zhaohq/PureRL-1.5B-v7-s2-async-l2-maskon-afew
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
7eaae3d38bc36ed153320de598afd4383ff4f2f8
PureRL-1.5B-v7-s2-async-l2-…/trainer_state.json
ModelHub XC 7eaae3d38b 初始化项目，由ModelHub XC社区提供模型 
Model: zhaohq/PureRL-1.5B-v7-s2-async-l2-maskon-afew
Source: Original Platform
2026-06-04 11:32:30 +08:00

12243 lines
502 KiB
JSON
Raw Blame History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.21333333333333335,
"eval_steps": 500,
"global_step": 200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"adv/mean_abs_final_conf": 0.47760647535324097,
"adv/mean_abs_reasoning": 0.4569147527217865,
"adv/mean_abs_step_conf": 0.7561872005462646,
"adv/ratio_final_to_reasoning": 1.0452857398632815,
"adv/ratio_step_to_reasoning": 1.6549853031484494,
"adv/std_final_conf": 0.7227410674095154,
"adv/std_reasoning": 0.7206857204437256,
"adv/std_step_conf": 0.9253036379814148,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5086206896551725,
"calib/avg_num_step_conf": 7.875,
"calib/ece": 0.2888991935483871,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0001713264989126051,
"calib/mean_conf": 0.9905120967741936,
"calib/mu_c": 0.9905632183908043,
"calib/mu_w": 0.9903918918918917,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2888991935483871,
"calib/std_conf": 0.0021794159006610276,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9119477557027226,
"calib/step_q_c_n": 1359.0,
"calib/step_q_gap": 0.0056311651395566376,
"calib/step_q_w": 0.9063165905631659,
"calib/step_q_w_n": 657.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2494.0,
"completions/max_terminated_length": 2494.0,
"completions/mean_length": 755.49609375,
"completions/mean_terminated_length": 776.7349243164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 397.0,
"epoch": 0.0010666666666666667,
"grad_norm": 0.025163671001791954,
"kl": 0.0005849599838256836,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.13,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.018737709149718285,
"mask/share_reasoning": 0.845859944820404,
"mask/share_step_conf": 0.10805858671665192,
"num_tokens": 300991.0,
"reward": 0.8751538991928101,
"reward_std": 0.2377150058746338,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/asymmetric_l2_reward": 0.7354698181152344,
"rewards/final_brier_reward_step": 0.6851503849029541,
"rewards/format_reward_step": 0.96875,
"step": 1
},
{
"adv/mean_abs_final_conf": 0.437887966632843,
"adv/mean_abs_reasoning": 0.4207462966442108,
"adv/mean_abs_step_conf": 0.7180101275444031,
"adv/ratio_final_to_reasoning": 1.0407411072310102,
"adv/ratio_step_to_reasoning": 1.7065156206272276,
"adv/std_final_conf": 0.6832791566848755,
"adv/std_reasoning": 0.6817297339439392,
"adv/std_step_conf": 0.9190711975097656,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.4872611464968153,
"calib/avg_num_step_conf": 7.6953125,
"calib/ece": 0.36465737051792824,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00024481637078155316,
"calib/mean_conf": 0.9901553784860557,
"calib/mu_c": 0.990063694267516,
"calib/mu_w": 0.9903085106382975,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36465737051792824,
"calib/std_conf": 0.001222205307190084,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9075405636208369,
"calib/step_q_c_n": 1171.0,
"calib/step_q_gap": -0.003804868168900244,
"calib/step_q_w": 0.9113454317897371,
"calib/step_q_w_n": 799.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2743.0,
"completions/max_terminated_length": 2743.0,
"completions/mean_length": 840.640625,
"completions/mean_terminated_length": 850.6087036132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 466.0,
"epoch": 0.0021333333333333334,
"grad_norm": 0.021795878186821938,
"kl": 0.0016820430755615234,
"learning_rate": 5.000000000000001e-07,
"loss": -0.0102,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.01788702979683876,
"mask/share_reasoning": 0.8706268668174744,
"mask/share_step_conf": 0.09976735711097717,
"num_tokens": 619483.0,
"reward": 0.8191705346107483,
"reward_std": 0.21779605746269226,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/asymmetric_l2_reward": 0.6992889642715454,
"rewards/final_brier_reward_step": 0.6203019618988037,
"rewards/format_reward_step": 0.98046875,
"step": 2
},
{
"adv/mean_abs_final_conf": 0.42324480414390564,
"adv/mean_abs_reasoning": 0.4121120572090149,
"adv/mean_abs_step_conf": 0.742616593837738,
"adv/ratio_final_to_reasoning": 1.0270138830935598,
"adv/ratio_step_to_reasoning": 1.8019773526332377,
"adv/std_final_conf": 0.7039074897766113,
"adv/std_reasoning": 0.7013108134269714,
"adv/std_step_conf": 0.9225370287895203,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.49612068965517236,
"calib/avg_num_step_conf": 7.6328125,
"calib/ece": 0.3051574803149606,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -7.758620689668483e-05,
"calib/mean_conf": 0.9901968503937008,
"calib/mu_c": 0.9901724137931033,
"calib/mu_w": 0.99025,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3051574803149606,
"calib/std_conf": 0.00138915580821869,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9130487804878049,
"calib/step_q_c_n": 1312.0,
"calib/step_q_gap": 0.01088367145353697,
"calib/step_q_w": 0.9021651090342679,
"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": 0.00390625,
"completions/max_length": 2662.0,
"completions/max_terminated_length": 2662.0,
"completions/mean_length": 802.8984375,
"completions/mean_terminated_length": 806.047119140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 431.0,
"epoch": 0.0032,
"grad_norm": 0.024119608104228973,
"kl": 0.0005092322826385498,
"learning_rate": 7.5e-07,
"loss": -0.0327,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.018682915717363358,
"mask/share_reasoning": 0.8749101758003235,
"mask/share_step_conf": 0.10250066965818405,
"num_tokens": 930281.0,
"reward": 0.8781776428222656,
"reward_std": 0.21866968274116516,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/asymmetric_l2_reward": 0.7362962961196899,
"rewards/final_brier_reward_step": 0.6856839656829834,
"rewards/format_reward_step": 0.9921875,
"step": 3
},
{
"adv/mean_abs_final_conf": 0.5237265825271606,
"adv/mean_abs_reasoning": 0.4569150507450104,
"adv/mean_abs_step_conf": 0.773900032043457,
"adv/ratio_final_to_reasoning": 1.1462230926147268,
"adv/ratio_step_to_reasoning": 1.6937503607762436,
"adv/std_final_conf": 0.756417453289032,
"adv/std_reasoning": 0.7015604376792908,
"adv/std_step_conf": 0.9241310358047485,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.4732142857142857,
"calib/avg_num_step_conf": 7.64453125,
"calib/ece": 0.2946245059288537,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.000535714285714084,
"calib/mean_conf": 0.9902766798418972,
"calib/mu_c": 0.9901136363636364,
"calib/mu_w": 0.9906493506493504,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2946245059288537,
"calib/std_conf": 0.0016401971479246202,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9098583146905294,
"calib/step_q_c_n": 1341.0,
"calib/step_q_gap": 0.0013031198853343806,
"calib/step_q_w": 0.908555194805195,
"calib/step_q_w_n": 616.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2449.0,
"completions/max_terminated_length": 2449.0,
"completions/mean_length": 782.640625,
"completions/mean_terminated_length": 791.9209594726562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 456.0,
"epoch": 0.004266666666666667,
"grad_norm": 0.027355490252375603,
"kl": 0.0005553364753723145,
"learning_rate": 1.0000000000000002e-06,
"loss": -0.0374,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.018635034561157227,
"mask/share_reasoning": 0.8674567937850952,
"mask/share_step_conf": 0.10218945145606995,
"num_tokens": 1236805.0,
"reward": 0.8771485090255737,
"reward_std": 0.25210636854171753,
"rewards/accuracy_reward_step": 0.6875,
"rewards/asymmetric_l2_reward": 0.7269708514213562,
"rewards/final_brier_reward_step": 0.6929511427879333,
"rewards/format_reward_step": 0.984375,
"step": 4
},
{
"adv/mean_abs_final_conf": 0.47414445877075195,
"adv/mean_abs_reasoning": 0.4151928126811981,
"adv/mean_abs_step_conf": 0.7436368465423584,
"adv/ratio_final_to_reasoning": 1.1419861912080336,
"adv/ratio_step_to_reasoning": 1.7910638716025966,
"adv/std_final_conf": 0.7365800738334656,
"adv/std_reasoning": 0.6815617680549622,
"adv/std_step_conf": 0.9012373089790344,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.49449053668654386,
"calib/avg_num_step_conf": 7.6796875,
"calib/ece": 0.45881392235609114,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9959839357429718,
"calib/gap": -0.004985437732261788,
"calib/mean_conf": 0.9875957161981259,
"calib/mu_c": 0.9852731829573933,
"calib/mu_w": 0.9902586206896551,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4561365461847391,
"calib/std_conf": 0.04157220198737621,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9131255901794145,
"calib/step_q_c_n": 1059.0,
"calib/step_q_gap": 0.01992823626541229,
"calib/step_q_w": 0.8931973539140022,
"calib/step_q_w_n": 907.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2956.0,
"completions/max_terminated_length": 2956.0,
"completions/mean_length": 806.9296875,
"completions/mean_terminated_length": 816.498046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.005333333333333333,
"grad_norm": 0.055844664573669434,
"kl": 0.0005984306335449219,
"learning_rate": 1.25e-06,
"loss": -0.1083,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.018338514491915703,
"mask/share_reasoning": 0.8661159873008728,
"mask/share_step_conf": 0.10382672399282455,
"num_tokens": 1550067.0,
"reward": 0.7309314012527466,
"reward_std": 0.18717044591903687,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/asymmetric_l2_reward": 0.6368963718414307,
"rewards/final_brier_reward_step": 0.5265287160873413,
"rewards/format_reward_step": 0.97265625,
"step": 5
},
{
"adv/mean_abs_final_conf": 0.40251004695892334,
"adv/mean_abs_reasoning": 0.374636173248291,
"adv/mean_abs_step_conf": 0.7238371968269348,
"adv/ratio_final_to_reasoning": 1.0744025155631698,
"adv/ratio_step_to_reasoning": 1.9321070641707896,
"adv/std_final_conf": 0.6800278425216675,
"adv/std_reasoning": 0.6612951159477234,
"adv/std_step_conf": 0.9232251048088074,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.4968253968253968,
"calib/avg_num_step_conf": 8.14453125,
"calib/ece": 0.42572580645161295,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -6.349206349198777e-05,
"calib/mean_conf": 0.9902419354838711,
"calib/mu_c": 0.9902142857142857,
"calib/mu_w": 0.9902777777777777,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42572580645161295,
"calib/std_conf": 0.0015364966841336825,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9111431143114312,
"calib/step_q_c_n": 1111.0,
"calib/step_q_gap": 0.0001369541471603064,
"calib/step_q_w": 0.9110061601642709,
"calib/step_q_w_n": 974.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2736.0,
"completions/max_terminated_length": 2736.0,
"completions/mean_length": 723.171875,
"completions/mean_terminated_length": 737.5776977539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.0064,
"grad_norm": 0.017985476180911064,
"kl": 0.0007612109184265137,
"learning_rate": 1.5e-06,
"loss": -0.1013,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.0195919219404459,
"mask/share_reasoning": 0.8470251560211182,
"mask/share_step_conf": 0.11385165899991989,
"num_tokens": 1841151.0,
"reward": 0.7441216111183167,
"reward_std": 0.19726765155792236,
"rewards/accuracy_reward_step": 0.546875,
"rewards/asymmetric_l2_reward": 0.6301347017288208,
"rewards/final_brier_reward_step": 0.5549836158752441,
"rewards/format_reward_step": 0.96875,
"step": 6
},
{
"adv/mean_abs_final_conf": 0.5039571523666382,
"adv/mean_abs_reasoning": 0.4919060170650482,
"adv/mean_abs_step_conf": 0.7596220374107361,
"adv/ratio_final_to_reasoning": 1.0244988572684939,
"adv/ratio_step_to_reasoning": 1.5442422150942827,
"adv/std_final_conf": 0.7577763199806213,
"adv/std_reasoning": 0.7576440572738647,
"adv/std_step_conf": 0.9329904317855835,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.49632380808004245,
"calib/avg_num_step_conf": 7.6171875,
"calib/ece": 0.3113414634146341,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -7.352383839875909e-05,
"calib/mean_conf": 0.9902032520325202,
"calib/mu_c": 0.9901796407185628,
"calib/mu_w": 0.9902531645569616,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3113414634146341,
"calib/std_conf": 0.0014111020290820972,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9102744148506859,
"calib/step_q_c_n": 1239.0,
"calib/step_q_gap": 0.007883416257155718,
"calib/step_q_w": 0.9023909985935302,
"calib/step_q_w_n": 711.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2887.0,
"completions/max_terminated_length": 2887.0,
"completions/mean_length": 847.07421875,
"completions/mean_terminated_length": 874.399169921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 436.0,
"epoch": 0.007466666666666667,
"grad_norm": 0.020187566056847572,
"kl": 0.0005465447902679443,
"learning_rate": 1.75e-06,
"loss": -0.1468,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.01694045029580593,
"mask/share_reasoning": 0.8599110841751099,
"mask/share_step_conf": 0.09189847111701965,
"num_tokens": 2165426.0,
"reward": 0.8446517586708069,
"reward_std": 0.24925880134105682,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/asymmetric_l2_reward": 0.7083820104598999,
"rewards/final_brier_reward_step": 0.658265233039856,
"rewards/format_reward_step": 0.9609375,
"step": 7
},
{
"adv/mean_abs_final_conf": 0.4072754383087158,
"adv/mean_abs_reasoning": 0.39883965253829956,
"adv/mean_abs_step_conf": 0.7444837093353271,
"adv/ratio_final_to_reasoning": 1.0211508201773047,
"adv/ratio_step_to_reasoning": 1.866624104692891,
"adv/std_final_conf": 0.6824573874473572,
"adv/std_reasoning": 0.6816043853759766,
"adv/std_step_conf": 0.9320701360702515,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5149176954732511,
"calib/avg_num_step_conf": 7.54296875,
"calib/ece": 0.33077235772357716,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.991869918699187,
"calib/gap": 0.0022045855379189128,
"calib/mean_conf": 0.9893089430894308,
"calib/mu_c": 0.9900617283950617,
"calib/mu_w": 0.9878571428571428,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33077235772357716,
"calib/std_conf": 0.008110647504853151,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9099438652766643,
"calib/step_q_c_n": 1247.0,
"calib/step_q_gap": 0.003028660598301647,
"calib/step_q_w": 0.9069152046783626,
"calib/step_q_w_n": 684.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2578.0,
"completions/max_terminated_length": 2578.0,
"completions/mean_length": 832.9765625,
"completions/mean_terminated_length": 849.5697631835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.008533333333333334,
"grad_norm": 0.030579403042793274,
"kl": 0.0005984306335449219,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.0817,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.017597010359168053,
"mask/share_reasoning": 0.8655336499214172,
"mask/share_step_conf": 0.09733809530735016,
"num_tokens": 2485180.0,
"reward": 0.8367857933044434,
"reward_std": 0.19032156467437744,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/asymmetric_l2_reward": 0.7142133712768555,
"rewards/final_brier_reward_step": 0.6406081914901733,
"rewards/format_reward_step": 0.9609375,
"step": 8
},
{
"adv/mean_abs_final_conf": 0.4530647397041321,
"adv/mean_abs_reasoning": 0.4173707962036133,
"adv/mean_abs_step_conf": 0.7631109952926636,
"adv/ratio_final_to_reasoning": 1.0855209416307738,
"adv/ratio_step_to_reasoning": 1.8283765951855957,
"adv/std_final_conf": 0.7208334803581238,
"adv/std_reasoning": 0.7015334367752075,
"adv/std_step_conf": 0.9326648712158203,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5064102564102564,
"calib/avg_num_step_conf": 7.609375,
"calib/ece": 0.30553278688524593,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9959016393442623,
"calib/gap": 0.012948717948718103,
"calib/mean_conf": 0.985860655737705,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9770512820512818,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.30553278688524593,
"calib/std_conf": 0.06327533064648865,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.910533015115354,
"calib/step_q_c_n": 1257.0,
"calib/step_q_gap": 0.0008658660560197928,
"calib/step_q_w": 0.9096671490593342,
"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": 0.03515625,
"completions/max_length": 3002.0,
"completions/max_terminated_length": 3002.0,
"completions/mean_length": 802.5078125,
"completions/mean_terminated_length": 831.7490234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 438.0,
"epoch": 0.0096,
"grad_norm": 0.0191666167229414,
"kl": 0.0006733536720275879,
"learning_rate": 2.25e-06,
"loss": -0.1468,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01763000525534153,
"mask/share_reasoning": 0.8491230607032776,
"mask/share_step_conf": 0.09809070080518723,
"num_tokens": 2798158.0,
"reward": 0.8323462009429932,
"reward_std": 0.24119850993156433,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/asymmetric_l2_reward": 0.6868234872817993,
"rewards/final_brier_reward_step": 0.6583374738693237,
"rewards/format_reward_step": 0.94921875,
"step": 9
},
{
"adv/mean_abs_final_conf": 0.46466493606567383,
"adv/mean_abs_reasoning": 0.4617326855659485,
"adv/mean_abs_step_conf": 0.76551353931427,
"adv/ratio_final_to_reasoning": 1.006350536991184,
"adv/ratio_step_to_reasoning": 1.6579149868413052,
"adv/std_final_conf": 0.7218580842018127,
"adv/std_reasoning": 0.7206411957740784,
"adv/std_step_conf": 0.9229851961135864,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.497684705231875,
"calib/avg_num_step_conf": 7.62109375,
"calib/ece": 0.35407600000000006,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -3.600801714020996e-05,
"calib/mean_conf": 0.9900760000000001,
"calib/mu_c": 0.9900628930817608,
"calib/mu_w": 0.990098901098901,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35407600000000006,
"calib/std_conf": 0.0008474809732377484,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.908442760942761,
"calib/step_q_c_n": 1188.0,
"calib/step_q_gap": 0.001635421493219713,
"calib/step_q_w": 0.9068073394495413,
"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": 0.015625,
"completions/max_length": 2895.0,
"completions/max_terminated_length": 2895.0,
"completions/mean_length": 826.671875,
"completions/mean_terminated_length": 839.793701171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 455.0,
"epoch": 0.010666666666666666,
"grad_norm": 0.019903983920812607,
"kl": 0.0007127523422241211,
"learning_rate": 2.5e-06,
"loss": -0.0049,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017885662615299225,
"mask/share_reasoning": 0.8675950765609741,
"mask/share_step_conf": 0.09889422357082367,
"num_tokens": 3116586.0,
"reward": 0.8278957009315491,
"reward_std": 0.24505752325057983,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/asymmetric_l2_reward": 0.7082241773605347,
"rewards/final_brier_reward_step": 0.6280359029769897,
"rewards/format_reward_step": 0.9765625,
"step": 10
},
{
"adv/mean_abs_final_conf": 0.4113199710845947,
"adv/mean_abs_reasoning": 0.4019410014152527,
"adv/mean_abs_step_conf": 0.7453023195266724,
"adv/ratio_final_to_reasoning": 1.023334194909995,
"adv/ratio_step_to_reasoning": 1.8542580052854243,
"adv/std_final_conf": 0.7009356021881104,
"adv/std_reasoning": 0.7014546394348145,
"adv/std_step_conf": 0.914635419845581,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.49439056166763173,
"calib/avg_num_step_conf": 7.4296875,
"calib/ece": 0.3488938775510203,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9959183673469387,
"calib/gap": -0.0006118268674003557,
"calib/mean_conf": 0.9897102040816326,
"calib/mu_c": 0.9894904458598724,
"calib/mu_w": 0.9901022727272728,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3488938775510203,
"calib/std_conf": 0.005806549605893123,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9094141914191419,
"calib/step_q_c_n": 1212.0,
"calib/step_q_gap": 0.0026721624336346617,
"calib/step_q_w": 0.9067420289855073,
"calib/step_q_w_n": 690.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2786.0,
"completions/max_terminated_length": 2786.0,
"completions/mean_length": 799.36328125,
"completions/mean_terminated_length": 831.857666015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 317.0,
"epoch": 0.011733333333333333,
"grad_norm": 0.03048000857234001,
"kl": 0.0008162856101989746,
"learning_rate": 2.7500000000000004e-06,
"loss": -0.1851,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.017573874443769455,
"mask/share_reasoning": 0.8479140996932983,
"mask/share_step_conf": 0.09544955939054489,
"num_tokens": 3425703.0,
"reward": 0.8215123414993286,
"reward_std": 0.20594213902950287,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/asymmetric_l2_reward": 0.7136968374252319,
"rewards/final_brier_reward_step": 0.6160464286804199,
"rewards/format_reward_step": 0.953125,
"step": 11
},
{
"adv/mean_abs_final_conf": 0.3859899342060089,
"adv/mean_abs_reasoning": 0.3481341600418091,
"adv/mean_abs_step_conf": 0.7245120406150818,
"adv/ratio_final_to_reasoning": 1.1087390394543688,
"adv/ratio_step_to_reasoning": 2.0811288398934242,
"adv/std_final_conf": 0.6618059873580933,
"adv/std_reasoning": 0.6404772996902466,
"adv/std_step_conf": 0.9084425568580627,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.4823494885124937,
"calib/avg_num_step_conf": 8.3125,
"calib/ece": 0.2634897959183673,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00028006037229555414,
"calib/mean_conf": 0.9900204081632653,
"calib/mu_c": 0.9899438202247192,
