48162105679512024,
"adv/mean_abs_step_conf": 0.740757405757904,
"adv/ratio_final_to_reasoning": 1.139483312450137,
"adv/ratio_step_to_reasoning": 1.538050289344013,
"adv/std_final_conf": 0.7968400120735168,
"adv/std_reasoning": 0.7392687797546387,
"adv/std_step_conf": 0.9352030158042908,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.770456240254898,
"calib/avg_num_step_conf": 6.73046875,
"calib/ece": 0.1746774193548388,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.657258064516129,
"calib/gap": 0.40739610873839066,
"calib/mean_conf": 0.7407258064516129,
"calib/mu_c": 0.9033557046979865,
"calib/mu_w": 0.4959595959595959,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.1572983870967743,
"calib/std_conf": 0.37258074116742235,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5475188781014024,
"calib/step_q_c_n": 927.0,
"calib/step_q_gap": 0.175898275086327,
"calib/step_q_w": 0.37162060301507543,
"calib/step_q_w_n": 796.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2862.0,
"completions/max_terminated_length": 2862.0,
"completions/mean_length": 566.06640625,
"completions/mean_terminated_length": 577.3426513671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 196.0,
"epoch": 0.19733333333333333,
"grad_norm": 0.03484448790550232,
"learning_rate": 4.1666666666666667e-07,
"loss": -0.0444,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.030003804713487625,
"mask/share_reasoning": 0.8277981877326965,
"mask/share_step_conf": 0.12266676127910614,
"num_tokens": 46566335.0,
"reward": 1.3510828018188477,
"reward_std": 0.2744816243648529,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.766825795173645,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8133730888366699,
"step": 185
},
{
"adv/mean_abs_final_conf": 0.5438560843467712,
"adv/mean_abs_reasoning": 0.4110875725746155,
"adv/mean_abs_step_conf": 0.758434534072876,
"adv/ratio_final_to_reasoning": 1.3229689259167698,
"adv/ratio_step_to_reasoning": 1.8449463926210379,
"adv/std_final_conf": 0.760007381439209,
"adv/std_reasoning": 0.6817102432250977,
"adv/std_step_conf": 0.9331895709037781,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7306553493880774,
"calib/avg_num_step_conf": 6.9921875,
"calib/ece": 0.2259760956175299,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.6613545816733067,
"calib/gap": 0.31593499144624304,
"calib/mean_conf": 0.753625498007968,
"calib/mu_c": 0.882013422818792,
"calib/mu_w": 0.566078431372549,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.19298804780876494,
"calib/std_conf": 0.3672379993879625,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5677588168373151,
"calib/step_q_c_n": 879.0,
"calib/step_q_gap": 0.18072259290756754,
"calib/step_q_w": 0.38703622392974757,
"calib/step_q_w_n": 911.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2909.0,
"completions/max_terminated_length": 2909.0,
"completions/mean_length": 564.75,
"completions/mean_terminated_length": 571.4466552734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 177.0,
"epoch": 0.1984,
"grad_norm": 0.04410063102841377,
"learning_rate": 3.8888888888888895e-07,
"loss": -0.037,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.030299216508865356,
"mask/share_reasoning": 0.832628607749939,
"mask/share_step_conf": 0.12535345554351807,
"num_tokens": 46815951.0,
"reward": 1.3567540645599365,
"reward_std": 0.24842774868011475,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.732350766658783,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8347193598747253,
"step": 186
},
{
"adv/mean_abs_final_conf": 0.6464817523956299,
"adv/mean_abs_reasoning": 0.5685774683952332,
"adv/mean_abs_step_conf": 0.7855064868927002,
"adv/ratio_final_to_reasoning": 1.137016129429602,
"adv/ratio_step_to_reasoning": 1.3815293967060158,
"adv/std_final_conf": 0.8438383936882019,
"adv/std_reasoning": 0.7929400205612183,
"adv/std_step_conf": 0.9357006549835205,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6160920320347497,
"calib/avg_num_step_conf": 7.67578125,
"calib/ece": 0.31314285714285706,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.6081632653061224,
"calib/gap": 0.13770598615447271,
"calib/mean_conf": 0.7271836734693878,
"calib/mu_c": 0.7867625899280575,