"calib/mu_w": 0.9902238805970147,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2634897959183673,
"calib/std_conf": 0.0009409026477730967,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.911249182472204,
"calib/step_q_c_n": 1529.0,
"calib/step_q_gap": 0.004579733390400964,
"calib/step_q_w": 0.906669449081803,
"calib/step_q_w_n": 599.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2865.0,
"completions/max_terminated_length": 2865.0,
"completions/mean_length": 755.1953125,
"completions/mean_terminated_length": 782.7125854492188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 423.0,
"epoch": 0.0128,
"grad_norm": 0.020890548825263977,
"kl": 0.0010690093040466309,
"learning_rate": 3e-06,
"loss": -0.2085,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.018778398633003235,
"mask/share_reasoning": 0.8324668407440186,
"mask/share_step_conf": 0.11359857022762299,
"num_tokens": 3723209.0,
"reward": 0.8852379322052002,
"reward_std": 0.18036311864852905,
"rewards/accuracy_reward_step": 0.6953125,
"rewards/asymmetric_l2_reward": 0.7404860258102417,
"rewards/final_brier_reward_step": 0.7003023624420166,
"rewards/format_reward_step": 0.953125,
"step": 12
},
{
"adv/mean_abs_final_conf": 0.3926801383495331,
"adv/mean_abs_reasoning": 0.38152408599853516,
"adv/mean_abs_step_conf": 0.7393319010734558,
"adv/ratio_final_to_reasoning": 1.029240755067403,
"adv/ratio_step_to_reasoning": 1.9378380768240526,
"adv/std_final_conf": 0.661359429359436,
"adv/std_reasoning": 0.6612658500671387,
"adv/std_step_conf": 0.9282354712486267,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.49732035822579507,
"calib/avg_num_step_conf": 8.1953125,
"calib/ece": 0.33808000000000005,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -5.359283548378446e-05,
"calib/mean_conf": 0.9900800000000001,
"calib/mu_c": 0.9900613496932515,
"calib/mu_w": 0.9901149425287353,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33808000000000005,
"calib/std_conf": 0.0008908422980528043,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9087509578544062,
"calib/step_q_c_n": 1305.0,
"calib/step_q_gap": 0.0046652075391475645,
"calib/step_q_w": 0.9040857503152586,
"calib/step_q_w_n": 793.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2774.0,
"completions/max_terminated_length": 2774.0,
"completions/mean_length": 809.140625,
"completions/mean_terminated_length": 828.56005859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 323.0,
"epoch": 0.013866666666666666,
"grad_norm": 0.024629024788737297,
"kl": 0.0017845630645751953,
"learning_rate": 3.2500000000000002e-06,
"loss": -0.0634,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.0179408211261034,
"mask/share_reasoning": 0.8538167476654053,
"mask/share_step_conf": 0.10480496287345886,
"num_tokens": 4034941.0,
"reward": 0.833107590675354,
"reward_std": 0.1895129680633545,
"rewards/accuracy_reward_step": 0.63671875,
"rewards/asymmetric_l2_reward": 0.7010772824287415,
"rewards/final_brier_reward_step": 0.6432628631591797,
"rewards/format_reward_step": 0.97265625,
"step": 13
},
{
"adv/mean_abs_final_conf": 0.4264124631881714,
"adv/mean_abs_reasoning": 0.42067527770996094,
"adv/mean_abs_step_conf": 0.7750456929206848,
"adv/ratio_final_to_reasoning": 1.0136380381310783,
"adv/ratio_step_to_reasoning": 1.8423846942939404,
"adv/std_final_conf": 0.6820985674858093,
"adv/std_reasoning": 0.681735098361969,
"adv/std_step_conf": 0.9328522682189941,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 7.88671875,
"calib/ece": 0.3588524590163934,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -1.1102230246251565e-16,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.99,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3588524590163934,
"calib/std_conf": 0.0,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9080388978930308,
"calib/step_q_c_n": 1234.0,
"calib/step_q_gap": 0.012459280058635946,
"calib/step_q_w": 0.8955796178343949,
"calib/step_q_w_n": 785.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3050.0,
"completions/max_terminated_length": 3050.0,
"completions/mean_length": 856.76953125,
"completions/mean_terminated_length": 880.8554077148438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 402.0,
"epoch": 0.014933333333333333,
"grad_norm": 0.016446553170681,
"kl": 0.0022586584091186523,
"learning_rate": 3.5e-06,
"loss": -0.0861,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01721174642443657,
"mask/share_reasoning": 0.8545247912406921,
"mask/share_step_conf": 0.10091973841190338,
"num_tokens": 4359674.0,
"reward": 0.8058513402938843,
"reward_std": 0.21509991586208344,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/asymmetric_l2_reward": 0.6914855241775513,
"rewards/final_brier_reward_step": 0.6084984540939331,
"rewards/format_reward_step": 0.953125,
"step": 14
},
{
"adv/mean_abs_final_conf": 0.3723694086074829,
"adv/mean_abs_reasoning": 0.3705359399318695,
"adv/mean_abs_step_conf": 0.7469418048858643,
"adv/ratio_final_to_reasoning": 1.0049481534124611,
"adv/ratio_step_to_reasoning": 2.015841715713743,
"adv/std_final_conf": 0.6613016724586487,
"adv/std_reasoning": 0.6611766219139099,
"adv/std_step_conf": 0.9211583137512207,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5081591847265221,
"calib/avg_num_step_conf": 7.75390625,
"calib/ece": 0.39157480314960624,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00016382868937081518,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.9900657894736843,
"calib/mu_w": 0.9899019607843135,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.39157480314960624,
"calib/std_conf": 0.0008873565094161146,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9095799011532125,
"calib/step_q_c_n": 1214.0,
"calib/step_q_gap": 0.0030040256668311116,
"calib/step_q_w": 0.9065758754863814,
"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": 1776.0,
"completions/max_terminated_length": 1776.0,
"completions/mean_length": 766.30859375,
"completions/mean_terminated_length": 769.3137817382812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 419.0,
"epoch": 0.016,
"grad_norm": 0.01988803967833519,
"kl": 0.002725958824157715,
"learning_rate": 3.7500000000000005e-06,
"loss": 0.0034,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.01912056654691696,
"mask/share_reasoning": 0.8694561719894409,
"mask/share_step_conf": 0.10751700401306152,
"num_tokens": 4663729.0,
"reward": 0.7999786138534546,
"reward_std": 0.19798709452152252,
"rewards/accuracy_reward_step": 0.59375,
"rewards/asymmetric_l2_reward": 0.6810728311538696,
"rewards/final_brier_reward_step": 0.6016968488693237,
"rewards/format_reward_step": 0.9921875,
"step": 15
},
{
"adv/mean_abs_final_conf": 0.4110031723976135,
"adv/mean_abs_reasoning": 0.41789019107818604,
"adv/mean_abs_step_conf": 0.7575955390930176,
"adv/ratio_final_to_reasoning": 0.9835195493275314,
"adv/ratio_step_to_reasoning": 1.8129057711030927,
"adv/std_final_conf": 0.7015405893325806,
"adv/std_reasoning": 0.7013484835624695,
"adv/std_step_conf": 0.9307181239128113,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.49444444444444446,
"calib/avg_num_step_conf": 7.8359375,
"calib/ece": 0.3529233870967743,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -5.555555555569747e-05,
"calib/mean_conf": 0.9900201612903227,
"calib/mu_c": 0.9899999999999998,
"calib/mu_w": 0.9900555555555555,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3529233870967743,
"calib/std_conf": 0.0003168595493044703,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.909328125,
"calib/step_q_c_n": 1280.0,
"calib/step_q_gap": 0.012289557506887094,
"calib/step_q_w": 0.897038567493113,
"calib/step_q_w_n": 726.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3010.0,
"completions/max_terminated_length": 3010.0,
"completions/mean_length": 906.58984375,
"completions/mean_terminated_length": 917.3399658203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 468.0,
"epoch": 0.017066666666666667,
"grad_norm": 0.028787488117814064,
"kl": 0.004323244094848633,
"learning_rate": 4.000000000000001e-06,
"loss": -0.0373,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.01627296581864357,
"mask/share_reasoning": 0.8792263269424438,
"mask/share_step_conf": 0.0927819162607193,
"num_tokens": 5004664.0,
"reward": 0.8178976774215698,
"reward_std": 0.20622055232524872,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/asymmetric_l2_reward": 0.6945248246192932,
"rewards/final_brier_reward_step": 0.6240831017494202,
"rewards/format_reward_step": 0.96875,
"step": 16
},
{
"adv/mean_abs_final_conf": 0.4864518642425537,
"adv/mean_abs_reasoning": 0.4650035500526428,
"adv/mean_abs_step_conf": 0.7391393184661865,
"adv/ratio_final_to_reasoning": 1.0461250547172871,
"adv/ratio_step_to_reasoning": 1.589534786094663,
"adv/std_final_conf": 0.7398475408554077,
"adv/std_reasoning": 0.7394839525222778,
"adv/std_step_conf": 0.9180856347084045,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.4942528735632184,
"calib/avg_num_step_conf": 9.015625,
"calib/ece": 0.28195121951219515,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.991869918699187,
"calib/gap": -0.001034482758620725,
"calib/mean_conf": 0.9892682926829268,
"calib/mu_c": 0.9889655172413793,
"calib/mu_w": 0.99,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.28195121951219515,
"calib/std_conf": 0.008081971475990674,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9013578595317726,
"calib/step_q_c_n": 1495.0,
"calib/step_q_gap": 0.004125387207049358,
"calib/step_q_w": 0.8972324723247233,
"calib/step_q_w_n": 813.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2895.0,
"completions/max_terminated_length": 2895.0,
"completions/mean_length": 839.33203125,
"completions/mean_terminated_length": 862.9276733398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 388.0,
"epoch": 0.018133333333333335,
"grad_norm": 0.025759395211935043,
"kl": 0.005819797515869141,
"learning_rate": 4.25e-06,
"loss": -0.0926,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.01730026863515377,
"mask/share_reasoning": 0.8481937646865845,
"mask/share_step_conf": 0.10716216266155243,
"num_tokens": 5323061.0,
"reward": 0.8808417320251465,
"reward_std": 0.2449316680431366,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/asymmetric_l2_reward": 0.7484192848205566,
"rewards/final_brier_reward_step": 0.6851390600204468,
"rewards/format_reward_step": 0.9609375,
"step": 17
},
{
"adv/mean_abs_final_conf": 0.4163411855697632,
"adv/mean_abs_reasoning": 0.4063420295715332,
"adv/mean_abs_step_conf": 0.7388330101966858,
"adv/ratio_final_to_reasoning": 1.0246077325763558,
"adv/ratio_step_to_reasoning": 1.818253974307672,
"adv/std_final_conf": 0.6988867521286011,
"adv/std_reasoning": 0.7013848423957825,
"adv/std_step_conf": 0.9287502765655518,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 8.3046875,
"calib/ece": 0.46139344262295084,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9959016393442623,
"calib/gap": 0.008448275862069199,
"calib/mean_conf": 0.9859836065573769,
"calib/mu_c": 0.9900000000000001,
"calib/mu_w": 0.9815517241379309,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.46139344262295084,
"calib/std_conf": 0.06325410420255871,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9047178988326848,
"calib/step_q_c_n": 1028.0,
"calib/step_q_gap": 0.07243101358678306,
"calib/step_q_w": 0.8322868852459018,
"calib/step_q_w_n": 1098.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2954.0,
"completions/max_terminated_length": 2954.0,
"completions/mean_length": 848.72265625,
"completions/mean_terminated_length": 876.1007690429688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 473.0,
"epoch": 0.0192,
"grad_norm": 0.02178441919386387,
"kl": 0.006207942962646484,
"learning_rate": 4.5e-06,
"loss": -0.1369,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01709761470556259,
"mask/share_reasoning": 0.852958083152771,
"mask/share_step_conf": 0.0986943170428276,
"num_tokens": 5651054.0,
"reward": 0.7043896317481995,
"reward_std": 0.19021376967430115,
"rewards/accuracy_reward_step": 0.5,
"rewards/asymmetric_l2_reward": 0.6054363250732422,
"rewards/final_brier_reward_step": 0.5127179622650146,
"rewards/format_reward_step": 0.953125,
"step": 18
},
{
"adv/mean_abs_final_conf": 0.40043365955352783,
"adv/mean_abs_reasoning": 0.398581862449646,
"adv/mean_abs_step_conf": 0.7589485049247742,
"adv/ratio_final_to_reasoning": 1.004645964300786,
"adv/ratio_step_to_reasoning": 1.9041220296888304,
"adv/std_final_conf": 0.6625997424125671,
"adv/std_reasoning": 0.6613322496414185,
"adv/std_step_conf": 0.9255523085594177,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5065441475654415,
"calib/avg_num_step_conf": 8.82421875,
"calib/ece": 0.3981376518218622,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.9919028340080972,
"calib/gap": 0.001713685067136983,
"calib/mean_conf": 0.9892307692307691,
"calib/mu_c": 0.9899315068493151,
"calib/mu_w": 0.9882178217821781,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3981376518218622,
"calib/std_conf": 0.008087038200582255,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9048542880932556,
"calib/step_q_c_n": 1201.0,
"calib/step_q_gap": -0.001251572020166014,
"calib/step_q_w": 0.9061058601134216,
"calib/step_q_w_n": 1058.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2829.0,
"completions/max_terminated_length": 2829.0,
"completions/mean_length": 788.2421875,
"completions/mean_terminated_length": 810.4015502929688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 504.0,
"epoch": 0.020266666666666665,
"grad_norm": 0.023767001926898956,
"kl": 0.010579109191894531,
"learning_rate": 4.75e-06,
"loss": -0.1017,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.0177196953445673,
"mask/share_reasoning": 0.8479148149490356,
"mask/share_step_conf": 0.1070217490196228,
"num_tokens": 5957604.0,
"reward": 0.787616491317749,
"reward_std": 0.1928047388792038,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/asymmetric_l2_reward": 0.6887671947479248,
"rewards/final_brier_reward_step": 0.5794343948364258,
"rewards/format_reward_step": 0.96484375,
"step": 19
},
{
"adv/mean_abs_final_conf": 0.41943609714508057,
"adv/mean_abs_reasoning": 0.38557207584381104,
"adv/mean_abs_step_conf": 0.7387959957122803,
"adv/ratio_final_to_reasoning": 1.0878279922817524,
"adv/ratio_step_to_reasoning": 1.9161034784364273,
"adv/std_final_conf": 0.700600802898407,
"adv/std_reasoning": 0.6614826321601868,
"adv/std_step_conf": 0.9205412268638611,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.5085093896713615,
"calib/avg_num_step_conf": 9.4453125,
"calib/ece": 0.3916344537815126,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.9831932773109243,
"calib/gap": 0.001138644366196817,
"calib/mean_conf": 0.9882731092436975,
"calib/mu_c": 0.988732394366197,
"calib/mu_w": 0.9875937500000002,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3916344537815126,
"calib/std_conf": 0.011876685709483406,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9072966692486446,
"calib/step_q_c_n": 1291.0,
"calib/step_q_gap": 0.02412275620516613,
"calib/step_q_w": 0.8831739130434785,
"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": 0.05078125,
"completions/max_length": 2846.0,
"completions/max_terminated_length": 2846.0,
"completions/mean_length": 806.140625,
"completions/mean_terminated_length": 849.2674560546875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 478.0,
"epoch": 0.021333333333333333,
"grad_norm": 0.028085876256227493,
"kl": 0.014958381652832031,
"learning_rate": 5e-06,
"loss": -0.1511,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.017452768981456757,
"mask/share_reasoning": 0.8182758688926697,
"mask/share_step_conf": 0.11349007487297058,
"num_tokens": 6268848.0,
"reward": 0.7662522792816162,
"reward_std": 0.20081782341003418,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/asymmetric_l2_reward": 0.67189621925354,
"rewards/final_brier_reward_step": 0.5637332201004028,
"rewards/format_reward_step": 0.9296875,
"step": 20
},
{
"adv/mean_abs_final_conf": 0.5309414267539978,
"adv/mean_abs_reasoning": 0.5210570096969604,
"adv/mean_abs_step_conf": 0.7509444952011108,
"adv/ratio_final_to_reasoning": 1.0189699339478917,
"adv/ratio_step_to_reasoning": 1.4411944973887785,
"adv/std_final_conf": 0.7589125633239746,
"adv/std_reasoning": 0.7577093839645386,
"adv/std_step_conf": 0.9168322682380676,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5154688418577308,
"calib/avg_num_step_conf": 9.46484375,
"calib/ece": 0.31445783132530125,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9919678714859438,
"calib/gap": 0.000945767195767222,
"calib/mean_conf": 0.9891566265060241,
"calib/mu_c": 0.9894642857142858,
"calib/mu_w": 0.9885185185185186,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.31445783132530125,
"calib/std_conf": 0.008145980250530215,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.90263933376041,
"calib/step_q_c_n": 1561.0,
"calib/step_q_gap": 0.010272744433263759,
"calib/step_q_w": 0.8923665893271462,
"calib/step_q_w_n": 862.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2760.0,
"completions/max_terminated_length": 2760.0,
"completions/mean_length": 831.56640625,
"completions/mean_terminated_length": 854.9437255859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 427.0,
"epoch": 0.0224,
"grad_norm": 0.0251162052154541,
"kl": 0.017168045043945312,
"learning_rate": 4.9722222222222224e-06,
"loss": -0.1365,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.017269905656576157,
"mask/share_reasoning": 0.8425857424736023,
"mask/share_step_conf": 0.11280064284801483,
"num_tokens": 6584689.0,
"reward": 0.8635594844818115,
"reward_std": 0.28787457942962646,
"rewards/accuracy_reward_step": 0.65625,
"rewards/asymmetric_l2_reward": 0.7380008101463318,
"rewards/final_brier_reward_step": 0.6633366942405701,
"rewards/format_reward_step": 0.97265625,
"step": 21
},
{
"adv/mean_abs_final_conf": 0.42734625935554504,
"adv/mean_abs_reasoning": 0.3793871998786926,
"adv/mean_abs_step_conf": 0.727765679359436,
"adv/ratio_final_to_reasoning": 1.1264119071286198,
"adv/ratio_step_to_reasoning": 1.9182662978406648,
"adv/std_final_conf": 0.7142421007156372,
"adv/std_reasoning": 0.6815976500511169,
"adv/std_step_conf": 0.9219359159469604,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.502267076203013,
"calib/avg_num_step_conf": 9.78515625,
"calib/ece": 0.3395102040816327,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9836734693877551,
"calib/gap": 0.0008980547023549335,
"calib/mean_conf": 0.9884897959183674,
"calib/mu_c": 0.9888050314465407,
"calib/mu_w": 0.9879069767441858,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3395102040816327,
"calib/std_conf": 0.011418070977897708,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.9052358168147641,
"calib/step_q_c_n": 1463.0,
"calib/step_q_gap": -0.0003112081372513842,
"calib/step_q_w": 0.9055470249520154,
"calib/step_q_w_n": 1042.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2533.0,
"completions/max_terminated_length": 2533.0,
"completions/mean_length": 796.98828125,
"completions/mean_terminated_length": 829.3861694335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 460.0,
"epoch": 0.023466666666666667,
"grad_norm": 0.05403800308704376,
"kl": 0.022701263427734375,
"learning_rate": 4.944444444444445e-06,
"loss": -0.2069,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.017237115651369095,
"mask/share_reasoning": 0.8285585045814514,
"mask/share_step_conf": 0.11514189094305038,
"num_tokens": 6890534.0,
"reward": 0.8221842646598816,
"reward_std": 0.17867985367774963,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/asymmetric_l2_reward": 0.699776291847229,
"rewards/final_brier_reward_step": 0.6289671659469604,
"rewards/format_reward_step": 0.95703125,
"step": 22
},
{
"adv/mean_abs_final_conf": 0.5448117852210999,
"adv/mean_abs_reasoning": 0.5245258808135986,
"adv/mean_abs_step_conf": 0.7671633958816528,
"adv/ratio_final_to_reasoning": 1.0386747444683482,
"adv/ratio_step_to_reasoning": 1.4625844480575412,
"adv/std_final_conf": 0.7926346063613892,
"adv/std_reasoning": 0.775558352470398,
"adv/std_step_conf": 0.9294148683547974,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.49233573926631446,
"calib/avg_num_step_conf": 9.94140625,
"calib/ece": 0.36487499999999995,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.9958333333333333,
"calib/gap": -0.006642607336855733,
"calib/mean_conf": 0.9857083333333334,
"calib/mu_c": 0.9832450331125826,
"calib/mu_w": 0.9898876404494383,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3607083333333333,
"calib/std_conf": 0.06377302144236928,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8959471050750537,
"calib/step_q_c_n": 1399.0,
"calib/step_q_gap": -0.0044455650820144355,
"calib/step_q_w": 0.9003926701570681,
"calib/step_q_w_n": 1146.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2893.0,
"completions/max_terminated_length": 2893.0,
"completions/mean_length": 843.62109375,
"completions/mean_terminated_length": 885.110595703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 434.0,
"epoch": 0.024533333333333334,
"grad_norm": 0.02799358405172825,
"kl": 0.025606155395507812,
"learning_rate": 4.9166666666666665e-06,
"loss": -0.2192,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.016374429687857628,
"mask/share_reasoning": 0.8253282308578491,
"mask/share_step_conf": 0.11142238974571228,
"num_tokens": 7210437.0,
"reward": 0.7847557067871094,
"reward_std": 0.2819891571998596,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/asymmetric_l2_reward": 0.671172022819519,
"rewards/final_brier_reward_step": 0.5928707122802734,
"rewards/format_reward_step": 0.9375,
"step": 23
},
{
"adv/mean_abs_final_conf": 0.49382612109184265,
"adv/mean_abs_reasoning": 0.4582400619983673,
"adv/mean_abs_step_conf": 0.7413434386253357,
"adv/ratio_final_to_reasoning": 1.0776581142606474,
"adv/ratio_step_to_reasoning": 1.6178058186190998,
"adv/std_final_conf": 0.7374116778373718,
"adv/std_reasoning": 0.7396427989006042,
"adv/std_step_conf": 0.9289793372154236,
"calib/answer_extract_rate": 0.90625,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 10.0234375,
"calib/ece": 0.4468965517241379,
"calib/final_conf_rate": 0.90625,
"calib/format_rate": 0.90234375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -1.1102230246251565e-16,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.99,
"calib/nonempty_final_conf_rate": 0.90625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4468965517241379,
"calib/std_conf": 0.0,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.9000912863070538,
"calib/step_q_c_n": 1205.0,
"calib/step_q_gap": 0.007982542736149956,
"calib/step_q_w": 0.8921087435709039,
"calib/step_q_w_n": 1361.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2990.0,
"completions/max_terminated_length": 2990.0,
"completions/mean_length": 908.16796875,
"completions/mean_terminated_length": 964.6929931640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 507.0,
"epoch": 0.0256,
"grad_norm": 0.01509635429829359,
"kl": 0.02335357666015625,
"learning_rate": 4.888888888888889e-06,
"loss": -0.2972,
"mask/has_final_conf_rate": 0.90625,
"mask/share_final_conf": 0.015077757649123669,
"mask/share_reasoning": 0.8210850954055786,
"mask/share_step_conf": 0.10524344444274902,
"num_tokens": 7547440.0,
"reward": 0.6930545568466187,
"reward_std": 0.24548444151878357,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/asymmetric_l2_reward": 0.6069023609161377,
"rewards/final_brier_reward_step": 0.5003003478050232,
"rewards/format_reward_step": 0.90234375,
"step": 24
},
{
"adv/mean_abs_final_conf": 0.48379945755004883,
"adv/mean_abs_reasoning": 0.4722136855125427,
"adv/mean_abs_step_conf": 0.7505365610122681,
"adv/ratio_final_to_reasoning": 1.0245350196170846,
"adv/ratio_step_to_reasoning": 1.5894002737291117,
"adv/std_final_conf": 0.7559794783592224,
"adv/std_reasoning": 0.7576834559440613,
"adv/std_step_conf": 0.9282037615776062,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 9.83203125,
"calib/ece": 0.3701652892561984,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -1.1102230246251565e-16,
"calib/mean_conf": 0.9900000000000001,
"calib/mu_c": 0.9899999999999998,
"calib/mu_w": 0.9899999999999999,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3701652892561984,
"calib/std_conf": 1.1102230246251565e-16,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8980571030640669,
"calib/step_q_c_n": 1436.0,
"calib/step_q_gap": -0.0016283733466638806,
"calib/step_q_w": 0.8996854764107308,
"calib/step_q_w_n": 1081.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2804.0,
"completions/max_terminated_length": 2804.0,
"completions/mean_length": 800.3515625,
"completions/mean_terminated_length": 836.2857055664062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 475.0,
"epoch": 0.02666666666666667,
"grad_norm": 0.023259948939085007,
"kl": 0.031223297119140625,
"learning_rate": 4.861111111111111e-06,
"loss": -0.2173,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016756581142544746,
"mask/share_reasoning": 0.8256217241287231,
"mask/share_step_conf": 0.11465294659137726,
"num_tokens": 7855554.0,
"reward": 0.7800040245056152,
"reward_std": 0.24854359030723572,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/asymmetric_l2_reward": 0.6599463820457458,
"rewards/final_brier_reward_step": 0.5930304527282715,
"rewards/format_reward_step": 0.9453125,
"step": 25
},
{
"adv/mean_abs_final_conf": 0.48266762495040894,
"adv/mean_abs_reasoning": 0.4587442874908447,
"adv/mean_abs_step_conf": 0.7445145845413208,
"adv/ratio_final_to_reasoning": 1.0521496138740292,
"adv/ratio_step_to_reasoning": 1.6229402846922192,
"adv/std_final_conf": 0.7365694046020508,
"adv/std_reasoning": 0.7395188212394714,
"adv/std_step_conf": 0.92665696144104,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.5117647058823529,
"calib/avg_num_step_conf": 10.3671875,
"calib/ece": 0.3440000000000001,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0004705882352943336,
"calib/mean_conf": 0.9898333333333335,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9895294117647055,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3440000000000001,
"calib/std_conf": 0.002034425935955619,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8974180041870202,
"calib/step_q_c_n": 1433.0,
"calib/step_q_gap": 0.006992123761139557,
"calib/step_q_w": 0.8904258804258807,
"calib/step_q_w_n": 1221.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 3021.0,
"completions/max_terminated_length": 3021.0,
"completions/mean_length": 839.625,
"completions/mean_terminated_length": 888.1983032226562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 581.0,
"epoch": 0.027733333333333332,
"grad_norm": 0.028885742649435997,
"kl": 0.0317230224609375,
"learning_rate": 4.833333333333333e-06,
"loss": -0.2952,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.015826698392629623,
"mask/share_reasoning": 0.8226677179336548,
"mask/share_step_conf": 0.10681814700365067,
"num_tokens": 8175738.0,
"reward": 0.800934374332428,
"reward_std": 0.22288797795772552,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/asymmetric_l2_reward": 0.680953860282898,
"rewards/final_brier_reward_step": 0.6123210787773132,
"rewards/format_reward_step": 0.9375,
"step": 26
},
{
"adv/mean_abs_final_conf": 0.46593204140663147,
"adv/mean_abs_reasoning": 0.4470588266849518,
"adv/mean_abs_step_conf": 0.7534698247909546,
"adv/ratio_final_to_reasoning": 1.0422164010531434,
"adv/ratio_step_to_reasoning": 1.685393017241408,
"adv/std_final_conf": 0.7386137843132019,
"adv/std_reasoning": 0.7393884658813477,
"adv/std_step_conf": 0.9277500510215759,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 10.34375,
"calib/ece": 0.39816326530612245,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 2.220446049250313e-16,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.9900000000000001,
"calib/mu_w": 0.9899999999999999,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.39816326530612245,
"calib/std_conf": 0.0,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8990035842293908,
"calib/step_q_c_n": 1395.0,
"calib/step_q_gap": 0.005883073455248744,
"calib/step_q_w": 0.893120510774142,
"calib/step_q_w_n": 1253.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2508.0,
"completions/max_terminated_length": 2508.0,
"completions/mean_length": 865.66796875,
"completions/mean_terminated_length": 893.5927124023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 560.0,
"epoch": 0.0288,
"grad_norm": 0.017182230949401855,
"kl": 0.03142356872558594,
"learning_rate": 4.805555555555556e-06,
"loss": -0.1524,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01604236289858818,
"mask/share_reasoning": 0.8432894945144653,
"mask/share_step_conf": 0.10941816121339798,
"num_tokens": 8502565.0,
"reward": 0.7764205932617188,
"reward_std": 0.23888157308101654,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/asymmetric_l2_reward": 0.6740305423736572,
"rewards/final_brier_reward_step": 0.5741230249404907,
"rewards/format_reward_step": 0.95703125,
"step": 27
},
{
"adv/mean_abs_final_conf": 0.3599877953529358,
"adv/mean_abs_reasoning": 0.3518277406692505,
"adv/mean_abs_step_conf": 0.7386596202850342,
"adv/ratio_final_to_reasoning": 1.0231933237218962,
"adv/ratio_step_to_reasoning": 2.099492265390864,
"adv/std_final_conf": 0.6624004244804382,
"adv/std_reasoning": 0.6616162061691284,
"adv/std_step_conf": 0.9197353720664978,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 9.81640625,
"calib/ece": 0.2569491525423728,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -4.440892098500626e-16,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9900000000000003,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2569491525423728,
"calib/std_conf": 0.0,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8916149068322982,
"calib/step_q_c_n": 1610.0,
"calib/step_q_gap": 0.009178583465742185,
"calib/step_q_w": 0.882436323366556,
"calib/step_q_w_n": 903.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2878.0,
"completions/max_terminated_length": 2878.0,
"completions/mean_length": 891.28125,
"completions/mean_terminated_length": 931.2979125976562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 502.0,
"epoch": 0.029866666666666666,
"grad_norm": 0.019003966823220253,
"kl": 0.029294967651367188,
"learning_rate": 4.777777777777778e-06,
"loss": -0.2184,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.015266390517354012,
"mask/share_reasoning": 0.8397279381752014,
"mask/share_step_conf": 0.10203687846660614,
"num_tokens": 8837677.0,
"reward": 0.8644207715988159,
"reward_std": 0.2272474765777588,
"rewards/accuracy_reward_step": 0.67578125,
"rewards/asymmetric_l2_reward": 0.7333863973617554,
"rewards/final_brier_reward_step": 0.6767050623893738,
"rewards/format_reward_step": 0.91796875,
"step": 28
},
{
"adv/mean_abs_final_conf": 0.46612828969955444,
"adv/mean_abs_reasoning": 0.4471941590309143,
"adv/mean_abs_step_conf": 0.7449893355369568,
"adv/ratio_final_to_reasoning": 1.0423398434131408,
"adv/ratio_step_to_reasoning": 1.6659191997305494,
"adv/std_final_conf": 0.7390708327293396,
"adv/std_reasoning": 0.7395082116127014,
"adv/std_step_conf": 0.9280605912208557,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.49645390070921985,
"calib/avg_num_step_conf": 9.35546875,
"calib/ece": 0.40727272727272734,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00014184397163097273,
"calib/mean_conf": 0.9899173553719008,
"calib/mu_c": 0.9898581560283688,
"calib/mu_w": 0.9899999999999998,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.40727272727272734,
"calib/std_conf": 0.0012829896443190113,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8957519379844963,
"calib/step_q_c_n": 1290.0,
"calib/step_q_gap": 0.0007293135501070802,
"calib/step_q_w": 0.8950226244343892,
"calib/step_q_w_n": 1105.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2735.0,
"completions/max_terminated_length": 2735.0,
"completions/mean_length": 947.33984375,
"completions/mean_terminated_length": 985.8495483398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 530.0,
"epoch": 0.030933333333333334,
"grad_norm": 0.020408691838383675,
"kl": 0.0307464599609375,
"learning_rate": 4.75e-06,
"loss": -0.2097,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.014833498746156693,
"mask/share_reasoning": 0.8490548133850098,
"mask/share_step_conf": 0.09704914689064026,
"num_tokens": 9187324.0,
"reward": 0.7522528171539307,
"reward_std": 0.23636189103126526,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/asymmetric_l2_reward": 0.6475716233253479,
"rewards/final_brier_reward_step": 0.5584964752197266,
"rewards/format_reward_step": 0.94140625,
"step": 29
},
{
"adv/mean_abs_final_conf": 0.6084608435630798,
"adv/mean_abs_reasoning": 0.5937298536300659,
"adv/mean_abs_step_conf": 0.77057945728302,
"adv/ratio_final_to_reasoning": 1.0248109301611643,
"adv/ratio_step_to_reasoning": 1.2978620707240762,
"adv/std_final_conf": 0.8092952370643616,
"adv/std_reasoning": 0.810136616230011,
"adv/std_step_conf": 0.9337138533592224,
"calib/answer_extract_rate": 0.89453125,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 10.41015625,
"calib/ece": 0.3611790393013099,
"calib/final_conf_rate": 0.89453125,
"calib/format_rate": 0.89453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 2.220446049250313e-16,
"calib/mean_conf": 0.9899999999999999,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9899999999999998,
"calib/nonempty_final_conf_rate": 0.89453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3611790393013099,
"calib/std_conf": 1.1102230246251565e-16,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8935240753663644,
"calib/step_q_c_n": 1433.0,
"calib/step_q_gap": 0.01962797147026052,
"calib/step_q_w": 0.8738961038961038,
"calib/step_q_w_n": 1232.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.07421875,
"completions/max_length": 2900.0,
"completions/max_terminated_length": 2900.0,
"completions/mean_length": 886.84375,
"completions/mean_terminated_length": 957.9408569335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 513.0,
"epoch": 0.032,
"grad_norm": 0.02599792182445526,
"kl": 0.031978607177734375,
"learning_rate": 4.722222222222222e-06,
"loss": -0.2959,
"mask/has_final_conf_rate": 0.89453125,
"mask/share_final_conf": 0.014466993510723114,
"mask/share_reasoning": 0.808133602142334,
"mask/share_step_conf": 0.1031806617975235,
"num_tokens": 9521340.0,
"reward": 0.7366952896118164,
"reward_std": 0.29571014642715454,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.6129330992698669,
"rewards/final_brier_reward_step": 0.5690511465072632,
"rewards/format_reward_step": 0.89453125,
"step": 30
},
{
"adv/mean_abs_final_conf": 0.5071862936019897,
"adv/mean_abs_reasoning": 0.49132299423217773,
"adv/mean_abs_step_conf": 0.7129656672477722,
"adv/ratio_final_to_reasoning": 1.0322869060801898,
"adv/ratio_step_to_reasoning": 1.45111398330129,
"adv/std_final_conf": 0.755790114402771,
"adv/std_reasoning": 0.757742166519165,
"adv/std_step_conf": 0.919549286365509,
"calib/answer_extract_rate": 0.8984375,
"calib/auroc": 0.5,
"calib/avg_num_step_conf": 10.78125,
"calib/ece": 0.4204347826086956,
"calib/final_conf_rate": 0.8984375,
"calib/format_rate": 0.8984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 1.1102230246251565e-16,
"calib/mean_conf": 0.99,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9899999999999999,
"calib/nonempty_final_conf_rate": 0.8984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4204347826086956,
"calib/std_conf": 0.0,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8882582123758594,
"calib/step_q_c_n": 1309.0,
"calib/step_q_gap": 0.02449529025318531,
"calib/step_q_w": 0.8637629221226741,
"calib/step_q_w_n": 1451.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.07421875,
"completions/max_length": 3061.0,
"completions/max_terminated_length": 3061.0,
"completions/mean_length": 974.890625,
"completions/mean_terminated_length": 1053.04638671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 574.0,
"epoch": 0.03306666666666667,
"grad_norm": 0.023567797616124153,
"kl": 0.033023834228515625,
"learning_rate": 4.694444444444445e-06,
"loss": -0.2323,
"mask/has_final_conf_rate": 0.8984375,
"mask/share_final_conf": 0.013692351058125496,
"mask/share_reasoning": 0.8139607906341553,
"mask/share_step_conf": 0.09812812507152557,
"num_tokens": 9876824.0,
"reward": 0.702761173248291,
"reward_std": 0.2406536340713501,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/asymmetric_l2_reward": 0.6041278839111328,
"rewards/final_brier_reward_step": 0.519363284111023,
"rewards/format_reward_step": 0.8984375,
"step": 31
},
{
"adv/mean_abs_final_conf": 0.41383790969848633,
"adv/mean_abs_reasoning": 0.4185928702354431,
"adv/mean_abs_step_conf": 0.7600575685501099,
"adv/ratio_final_to_reasoning": 0.9886406079150791,
"adv/ratio_step_to_reasoning": 1.8157441815064967,
"adv/std_final_conf": 0.6787320375442505,
"adv/std_reasoning": 0.6817558407783508,
"adv/std_step_conf": 0.9247509837150574,
"calib/answer_extract_rate": 0.8984375,
"calib/auroc": 0.50185785437584,
"calib/avg_num_step_conf": 10.3359375,
"calib/ece": 0.3854347826086957,
"calib/final_conf_rate": 0.8984375,
"calib/format_rate": 0.8984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 3.952881650670648e-06,
"calib/mean_conf": 0.9897826086956522,
"calib/mu_c": 0.9897841726618704,
"calib/mu_w": 0.9897802197802197,
"calib/nonempty_final_conf_rate": 0.8984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3854347826086957,
"calib/std_conf": 0.0023674701547933895,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8834646840148699,
"calib/step_q_c_n": 1345.0,
"calib/step_q_gap": 0.03677752029465464,
"calib/step_q_w": 0.8466871637202152,
"calib/step_q_w_n": 1301.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.08203125,
"completions/max_length": 2773.0,
"completions/max_terminated_length": 2773.0,
"completions/mean_length": 847.02734375,
"completions/mean_terminated_length": 922.7191162109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 507.0,
"epoch": 0.034133333333333335,
"grad_norm": 0.024691706523299217,
"kl": 0.036327362060546875,
"learning_rate": 4.666666666666667e-06,
"loss": -0.2709,
"mask/has_final_conf_rate": 0.8984375,
"mask/share_final_conf": 0.015017786994576454,
"mask/share_reasoning": 0.7982213497161865,
"mask/share_step_conf": 0.10472959280014038,
"num_tokens": 10200367.0,
"reward": 0.7324390411376953,
"reward_std": 0.18825767934322357,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/asymmetric_l2_reward": 0.6264613270759583,
"rewards/final_brier_reward_step": 0.5501354932785034,
"rewards/format_reward_step": 0.8984375,
"step": 32
},
{
"adv/mean_abs_final_conf": 0.5061349868774414,
"adv/mean_abs_reasoning": 0.47032085061073303,
"adv/mean_abs_step_conf": 0.7482904195785522,
"adv/ratio_final_to_reasoning": 1.076148306459733,
"adv/ratio_step_to_reasoning": 1.5910211478118885,
"adv/std_final_conf": 0.7739526033401489,
"adv/std_reasoning": 0.7576538920402527,
"adv/std_step_conf": 0.9329660534858704,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.4927536231884058,
"calib/avg_num_step_conf": 9.43359375,
"calib/ece": 0.43089068825910926,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.9959514170040485,
"calib/gap": -0.0007246376811593791,
"calib/mean_conf": 0.9895951417004049,
"calib/mu_c": 0.9892753623188405,
"calib/mu_w": 0.9899999999999999,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43089068825910926,
"calib/std_conf": 0.005747562120068867,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8886196319018405,
"calib/step_q_c_n": 1304.0,
"calib/step_q_gap": 0.007719541892839521,
"calib/step_q_w": 0.8809000900090009,
"calib/step_q_w_n": 1111.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2721.0,
"completions/max_terminated_length": 2721.0,
"completions/mean_length": 862.5625,
"completions/mean_terminated_length": 883.2640380859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 474.0,
"epoch": 0.0352,
"grad_norm": 0.02115471102297306,
"kl": 0.04196929931640625,
"learning_rate": 4.638888888888889e-06,
"loss": -0.1091,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.016444312408566475,
"mask/share_reasoning": 0.84872967004776,
"mask/share_step_conf": 0.11138851940631866,
"num_tokens": 10528055.0,
"reward": 0.7556190490722656,
"reward_std": 0.2505682706832886,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/asymmetric_l2_reward": 0.6630150079727173,
"rewards/final_brier_reward_step": 0.5474417805671692,
"rewards/format_reward_step": 0.96484375,
"step": 33
},
{
"adv/mean_abs_final_conf": 0.6398830413818359,
"adv/mean_abs_reasoning": 0.6042325496673584,
"adv/mean_abs_step_conf": 0.7579690217971802,
"adv/ratio_final_to_reasoning": 1.0590012764689751,
"adv/ratio_step_to_reasoning": 1.2544326223644466,
"adv/std_final_conf": 0.8544065356254578,
"adv/std_reasoning": 0.8431804180145264,
"adv/std_step_conf": 0.9319725036621094,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.49333333333333335,
"calib/avg_num_step_conf": 10.87109375,
"calib/ece": 0.3567510548523207,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.0005333333333330526,
"calib/mean_conf": 0.9896624472573841,
"calib/mu_c": 0.9894666666666666,
"calib/mu_w": 0.9899999999999997,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3567510548523207,
"calib/std_conf": 0.0036589845381736407,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8843560872624913,
"calib/step_q_c_n": 1421.0,
"calib/step_q_gap": 0.011830389758820115,
"calib/step_q_w": 0.8725256975036711,
"calib/step_q_w_n": 1362.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.06640625,
"completions/max_length": 2696.0,
"completions/max_terminated_length": 2696.0,
"completions/mean_length": 819.0859375,
"completions/mean_terminated_length": 877.3472290039062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 521.0,
"epoch": 0.03626666666666667,
"grad_norm": 0.024255702272057533,
"kl": 0.039142608642578125,
"learning_rate": 4.611111111111112e-06,
"loss": -0.3183,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.015937991440296173,
"mask/share_reasoning": 0.8053004741668701,
"mask/share_step_conf": 0.11235527694225311,
"num_tokens": 10842853.0,
"reward": 0.7982138395309448,
"reward_std": 0.3219480514526367,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/asymmetric_l2_reward": 0.7014608979225159,
"rewards/final_brier_reward_step": 0.5926230549812317,
"rewards/format_reward_step": 0.92578125,
"step": 34
},
{
"adv/mean_abs_final_conf": 0.5378986597061157,
"adv/mean_abs_reasoning": 0.5105568170547485,
"adv/mean_abs_step_conf": 0.7341263294219971,
"adv/ratio_final_to_reasoning": 1.053552987127063,
"adv/ratio_step_to_reasoning": 1.4378935015635577,
"adv/std_final_conf": 0.792346179485321,
"adv/std_reasoning": 0.7756903171539307,
"adv/std_step_conf": 0.9303335547447205,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.5099009900990099,
"calib/avg_num_step_conf": 10.52734375,
"calib/ece": 0.4197021276595744,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00019801980198030922,
"calib/mean_conf": 0.9899148936170212,
"calib/mu_c": 0.99,
"calib/mu_w": 0.9898019801980197,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4197021276595744,
"calib/std_conf": 0.000918597155103103,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8725444264943457,
"calib/step_q_c_n": 1238.0,
"calib/step_q_gap": 0.027980253536212474,
"calib/step_q_w": 0.8445641729581332,
"calib/step_q_w_n": 1457.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0703125,
"completions/max_length": 2928.0,