"calib/mu_w": 0.6490566037735848,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.23648979591836725,
"calib/std_conf": 0.36267237553795345,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.5256306306306306,
"calib/step_q_c_n": 888.0,
"calib/step_q_gap": 0.1536343446510578,
"calib/step_q_w": 0.37199628597957285,
"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": 0.03125,
"completions/max_length": 2661.0,
"completions/max_terminated_length": 2661.0,
"completions/mean_length": 575.4140625,
"completions/mean_terminated_length": 593.9757690429688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 207.0,
"epoch": 0.19946666666666665,
"grad_norm": 0.03723229467868805,
"learning_rate": 3.611111111111111e-07,
"loss": -0.2321,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.027325410395860672,
"mask/share_reasoning": 0.8210077285766602,
"mask/share_step_conf": 0.12041684985160828,
"num_tokens": 47064801.0,
"reward": 1.249746561050415,
"reward_std": 0.3241482675075531,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.6286714673042297,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.786191999912262,
"step": 187
},
{
"adv/mean_abs_final_conf": 0.5331639647483826,
"adv/mean_abs_reasoning": 0.4191775321960449,
"adv/mean_abs_step_conf": 0.7717061042785645,
"adv/ratio_final_to_reasoning": 1.2719287743194867,
"adv/ratio_step_to_reasoning": 1.8410006381679007,
"adv/std_final_conf": 0.7863588929176331,
"adv/std_reasoning": 0.7014192342758179,
"adv/std_step_conf": 0.9349312782287598,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7297077922077922,
"calib/avg_num_step_conf": 7.03125,
"calib/ece": 0.20650602409638558,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.7188755020080321,
"calib/gap": 0.3225232919254659,
"calib/mean_conf": 0.812289156626506,
"calib/mu_c": 0.9262732919254658,
"calib/mu_w": 0.6037499999999999,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.18610441767068275,
"calib/std_conf": 0.3238820449469957,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5754664179104477,
"calib/step_q_c_n": 1072.0,
"calib/step_q_gap": 0.17839224208627186,
"calib/step_q_w": 0.39707417582417587,
"calib/step_q_w_n": 728.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2316.0,
"completions/max_terminated_length": 2316.0,
"completions/mean_length": 551.86328125,
"completions/mean_terminated_length": 560.623046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 192.0,
"epoch": 0.20053333333333334,
"grad_norm": 0.041271813213825226,
"learning_rate": 3.3333333333333335e-07,
"loss": -0.0684,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.0306197889149189,
"mask/share_reasoning": 0.8181106448173523,
"mask/share_step_conf": 0.1356445550918579,
"num_tokens": 47310150.0,
"reward": 1.390303134918213,
"reward_std": 0.2478947937488556,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/final_brier_reward_step": 0.7650250196456909,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8476343154907227,
"step": 188
},
{
"adv/mean_abs_final_conf": 0.5081390738487244,
"adv/mean_abs_reasoning": 0.32521361112594604,
"adv/mean_abs_step_conf": 0.7686993479728699,
"adv/ratio_final_to_reasoning": 1.5624778805827302,
"adv/ratio_step_to_reasoning": 2.3636752020049197,
"adv/std_final_conf": 0.745349645614624,
"adv/std_reasoning": 0.5961154699325562,
"adv/std_step_conf": 0.9345230460166931,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.7175714472138849,
"calib/avg_num_step_conf": 6.1796875,
"calib/ece": 0.22227450980392155,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.5764705882352941,
"calib/gap": 0.3352218452303275,
"calib/mean_conf": 0.6709019607843137,
"calib/mu_c": 0.7984177215189874,
"calib/mu_w": 0.4631958762886599,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.13678431372549016,
"calib/std_conf": 0.40453746305597355,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.530032119914347,
"calib/step_q_c_n": 934.0,
"calib/step_q_gap": 0.18185310756866796,
"calib/step_q_w": 0.348179012345679,
"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": 0.0,
"completions/max_length": 2624.0,
"completions/max_terminated_length": 2624.0,
"completions/mean_length": 544.19921875,
"completions/mean_terminated_length": 544.19921875,
"completions/min_length": 147.0,
"completions/min_terminated_length": 147.0,
"epoch": 0.2016,
"grad_norm": 0.03426051139831543,
"learning_rate": 3.055555555555556e-07,
"loss": 0.0191,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03142692893743515,
"mask/share_reasoning": 0.841963529586792,
"mask/share_step_conf": 0.12660948932170868,
"num_tokens": 47557233.0,
"reward": 1.3983986377716064,
"reward_std": 0.1729370504617691,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/final_brier_reward_step": 0.7530906200408936,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.8605251312255859,
"step": 189
},
{
"adv/mean_abs_final_conf": 0.6338986754417419,
"adv/mean_abs_reasoning": 0.41939041018486023,
"adv/mean_abs_step_conf": 0.7677431106567383,
"adv/ratio_final_to_reasoning": 1.5114763238442435,
"adv/ratio_step_to_reasoning": 1.8306167523437888,
"adv/std_final_conf": 0.8425524234771729,
"adv/std_reasoning": 0.7013546228408813,
"adv/std_step_conf": 0.9354836344718933,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7676586003924133,
"calib/avg_num_step_conf": 7.09765625,
"calib/ece": 0.19530120481927704,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.5943775100401606,
"calib/gap": 0.3700019620667102,
"calib/mean_conf": 0.7113654618473896,
"calib/mu_c": 0.874820143884892,
"calib/mu_w": 0.5048181818181818,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.17421686746987947,
"calib/std_conf": 0.3776525237392623,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5215325248070563,
"calib/step_q_c_n": 907.0,
"calib/step_q_gap": 0.1547523050268365,
"calib/step_q_w": 0.36678021978021974,
"calib/step_q_w_n": 910.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2315.0,
"completions/max_terminated_length": 2315.0,
"completions/mean_length": 608.9609375,
"completions/mean_terminated_length": 616.1818237304688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 195.0,
"epoch": 0.20266666666666666,
"grad_norm": 0.04120016098022461,
"learning_rate": 2.7777777777777776e-07,
"loss": -0.0373,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.027381490916013718,
"mask/share_reasoning": 0.8419528007507324,
"mask/share_step_conf": 0.11894698441028595,
"num_tokens": 47818735.0,
"reward": 1.3498507738113403,
"reward_std": 0.25871074199676514,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.7487621307373047,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8239072561264038,
"step": 190
},
{
"adv/mean_abs_final_conf": 0.6130039691925049,
"adv/mean_abs_reasoning": 0.45739710330963135,
"adv/mean_abs_step_conf": 0.7466780543327332,
"adv/ratio_final_to_reasoning": 1.3402008118480293,
"adv/ratio_step_to_reasoning": 1.6324503345778194,
"adv/std_final_conf": 0.8566485643386841,
"adv/std_reasoning": 0.7392563223838806,
"adv/std_step_conf": 0.9357996582984924,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6787549407114625,
"calib/avg_num_step_conf": 7.6796875,
"calib/ece": 0.2614979757085019,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.659919028340081,
"calib/gap": 0.3105711462450593,
"calib/mean_conf": 0.7515384615384615,
"calib/mu_c": 0.8961363636363637,
"calib/mu_w": 0.5855652173913044,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23931174089068816,
"calib/std_conf": 0.36746431364155935,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.544176072234763,
"calib/step_q_c_n": 886.0,
"calib/step_q_gap": 0.18581496112365198,
"calib/step_q_w": 0.3583611111111111,
"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": 0.0234375,
"completions/max_length": 2777.0,
"completions/max_terminated_length": 2777.0,
"completions/mean_length": 549.3671875,
"completions/mean_terminated_length": 562.552001953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 153.0,
"epoch": 0.20373333333333332,
"grad_norm": 0.06811851263046265,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.1334,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.030976133421063423,
"mask/share_reasoning": 0.8067940473556519,
"mask/share_step_conf": 0.1387922763824463,
"num_tokens": 48063541.0,
"reward": 1.3004114627838135,
"reward_std": 0.2827233374118805,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.698123037815094,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.803303062915802,