"completions/max_terminated_length": 2928.0,
"completions/mean_length": 894.0,
"completions/mean_terminated_length": 961.6134643554688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 543.0,
"epoch": 0.037333333333333336,
"grad_norm": 0.022384105250239372,
"kl": 0.038501739501953125,
"learning_rate": 4.583333333333333e-06,
"loss": -0.335,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.014649630524218082,
"mask/share_reasoning": 0.8177850842475891,
"mask/share_step_conf": 0.09725277125835419,
"num_tokens": 11180973.0,
"reward": 0.7150118350982666,
"reward_std": 0.280878484249115,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/asymmetric_l2_reward": 0.6095709204673767,
"rewards/final_brier_reward_step": 0.5313901901245117,
"rewards/format_reward_step": 0.91796875,
"step": 35
},
{
"adv/mean_abs_final_conf": 0.3458389937877655,
"adv/mean_abs_reasoning": 0.3302667737007141,
"adv/mean_abs_step_conf": 0.7599014639854431,
"adv/ratio_final_to_reasoning": 1.0471504290684803,
"adv/ratio_step_to_reasoning": 2.300871672528771,
"adv/std_final_conf": 0.6069803833961487,
"adv/std_reasoning": 0.5962504148483276,
"adv/std_step_conf": 0.9211536049842834,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.49728260869565216,
"calib/avg_num_step_conf": 11.12109375,
"calib/ece": 0.2419105691056911,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00016304347826123777,
"calib/mean_conf": 0.9898780487804878,
"calib/mu_c": 0.9898369565217391,
"calib/mu_w": 0.9900000000000003,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2419105691056911,
"calib/std_conf": 0.0019088385173778714,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8667872340425532,
"calib/step_q_c_n": 1880.0,
"calib/step_q_gap": -0.013274813527250351,
"calib/step_q_w": 0.8800620475698036,
"calib/step_q_w_n": 967.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2549.0,
"completions/max_terminated_length": 2549.0,
"completions/mean_length": 861.90625,
"completions/mean_terminated_length": 882.592041015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 449.0,
"epoch": 0.0384,
"grad_norm": 684.32421875,
"kl": 3056.043258666992,
"learning_rate": 4.555555555555556e-06,
"loss": 33.7113,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.016752172261476517,
"mask/share_reasoning": 0.8371760249137878,
"mask/share_step_conf": 0.12263430655002594,
"num_tokens": 11504333.0,
"reward": 0.917310357093811,
"reward_std": 0.16187608242034912,
"rewards/accuracy_reward_step": 0.71875,
"rewards/asymmetric_l2_reward": 0.7751913666725159,
"rewards/final_brier_reward_step": 0.7234917879104614,
"rewards/format_reward_step": 0.9609375,
"step": 36
},
{
"adv/mean_abs_final_conf": 0.3826274275779724,
"adv/mean_abs_reasoning": 0.3682462275028229,
"adv/mean_abs_step_conf": 0.7657290101051331,
"adv/ratio_final_to_reasoning": 1.0390532176600216,
"adv/ratio_step_to_reasoning": 2.0793940383252485,
"adv/std_final_conf": 0.6385846734046936,
"adv/std_reasoning": 0.6405808329582214,
"adv/std_step_conf": 0.9296008348464966,
"calib/answer_extract_rate": 0.91015625,
"calib/auroc": 0.5043478260869565,
"calib/avg_num_step_conf": 10.8515625,
"calib/ece": 0.48317596566523624,
"calib/final_conf_rate": 0.91015625,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.9957081545064378,
"calib/gap": 0.0007826086956519429,
"calib/mean_conf": 0.9896137339055796,
"calib/mu_c": 0.9899999999999998,
"calib/mu_w": 0.9892173913043478,
"calib/nonempty_final_conf_rate": 0.91015625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.48317596566523624,
"calib/std_conf": 0.005883429867191,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8636806722689075,
"calib/step_q_c_n": 1190.0,
"calib/step_q_gap": 0.02601064707999068,
"calib/step_q_w": 0.8376700251889169,
"calib/step_q_w_n": 1588.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0703125,
"completions/max_length": 3059.0,
"completions/max_terminated_length": 3059.0,
"completions/mean_length": 896.26171875,
"completions/mean_terminated_length": 964.0462646484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 532.0,
"epoch": 0.039466666666666664,
"grad_norm": 0.023465532809495926,
"kl": 0.043636322021484375,
"learning_rate": 4.527777777777778e-06,
"loss": -0.2311,
"mask/has_final_conf_rate": 0.91015625,
"mask/share_final_conf": 0.015032488852739334,
"mask/share_reasoning": 0.807826042175293,
"mask/share_step_conf": 0.106828972697258,
"num_tokens": 11840872.0,
"reward": 0.6838352680206299,
"reward_std": 0.19802510738372803,
"rewards/accuracy_reward_step": 0.4609375,
"rewards/asymmetric_l2_reward": 0.6229565143585205,
"rewards/final_brier_reward_step": 0.4704952836036682,
"rewards/format_reward_step": 0.91015625,
"step": 37
},
{
"adv/mean_abs_final_conf": 0.5782315135002136,
"adv/mean_abs_reasoning": 0.5265728235244751,
"adv/mean_abs_step_conf": 0.7494425773620605,
"adv/ratio_final_to_reasoning": 1.0981036006187612,
"adv/ratio_step_to_reasoning": 1.4232458339681604,
"adv/std_final_conf": 0.8226618766784668,
"adv/std_reasoning": 0.7931339740753174,
"adv/std_step_conf": 0.9324886798858643,
"calib/answer_extract_rate": 0.890625,
"calib/auroc": 0.5159055118110236,
"calib/avg_num_step_conf": 11.44140625,
"calib/ece": 0.42969162995594723,
"calib/final_conf_rate": 0.88671875,
"calib/format_rate": 0.88671875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0006488188976377929,
"calib/mean_conf": 0.9891629955947138,
"calib/mu_c": 0.9894488188976377,
"calib/mu_w": 0.9887999999999999,
"calib/nonempty_final_conf_rate": 0.88671875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42969162995594723,
"calib/std_conf": 0.006133955886290556,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8472820109976434,
"calib/step_q_c_n": 1273.0,
"calib/step_q_gap": 0.021545296021798,
"calib/step_q_w": 0.8257367149758454,
"calib/step_q_w_n": 1656.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.07421875,
"completions/max_length": 3054.0,
"completions/max_terminated_length": 3054.0,
"completions/mean_length": 911.984375,
"completions/mean_terminated_length": 985.0969848632812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 585.0,
"epoch": 0.04053333333333333,
"grad_norm": 0.029373837634921074,
"kl": 0.053211212158203125,
"learning_rate": 4.5e-06,
"loss": -0.1981,
"mask/has_final_conf_rate": 0.88671875,
"mask/share_final_conf": 0.014381938613951206,
"mask/share_reasoning": 0.805593729019165,
"mask/share_step_conf": 0.10580561310052872,
"num_tokens": 12181228.0,
"reward": 0.692814826965332,
"reward_std": 0.26417291164398193,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/asymmetric_l2_reward": 0.604360818862915,
"rewards/final_brier_reward_step": 0.5047062635421753,
"rewards/format_reward_step": 0.88671875,
"step": 38
},
{
"adv/mean_abs_final_conf": 0.6115273237228394,
"adv/mean_abs_reasoning": 0.5294957160949707,
"adv/mean_abs_step_conf": 0.7513197064399719,
"adv/ratio_final_to_reasoning": 1.1549240251325383,
"adv/ratio_step_to_reasoning": 1.4189344381876259,
"adv/std_final_conf": 0.8244526386260986,
"adv/std_reasoning": 0.7757487297058105,
"adv/std_step_conf": 0.9348276853561401,
"calib/answer_extract_rate": 0.89453125,
"calib/auroc": 0.529126213592233,
"calib/avg_num_step_conf": 13.20703125,
"calib/ece": 0.4389082969432314,
"calib/final_conf_rate": 0.89453125,
"calib/format_rate": 0.89453125,
"calib/frac_conf_gt_0.9": 0.9956331877729258,
"calib/gap": 0.001941747572815733,
"calib/mean_conf": 0.9891266375545851,
"calib/mu_c": 0.9899999999999999,
"calib/mu_w": 0.9880582524271841,
"calib/nonempty_final_conf_rate": 0.89453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4389082969432314,
"calib/std_conf": 0.006938698363682519,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8546702870442204,
"calib/step_q_c_n": 1289.0,
"calib/step_q_gap": 0.03671139603083606,
"calib/step_q_w": 0.8179588910133844,
"calib/step_q_w_n": 2092.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.09375,
"completions/max_length": 2726.0,
"completions/max_terminated_length": 2726.0,
"completions/mean_length": 853.80078125,
"completions/mean_terminated_length": 942.125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 563.0,
"epoch": 0.0416,
"grad_norm": 0.02601790241897106,
"kl": 0.05804443359375,
"learning_rate": 4.472222222222223e-06,
"loss": -0.3918,
"mask/has_final_conf_rate": 0.89453125,
"mask/share_final_conf": 0.014618676155805588,
"mask/share_reasoning": 0.7844524383544922,
"mask/share_step_conf": 0.10717888176441193,
"num_tokens": 12505889.0,
"reward": 0.7072185277938843,
"reward_std": 0.2773115038871765,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/asymmetric_l2_reward": 0.635445237159729,
"rewards/final_brier_reward_step": 0.5016480088233948,
"rewards/format_reward_step": 0.89453125,
"step": 39
},
{
"adv/mean_abs_final_conf": 0.5351214408874512,
"adv/mean_abs_reasoning": 0.5108170509338379,
"adv/mean_abs_step_conf": 0.7385743260383606,
"adv/ratio_final_to_reasoning": 1.0475794414246389,
"adv/ratio_step_to_reasoning": 1.4458685838465135,
"adv/std_final_conf": 0.7758241295814514,
"adv/std_reasoning": 0.7756544947624207,
"adv/std_step_conf": 0.9328163862228394,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.5075821104122991,
"calib/avg_num_step_conf": 10.9765625,
"calib/ece": 0.42883817427385895,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0004116002795245821,
"calib/mean_conf": 0.9890041493775934,
"calib/mu_c": 0.989185185185185,
"calib/mu_w": 0.9887735849056604,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42883817427385895,
"calib/std_conf": 0.005819196965396921,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8614264264264265,
"calib/step_q_c_n": 1332.0,
"calib/step_q_gap": 0.015364180147671491,
"calib/step_q_w": 0.846062246278755,
"calib/step_q_w_n": 1478.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 3031.0,
"completions/max_terminated_length": 3031.0,
"completions/mean_length": 921.484375,
"completions/mean_terminated_length": 962.8571166992188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 472.0,
"epoch": 0.042666666666666665,
"grad_norm": 0.022954532876610756,
"kl": 0.06566619873046875,
"learning_rate": 4.444444444444444e-06,
"loss": -0.187,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.015071472153067589,
"mask/share_reasoning": 0.8308209180831909,
"mask/share_step_conf": 0.11113885045051575,
"num_tokens": 12848549.0,
"reward": 0.7284168004989624,
"reward_std": 0.27028942108154297,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/asymmetric_l2_reward": 0.6276761889457703,
"rewards/final_brier_reward_step": 0.5361886620521545,
"rewards/format_reward_step": 0.9375,
"step": 40
},
{
"adv/mean_abs_final_conf": 0.5005245804786682,
"adv/mean_abs_reasoning": 0.4871232509613037,
"adv/mean_abs_step_conf": 0.7367668747901917,
"adv/ratio_final_to_reasoning": 1.0275111678428774,
"adv/ratio_step_to_reasoning": 1.5124855431068702,
"adv/std_final_conf": 0.7579271197319031,
"adv/std_reasoning": 0.757828950881958,
"adv/std_step_conf": 0.9302553534507751,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.5046875000000001,
"calib/avg_num_step_conf": 11.92578125,
"calib/ece": 0.1963223140495869,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.00011250000000007088,
"calib/mean_conf": 0.989710743801653,
"calib/mu_c": 0.9896875000000002,
"calib/mu_w": 0.9898000000000002,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1963223140495869,
"calib/std_conf": 0.002787036153540043,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8462296650717704,
"calib/step_q_c_n": 2090.0,
"calib/step_q_gap": 0.03025874087654723,
"calib/step_q_w": 0.8159709241952232,
"calib/step_q_w_n": 963.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 3031.0,
"completions/max_terminated_length": 3031.0,
"completions/mean_length": 869.15234375,
"completions/mean_terminated_length": 908.1754760742188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 477.0,
"epoch": 0.04373333333333333,
"grad_norm": 0.015356858260929585,
"kl": 0.07578277587890625,
"learning_rate": 4.416666666666667e-06,
"loss": -0.1941,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.015964847058057785,
"mask/share_reasoning": 0.8180328011512756,
"mask/share_step_conf": 0.12303359061479568,
"num_tokens": 13178300.0,
"reward": 0.9321384429931641,
"reward_std": 0.26598113775253296,
"rewards/accuracy_reward_step": 0.75,
"rewards/asymmetric_l2_reward": 0.7713373899459839,
"rewards/final_brier_reward_step": 0.7538769245147705,
"rewards/format_reward_step": 0.9453125,
"step": 41
},
{
"adv/mean_abs_final_conf": 0.3991064131259918,
"adv/mean_abs_reasoning": 0.40221142768859863,
"adv/mean_abs_step_conf": 0.7366739511489868,
"adv/ratio_final_to_reasoning": 0.9922801433553231,
"adv/ratio_step_to_reasoning": 1.8315589772833027,
"adv/std_final_conf": 0.6996758580207825,
"adv/std_reasoning": 0.7014210224151611,
"adv/std_step_conf": 0.9316682815551758,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.5180704898446834,
"calib/avg_num_step_conf": 12.43359375,
"calib/ece": 0.38135021097046407,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.9957805907172996,
"calib/gap": 0.001980286738351489,
"calib/mean_conf": 0.9889451476793248,
"calib/mu_c": 0.9897222222222223,
"calib/mu_w": 0.9877419354838708,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.38135021097046407,
"calib/std_conf": 0.007750401356105529,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8366325224071705,
"calib/step_q_c_n": 1562.0,
"calib/step_q_gap": -0.0004865399000473225,
"calib/step_q_w": 0.8371190623072178,
"calib/step_q_w_n": 1621.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2812.0,
"completions/max_terminated_length": 2812.0,
"completions/mean_length": 860.84765625,
"completions/mean_terminated_length": 903.1843872070312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 518.0,
"epoch": 0.0448,
"grad_norm": 0.018000788986682892,
"kl": 0.0889892578125,
"learning_rate": 4.388888888888889e-06,
"loss": -0.1293,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.016016989946365356,
"mask/share_reasoning": 0.8085422515869141,
"mask/share_step_conf": 0.1285657286643982,
"num_tokens": 13503045.0,
"reward": 0.7686011791229248,
"reward_std": 0.19247505068778992,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.6683084964752197,
"rewards/final_brier_reward_step": 0.5712375044822693,
"rewards/format_reward_step": 0.92578125,
"step": 42
},
{
"adv/mean_abs_final_conf": 0.5601272583007812,
"adv/mean_abs_reasoning": 0.5113959908485413,
"adv/mean_abs_step_conf": 0.7792887687683105,
"adv/ratio_final_to_reasoning": 1.0952906716601003,
"adv/ratio_step_to_reasoning": 1.5238460658936028,
"adv/std_final_conf": 0.7849826812744141,
"adv/std_reasoning": 0.7578847408294678,
"adv/std_step_conf": 0.9340676069259644,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5139602319801161,
"calib/avg_num_step_conf": 12.08203125,
"calib/ece": 0.2834439834024896,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.983402489626556,
"calib/gap": -0.001004971002485311,
"calib/mean_conf": 0.9871784232365146,
"calib/mu_c": 0.9868823529411763,
"calib/mu_w": 0.9878873239436616,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2826141078838174,
"calib/std_conf": 0.01650868346897965,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8531187010078388,
"calib/step_q_c_n": 1786.0,
"calib/step_q_gap": 0.05383790529246024,
"calib/step_q_w": 0.7992807957153786,
"calib/step_q_w_n": 1307.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2688.0,
"completions/max_terminated_length": 2688.0,
"completions/mean_length": 865.765625,
"completions/mean_terminated_length": 900.9592895507812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 501.0,
"epoch": 0.04586666666666667,
"grad_norm": 0.018080631271004677,
"kl": 0.095123291015625,
"learning_rate": 4.361111111111112e-06,
"loss": -0.1282,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016060657799243927,
"mask/share_reasoning": 0.8252454996109009,
"mask/share_step_conf": 0.11963135004043579,
"num_tokens": 13829905.0,
"reward": 0.859089732170105,
"reward_std": 0.2762579023838043,
"rewards/accuracy_reward_step": 0.6640625,
"rewards/asymmetric_l2_reward": 0.7275747060775757,
"rewards/final_brier_reward_step": 0.6702921390533447,
"rewards/format_reward_step": 0.9375,
"step": 43
},
{
"adv/mean_abs_final_conf": 0.6635046005249023,
"adv/mean_abs_reasoning": 0.5385121703147888,
"adv/mean_abs_step_conf": 0.7993447780609131,
"adv/ratio_final_to_reasoning": 1.2321069738072008,
"adv/ratio_step_to_reasoning": 1.4843578699319901,
"adv/std_final_conf": 0.8506072163581848,
"adv/std_reasoning": 0.7756995558738708,
"adv/std_step_conf": 0.9337899088859558,
"calib/answer_extract_rate": 0.890625,
"calib/auroc": 0.4961926961926962,
"calib/avg_num_step_conf": 13.08984375,
"calib/ece": 0.5044933920704846,
"calib/final_conf_rate": 0.88671875,
"calib/format_rate": 0.87890625,
"calib/frac_conf_gt_0.9": 0.986784140969163,
"calib/gap": -0.004595959595959709,
"calib/mean_conf": 0.9855506607929516,
"calib/mu_c": 0.9831818181818179,
"calib/mu_w": 0.9877777777777776,
"calib/nonempty_final_conf_rate": 0.88671875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.5027312775330397,
"calib/std_conf": 0.027861059256808065,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8404388185654009,
"calib/step_q_c_n": 1185.0,
"calib/step_q_gap": 0.005138726229297452,
"calib/step_q_w": 0.8353000923361035,
"calib/step_q_w_n": 2166.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0859375,
"completions/max_length": 2944.0,
"completions/max_terminated_length": 2944.0,
"completions/mean_length": 913.73828125,
"completions/mean_terminated_length": 999.6453247070312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 517.0,
"epoch": 0.046933333333333334,
"grad_norm": 0.029077230021357536,
"kl": 0.0980377197265625,
"learning_rate": 4.333333333333334e-06,
"loss": -0.3187,
"mask/has_final_conf_rate": 0.88671875,
"mask/share_final_conf": 0.014052234590053558,
"mask/share_reasoning": 0.7942177057266235,
"mask/share_step_conf": 0.10579255223274231,
"num_tokens": 14170142.0,
"reward": 0.6271952390670776,
"reward_std": 0.26996076107025146,
"rewards/accuracy_reward_step": 0.43359375,
"rewards/asymmetric_l2_reward": 0.5590915679931641,
"rewards/final_brier_reward_step": 0.4327988028526306,
"rewards/format_reward_step": 0.87890625,
"step": 44
},
{
"adv/mean_abs_final_conf": 0.5559475421905518,
"adv/mean_abs_reasoning": 0.5480928421020508,
"adv/mean_abs_step_conf": 0.751115083694458,
"adv/ratio_final_to_reasoning": 1.014330966371275,
"adv/ratio_step_to_reasoning": 1.3704157872483327,
"adv/std_final_conf": 0.8367322683334351,
"adv/std_reasoning": 0.8102704882621765,
"adv/std_step_conf": 0.9335227608680725,
"calib/answer_extract_rate": 0.83984375,
"calib/auroc": 0.5272909821102592,
"calib/avg_num_step_conf": 15.546875,
"calib/ece": 0.373860465116279,
"calib/final_conf_rate": 0.83984375,
"calib/format_rate": 0.83203125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0013453815261041946,
"calib/mean_conf": 0.9878139534883721,
"calib/mu_c": 0.9883333333333332,
"calib/mu_w": 0.986987951807229,
"calib/nonempty_final_conf_rate": 0.83984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.373860465116279,
"calib/std_conf": 0.008090349732152157,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8533936955063716,
"calib/step_q_c_n": 1491.0,
"calib/step_q_gap": 0.029629131424410815,
"calib/step_q_w": 0.8237645640819607,
"calib/step_q_w_n": 2489.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.12109375,
"completions/max_length": 3026.0,
"completions/max_terminated_length": 3026.0,
"completions/mean_length": 871.33203125,
"completions/mean_terminated_length": 991.3822631835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 479.0,
"epoch": 0.048,
"grad_norm": 0.035998519510030746,
"kl": 0.10224151611328125,
"learning_rate": 4.305555555555556e-06,
"loss": -0.4043,
"mask/has_final_conf_rate": 0.83984375,
"mask/share_final_conf": 0.013904090970754623,
"mask/share_reasoning": 0.7514083385467529,
"mask/share_step_conf": 0.11359383165836334,
"num_tokens": 14498251.0,
"reward": 0.6943749189376831,
"reward_std": 0.2721937298774719,
"rewards/accuracy_reward_step": 0.515625,
"rewards/asymmetric_l2_reward": 0.599319338798523,
"rewards/final_brier_reward_step": 0.5198991894721985,
"rewards/format_reward_step": 0.83203125,
"step": 45
},
{
"adv/mean_abs_final_conf": 0.5597379207611084,
"adv/mean_abs_reasoning": 0.5508193373680115,
"adv/mean_abs_step_conf": 0.7796428203582764,
"adv/ratio_final_to_reasoning": 1.0161914856433922,
"adv/ratio_step_to_reasoning": 1.4154238376663675,
"adv/std_final_conf": 0.806285560131073,
"adv/std_reasoning": 0.7932658195495605,
"adv/std_step_conf": 0.9322901368141174,
"calib/answer_extract_rate": 0.85546875,
"calib/auroc": 0.5066286289645913,
"calib/avg_num_step_conf": 14.1875,
"calib/ece": 0.44771689497716904,
"calib/final_conf_rate": 0.85546875,
"calib/format_rate": 0.84765625,
"calib/frac_conf_gt_0.9": 0.9908675799086758,
"calib/gap": 0.0003582815908710346,
"calib/mean_conf": 0.9865296803652969,
"calib/mu_c": 0.9866949152542372,
"calib/mu_w": 0.9863366336633662,
"calib/nonempty_final_conf_rate": 0.85546875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44771689497716904,
"calib/std_conf": 0.01245311885851197,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8513843774168601,
"calib/step_q_c_n": 1293.0,
"calib/step_q_gap": 0.0327140054630336,
"calib/step_q_w": 0.8186703719538265,
"calib/step_q_w_n": 2339.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.09765625,
"completions/max_length": 2961.0,
"completions/max_terminated_length": 2961.0,
"completions/mean_length": 917.60546875,
"completions/mean_terminated_length": 1016.9133911132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 504.0,
"epoch": 0.04906666666666667,