"step": 191
},
{
"adv/mean_abs_final_conf": 0.5427058935165405,
"adv/mean_abs_reasoning": 0.3278455138206482,
"adv/mean_abs_step_conf": 0.7394140958786011,
"adv/ratio_final_to_reasoning": 1.6553708092325286,
"adv/ratio_step_to_reasoning": 2.25537353633915,
"adv/std_final_conf": 0.7968785762786865,
"adv/std_reasoning": 0.6404013633728027,
"adv/std_step_conf": 0.9348773956298828,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.8160141301148072,
"calib/avg_num_step_conf": 7.12890625,
"calib/ece": 0.14954918032786899,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.6680327868852459,
"calib/gap": 0.43998086546953197,
"calib/mean_conf": 0.7689754098360656,
"calib/mu_c": 0.9240506329113924,
"calib/mu_w": 0.4840697674418605,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.13549180327868865,
"calib/std_conf": 0.35132561523452,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5471717171717172,
"calib/step_q_c_n": 990.0,
"calib/step_q_gap": 0.2089202201657292,
"calib/step_q_w": 0.338251497005988,
"calib/step_q_w_n": 835.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2617.0,
"completions/max_terminated_length": 2617.0,
"completions/mean_length": 587.4296875,
"completions/mean_terminated_length": 599.1314697265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 156.0,
"epoch": 0.2048,
"grad_norm": 0.04900625720620155,
"learning_rate": 2.2222222222222224e-07,
"loss": -0.1622,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.030505452305078506,
"mask/share_reasoning": 0.8191900253295898,
"mask/share_step_conf": 0.13077324628829956,
"num_tokens": 48318899.0,
"reward": 1.398130178451538,
"reward_std": 0.27167192101478577,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.7914214134216309,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8453881740570068,
"step": 192
},
{
"adv/mean_abs_final_conf": 0.701214075088501,
"adv/mean_abs_reasoning": 0.630457878112793,
"adv/mean_abs_step_conf": 0.7696429491043091,
"adv/ratio_final_to_reasoning": 1.1122298561602704,
"adv/ratio_step_to_reasoning": 1.2207682318256559,
"adv/std_final_conf": 0.8996307253837585,
"adv/std_reasoning": 0.8267195820808411,
"adv/std_step_conf": 0.9354685544967651,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7785456730769231,
"calib/avg_num_step_conf": 6.60546875,
"calib/ece": 0.18568548387096784,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.5604838709677419,
"calib/gap": 0.3694871794871795,
"calib/mean_conf": 0.6846370967741936,
"calib/mu_c": 0.8395833333333333,
"calib/mu_w": 0.47009615384615383,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.14483870967741946,
"calib/std_conf": 0.37652039816819033,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.525945945945946,
"calib/step_q_c_n": 888.0,
"calib/step_q_gap": 0.14681020497458858,
"calib/step_q_w": 0.37913574097135744,
"calib/step_q_w_n": 803.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2824.0,
"completions/max_terminated_length": 2824.0,
"completions/mean_length": 587.8359375,
"completions/mean_terminated_length": 597.1666870117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 233.0,
"epoch": 0.20586666666666667,
"grad_norm": 0.04521409794688225,
"learning_rate": 1.9444444444444447e-07,
"loss": 0.0169,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.027636289596557617,
"mask/share_reasoning": 0.8352762460708618,
"mask/share_step_conf": 0.12146242707967758,
"num_tokens": 48575097.0,
"reward": 1.3805063962936401,
"reward_std": 0.27369970083236694,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.7593636512756348,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8476996421813965,
"step": 193
},
{
"adv/mean_abs_final_conf": 0.6136116981506348,
"adv/mean_abs_reasoning": 0.4309585988521576,
"adv/mean_abs_step_conf": 0.7773469686508179,
"adv/ratio_final_to_reasoning": 1.423829805890791,
"adv/ratio_step_to_reasoning": 1.803762520857579,
"adv/std_final_conf": 0.8142480254173279,
"adv/std_reasoning": 0.701364278793335,
"adv/std_step_conf": 0.9353013634681702,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7455775577557754,
"calib/avg_num_step_conf": 6.56640625,
"calib/ece": 0.21454183266932267,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.6334661354581673,