"grad_norm": 0.022643912583589554,
"kl": 0.1025543212890625,
"learning_rate": 4.277777777777778e-06,
"loss": -0.3343,
"mask/has_final_conf_rate": 0.85546875,
"mask/share_final_conf": 0.014070061035454273,
"mask/share_reasoning": 0.7752383947372437,
"mask/share_step_conf": 0.11303532123565674,
"num_tokens": 14837926.0,
"reward": 0.6594477295875549,
"reward_std": 0.27707844972610474,
"rewards/accuracy_reward_step": 0.4609375,
"rewards/asymmetric_l2_reward": 0.5859012603759766,
"rewards/final_brier_reward_step": 0.47127535939216614,
"rewards/format_reward_step": 0.84765625,
"step": 46
},
{
"adv/mean_abs_final_conf": 0.6609259843826294,
"adv/mean_abs_reasoning": 0.5075252056121826,
"adv/mean_abs_step_conf": 0.7495334148406982,
"adv/ratio_final_to_reasoning": 1.3022525326311882,
"adv/ratio_step_to_reasoning": 1.4768397836253326,
"adv/std_final_conf": 0.8575954437255859,
"adv/std_reasoning": 0.7579585909843445,
"adv/std_step_conf": 0.9317649006843567,
"calib/answer_extract_rate": 0.859375,
"calib/auroc": 0.5688089727612604,
"calib/avg_num_step_conf": 13.17578125,
"calib/ece": 0.35684931506849316,
"calib/final_conf_rate": 0.85546875,
"calib/format_rate": 0.8515625,
"calib/frac_conf_gt_0.9": 0.9863013698630136,
"calib/gap": 0.005038276660139229,
"calib/mean_conf": 0.9824200913242009,
"calib/mu_c": 0.9843065693430658,
"calib/mu_w": 0.9792682926829266,
"calib/nonempty_final_conf_rate": 0.85546875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35684931506849316,
"calib/std_conf": 0.016360291553566133,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8437517730496452,
"calib/step_q_c_n": 1410.0,
"calib/step_q_gap": 0.007088502545315123,
"calib/step_q_w": 0.8366632705043301,
"calib/step_q_w_n": 1963.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.09375,
"completions/max_length": 2975.0,
"completions/max_terminated_length": 2975.0,
"completions/mean_length": 917.69140625,
"completions/mean_terminated_length": 1012.625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 492.0,
"epoch": 0.050133333333333335,
"grad_norm": 0.023026349022984505,
"kl": 0.120025634765625,
"learning_rate": 4.25e-06,
"loss": -0.3556,
"mask/has_final_conf_rate": 0.85546875,
"mask/share_final_conf": 0.013789718970656395,
"mask/share_reasoning": 0.787351131439209,
"mask/share_step_conf": 0.10510917007923126,
"num_tokens": 15178831.0,
"reward": 0.7321069240570068,
"reward_std": 0.24354524910449982,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/asymmetric_l2_reward": 0.6390807032585144,
"rewards/final_brier_reward_step": 0.547789454460144,
"rewards/format_reward_step": 0.8515625,
"step": 47
},
{
"adv/mean_abs_final_conf": 0.7052232027053833,
"adv/mean_abs_reasoning": 0.6574268341064453,
"adv/mean_abs_step_conf": 0.7477045059204102,
"adv/ratio_final_to_reasoning": 1.0727021869496722,
"adv/ratio_step_to_reasoning": 1.1373197246149338,
"adv/std_final_conf": 0.8916906714439392,
"adv/std_reasoning": 0.8751892447471619,
"adv/std_step_conf": 0.9354519844055176,
"calib/answer_extract_rate": 0.8515625,
"calib/auroc": 0.4789540816326531,
"calib/avg_num_step_conf": 14.9296875,
"calib/ece": 0.43339449541284414,
"calib/final_conf_rate": 0.8515625,
"calib/format_rate": 0.8515625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.001341836734693791,
"calib/mean_conf": 0.9838532110091744,
"calib/mu_c": 0.9832500000000001,
"calib/mu_w": 0.9845918367346939,
"calib/nonempty_final_conf_rate": 0.8515625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43339449541284414,
"calib/std_conf": 0.012371041717481143,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8498358208955226,
"calib/step_q_c_n": 1340.0,
"calib/step_q_gap": 0.016495772547416077,
"calib/step_q_w": 0.8333400483481065,
"calib/step_q_w_n": 2482.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.12109375,
"completions/max_length": 3072.0,
"completions/max_terminated_length": 3072.0,
"completions/mean_length": 853.24609375,
"completions/mean_terminated_length": 970.804443359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 534.0,
"epoch": 0.0512,
"grad_norm": 0.040837667882442474,
"kl": 0.13056182861328125,
"learning_rate": 4.222222222222223e-06,
"loss": -0.4883,
"mask/has_final_conf_rate": 0.8515625,
"mask/share_final_conf": 0.014190906658768654,
"mask/share_reasoning": 0.7518985271453857,
"mask/share_step_conf": 0.11281684041023254,
"num_tokens": 15500950.0,
"reward": 0.67741858959198,
"reward_std": 0.3460836708545685,
"rewards/accuracy_reward_step": 0.46875,
"rewards/asymmetric_l2_reward": 0.6105802059173584,
"rewards/final_brier_reward_step": 0.4801945090293884,
"rewards/format_reward_step": 0.8515625,
"step": 48
},
{
"adv/mean_abs_final_conf": 0.70012366771698,
"adv/mean_abs_reasoning": 0.4757267236709595,
"adv/mean_abs_step_conf": 0.7478195428848267,
"adv/ratio_final_to_reasoning": 1.4716929549688837,
"adv/ratio_step_to_reasoning": 1.5719519330641232,
"adv/std_final_conf": 0.8788543939590454,
"adv/std_reasoning": 0.7578703761100769,
"adv/std_step_conf": 0.9321491718292236,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.5525631939522797,
"calib/avg_num_step_conf": 11.640625,
"calib/ece": 0.3327542372881357,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.9957627118644068,
"calib/gap": 0.0031207181667846084,
"calib/mean_conf": 0.9810593220338985,
"calib/mu_c": 0.982156862745098,
"calib/mu_w": 0.9790361445783133,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3327542372881357,
"calib/std_conf": 0.015047265854960285,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8497086801426873,
"calib/step_q_c_n": 1682.0,
"calib/step_q_gap": 0.0004097587251219448,
"calib/step_q_w": 0.8492989214175654,
"calib/step_q_w_n": 1298.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 3035.0,
"completions/max_terminated_length": 3035.0,
"completions/mean_length": 868.53515625,
"completions/mean_terminated_length": 915.0,
"completions/min_length": 0.0,
"completions/min_terminated_length": 344.0,
"epoch": 0.05226666666666667,
"grad_norm": 0.024042055010795593,
"kl": 0.1303863525390625,
"learning_rate": 4.194444444444445e-06,
"loss": -0.1904,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.015961725264787674,
"mask/share_reasoning": 0.8114060163497925,
"mask/share_step_conf": 0.12185099720954895,
"num_tokens": 15827831.0,
"reward": 0.7799708843231201,
"reward_std": 0.2652823328971863,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/asymmetric_l2_reward": 0.6593862771987915,
"rewards/final_brier_reward_step": 0.5989929437637329,
"rewards/format_reward_step": 0.91015625,
"step": 49
},
{
"adv/mean_abs_final_conf": 0.755120575428009,
"adv/mean_abs_reasoning": 0.6018161773681641,
"adv/mean_abs_step_conf": 0.7432763576507568,
"adv/ratio_final_to_reasoning": 1.2547362530702797,
"adv/ratio_step_to_reasoning": 1.2350554631170272,
"adv/std_final_conf": 0.9192172884941101,
"adv/std_reasoning": 0.8434169292449951,
"adv/std_step_conf": 0.9341980814933777,
"calib/answer_extract_rate": 0.8984375,
"calib/auroc": 0.5903263403263402,
"calib/avg_num_step_conf": 12.55078125,
"calib/ece": 0.3453303964757709,
"calib/final_conf_rate": 0.88671875,
"calib/format_rate": 0.87890625,
"calib/frac_conf_gt_0.9": 0.9911894273127754,
"calib/gap": 0.0040825840825842175,
"calib/mean_conf": 0.9735242290748899,
"calib/mu_c": 0.9750349650349649,
"calib/mu_w": 0.9709523809523807,
"calib/nonempty_final_conf_rate": 0.88671875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34444933920704845,
"calib/std_conf": 0.019754062398247826,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8568041237113403,
"calib/step_q_c_n": 1552.0,
"calib/step_q_gap": 0.019308639063537636,
"calib/step_q_w": 0.8374954846478027,
"calib/step_q_w_n": 1661.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0703125,
"completions/max_length": 2516.0,
"completions/max_terminated_length": 2516.0,
"completions/mean_length": 892.41796875,
"completions/mean_terminated_length": 959.9118041992188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 499.0,
"epoch": 0.05333333333333334,
"grad_norm": 0.0485478974878788,
"kl": 0.1522216796875,
"learning_rate": 4.166666666666667e-06,
"loss": -0.3244,
"mask/has_final_conf_rate": 0.88671875,
"mask/share_final_conf": 0.01499384269118309,
"mask/share_reasoning": 0.8000199794769287,
"mask/share_step_conf": 0.1146736890077591,
"num_tokens": 16161650.0,
"reward": 0.7477984428405762,
"reward_std": 0.3347986936569214,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.6312395334243774,
"rewards/final_brier_reward_step": 0.5760761499404907,
"rewards/format_reward_step": 0.87890625,
"step": 50
},
{
"adv/mean_abs_final_conf": 0.7460015416145325,
"adv/mean_abs_reasoning": 0.49163341522216797,
"adv/mean_abs_step_conf": 0.7601824402809143,
"adv/ratio_final_to_reasoning": 1.5173938925152517,
"adv/ratio_step_to_reasoning": 1.5462383490296112,
"adv/std_final_conf": 0.9048994183540344,
"adv/std_reasoning": 0.7755351662635803,
"adv/std_step_conf": 0.9343783855438232,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.5319148936170213,
"calib/avg_num_step_conf": 12.09375,
"calib/ece": 0.3700000000000001,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 0.9872881355932204,
"calib/gap": 0.002463290380581462,
"calib/mean_conf": 0.9716949152542372,
"calib/mu_c": 0.972676056338028,
"calib/mu_w": 0.9702127659574465,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.3700000000000001,
"calib/std_conf": 0.018124105343005043,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.8501683501683501,
"calib/step_q_c_n": 1485.0,
"calib/step_q_gap": -0.0004834189191730287,
"calib/step_q_w": 0.8506517690875232,
"calib/step_q_w_n": 1611.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2702.0,
"completions/max_terminated_length": 2702.0,
"completions/mean_length": 851.8828125,
"completions/mean_terminated_length": 901.1652221679688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 71.0,
"epoch": 0.0544,
"grad_norm": 0.04129322990775108,
"kl": 0.1435089111328125,
"learning_rate": 4.138888888888889e-06,
"loss": -0.2908,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.0158391110599041,
"mask/share_reasoning": 0.8140634894371033,
"mask/share_step_conf": 0.11540989577770233,
"num_tokens": 16489028.0,
"reward": 0.7558123469352722,
"reward_std": 0.2396428883075714,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/asymmetric_l2_reward": 0.6501593589782715,
"rewards/final_brier_reward_step": 0.5677152276039124,
"rewards/format_reward_step": 0.9140625,
"step": 51
},
{
"adv/mean_abs_final_conf": 0.7636961936950684,
"adv/mean_abs_reasoning": 0.515580415725708,
"adv/mean_abs_step_conf": 0.7763593792915344,
"adv/ratio_final_to_reasoning": 1.481235846827354,
"adv/ratio_step_to_reasoning": 1.5057968759320806,
"adv/std_final_conf": 0.8879469633102417,
"adv/std_reasoning": 0.7578575015068054,
"adv/std_step_conf": 0.9332748651504517,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.5637815278533842,
"calib/avg_num_step_conf": 11.47265625,
"calib/ece": 0.2529184549356223,
"calib/final_conf_rate": 0.91015625,
"calib/format_rate": 0.90625,
"calib/frac_conf_gt_0.9": 0.9828326180257511,
"calib/gap": 0.004101796407185443,
"calib/mean_conf": 0.9679399141630902,
"calib/mu_c": 0.9691017964071855,
"calib/mu_w": 0.9650000000000001,
"calib/nonempty_final_conf_rate": 0.91015625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.25206008583690986,
"calib/std_conf": 0.02012953318957994,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8531514084507043,
"calib/step_q_c_n": 1704.0,
"calib/step_q_gap": 0.034789689067087126,
"calib/step_q_w": 0.8183617193836171,
"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": 0.046875,
"completions/max_length": 2878.0,
"completions/max_terminated_length": 2878.0,
"completions/mean_length": 899.96484375,
"completions/mean_terminated_length": 944.225341796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.055466666666666664,
"grad_norm": 0.03694462776184082,
"kl": 0.1541900634765625,
"learning_rate": 4.111111111111111e-06,
"loss": -0.2186,
"mask/has_final_conf_rate": 0.91015625,
"mask/share_final_conf": 0.01521845068782568,
"mask/share_reasoning": 0.8235276937484741,
"mask/share_step_conf": 0.11437886953353882,
"num_tokens": 16827371.0,
"reward": 0.8517237901687622,
"reward_std": 0.2654600739479065,
"rewards/accuracy_reward_step": 0.65625,
"rewards/asymmetric_l2_reward": 0.7257670760154724,
"rewards/final_brier_reward_step": 0.6651804447174072,
"rewards/format_reward_step": 0.90625,
"step": 52
},
{
"adv/mean_abs_final_conf": 0.7440406084060669,
"adv/mean_abs_reasoning": 0.5574250221252441,
"adv/mean_abs_step_conf": 0.739154577255249,
"adv/ratio_final_to_reasoning": 1.3347815022176979,
"adv/ratio_step_to_reasoning": 1.3260161419326693,
"adv/std_final_conf": 0.8806307911872864,
"adv/std_reasoning": 0.8100613355636597,
"adv/std_step_conf": 0.9332544803619385,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5653889943074004,
"calib/avg_num_step_conf": 10.86328125,
"calib/ece": 0.3233749999999999,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.9916666666666667,
"calib/gap": 0.001628083491461041,
"calib/mean_conf": 0.9658750000000002,
"calib/mu_c": 0.9664516129032257,
"calib/mu_w": 0.9648235294117646,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.32170833333333326,
"calib/std_conf": 0.027780107541188533,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.8536102626756261,
"calib/step_q_c_n": 1637.0,
"calib/step_q_gap": 0.0011190039343673686,
"calib/step_q_w": 0.8524912587412588,
"calib/step_q_w_n": 1144.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 3010.0,
"completions/max_terminated_length": 3010.0,
"completions/mean_length": 982.72265625,
"completions/mean_terminated_length": 1002.298828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 472.0,
"epoch": 0.05653333333333333,
"grad_norm": 0.06327595561742783,
"kl": 0.1549835205078125,
"learning_rate": 4.083333333333334e-06,
"loss": -0.0358,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.014779354445636272,
"mask/share_reasoning": 0.852211058139801,
"mask/share_step_conf": 0.11347834765911102,
"num_tokens": 17184772.0,
"reward": 0.8161087036132812,
"reward_std": 0.29630500078201294,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/asymmetric_l2_reward": 0.7033169865608215,
"rewards/final_brier_reward_step": 0.6195253729820251,
"rewards/format_reward_step": 0.92578125,
"step": 53
},
{
"adv/mean_abs_final_conf": 0.7146573662757874,
"adv/mean_abs_reasoning": 0.5098264813423157,
"adv/mean_abs_step_conf": 0.7374184131622314,
"adv/ratio_final_to_reasoning": 1.4017658800189725,
"adv/ratio_step_to_reasoning": 1.4464105733007275,
"adv/std_final_conf": 0.8893719911575317,
"adv/std_reasoning": 0.7930194139480591,
"adv/std_step_conf": 0.9324125647544861,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.45959719504240065,
"calib/avg_num_step_conf": 12.3515625,
"calib/ece": 0.27360995850622416,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.9875518672199171,
"calib/gap": -0.0023287671232876672,
"calib/mean_conf": 0.9707053941908714,
"calib/mu_c": 0.9700000000000001,
"calib/mu_w": 0.9723287671232878,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.27360995850622416,
"calib/std_conf": 0.017015989915706308,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8597100685292567,
"calib/step_q_c_n": 1897.0,
"calib/step_q_gap": 0.015520345209098507,
"calib/step_q_w": 0.8441897233201582,
"calib/step_q_w_n": 1265.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 3008.0,
"completions/max_terminated_length": 3008.0,
"completions/mean_length": 891.2109375,
"completions/mean_terminated_length": 919.9596557617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 454.0,
"epoch": 0.0576,
"grad_norm": 0.025256404653191566,
"kl": 0.159423828125,
"learning_rate": 4.055555555555556e-06,
"loss": -0.0848,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01595216616988182,
"mask/share_reasoning": 0.8246407508850098,
"mask/share_step_conf": 0.1281571239233017,
"num_tokens": 17519154.0,
"reward": 0.8532845973968506,
"reward_std": 0.26809757947921753,
"rewards/accuracy_reward_step": 0.6640625,
"rewards/asymmetric_l2_reward": 0.71662437915802,
"rewards/final_brier_reward_step": 0.6704136729240417,
"rewards/format_reward_step": 0.93359375,
"step": 54
},
{
"adv/mean_abs_final_conf": 0.7599447965621948,
"adv/mean_abs_reasoning": 0.6141306757926941,
"adv/mean_abs_step_conf": 0.7744641304016113,
"adv/ratio_final_to_reasoning": 1.2374317494909206,
"adv/ratio_step_to_reasoning": 1.2610738413318396,
"adv/std_final_conf": 0.8942292332649231,
"adv/std_reasoning": 0.8267177939414978,
"adv/std_step_conf": 0.9326158761978149,
"calib/answer_extract_rate": 0.89453125,
"calib/auroc": 0.5516528925619836,
"calib/avg_num_step_conf": 11.890625,
"calib/ece": 0.4322869955156952,
"calib/final_conf_rate": 0.87109375,
"calib/format_rate": 0.87109375,
"calib/frac_conf_gt_0.9": 0.9955156950672646,
"calib/gap": 0.004671852211959759,
"calib/mean_conf": 0.9748878923766816,
"calib/mu_c": 0.97702479338843,
"calib/mu_w": 0.9723529411764702,
"calib/nonempty_final_conf_rate": 0.87109375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4322869955156952,
"calib/std_conf": 0.019611952668874544,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8601275318829709,
"calib/step_q_c_n": 1333.0,
"calib/step_q_gap": -0.006073520133393906,
"calib/step_q_w": 0.8662010520163648,
"calib/step_q_w_n": 1711.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0625,
"completions/max_length": 2832.0,
"completions/max_terminated_length": 2832.0,
"completions/mean_length": 880.0546875,
"completions/mean_terminated_length": 938.7250366210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 496.0,
"epoch": 0.058666666666666666,
"grad_norm": 0.04848377779126167,
"kl": 0.155914306640625,
"learning_rate": 4.027777777777779e-06,
"loss": -0.2427,
"mask/has_final_conf_rate": 0.87109375,
"mask/share_final_conf": 0.015094468370079994,
"mask/share_reasoning": 0.8024145364761353,
"mask/share_step_conf": 0.1199910044670105,
"num_tokens": 17852272.0,
"reward": 0.6880007982254028,
"reward_std": 0.25061482191085815,
"rewards/accuracy_reward_step": 0.48828125,
"rewards/asymmetric_l2_reward": 0.610323429107666,
"rewards/final_brier_reward_step": 0.49380311369895935,
"rewards/format_reward_step": 0.87109375,
"step": 55
},
{
"adv/mean_abs_final_conf": 0.7145578861236572,
"adv/mean_abs_reasoning": 0.6271743774414062,
"adv/mean_abs_step_conf": 0.7482906579971313,
"adv/ratio_final_to_reasoning": 1.13932888814549,
"adv/ratio_step_to_reasoning": 1.1931142038197189,
"adv/std_final_conf": 0.906951904296875,
"adv/std_reasoning": 0.8591962456703186,
"adv/std_step_conf": 0.9356598258018494,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.6712661161813703,
"calib/avg_num_step_conf": 11.71875,
"calib/ece": 0.4738297872340427,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.9957446808510638,
"calib/gap": 0.010332464146023512,
"calib/mean_conf": 0.9759574468085107,
"calib/mu_c": 0.9811016949152542,
"calib/mu_w": 0.9707692307692307,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4738297872340427,
"calib/std_conf": 0.01582612632112727,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8583504340962904,
"calib/step_q_c_n": 1267.0,
"calib/step_q_gap": -0.012578590716173621,
"calib/step_q_w": 0.870929024812464,
"calib/step_q_w_n": 1733.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 3033.0,
"completions/max_terminated_length": 3033.0,
"completions/mean_length": 957.1796875,
"completions/mean_terminated_length": 1000.155029296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 449.0,
"epoch": 0.05973333333333333,
"grad_norm": 0.02709430269896984,
"kl": 0.1496734619140625,
"learning_rate": 4.000000000000001e-06,
"loss": -0.1615,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.014955723658204079,
"mask/share_reasoning": 0.830254852771759,
"mask/share_step_conf": 0.11182068288326263,
"num_tokens": 18204150.0,
"reward": 0.6914936304092407,
"reward_std": 0.304796040058136,
"rewards/accuracy_reward_step": 0.46484375,
"rewards/asymmetric_l2_reward": 0.6214755773544312,
"rewards/final_brier_reward_step": 0.4865117073059082,
"rewards/format_reward_step": 0.91015625,
"step": 56
},
{
"adv/mean_abs_final_conf": 0.7080997228622437,
"adv/mean_abs_reasoning": 0.5576961040496826,
"adv/mean_abs_step_conf": 0.7475917935371399,
"adv/ratio_final_to_reasoning": 1.2696874116932368,
"adv/ratio_step_to_reasoning": 1.3405002977581144,
"adv/std_final_conf": 0.9036187529563904,
"adv/std_reasoning": 0.8101306557655334,
"adv/std_step_conf": 0.9331084489822388,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.5666931637519873,
"calib/avg_num_step_conf": 11.23828125,
"calib/ece": 0.34583690987124477,
"calib/final_conf_rate": 0.91015625,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.9957081545064378,
"calib/gap": 0.0029936406995230147,
"calib/mean_conf": 0.9793133047210301,
"calib/mu_c": 0.9804054054054052,
"calib/mu_w": 0.9774117647058822,
"calib/nonempty_final_conf_rate": 0.91015625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3449785407725323,
"calib/std_conf": 0.018353950675204293,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8543750000000001,
"calib/step_q_c_n": 1536.0,
"calib/step_q_gap": -0.004916573452647177,
"calib/step_q_w": 0.8592915734526473,
"calib/step_q_w_n": 1341.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2920.0,
"completions/max_terminated_length": 2920.0,
"completions/mean_length": 880.0859375,
"completions/mean_terminated_length": 930.9999389648438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 517.0,
"epoch": 0.0608,
"grad_norm": 0.028827087953686714,
"kl": 0.15277099609375,
"learning_rate": 3.972222222222223e-06,
"loss": -0.2121,
"mask/has_final_conf_rate": 0.91015625,
"mask/share_final_conf": 0.015749987214803696,
"mask/share_reasoning": 0.8137336373329163,
"mask/share_step_conf": 0.11582887172698975,
"num_tokens": 18536244.0,
"reward": 0.7834737300872803,
"reward_std": 0.30211466550827026,
"rewards/accuracy_reward_step": 0.578125,
"rewards/asymmetric_l2_reward": 0.676862359046936,
"rewards/final_brier_reward_step": 0.5924289226531982,
"rewards/format_reward_step": 0.91015625,
"step": 57
},
{
"adv/mean_abs_final_conf": 0.7426167726516724,
"adv/mean_abs_reasoning": 0.5847933888435364,
"adv/mean_abs_step_conf": 0.7576659917831421,
"adv/ratio_final_to_reasoning": 1.2698788782825352,
"adv/ratio_step_to_reasoning": 1.2956131280510397,
"adv/std_final_conf": 0.9091824293136597,
"adv/std_reasoning": 0.8431410789489746,
"adv/std_step_conf": 0.9340181946754456,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.5531134226275293,
"calib/avg_num_step_conf": 11.6015625,
"calib/ece": 0.4662025316455697,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.9957805907172996,
"calib/gap": 0.005793673411228273,
"calib/mean_conf": 0.9767510548523207,
"calib/mu_c": 0.9795867768595041,
"calib/mu_w": 0.9737931034482759,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4662025316455697,
"calib/std_conf": 0.02817952635581979,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8562822458270105,
"calib/step_q_c_n": 1318.0,
"calib/step_q_gap": -0.0005579478775900037,
"calib/step_q_w": 0.8568401937046005,
"calib/step_q_w_n": 1652.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2634.0,
"completions/max_terminated_length": 2634.0,
"completions/mean_length": 985.74609375,
"completions/mean_terminated_length": 1025.8170166015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 511.0,
"epoch": 0.06186666666666667,
"grad_norm": 0.020769745111465454,
"kl": 0.148895263671875,
"learning_rate": 3.944444444444445e-06,
"loss": -0.2007,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.014388788491487503,
"mask/share_reasoning": 0.8343918323516846,
"mask/share_step_conf": 0.11215688288211823,
"num_tokens": 18894915.0,
"reward": 0.6976282596588135,
"reward_std": 0.2969515323638916,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/asymmetric_l2_reward": 0.6196170449256897,
"rewards/final_brier_reward_step": 0.4951706826686859,
"rewards/format_reward_step": 0.92578125,
"step": 58
},
{
"adv/mean_abs_final_conf": 0.6896352171897888,
"adv/mean_abs_reasoning": 0.6210002899169922,
"adv/mean_abs_step_conf": 0.7944797873497009,
"adv/ratio_final_to_reasoning": 1.1105231807250378,
"adv/ratio_step_to_reasoning": 1.2793549379113773,
"adv/std_final_conf": 0.8672558069229126,
"adv/std_reasoning": 0.8269534707069397,
"adv/std_step_conf": 0.9332959055900574,
"calib/answer_extract_rate": 0.90625,
"calib/auroc": 0.5547254247572815,
"calib/avg_num_step_conf": 11.2890625,
"calib/ece": 0.4279220779220779,
"calib/final_conf_rate": 0.90234375,
"calib/format_rate": 0.8984375,
"calib/frac_conf_gt_0.9": 0.9913419913419913,
"calib/gap": 0.003496662621359614,
"calib/mean_conf": 0.982034632034632,
"calib/mu_c": 0.9835937500000002,
"calib/mu_w": 0.9800970873786405,
"calib/nonempty_final_conf_rate": 0.90234375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4279220779220779,
"calib/std_conf": 0.015024327018331942,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8552635599694423,
"calib/step_q_c_n": 1309.0,
"calib/step_q_gap": 0.010437500513401754,
"calib/step_q_w": 0.8448260594560405,
"calib/step_q_w_n": 1581.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0625,
"completions/max_length": 3043.0,
"completions/max_terminated_length": 3043.0,
"completions/mean_length": 881.99609375,
"completions/mean_terminated_length": 940.7958984375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 230.0,
"epoch": 0.06293333333333333,
"grad_norm": 0.018716448917984962,
"kl": 0.161285400390625,
"learning_rate": 3.916666666666667e-06,
"loss": -0.2376,
"mask/has_final_conf_rate": 0.90234375,
"mask/share_final_conf": 0.015686459839344025,
"mask/share_reasoning": 0.8076034188270569,
"mask/share_step_conf": 0.1142100840806961,
"num_tokens": 19226954.0,
"reward": 0.7167990207672119,
"reward_std": 0.32347404956817627,
"rewards/accuracy_reward_step": 0.5,
"rewards/asymmetric_l2_reward": 0.6384674906730652,
"rewards/final_brier_reward_step": 0.515442967414856,
"rewards/format_reward_step": 0.8984375,
"step": 59
},
{
"adv/mean_abs_final_conf": 0.6956037282943726,
"adv/mean_abs_reasoning": 0.5651669502258301,
"adv/mean_abs_step_conf": 0.7514952421188354,
"adv/ratio_final_to_reasoning": 1.2307933576378138,
"adv/ratio_step_to_reasoning": 1.32968716910738,
"adv/std_final_conf": 0.8582842946052551,
"adv/std_reasoning": 0.8101160526275635,
"adv/std_step_conf": 0.9319243431091309,
"calib/answer_extract_rate": 0.89453125,
"calib/auroc": 0.6115804976358422,
"calib/avg_num_step_conf": 10.86328125,
"calib/ece": 0.40213043478260874,
"calib/final_conf_rate": 0.8984375,
"calib/format_rate": 0.89453125,
"calib/frac_conf_gt_0.9": 0.9956521739130435,
"calib/gap": 0.010303852414541126,
"calib/mean_conf": 0.9803913043478262,
"calib/mu_c": 0.9847368421052629,
"calib/mu_w": 0.9744329896907218,
"calib/nonempty_final_conf_rate": 0.8984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40213043478260874,
"calib/std_conf": 0.027208117603319572,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8575264750378215,
"calib/step_q_c_n": 1322.0,
"calib/step_q_gap": 0.018479182371611724,
"calib/step_q_w": 0.8390472926662098,
"calib/step_q_w_n": 1459.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2964.0,
"completions/max_terminated_length": 2964.0,
"completions/mean_length": 936.8046875,
"completions/mean_terminated_length": 990.9999389648438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 405.0,
"epoch": 0.064,
"grad_norm": 0.027904795482754707,
"kl": 0.146087646484375,
"learning_rate": 3.88888888888889e-06,
"loss": -0.1497,
"mask/has_final_conf_rate": 0.8984375,
"mask/share_final_conf": 0.015355521813035011,
"mask/share_reasoning": 0.8180770874023438,
"mask/share_step_conf": 0.11187991499900818,
"num_tokens": 19575632.0,
"reward": 0.7286934852600098,
"reward_std": 0.2839881181716919,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/asymmetric_l2_reward": 0.6367565393447876,
"rewards/final_brier_reward_step": 0.5378179550170898,
"rewards/format_reward_step": 0.89453125,
"step": 60
},
{
"adv/mean_abs_final_conf": 0.5476254820823669,
"adv/mean_abs_reasoning": 0.46652138233184814,
"adv/mean_abs_step_conf": 0.7672770023345947,
"adv/ratio_final_to_reasoning": 1.1738486226400389,
"adv/ratio_step_to_reasoning": 1.6446770317353037,
"adv/std_final_conf": 0.7907182574272156,
"adv/std_reasoning": 0.7393697500228882,
"adv/std_step_conf": 0.9312024712562561,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.561746437705517,
"calib/avg_num_step_conf": 11.578125,
"calib/ece": 0.33097560975609763,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9959349593495935,
"calib/gap": 0.0031099744245526306,
"calib/mean_conf": 0.9854471544715447,
"calib/mu_c": 0.9865217391304347,
"calib/mu_w": 0.9834117647058821,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33097560975609763,
"calib/std_conf": 0.011634825248779398,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.849314606741573,
"calib/step_q_c_n": 1780.0,
"calib/step_q_gap": 0.00033656620103239554,
"calib/step_q_w": 0.8489780405405406,
"calib/step_q_w_n": 1184.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2979.0,
"completions/max_terminated_length": 2979.0,
"completions/mean_length": 852.55859375,
"completions/mean_terminated_length": 866.09130859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.06506666666666666,
"grad_norm": 0.021844208240509033,
"kl": 0.1577911376953125,
"learning_rate": 3.861111111111112e-06,
"loss": -0.0096,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.017297610640525818,
"mask/share_reasoning": 0.8325221538543701,
"mask/share_step_conf": 0.13455525040626526,
"num_tokens": 19897951.0,
"reward": 0.8288745880126953,
"reward_std": 0.22828470170497894,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/asymmetric_l2_reward": 0.701357364654541,
"rewards/final_brier_reward_step": 0.6392042636871338,
"rewards/format_reward_step": 0.953125,
"step": 61
},
{
"adv/mean_abs_final_conf": 0.7341781258583069,
"adv/mean_abs_reasoning": 0.661810040473938,
"adv/mean_abs_step_conf": 0.7774771451950073,
"adv/ratio_final_to_reasoning": 1.1093487269134574,
"adv/ratio_step_to_reasoning": 1.174773874144063,
"adv/std_final_conf": 0.8844819664955139,
"adv/std_reasoning": 0.8593194484710693,
"adv/std_step_conf": 0.934329092502594,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.561044003451251,
"calib/avg_num_step_conf": 11.546875,
"calib/ece": 0.46584745762711866,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.90625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0035432844406099173,
"calib/mean_conf": 0.9827966101694915,
"calib/mu_c": 0.9845081967213115,
"calib/mu_w": 0.9809649122807016,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.46584745762711866,
"calib/std_conf": 0.013522460488446315,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.8506163828061638,
"calib/step_q_c_n": 1233.0,
"calib/step_q_gap": -0.002302943949494929,
"calib/step_q_w": 0.8529193267556587,
"calib/step_q_w_n": 1723.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2911.0,
"completions/max_terminated_length": 2911.0,
"completions/mean_length": 880.30859375,
"completions/mean_terminated_length": 927.4032592773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 450.0,
"epoch": 0.06613333333333334,
"grad_norm": 0.0222491268068552,
"kl": 0.1569976806640625,
"learning_rate": 3.833333333333334e-06,
"loss": -0.2313,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.015586305409669876,
"mask/share_reasoning": 0.8132479190826416,
"mask/share_step_conf": 0.1203845664858818,
"num_tokens": 20230390.0,
"reward": 0.6956356763839722,
"reward_std": 0.3364129066467285,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/asymmetric_l2_reward": 0.6262305974960327,
"rewards/final_brier_reward_step": 0.4884781241416931,
"rewards/format_reward_step": 0.90625,
"step": 62
},
{
"adv/mean_abs_final_conf": 0.627443790435791,
"adv/mean_abs_reasoning": 0.4884450137615204,
"adv/mean_abs_step_conf": 0.7527927756309509,
"adv/ratio_final_to_reasoning": 1.2845740518545568,
"adv/ratio_step_to_reasoning": 1.5412027033169722,
"adv/std_final_conf": 0.8280969858169556,
"adv/std_reasoning": 0.7576963305473328,
"adv/std_step_conf": 0.9316011071205139,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6043795620437957,
"calib/avg_num_step_conf": 10.62890625,
"calib/ece": 0.42854251012145755,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9959514170040485,
"calib/gap": 0.007569343065693679,
"calib/mean_conf": 0.9831983805668015,
"calib/mu_c": 0.9865693430656934,
"calib/mu_w": 0.9789999999999998,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42854251012145755,
"calib/std_conf": 0.016710596816404612,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8481534090909091,
"calib/step_q_c_n": 1408.0,
"calib/step_q_gap": 0.020643888908121633,
"calib/step_q_w": 0.8275095201827875,
"calib/step_q_w_n": 1313.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2990.0,
"completions/max_terminated_length": 2990.0,
"completions/mean_length": 946.09765625,
"completions/mean_terminated_length": 957.3162231445312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 478.0,
"epoch": 0.0672,
"grad_norm": 0.025082435458898544,
"kl": 0.1472930908203125,
"learning_rate": 3.8055555555555556e-06,
"loss": -0.0259,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.015975866466760635,
"mask/share_reasoning": 0.8547379374504089,
"mask/share_step_conf": 0.11756742745637894,
"num_tokens": 20581231.0,
"reward": 0.7615793347358704,
"reward_std": 0.25341373682022095,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/asymmetric_l2_reward": 0.6730742454528809,
"rewards/final_brier_reward_step": 0.552428126335144,
"rewards/format_reward_step": 0.953125,
"step": 63
},
{
"adv/mean_abs_final_conf": 0.7004935145378113,
"adv/mean_abs_reasoning": 0.603089451789856,
"adv/mean_abs_step_conf": 0.7326570749282837,
"adv/ratio_final_to_reasoning": 1.1615084834577662,
"adv/ratio_step_to_reasoning": 1.2148398098389805,
"adv/std_final_conf": 0.896488606929779,
"adv/std_reasoning": 0.8430898189544678,
"adv/std_step_conf": 0.932219922542572,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.5474950396825398,
"calib/avg_num_step_conf": 11.20703125,
"calib/ece": 0.28637500000000016,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0027579365079367646,
"calib/mean_conf": 0.9863750000000001,
"calib/mu_c": 0.9872023809523811,
"calib/mu_w": 0.9844444444444443,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.28637500000000016,
"calib/std_conf": 0.009692404672388246,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.844783362218371,
"calib/step_q_c_n": 1731.0,
"calib/step_q_gap": 0.024844873641921117,
"calib/step_q_w": 0.8199384885764499,
"calib/step_q_w_n": 1138.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2966.0,
"completions/max_terminated_length": 2966.0,
"completions/mean_length": 880.98046875,
"completions/mean_terminated_length": 913.0809936523438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 515.0,
"epoch": 0.06826666666666667,
"grad_norm": 0.02380029298365116,
"kl": 0.1593780517578125,
"learning_rate": 3.777777777777778e-06,
"loss": -0.097,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.0161864273250103,
"mask/share_reasoning": 0.8278210163116455,
"mask/share_step_conf": 0.12083625793457031,
"num_tokens": 20910538.0,
"reward": 0.8677514791488647,
"reward_std": 0.2944144904613495,
"rewards/accuracy_reward_step": 0.65625,
"rewards/asymmetric_l2_reward": 0.7520149946212769,
"rewards/final_brier_reward_step": 0.66473788022995,
"rewards/format_reward_step": 0.9375,
"step": 64
},
{
"adv/mean_abs_final_conf": 0.5200560688972473,
"adv/mean_abs_reasoning": 0.3561217188835144,
"adv/mean_abs_step_conf": 0.756298303604126,
"adv/ratio_final_to_reasoning": 1.4603323563855846,
"adv/ratio_step_to_reasoning": 2.1237073267398983,
"adv/std_final_conf": 0.7643717527389526,
"adv/std_reasoning": 0.6403516530990601,
"adv/std_step_conf": 0.9288119673728943,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5231881214024071,
"calib/avg_num_step_conf": 10.87109375,
"calib/ece": 0.40035856573705186,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.9960159362549801,
"calib/gap": 0.004525771847200155,
"calib/mean_conf": 0.9860159362549801,
"calib/mu_c": 0.9878911564625849,
"calib/mu_w": 0.9833653846153847,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.40035856573705186,
"calib/std_conf": 0.024835389241660396,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8443627140139505,
"calib/step_q_c_n": 1577.0,
"calib/step_q_gap": 0.00232291301892551,
"calib/step_q_w": 0.842039800995025,
"calib/step_q_w_n": 1206.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2816.0,
"completions/max_terminated_length": 2816.0,
"completions/mean_length": 820.27734375,
"completions/mean_terminated_length": 823.494140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 473.0,
"epoch": 0.06933333333333333,
"grad_norm": 0.01668195053935051,
"kl": 0.1640777587890625,
"learning_rate": 3.7500000000000005e-06,
"loss": 0.0244,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.018098723143339157,
"mask/share_reasoning": 0.8402106761932373,
"mask/share_step_conf": 0.13778437674045563,
"num_tokens": 21225553.0,
"reward": 0.7908812761306763,
"reward_std": 0.18446239829063416,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/asymmetric_l2_reward": 0.6852684617042542,
"rewards/final_brier_reward_step": 0.5863378643989563,
"rewards/format_reward_step": 0.9765625,
"step": 65
},
{
"adv/mean_abs_final_conf": 0.6290441751480103,
"adv/mean_abs_reasoning": 0.5717591047286987,
"adv/mean_abs_step_conf": 0.7339279651641846,
"adv/ratio_final_to_reasoning": 1.1001909194720974,
"adv/ratio_step_to_reasoning": 1.2836314438970506,
"adv/std_final_conf": 0.8466580510139465,
"adv/std_reasoning": 0.8266621828079224,
"adv/std_step_conf": 0.9322919249534607,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.5569719413189005,
"calib/avg_num_step_conf": 10.75,
"calib/ece": 0.4869198312236286,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.003078621279020144,
"calib/mean_conf": 0.9848101265822784,
"calib/mu_c": 0.9863559322033897,
"calib/mu_w": 0.9832773109243695,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4869198312236286,
"calib/std_conf": 0.011421435117915504,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8389644012944985,
"calib/step_q_c_n": 1236.0,
"calib/step_q_gap": 0.003054111057031461,
"calib/step_q_w": 0.835910290237467,
"calib/step_q_w_n": 1516.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2987.0,
"completions/max_terminated_length": 2987.0,
"completions/mean_length": 975.1328125,
"completions/mean_terminated_length": 1002.546142578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 490.0,
"epoch": 0.0704,
"grad_norm": 0.0274101160466671,
"kl": 0.1490325927734375,
"learning_rate": 3.7222222222222225e-06,
"loss": -0.0766,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.01519216038286686,
"mask/share_reasoning": 0.8440859317779541,
"mask/share_step_conf": 0.113378144800663,
"num_tokens": 21581539.0,
"reward": 0.6924327611923218,
"reward_std": 0.2524971663951874,
"rewards/accuracy_reward_step": 0.4609375,
"rewards/asymmetric_l2_reward": 0.6322306394577026,
"rewards/final_brier_reward_step": 0.4760722517967224,
"rewards/format_reward_step": 0.921875,
"step": 66
},
{
"adv/mean_abs_final_conf": 0.5527406334877014,
"adv/mean_abs_reasoning": 0.4371800720691681,
"adv/mean_abs_step_conf": 0.7184213995933533,
"adv/ratio_final_to_reasoning": 1.2643317223303125,
"adv/ratio_step_to_reasoning": 1.6433077477504254,
"adv/std_final_conf": 0.795620858669281,
"adv/std_reasoning": 0.7393509149551392,
"adv/std_step_conf": 0.9301301836967468,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5891039426523297,
"calib/avg_num_step_conf": 11.0078125,
"calib/ece": 0.36938271604938266,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005051612903225977,
"calib/mean_conf": 0.9866666666666666,
"calib/mu_c": 0.9885999999999998,
"calib/mu_w": 0.9835483870967738,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36938271604938266,
"calib/std_conf": 0.009428090415820642,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8425621414913959,
"calib/step_q_c_n": 1569.0,
"calib/step_q_gap": 0.021385199938153243,
"calib/step_q_w": 0.8211769415532426,
"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": 0.03515625,
"completions/max_length": 2512.0,
"completions/max_terminated_length": 2512.0,
"completions/mean_length": 868.9375,
"completions/mean_terminated_length": 900.5992431640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 487.0,
"epoch": 0.07146666666666666,
"grad_norm": 0.020727286115288734,
"kl": 0.1535186767578125,
"learning_rate": 3.694444444444445e-06,
"loss": -0.172,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01617474853992462,
"mask/share_reasoning": 0.8293904066085815,
"mask/share_step_conf": 0.11927864700555801,
"num_tokens": 21908995.0,
"reward": 0.7993296384811401,
"reward_std": 0.21487489342689514,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/asymmetric_l2_reward": 0.6986775398254395,
"rewards/final_brier_reward_step": 0.5937316417694092,
"rewards/format_reward_step": 0.9453125,
"step": 67
},
{
"adv/mean_abs_final_conf": 0.4896950125694275,
"adv/mean_abs_reasoning": 0.4402713477611542,
"adv/mean_abs_step_conf": 0.7603869438171387,
"adv/ratio_final_to_reasoning": 1.1122572819230687,
"adv/ratio_step_to_reasoning": 1.7270870513918752,
"adv/std_final_conf": 0.7536525130271912,
"adv/std_reasoning": 0.7206746935844421,
"adv/std_step_conf": 0.9291406273841858,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.5731757877280265,
"calib/avg_num_step_conf": 10.5078125,
"calib/ece": 0.4336363636363636,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.9958677685950413,
"calib/gap": 0.0027736318407960425,
"calib/mean_conf": 0.985702479338843,
"calib/mu_c": 0.9869402985074627,
"calib/mu_w": 0.9841666666666666,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4328099173553719,
"calib/std_conf": 0.015604960340907914,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.838406827880512,
"calib/step_q_c_n": 1406.0,
"calib/step_q_gap": 0.012269756229421502,
"calib/step_q_w": 0.8261370716510905,
"calib/step_q_w_n": 1284.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3019.0,
"completions/max_terminated_length": 3019.0,
"completions/mean_length": 862.625,
"completions/mean_terminated_length": 890.4515991210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 402.0,
"epoch": 0.07253333333333334,
"grad_norm": 0.015150300227105618,
"kl": 0.15191650390625,
"learning_rate": 3.6666666666666666e-06,
"loss": -0.1352,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016995739191770554,
"mask/share_reasoning": 0.8243101835250854,
"mask/share_step_conf": 0.1274440884590149,
"num_tokens": 22233915.0,
"reward": 0.7585927844047546,
"reward_std": 0.20454856753349304,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/asymmetric_l2_reward": 0.6879211068153381,
"rewards/final_brier_reward_step": 0.5362957119941711,
"rewards/format_reward_step": 0.94140625,
"step": 68
},
{
"adv/mean_abs_final_conf": 0.7056964039802551,
"adv/mean_abs_reasoning": 0.6312321424484253,
"adv/mean_abs_step_conf": 0.7472624778747559,
"adv/ratio_final_to_reasoning": 1.1179665237625538,
"adv/ratio_step_to_reasoning": 1.1838156323540048,
"adv/std_final_conf": 0.8818315863609314,
"adv/std_reasoning": 0.8590497970581055,
"adv/std_step_conf": 0.9349477887153625,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.5457418727424284,