"calib/gap": 0.3424349834983498,
"calib/mean_conf": 0.7214741035856573,
"calib/mu_c": 0.8592666666666666,
"calib/mu_w": 0.5168316831683168,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.16920318725099598,
"calib/std_conf": 0.3844339481429212,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5600668151447662,
"calib/step_q_c_n": 898.0,
"calib/step_q_gap": 0.16935161718818897,
"calib/step_q_w": 0.3907151979565772,
"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": 0.01953125,
"completions/max_length": 1768.0,
"completions/max_terminated_length": 1768.0,
"completions/mean_length": 536.14453125,
"completions/mean_terminated_length": 546.82470703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 189.0,
"epoch": 0.20693333333333333,
"grad_norm": 0.036717407405376434,
"learning_rate": 1.6666666666666668e-07,
"loss": -0.0308,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.030489569529891014,
"mask/share_reasoning": 0.8255350589752197,
"mask/share_step_conf": 0.12444411218166351,
"num_tokens": 48818294.0,
"reward": 1.3486157655715942,
"reward_std": 0.26706817746162415,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.7385668158531189,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8234729766845703,
"step": 194
},
{
"adv/mean_abs_final_conf": 0.5755367279052734,
"adv/mean_abs_reasoning": 0.46918901801109314,
"adv/mean_abs_step_conf": 0.7148308753967285,
"adv/ratio_final_to_reasoning": 1.2266628284374421,
"adv/ratio_step_to_reasoning": 1.5235456243773966,
"adv/std_final_conf": 0.8129864931106567,
"adv/std_reasoning": 0.7393112182617188,
"adv/std_step_conf": 0.9357689023017883,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.6910894660894662,
"calib/avg_num_step_conf": 6.88671875,
"calib/ece": 0.24008368200836827,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 0.6317991631799164,
"calib/gap": 0.2835815295815297,
"calib/mean_conf": 0.7251046025104603,
"calib/mu_c": 0.8425714285714286,
"calib/mu_w": 0.5589898989898989,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 0.96875,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.18970711297071136,
"calib/std_conf": 0.37731917382077446,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.5250125142207055,
"calib/step_q_c_n": 879.0,
"calib/step_q_gap": 0.15727608888133898,
"calib/step_q_w": 0.3677364253393665,
"calib/step_q_w_n": 884.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3051.0,
"completions/max_terminated_length": 3051.0,
"completions/mean_length": 575.69921875,
"completions/mean_terminated_length": 594.2701416015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 158.0,
"epoch": 0.208,
"grad_norm": 0.04009755328297615,
"learning_rate": 1.3888888888888888e-07,
"loss": -0.1084,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.028049079701304436,
"mask/share_reasoning": 0.817302942276001,
"mask/share_step_conf": 0.12339800596237183,
"num_tokens": 49071657.0,
"reward": 1.2790379524230957,
"reward_std": 0.27997830510139465,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.679855465888977,
"rewards/format_reward_step": 0.921875,
"rewards/step_l2_reward": 0.7922351360321045,
"step": 195
},
{
"adv/mean_abs_final_conf": 0.526882529258728,
"adv/mean_abs_reasoning": 0.3654212951660156,
"adv/mean_abs_step_conf": 0.7486469745635986,
"adv/ratio_final_to_reasoning": 1.441849547983673,
"adv/ratio_step_to_reasoning": 2.048722897288946,
"adv/std_final_conf": 0.7710081338882446,
"adv/std_reasoning": 0.6402504444122314,
"adv/std_step_conf": 0.9332286715507507,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7198146372643017,
"calib/avg_num_step_conf": 6.4921875,
"calib/ece": 0.24598425196850385,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.7086614173228346,
"calib/gap": 0.3077494407158837,
"calib/mean_conf": 0.7851968503937008,
"calib/mu_c": 0.9124161073825504,
"calib/mu_w": 0.6046666666666667,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.22228346456692905,
"calib/std_conf": 0.3452862566700246,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6140914036996734,
"calib/step_q_c_n": 919.0,
"calib/step_q_gap": 0.19530270921784298,
"calib/step_q_w": 0.4187886944818304,
"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": 0.00390625,
"completions/max_length": 2571.0,
"completions/max_terminated_length": 2571.0,
"completions/mean_length": 482.88671875,
"completions/mean_terminated_length": 484.7804260253906,
"completions/min_length": 0.0,
"completions/min_terminated_length": 171.0,
"epoch": 0.20906666666666668,
"grad_norm": 0.033003684133291245,
"learning_rate": 1.1111111111111112e-07,
"loss": 0.0064,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03300885111093521,
"mask/share_reasoning": 0.8257684707641602,
"mask/share_step_conf": 0.13731642067432404,
"num_tokens": 49297820.0,
"reward": 1.3620078563690186,
"reward_std": 0.1846241056919098,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.7422569990158081,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8334574699401855,
"step": 196
},
{
"adv/mean_abs_final_conf": 0.6287498474121094,
"adv/mean_abs_reasoning": 0.4730614423751831,
"adv/mean_abs_step_conf": 0.7427409291267395,
"adv/ratio_final_to_reasoning": 1.329108211092483,
"adv/ratio_step_to_reasoning": 1.570072854379188,
"adv/std_final_conf": 0.8411765098571777,
"adv/std_reasoning": 0.7207943201065063,
"adv/std_step_conf": 0.9351040720939636,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7412587412587412,
"calib/avg_num_step_conf": 7.07421875,
"calib/ece": 0.21534412955465593,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.5951417004048583,
"calib/gap": 0.2999125874125874,
"calib/mean_conf": 0.7295951417004047,
"calib/mu_c": 0.8558741258741259,
"calib/mu_w": 0.5559615384615385,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.18299595141700412,
"calib/std_conf": 0.35663269707665646,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5274426229508197,
"calib/step_q_c_n": 915.0,
"calib/step_q_gap": 0.14353637295081967,
"calib/step_q_w": 0.38390625,
"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": 0.03125,
"completions/max_length": 1876.0,
"completions/max_terminated_length": 1876.0,
"completions/mean_length": 536.7734375,
"completions/mean_terminated_length": 554.0886840820312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 202.0,
"epoch": 0.21013333333333334,
"grad_norm": 0.03876349329948425,
"learning_rate": 8.333333333333334e-08,
"loss": -0.2178,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.029733015224337578,
"mask/share_reasoning": 0.810360312461853,
"mask/share_step_conf": 0.12865662574768066,
"num_tokens": 49540290.0,
"reward": 1.3443280458450317,
"reward_std": 0.2846136689186096,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/final_brier_reward_step": 0.7261113524436951,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8289285898208618,
"step": 197
},
{
"adv/mean_abs_final_conf": 0.5739809274673462,
"adv/mean_abs_reasoning": 0.41014930605888367,
"adv/mean_abs_step_conf": 0.7451508045196533,
"adv/ratio_final_to_reasoning": 1.3994438585858335,
"adv/ratio_step_to_reasoning": 1.8167793862186241,
"adv/std_final_conf": 0.8088400363922119,
"adv/std_reasoning": 0.6613656878471375,
"adv/std_step_conf": 0.9343599677085876,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7693077230892357,
"calib/avg_num_step_conf": 7.0234375,
"calib/ece": 0.17290836653386443,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.6055776892430279,
"calib/gap": 0.3876623982926503,
"calib/mean_conf": 0.7000796812749004,
"calib/mu_c": 0.851437908496732,
"calib/mu_w": 0.46377551020408164,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.13171314741035847,
"calib/std_conf": 0.38553940278665894,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5258260869565218,
"calib/step_q_c_n": 1035.0,
"calib/step_q_gap": 0.122103937546299,
"calib/step_q_w": 0.4037221494102228,
"calib/step_q_w_n": 763.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 1477.0,
"completions/max_terminated_length": 1477.0,
"completions/mean_length": 516.171875,
"completions/mean_terminated_length": 526.4542236328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 185.0,
"epoch": 0.2112,
"grad_norm": 0.026938652619719505,
"learning_rate": 5.555555555555556e-08,
"loss": -0.0684,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03219270706176758,
"mask/share_reasoning": 0.8046937584877014,
"mask/share_step_conf": 0.1435822695493698,
"num_tokens": 49777814.0,