"calib/avg_num_step_conf": 10.9765625,
"calib/ece": 0.47625000000000006,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0028480133370382132,
"calib/mean_conf": 0.9845833333333334,
"calib/mu_c": 0.985983606557377,
"calib/mu_w": 0.9831355932203388,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.47625000000000006,
"calib/std_conf": 0.011430648372783693,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8321756647864625,
"calib/step_q_c_n": 1241.0,
"calib/step_q_gap": 0.006898418132542883,
"calib/step_q_w": 0.8252772466539197,
"calib/step_q_w_n": 1569.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2630.0,
"completions/max_terminated_length": 2630.0,
"completions/mean_length": 934.42578125,
"completions/mean_terminated_length": 972.4105224609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 532.0,
"epoch": 0.0736,
"grad_norm": 0.01587671972811222,
"kl": 0.1492156982421875,
"learning_rate": 3.638888888888889e-06,
"loss": -0.1755,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.015048105269670486,
"mask/share_reasoning": 0.8365316390991211,
"mask/share_step_conf": 0.10935773700475693,
"num_tokens": 22577624.0,
"reward": 0.7187528610229492,
"reward_std": 0.2709349989891052,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/asymmetric_l2_reward": 0.6646471619606018,
"rewards/final_brier_reward_step": 0.49160856008529663,
"rewards/format_reward_step": 0.9296875,
"step": 69
},
{
"adv/mean_abs_final_conf": 0.5599315762519836,
"adv/mean_abs_reasoning": 0.4661853313446045,
"adv/mean_abs_step_conf": 0.7805700898170471,
"adv/ratio_final_to_reasoning": 1.2010922236377315,
"adv/ratio_step_to_reasoning": 1.6743772000840782,
"adv/std_final_conf": 0.8068824410438538,
"adv/std_reasoning": 0.7208148837089539,
"adv/std_step_conf": 0.932051420211792,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.5884371121979707,
"calib/avg_num_step_conf": 10.9375,
"calib/ece": 0.4188983050847458,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 0.9915254237288136,
"calib/gap": 0.007489597780860113,
"calib/mean_conf": 0.9810169491525423,
"calib/mu_c": 0.9842857142857142,
"calib/mu_w": 0.9767961165048541,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.41817796610169494,
"calib/std_conf": 0.027888696164273755,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.839237958303379,
"calib/step_q_c_n": 1391.0,
"calib/step_q_gap": 0.02184973970863091,
"calib/step_q_w": 0.8173882185947481,
"calib/step_q_w_n": 1409.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2930.0,
"completions/max_terminated_length": 2930.0,
"completions/mean_length": 915.4296875,
"completions/mean_terminated_length": 960.4507446289062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 390.0,
"epoch": 0.07466666666666667,
"grad_norm": 0.03501075878739357,
"kl": 0.1421356201171875,
"learning_rate": 3.6111111111111115e-06,
"loss": -0.1519,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.015805684030056,
"mask/share_reasoning": 0.8204933404922485,
"mask/share_step_conf": 0.11682601273059845,
"num_tokens": 22918966.0,
"reward": 0.7455442547798157,
"reward_std": 0.22053103148937225,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/asymmetric_l2_reward": 0.6656532883644104,
"rewards/final_brier_reward_step": 0.5371538996696472,
"rewards/format_reward_step": 0.921875,
"step": 70
},
{
"adv/mean_abs_final_conf": 0.6588914394378662,
"adv/mean_abs_reasoning": 0.5920547246932983,
"adv/mean_abs_step_conf": 0.7465313673019409,
"adv/ratio_final_to_reasoning": 1.1128894204487452,
"adv/ratio_step_to_reasoning": 1.2609161554933388,
"adv/std_final_conf": 0.860008180141449,
"adv/std_reasoning": 0.8267221450805664,
"adv/std_step_conf": 0.9302997589111328,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5749336870026525,
"calib/avg_num_step_conf": 11.23046875,
"calib/ece": 0.45276422764227625,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9959349593495935,
"calib/gap": 0.006875331564986875,
"calib/mean_conf": 0.9812195121951219,
"calib/mu_c": 0.9844615384615385,
"calib/mu_w": 0.9775862068965516,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.45276422764227625,
"calib/std_conf": 0.027735257296859942,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8306344628695026,
"calib/step_q_c_n": 1387.0,
"calib/step_q_gap": -7.790272189522884e-05,
"calib/step_q_w": 0.8307123655913978,
"calib/step_q_w_n": 1488.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2964.0,
"completions/max_terminated_length": 2964.0,
"completions/mean_length": 918.62890625,
"completions/mean_terminated_length": 940.676025390625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 501.0,
"epoch": 0.07573333333333333,
"grad_norm": 0.031742144376039505,
"kl": 0.1444549560546875,
"learning_rate": 3.5833333333333335e-06,
"loss": -0.0708,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.01613161340355873,
"mask/share_reasoning": 0.8374041318893433,
"mask/share_step_conf": 0.1230267882347107,
"num_tokens": 23258543.0,
"reward": 0.7496272325515747,
"reward_std": 0.3023836016654968,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/asymmetric_l2_reward": 0.678457498550415,
"rewards/final_brier_reward_step": 0.527046799659729,
"rewards/format_reward_step": 0.9609375,
"step": 71
},
{
"adv/mean_abs_final_conf": 0.664090633392334,
"adv/mean_abs_reasoning": 0.5109577178955078,
"adv/mean_abs_step_conf": 0.7358330488204956,
"adv/ratio_final_to_reasoning": 1.2996978226056315,
"adv/ratio_step_to_reasoning": 1.4401055567791137,
"adv/std_final_conf": 0.8751429319381714,
"adv/std_reasoning": 0.7929494976997375,
"adv/std_step_conf": 0.9357478022575378,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6445384211906635,
"calib/avg_num_step_conf": 11.00390625,
"calib/ece": 0.4796761133603239,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008731313926042317,
"calib/mean_conf": 0.9817004048582996,
"calib/mu_c": 0.986048387096774,
"calib/mu_w": 0.9773170731707317,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4796761133603239,
"calib/std_conf": 0.01339010299197453,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8339120370370371,
"calib/step_q_c_n": 1296.0,
"calib/step_q_gap": 0.011755561034407225,
"calib/step_q_w": 0.8221564760026299,
"calib/step_q_w_n": 1521.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2715.0,
"completions/max_terminated_length": 2715.0,
"completions/mean_length": 843.8828125,
"completions/mean_terminated_length": 864.1360473632812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 381.0,
"epoch": 0.0768,
"grad_norm": 0.027199726551771164,
"kl": 0.15850830078125,
"learning_rate": 3.555555555555556e-06,
"loss": -0.1048,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.016982506960630417,
"mask/share_reasoning": 0.8334957361221313,
"mask/share_step_conf": 0.12608429789543152,
"num_tokens": 23578985.0,
"reward": 0.7423896789550781,
"reward_std": 0.2794683873653412,
"rewards/accuracy_reward_step": 0.484375,
"rewards/asymmetric_l2_reward": 0.6939469575881958,
"rewards/final_brier_reward_step": 0.5017699003219604,
"rewards/format_reward_step": 0.9609375,
"step": 72
},
{
"adv/mean_abs_final_conf": 0.7219201922416687,
"adv/mean_abs_reasoning": 0.4985666275024414,
"adv/mean_abs_step_conf": 0.7653075456619263,
"adv/ratio_final_to_reasoning": 1.447991406601184,
"adv/ratio_step_to_reasoning": 1.5350155895827156,
"adv/std_final_conf": 0.8606438636779785,
"adv/std_reasoning": 0.7395066022872925,
"adv/std_step_conf": 0.9299985766410828,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5828644501278772,
"calib/avg_num_step_conf": 11.06640625,
"calib/ece": 0.3224796747967481,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.004689075630252448,
"calib/mean_conf": 0.9769512195121952,
"calib/mu_c": 0.9785714285714286,
"calib/mu_w": 0.9738823529411762,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3224796747967481,
"calib/std_conf": 0.014707640654936523,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8299045346062053,
"calib/step_q_c_n": 1676.0,
"calib/step_q_gap": 0.0015899278646321902,
"calib/step_q_w": 0.8283146067415731,
"calib/step_q_w_n": 1157.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2846.0,
"completions/max_terminated_length": 2846.0,
"completions/mean_length": 861.546875,
"completions/mean_terminated_length": 882.2240600585938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 453.0,
"epoch": 0.07786666666666667,
"grad_norm": 0.03334764018654823,
"kl": 0.1400146484375,
"learning_rate": 3.5277777777777784e-06,
"loss": -0.0773,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.016731340438127518,
"mask/share_reasoning": 0.8360512852668762,
"mask/share_step_conf": 0.12377987056970596,
"num_tokens": 23906573.0,
"reward": 0.8440332412719727,
"reward_std": 0.2482633739709854,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/asymmetric_l2_reward": 0.7245661020278931,
"rewards/final_brier_reward_step": 0.6455316543579102,
"rewards/format_reward_step": 0.9609375,
"step": 73
},
{
"adv/mean_abs_final_conf": 0.7695039510726929,
"adv/mean_abs_reasoning": 0.5811711549758911,
"adv/mean_abs_step_conf": 0.7675015926361084,
"adv/ratio_final_to_reasoning": 1.3240573701642409,
"adv/ratio_step_to_reasoning": 1.3206119850665108,
"adv/std_final_conf": 0.8966401815414429,
"adv/std_reasoning": 0.8101295828819275,
"adv/std_step_conf": 0.9333306550979614,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.5982724440668366,
"calib/avg_num_step_conf": 11.66796875,
"calib/ece": 0.4176569037656903,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005846785613140892,
"calib/mean_conf": 0.9699581589958158,
"calib/mu_c": 0.9725757575757575,
"calib/mu_w": 0.9667289719626166,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4176569037656903,
"calib/std_conf": 0.014067913815529444,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8376606145251396,
"calib/step_q_c_n": 1432.0,
"calib/step_q_gap": 0.02402074314250291,
"calib/step_q_w": 0.8136398713826367,
"calib/step_q_w_n": 1555.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2779.0,
"completions/max_terminated_length": 2779.0,
"completions/mean_length": 911.06640625,
"completions/mean_terminated_length": 955.8729248046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 436.0,
"epoch": 0.07893333333333333,
"grad_norm": 0.01728581264615059,
"kl": 0.1635284423828125,
"learning_rate": 3.5e-06,
"loss": -0.1305,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.015495438128709793,
"mask/share_reasoning": 0.814774751663208,
"mask/share_step_conf": 0.12285478413105011,
"num_tokens": 24243734.0,
"reward": 0.7644367218017578,
"reward_std": 0.29110610485076904,
"rewards/accuracy_reward_step": 0.515625,
"rewards/asymmetric_l2_reward": 0.6974785327911377,
"rewards/final_brier_reward_step": 0.5423324108123779,
"rewards/format_reward_step": 0.9296875,
"step": 74
},
{
"adv/mean_abs_final_conf": 0.7328017950057983,
"adv/mean_abs_reasoning": 0.49293094873428345,
"adv/mean_abs_step_conf": 0.751336932182312,
"adv/ratio_final_to_reasoning": 1.4866215985980185,
"adv/ratio_step_to_reasoning": 1.5242234923807216,
"adv/std_final_conf": 0.8824196457862854,
"adv/std_reasoning": 0.7577341794967651,
"adv/std_step_conf": 0.929182767868042,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.5827731092436974,
"calib/avg_num_step_conf": 10.87109375,
"calib/ece": 0.2505349794238683,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.9958847736625515,
"calib/gap": 0.0043621848739494595,
"calib/mean_conf": 0.9694650205761317,
"calib/mu_c": 0.970685714285714,
"calib/mu_w": 0.9663235294117646,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.24991769547325102,
"calib/std_conf": 0.01597715095289499,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.8350551267916207,
"calib/step_q_c_n": 1814.0,
"calib/step_q_gap": 0.006159357957771339,
"calib/step_q_w": 0.8288957688338494,
"calib/step_q_w_n": 969.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2881.0,
"completions/max_terminated_length": 2881.0,
"completions/mean_length": 850.83984375,
"completions/mean_terminated_length": 874.759033203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 435.0,
"epoch": 0.08,
"grad_norm": 0.06992490589618683,
"kl": 0.1568603515625,
"learning_rate": 3.4722222222222224e-06,
"loss": -0.1116,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01692769303917885,
"mask/share_reasoning": 0.8271828889846802,
"mask/share_step_conf": 0.12854570150375366,
"num_tokens": 24566301.0,
"reward": 0.8857764005661011,
"reward_std": 0.25138625502586365,
"rewards/accuracy_reward_step": 0.68359375,
"rewards/asymmetric_l2_reward": 0.754860520362854,
"rewards/final_brier_reward_step": 0.6924734115600586,
"rewards/format_reward_step": 0.9375,
"step": 75
},
{
"adv/mean_abs_final_conf": 0.7342768311500549,
"adv/mean_abs_reasoning": 0.4722954034805298,
"adv/mean_abs_step_conf": 0.749976634979248,
"adv/ratio_final_to_reasoning": 1.5546982370331819,
"adv/ratio_step_to_reasoning": 1.587939728933156,
"adv/std_final_conf": 0.8730382323265076,
"adv/std_reasoning": 0.7395159602165222,
"adv/std_step_conf": 0.9327054619789124,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6478126380910296,
"calib/avg_num_step_conf": 10.921875,
"calib/ece": 0.36020920502092035,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008645603181617445,
"calib/mean_conf": 0.9710878661087865,
"calib/mu_c": 0.9744520547945205,
"calib/mu_w": 0.965806451612903,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36020920502092035,
"calib/std_conf": 0.014422952826999605,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8340337224383917,
"calib/step_q_c_n": 1542.0,
"calib/step_q_gap": 0.018419687350672453,
"calib/step_q_w": 0.8156140350877192,
"calib/step_q_w_n": 1254.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2819.0,
"completions/max_terminated_length": 2819.0,
"completions/mean_length": 876.48828125,
"completions/mean_terminated_length": 915.8407592773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 432.0,
"epoch": 0.08106666666666666,
"grad_norm": 0.02016083337366581,
"kl": 0.153961181640625,
"learning_rate": 3.444444444444445e-06,
"loss": -0.2091,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.015996340662240982,
"mask/share_reasoning": 0.8220898509025574,
"mask/share_step_conf": 0.11894506961107254,
"num_tokens": 24893738.0,
"reward": 0.7962677478790283,
"reward_std": 0.24296647310256958,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/asymmetric_l2_reward": 0.6975725889205933,
"rewards/final_brier_reward_step": 0.5941816568374634,
"rewards/format_reward_step": 0.93359375,
"step": 76
},
{
"adv/mean_abs_final_conf": 0.7706581354141235,
"adv/mean_abs_reasoning": 0.5369069576263428,
"adv/mean_abs_step_conf": 0.7580994963645935,
"adv/ratio_final_to_reasoning": 1.4353662668503888,
"adv/ratio_step_to_reasoning": 1.411975549201559,
"adv/std_final_conf": 0.8862397074699402,
"adv/std_reasoning": 0.7756330370903015,
"adv/std_step_conf": 0.9303173422813416,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.617000469410108,
"calib/avg_num_step_conf": 12.078125,
"calib/ece": 0.28868312757201664,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006833046471601212,
"calib/mean_conf": 0.9718106995884775,
"calib/mu_c": 0.9739759036144577,
"calib/mu_w": 0.9671428571428565,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.28868312757201664,
"calib/std_conf": 0.014544283475479397,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.834141914191419,
"calib/step_q_c_n": 1818.0,
"calib/step_q_gap": -0.0021767671272623357,
"calib/step_q_w": 0.8363186813186814,
"calib/step_q_w_n": 1274.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2651.0,
"completions/max_terminated_length": 2651.0,
"completions/mean_length": 891.25,
"completions/mean_terminated_length": 923.7247314453125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 479.0,
"epoch": 0.08213333333333334,
"grad_norm": 0.05896781384944916,
"kl": 0.1526947021484375,
"learning_rate": 3.416666666666667e-06,
"loss": -0.1115,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.015972012653946877,
"mask/share_reasoning": 0.8222805261611938,
"mask/share_step_conf": 0.12659120559692383,
"num_tokens": 25226562.0,
"reward": 0.8617645502090454,
"reward_std": 0.26767075061798096,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/asymmetric_l2_reward": 0.7367497682571411,
"rewards/final_brier_reward_step": 0.667248010635376,
"rewards/format_reward_step": 0.94921875,
"step": 77
},
{
"adv/mean_abs_final_conf": 0.7285124063491821,
"adv/mean_abs_reasoning": 0.47509482502937317,
"adv/mean_abs_step_conf": 0.7568221688270569,
"adv/ratio_final_to_reasoning": 1.5334042131571128,
"adv/ratio_step_to_reasoning": 1.5929918175393,
"adv/std_final_conf": 0.8647627830505371,
"adv/std_reasoning": 0.7206560969352722,
"adv/std_step_conf": 0.9324377775192261,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5533693415637861,
"calib/avg_num_step_conf": 10.16796875,
"calib/ece": 0.4031349206349205,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0031250000000002665,
"calib/mean_conf": 0.9745634920634919,
"calib/mu_c": 0.9759027777777778,
"calib/mu_w": 0.9727777777777775,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4031349206349205,
"calib/std_conf": 0.014993647348839642,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8269742198100407,
"calib/step_q_c_n": 1474.0,
"calib/step_q_gap": 0.0038032720686768107,
"calib/step_q_w": 0.8231709477413639,
"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": 0.0078125,
"completions/max_length": 2514.0,
"completions/max_terminated_length": 2514.0,
"completions/mean_length": 942.19921875,
"completions/mean_terminated_length": 949.6181030273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 500.0,
"epoch": 0.0832,
"grad_norm": 0.021009007468819618,
"kl": 0.1553192138671875,
"learning_rate": 3.3888888888888893e-06,
"loss": -0.0389,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.015865862369537354,
"mask/share_reasoning": 0.8624191284179688,
"mask/share_step_conf": 0.11390247195959091,
"num_tokens": 25575789.0,
"reward": 0.8056474328041077,
"reward_std": 0.2189893126487732,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.7173092365264893,
"rewards/final_brier_reward_step": 0.5846105217933655,
"rewards/format_reward_step": 0.984375,
"step": 78
},
{
"adv/mean_abs_final_conf": 0.6996117234230042,
"adv/mean_abs_reasoning": 0.5063235759735107,
"adv/mean_abs_step_conf": 0.7326298356056213,
"adv/ratio_final_to_reasoning": 1.381748266566212,
"adv/ratio_step_to_reasoning": 1.4469597513743864,
"adv/std_final_conf": 0.8834125399589539,
"adv/std_reasoning": 0.7755327224731445,
"adv/std_step_conf": 0.9304167032241821,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5155382907880133,
"calib/avg_num_step_conf": 10.75390625,
"calib/ece": 0.3303688524590165,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.9918032786885246,
"calib/gap": -0.0016374398816130231,
"calib/mean_conf": 0.9779918032786885,
"calib/mu_c": 0.9774213836477985,
"calib/mu_w": 0.9790588235294115,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.328360655737705,
"calib/std_conf": 0.025099472890304626,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8302879424115178,
"calib/step_q_c_n": 1667.0,
"calib/step_q_gap": 0.013160870588313278,
"calib/step_q_w": 0.8171270718232045,
"calib/step_q_w_n": 1086.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2940.0,
"completions/max_terminated_length": 2940.0,
"completions/mean_length": 880.53125,
"completions/mean_terminated_length": 912.6154174804688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 497.0,
"epoch": 0.08426666666666667,
"grad_norm": 0.019594348967075348,
"kl": 0.1408843994140625,
"learning_rate": 3.3611111111111117e-06,
"loss": -0.1907,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01564885675907135,
"mask/share_reasoning": 0.8325054049491882,
"mask/share_step_conf": 0.11668950319290161,
"num_tokens": 25907581.0,
"reward": 0.8450396060943604,
"reward_std": 0.25687703490257263,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/asymmetric_l2_reward": 0.7405940294265747,
"rewards/final_brier_reward_step": 0.6338601112365723,
"rewards/format_reward_step": 0.94921875,
"step": 79
},
{
"adv/mean_abs_final_conf": 0.6465312242507935,
"adv/mean_abs_reasoning": 0.5463018417358398,
"adv/mean_abs_step_conf": 0.7440537214279175,
"adv/ratio_final_to_reasoning": 1.1834688717074082,
"adv/ratio_step_to_reasoning": 1.3619828171615411,
"adv/std_final_conf": 0.8512998819351196,
"adv/std_reasoning": 0.7928522825241089,
"adv/std_step_conf": 0.9313137531280518,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5054459817486017,
"calib/avg_num_step_conf": 11.00390625,
"calib/ece": 0.2996015936254981,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0004430379746838131,
"calib/mean_conf": 0.9848605577689243,
"calib/mu_c": 0.9849999999999999,
"calib/mu_w": 0.9845569620253161,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2996015936254981,
"calib/std_conf": 0.011054035970751991,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8219946666666665,
"calib/step_q_c_n": 1875.0,
"calib/step_q_gap": 0.022281290870488246,
"calib/step_q_w": 0.7997133757961783,
"calib/step_q_w_n": 942.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2486.0,
"completions/max_terminated_length": 2486.0,
"completions/mean_length": 858.70703125,
"completions/mean_terminated_length": 868.8893432617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 484.0,
"epoch": 0.08533333333333333,
"grad_norm": 0.016964834183454514,
"kl": 0.1446990966796875,
"learning_rate": 3.3333333333333333e-06,
"loss": -0.0618,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.01695704460144043,
"mask/share_reasoning": 0.8412275910377502,
"mask/share_step_conf": 0.13009659945964813,
"num_tokens": 26229570.0,
"reward": 0.8783482313156128,