"reward": 1.4053709506988525,
"reward_std": 0.23416009545326233,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.7742702960968018,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8604233264923096,
"step": 198
},
{
"adv/mean_abs_final_conf": 0.6629692316055298,
"adv/mean_abs_reasoning": 0.5119156837463379,
"adv/mean_abs_step_conf": 0.7785393595695496,
"adv/ratio_final_to_reasoning": 1.2950750536763027,
"adv/ratio_step_to_reasoning": 1.5208351380680256,
"adv/std_final_conf": 0.8606733083724976,
"adv/std_reasoning": 0.7576802372932434,
"adv/std_step_conf": 0.9351163506507874,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7201812647018125,
"calib/avg_num_step_conf": 7.4140625,
"calib/ece": 0.23914285714285713,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.6530612244897959,
"calib/gap": 0.26443960149439594,
"calib/mean_conf": 0.7648571428571428,
"calib/mu_c": 0.8717123287671232,
"calib/mu_w": 0.6072727272727273,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2040408163265306,
"calib/std_conf": 0.3489361089988614,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5189263157894737,
"calib/step_q_c_n": 950.0,
"calib/step_q_gap": 0.15909509216078171,
"calib/step_q_w": 0.35983122362869197,
"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": 0.03125,
"completions/max_length": 3031.0,
"completions/max_terminated_length": 3031.0,
"completions/mean_length": 599.40234375,
"completions/mean_terminated_length": 618.7379150390625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 114.0,
"epoch": 0.21226666666666666,
"grad_norm": 0.045903194695711136,
"learning_rate": 2.777777777777778e-08,
"loss": -0.0885,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.029736265540122986,
"mask/share_reasoning": 0.8093239068984985,
"mask/share_step_conf": 0.12968984246253967,
"num_tokens": 50035461.0,
"reward": 1.333052635192871,
"reward_std": 0.30268052220344543,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.704621434211731,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8280073404312134,
"step": 199
},
{
"adv/mean_abs_final_conf": 0.5917401313781738,
"adv/mean_abs_reasoning": 0.4564133584499359,
"adv/mean_abs_step_conf": 0.753562331199646,
"adv/ratio_final_to_reasoning": 1.2965004648151242,
"adv/ratio_step_to_reasoning": 1.6510523131024974,
"adv/std_final_conf": 0.8196789026260376,
"adv/std_reasoning": 0.7206570506095886,
"adv/std_step_conf": 0.9346625804901123,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.7932464886107271,
"calib/avg_num_step_conf": 6.26171875,
"calib/ece": 0.17487603305785127,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.6446280991735537,
"calib/gap": 0.3713557965213594,
"calib/mean_conf": 0.743801652892562,
"calib/mu_c": 0.8834437086092716,
"calib/mu_w": 0.5120879120879122,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 0.9765625,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.14735537190082648,
"calib/std_conf": 0.36617919155496315,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.5808118214716526,
"calib/step_q_c_n": 829.0,
"calib/step_q_gap": 0.22709089123909437,
"calib/step_q_w": 0.3537209302325582,
"calib/step_q_w_n": 774.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2845.0,
"completions/max_terminated_length": 2845.0,
"completions/mean_length": 552.8046875,
"completions/mean_terminated_length": 568.3453369140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.21333333333333335,
"grad_norm": 0.04398513212800026,
"learning_rate": 0.0,
"loss": -0.1976,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.030634820461273193,
"mask/share_reasoning": 0.8216097354888916,
"mask/share_step_conf": 0.12041172385215759,
"num_tokens": 50285027.0,
"reward": 1.3560163974761963,
"reward_std": 0.26149389147758484,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.7444355487823486,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.8306736946105957,
"step": 200
},
{
"epoch": 0.21333333333333335,
"step": 200,
"total_flos": 0.0,
"train_loss": -0.03754158863797784,
"train_runtime": 17053.3275,
"train_samples_per_second": 3.002,
"train_steps_per_second": 0.012
}
],
"logging_steps": 1,
"max_steps": 200,
"num_input_tokens_seen": 50285027,
"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