"reward_std": 0.2699282169342041,
"rewards/accuracy_reward_step": 0.671875,
"rewards/asymmetric_l2_reward": 0.7498530149459839,
"rewards/final_brier_reward_step": 0.6771558523178101,
"rewards/format_reward_step": 0.9765625,
"step": 80
},
{
"adv/mean_abs_final_conf": 0.6169295907020569,
"adv/mean_abs_reasoning": 0.5010291934013367,
"adv/mean_abs_step_conf": 0.7242743968963623,
"adv/ratio_final_to_reasoning": 1.231324639017354,
"adv/ratio_step_to_reasoning": 1.4455732449031184,
"adv/std_final_conf": 0.8448789119720459,
"adv/std_reasoning": 0.7930719256401062,
"adv/std_step_conf": 0.9312515258789062,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.6767689061992859,
"calib/avg_num_step_conf": 11.25390625,
"calib/ece": 0.313135593220339,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 0.9957627118644068,
"calib/gap": 0.010243427458617727,
"calib/mean_conf": 0.9826271186440678,
"calib/mu_c": 0.9860126582278479,
"calib/mu_w": 0.9757692307692302,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.313135593220339,
"calib/std_conf": 0.013711801554401222,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8287856257744732,
"calib/step_q_c_n": 1614.0,
"calib/step_q_gap": 0.01665460762135562,
"calib/step_q_w": 0.8121310181531176,
"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": 0.046875,
"completions/max_length": 2737.0,
"completions/max_terminated_length": 2737.0,
"completions/mean_length": 894.08203125,
"completions/mean_terminated_length": 938.05322265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 503.0,
"epoch": 0.0864,
"grad_norm": 0.01762450486421585,
"kl": 0.13702392578125,
"learning_rate": 3.3055555555555558e-06,
"loss": -0.1868,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.016055608168244362,
"mask/share_reasoning": 0.8190659284591675,
"mask/share_step_conf": 0.1180034726858139,
"num_tokens": 26564703.0,
"reward": 0.8258570432662964,
"reward_std": 0.27940231561660767,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/asymmetric_l2_reward": 0.7132589221000671,
"rewards/final_brier_reward_step": 0.631423830986023,
"rewards/format_reward_step": 0.91796875,
"step": 81
},
{
"adv/mean_abs_final_conf": 0.5887628793716431,
"adv/mean_abs_reasoning": 0.4778270125389099,
"adv/mean_abs_step_conf": 0.7516727447509766,
"adv/ratio_final_to_reasoning": 1.2321674244477745,
"adv/ratio_step_to_reasoning": 1.5731064276944098,
"adv/std_final_conf": 0.7913881540298462,
"adv/std_reasoning": 0.7394364476203918,
"adv/std_step_conf": 0.929599940776825,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5940373363053776,
"calib/avg_num_step_conf": 11.39453125,
"calib/ece": 0.38155102040816335,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9959183673469387,
"calib/gap": 0.005399136249651604,
"calib/mean_conf": 0.9856326530612246,
"calib/mu_c": 0.9877702702702702,
"calib/mu_w": 0.9823711340206186,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.38155102040816335,
"calib/std_conf": 0.011397625741929896,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.833109296482412,
"calib/step_q_c_n": 1592.0,
"calib/step_q_gap": 0.018347032331468727,
"calib/step_q_w": 0.8147622641509433,
"calib/step_q_w_n": 1325.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2572.0,
"completions/max_terminated_length": 2572.0,
"completions/mean_length": 838.4921875,
"completions/mean_terminated_length": 858.6160278320312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 517.0,
"epoch": 0.08746666666666666,
"grad_norm": 0.020330313593149185,
"kl": 0.1495513916015625,
"learning_rate": 3.277777777777778e-06,
"loss": -0.1235,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.017101511359214783,
"mask/share_reasoning": 0.8278982043266296,
"mask/share_step_conf": 0.13156276941299438,
"num_tokens": 26884909.0,
"reward": 0.8035461902618408,
"reward_std": 0.219723179936409,
"rewards/accuracy_reward_step": 0.578125,
"rewards/asymmetric_l2_reward": 0.7088985443115234,
"rewards/final_brier_reward_step": 0.5911625027656555,
"rewards/format_reward_step": 0.95703125,
"step": 82
},
{
"adv/mean_abs_final_conf": 0.6281593441963196,
"adv/mean_abs_reasoning": 0.469433069229126,
"adv/mean_abs_step_conf": 0.7515369653701782,
"adv/ratio_final_to_reasoning": 1.3381233350855408,
"adv/ratio_step_to_reasoning": 1.60094593805312,
"adv/std_final_conf": 0.841602087020874,
"adv/std_reasoning": 0.7394808530807495,
"adv/std_step_conf": 0.9299801588058472,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.5899823633156966,
"calib/avg_num_step_conf": 10.578125,
"calib/ece": 0.4223333333333332,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.9958333333333333,
"calib/gap": 0.0019894179894178743,
"calib/mean_conf": 0.9808333333333334,
"calib/mu_c": 0.9817037037037035,
"calib/mu_w": 0.9797142857142856,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4203333333333332,
"calib/std_conf": 0.03284010082539673,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8315280898876406,
"calib/step_q_c_n": 1335.0,
"calib/step_q_gap": 0.01302117073250586,
"calib/step_q_w": 0.8185069191551347,
"calib/step_q_w_n": 1373.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2912.0,
"completions/max_terminated_length": 2912.0,
"completions/mean_length": 923.578125,
"completions/mean_terminated_length": 949.5421142578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 423.0,
"epoch": 0.08853333333333334,
"grad_norm": 0.019025607034564018,
"kl": 0.1397857666015625,
"learning_rate": 3.2500000000000002e-06,
"loss": -0.1334,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.015944354236125946,
"mask/share_reasoning": 0.8402930498123169,
"mask/share_step_conf": 0.11641887575387955,
"num_tokens": 27228609.0,
"reward": 0.7492298483848572,
"reward_std": 0.22660614550113678,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/asymmetric_l2_reward": 0.6692612171173096,
"rewards/final_brier_reward_step": 0.5385733842849731,
"rewards/format_reward_step": 0.92578125,
"step": 83
},
{
"adv/mean_abs_final_conf": 0.6215704679489136,
"adv/mean_abs_reasoning": 0.4951096475124359,
"adv/mean_abs_step_conf": 0.7436130046844482,
"adv/ratio_final_to_reasoning": 1.255419826844116,
"adv/ratio_step_to_reasoning": 1.5019158047526644,
"adv/std_final_conf": 0.8289499282836914,
"adv/std_reasoning": 0.7576406002044678,
"adv/std_step_conf": 0.9340832233428955,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.576388888888889,
"calib/avg_num_step_conf": 11.8359375,
"calib/ece": 0.3599586776859504,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.987603305785124,
"calib/gap": 0.0008771929824562541,
"calib/mean_conf": 0.9788842975206613,
"calib/mu_c": 0.9792105263157895,
"calib/mu_w": 0.9783333333333333,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35537190082644626,
"calib/std_conf": 0.05506565428974546,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.829752504419564,
"calib/step_q_c_n": 1697.0,
"calib/step_q_gap": 0.025296390391057022,
"calib/step_q_w": 0.804456114028507,
"calib/step_q_w_n": 1333.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2814.0,
"completions/max_terminated_length": 2814.0,
"completions/mean_length": 848.2734375,
"completions/mean_terminated_length": 886.359130859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 441.0,
"epoch": 0.0896,
"grad_norm": 0.017854604870080948,
"kl": 0.14837646484375,
"learning_rate": 3.2222222222222227e-06,
"loss": -0.1635,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016548432409763336,
"mask/share_reasoning": 0.8113114833831787,
"mask/share_step_conf": 0.12917135655879974,
"num_tokens": 27551687.0,
"reward": 0.8223565816879272,
"reward_std": 0.24511127173900604,
"rewards/accuracy_reward_step": 0.59375,
"rewards/asymmetric_l2_reward": 0.7312041521072388,
"rewards/final_brier_reward_step": 0.605696439743042,
"rewards/format_reward_step": 0.9453125,
"step": 84
},
{
"adv/mean_abs_final_conf": 0.6229391694068909,
"adv/mean_abs_reasoning": 0.5074774026870728,
"adv/mean_abs_step_conf": 0.7801123261451721,
"adv/ratio_final_to_reasoning": 1.2275210011489233,
"adv/ratio_step_to_reasoning": 1.5372355931801263,
"adv/std_final_conf": 0.8200711011886597,
"adv/std_reasoning": 0.739483654499054,
"adv/std_step_conf": 0.9316701292991638,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6454227812718379,
"calib/avg_num_step_conf": 10.38671875,
"calib/ece": 0.42203319502074677,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.9875518672199171,
"calib/gap": 0.013357092941998605,
"calib/mean_conf": 0.9821991701244813,
"calib/mu_c": 0.988074074074074,
"calib/mu_w": 0.9747169811320754,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42203319502074677,
"calib/std_conf": 0.025599406748347184,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8299927272727274,
"calib/step_q_c_n": 1375.0,
"calib/step_q_gap": 0.008754409515718042,
"calib/step_q_w": 0.8212383177570094,
"calib/step_q_w_n": 1284.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2992.0,
"completions/max_terminated_length": 2992.0,
"completions/mean_length": 896.77734375,
"completions/mean_terminated_length": 929.4534912109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 476.0,
"epoch": 0.09066666666666667,
"grad_norm": 0.030835958197712898,
"kl": 0.145050048828125,
"learning_rate": 3.1944444444444443e-06,
"loss": -0.0587,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.015947245061397552,
"mask/share_reasoning": 0.8292930126190186,
"mask/share_step_conf": 0.11960351467132568,
"num_tokens": 27889086.0,
"reward": 0.7618950605392456,
"reward_std": 0.23239392042160034,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/asymmetric_l2_reward": 0.6826740503311157,
"rewards/final_brier_reward_step": 0.5473659634590149,
"rewards/format_reward_step": 0.94140625,
"step": 85
},
{
"adv/mean_abs_final_conf": 0.5596708059310913,
"adv/mean_abs_reasoning": 0.398525208234787,
"adv/mean_abs_step_conf": 0.7483077049255371,
"adv/ratio_final_to_reasoning": 1.4043548422195844,
"adv/ratio_step_to_reasoning": 1.8776922750760585,
"adv/std_final_conf": 0.7832596898078918,
"adv/std_reasoning": 0.6817492246627808,
"adv/std_step_conf": 0.9316436648368835,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5768398268398268,
"calib/avg_num_step_conf": 11.32421875,
"calib/ece": 0.43963114754098376,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9877049180327869,
"calib/gap": 0.007843614718614389,
"calib/mean_conf": 0.9789754098360656,
"calib/mu_c": 0.9825757575757574,
"calib/mu_w": 0.974732142857143,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4388114754098362,
"calib/std_conf": 0.03275867126219898,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.82452216066482,
"calib/step_q_c_n": 1444.0,
"calib/step_q_gap": 0.017264428706057156,
"calib/step_q_w": 0.8072577319587628,
"calib/step_q_w_n": 1455.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2710.0,
"completions/max_terminated_length": 2710.0,
"completions/mean_length": 880.98828125,
"completions/mean_terminated_length": 920.5428466796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 461.0,
"epoch": 0.09173333333333333,
"grad_norm": 0.021164577454328537,
"kl": 0.142791748046875,
"learning_rate": 3.1666666666666667e-06,
"loss": -0.1413,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.016112584620714188,
"mask/share_reasoning": 0.8145743608474731,
"mask/share_step_conf": 0.12634429335594177,
"num_tokens": 28220131.0,
"reward": 0.7546988725662231,
"reward_std": 0.1984507292509079,
"rewards/accuracy_reward_step": 0.515625,
"rewards/asymmetric_l2_reward": 0.6793574690818787,
"rewards/final_brier_reward_step": 0.5362902283668518,
"rewards/format_reward_step": 0.953125,
"step": 86
},
{
"adv/mean_abs_final_conf": 0.6093852519989014,
"adv/mean_abs_reasoning": 0.5044052600860596,
"adv/mean_abs_step_conf": 0.754356324672699,
"adv/ratio_final_to_reasoning": 1.208126283010868,
"adv/ratio_step_to_reasoning": 1.4955361975091095,
"adv/std_final_conf": 0.8310155272483826,
"adv/std_reasoning": 0.7755122780799866,
"adv/std_step_conf": 0.9313125014305115,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6357083678541839,
"calib/avg_num_step_conf": 10.44140625,
"calib/ece": 0.2630290456431535,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.9377593360995851,
"calib/gap": 0.03550207125103577,
"calib/mean_conf": 0.9684232365145228,
"calib/mu_c": 0.9788823529411764,
"calib/mu_w": 0.9433802816901407,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2630290456431535,
"calib/std_conf": 0.07947965709286663,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8273033067274801,
"calib/step_q_c_n": 1754.0,
"calib/step_q_gap": 0.01316837745653332,
"calib/step_q_w": 0.8141349292709468,
"calib/step_q_w_n": 919.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2499.0,
"completions/max_terminated_length": 2499.0,
"completions/mean_length": 812.0078125,
"completions/mean_terminated_length": 855.4485473632812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 397.0,
"epoch": 0.0928,
"grad_norm": 0.026130422949790955,
"kl": 0.150726318359375,
"learning_rate": 3.138888888888889e-06,
"loss": -0.19,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016727639362215996,
"mask/share_reasoning": 0.8078656792640686,
"mask/share_step_conf": 0.12462540715932846,
"num_tokens": 28533501.0,
"reward": 0.8941546678543091,
"reward_std": 0.24176597595214844,
"rewards/accuracy_reward_step": 0.6640625,
"rewards/asymmetric_l2_reward": 0.7786323428153992,
"rewards/final_brier_reward_step": 0.6885831952095032,
"rewards/format_reward_step": 0.94140625,
"step": 87
},
{
"adv/mean_abs_final_conf": 0.6701643466949463,
"adv/mean_abs_reasoning": 0.46769726276397705,
"adv/mean_abs_step_conf": 0.7661531567573547,
"adv/ratio_final_to_reasoning": 1.4329020074533643,
"adv/ratio_step_to_reasoning": 1.638139065064388,
"adv/std_final_conf": 0.8791421055793762,
"adv/std_reasoning": 0.7394110560417175,
"adv/std_step_conf": 0.9323065876960754,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5340203274985884,
"calib/avg_num_step_conf": 10.29296875,
"calib/ece": 0.3519105691056911,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9390243902439024,
"calib/gap": -0.000447487295313187,
"calib/mean_conf": 0.968089430894309,
"calib/mu_c": 0.967922077922078,
"calib/mu_w": 0.9683695652173911,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34699186991869924,
"calib/std_conf": 0.06487643140392849,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8267061923583663,
"calib/step_q_c_n": 1518.0,
"calib/step_q_gap": 0.013993569260783545,
"calib/step_q_w": 0.8127126230975827,
"calib/step_q_w_n": 1117.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2781.0,
"completions/max_terminated_length": 2781.0,
"completions/mean_length": 887.2734375,
"completions/mean_terminated_length": 912.2168579101562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 413.0,
"epoch": 0.09386666666666667,
"grad_norm": 0.030685633420944214,
"kl": 0.137176513671875,
"learning_rate": 3.1111111111111116e-06,
"loss": -0.0492,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.016321662813425064,
"mask/share_reasoning": 0.8399770259857178,
"mask/share_step_conf": 0.11635757982730865,
"num_tokens": 28870491.0,
"reward": 0.8365731835365295,
"reward_std": 0.23255041241645813,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/asymmetric_l2_reward": 0.7413724660873413,
"rewards/final_brier_reward_step": 0.619273841381073,
"rewards/format_reward_step": 0.9609375,
"step": 88
},
{
"adv/mean_abs_final_conf": 0.691696047782898,
"adv/mean_abs_reasoning": 0.44166111946105957,
"adv/mean_abs_step_conf": 0.7274096012115479,
"adv/ratio_final_to_reasoning": 1.5661239291947306,
"adv/ratio_step_to_reasoning": 1.6469858204842098,
"adv/std_final_conf": 0.8953777551651001,
"adv/std_reasoning": 0.7394091486930847,
"adv/std_step_conf": 0.9307010769844055,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.5642153781688665,
"calib/avg_num_step_conf": 10.90625,
"calib/ece": 0.4304583333333334,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.9291666666666667,
"calib/gap": 0.0010580347789652667,
"calib/mean_conf": 0.9635416666666666,
"calib/mu_c": 0.9640310077519381,
"calib/mu_w": 0.9629729729729728,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4282500000000001,
"calib/std_conf": 0.07483007370406335,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8260374531835207,
"calib/step_q_c_n": 1335.0,
"calib/step_q_gap": 0.013703891069588026,
"calib/step_q_w": 0.8123335621139327,
"calib/step_q_w_n": 1457.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3033.0,
"completions/max_terminated_length": 3033.0,
"completions/mean_length": 898.11328125,
"completions/mean_terminated_length": 927.0846557617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 491.0,
"epoch": 0.09493333333333333,
"grad_norm": 0.02352975867688656,
"kl": 0.142669677734375,
"learning_rate": 3.0833333333333336e-06,
"loss": -0.0897,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.01606300100684166,
"mask/share_reasoning": 0.8333423137664795,
"mask/share_step_conf": 0.11934471130371094,
"num_tokens": 29209296.0,
"reward": 0.7457454204559326,
"reward_std": 0.2067442238330841,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/asymmetric_l2_reward": 0.6747016906738281,
"rewards/final_brier_reward_step": 0.5292890071868896,
"rewards/format_reward_step": 0.93359375,
"step": 89
},
{
"adv/mean_abs_final_conf": 0.6971824169158936,
"adv/mean_abs_reasoning": 0.46878379583358765,
"adv/mean_abs_step_conf": 0.748717188835144,
"adv/ratio_final_to_reasoning": 1.4872152645041181,
"adv/ratio_step_to_reasoning": 1.597148185345829,
"adv/std_final_conf": 0.8981074094772339,
"adv/std_reasoning": 0.7394995093345642,
"adv/std_step_conf": 0.9299719929695129,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5588279042094303,
"calib/avg_num_step_conf": 10.8828125,
"calib/ece": 0.31391836734693884,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.926530612244898,
"calib/gap": -0.0013595121225642437,
"calib/mean_conf": 0.9640408163265306,
"calib/mu_c": 0.9635802469135802,
"calib/mu_w": 0.9649397590361445,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3083673469387756,
"calib/std_conf": 0.07823824399757069,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8285641605426795,
"calib/step_q_c_n": 1769.0,
"calib/step_q_gap": 0.015279991417802341,
"calib/step_q_w": 0.8132841691248771,
"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": 0.01953125,
"completions/max_length": 2880.0,
"completions/max_terminated_length": 2880.0,
"completions/mean_length": 905.1875,
"completions/mean_terminated_length": 923.2191772460938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 423.0,
"epoch": 0.096,
"grad_norm": 0.058208346366882324,
"kl": 0.141021728515625,
"learning_rate": 3.055555555555556e-06,
"loss": -0.073,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016423799097537994,
"mask/share_reasoning": 0.8371864557266235,
"mask/share_step_conf": 0.12685847282409668,
"num_tokens": 29544344.0,
"reward": 0.8589211702346802,
"reward_std": 0.24496112763881683,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/asymmetric_l2_reward": 0.7514224052429199,
"rewards/final_brier_reward_step": 0.6484510898590088,
"rewards/format_reward_step": 0.95703125,
"step": 90
},
{
"adv/mean_abs_final_conf": 0.6069288849830627,
"adv/mean_abs_reasoning": 0.4890816807746887,
"adv/mean_abs_step_conf": 0.7374441623687744,
"adv/ratio_final_to_reasoning": 1.2409560792007341,
"adv/ratio_step_to_reasoning": 1.5078139119843703,
"adv/std_final_conf": 0.8355866074562073,
"adv/std_reasoning": 0.7576850652694702,
"adv/std_step_conf": 0.9319037795066833,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5358258928571429,
"calib/avg_num_step_conf": 11.01953125,
"calib/ece": 0.3284836065573772,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.9467213114754098,
"calib/gap": -0.0008928571428568066,
"calib/mean_conf": 0.9690573770491804,
"calib/mu_c": 0.96875,
"calib/mu_w": 0.9696428571428568,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.32090163934426247,
"calib/std_conf": 0.06612467102156572,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8274085952533674,
"calib/step_q_c_n": 1559.0,
"calib/step_q_gap": 0.01514235119631513,
"calib/step_q_w": 0.8122662440570523,
"calib/step_q_w_n": 1262.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2798.0,
"completions/max_terminated_length": 2798.0,
"completions/mean_length": 904.19140625,
"completions/mean_terminated_length": 937.1376953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 541.0,
"epoch": 0.09706666666666666,
"grad_norm": 0.037474095821380615,
"kl": 0.14019775390625,
"learning_rate": 3.0277777777777776e-06,
"loss": -0.1753,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.015775000676512718,
"mask/share_reasoning": 0.8341672420501709,
"mask/share_step_conf": 0.11490151286125183,
"num_tokens": 29883529.0,
"reward": 0.8479245901107788,
"reward_std": 0.23241955041885376,
"rewards/accuracy_reward_step": 0.625,
"rewards/asymmetric_l2_reward": 0.7403823137283325,
"rewards/final_brier_reward_step": 0.6398417949676514,
"rewards/format_reward_step": 0.953125,
"step": 91
},
{
"adv/mean_abs_final_conf": 0.6295256018638611,
"adv/mean_abs_reasoning": 0.48855841159820557,
"adv/mean_abs_step_conf": 0.7306718230247498,
"adv/ratio_final_to_reasoning": 1.2885370242721113,
"adv/ratio_step_to_reasoning": 1.4955669694326341,
"adv/std_final_conf": 0.8479897975921631,
"adv/std_reasoning": 0.7754259705543518,
"adv/std_step_conf": 0.929085910320282,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5820178448867536,
"calib/avg_num_step_conf": 10.65234375,
"calib/ece": 0.3611646586345382,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9919678714859438,
"calib/gap": 0.007409059711736665,
"calib/mean_conf": 0.9836546184738956,
"calib/mu_c": 0.9864516129032257,
"calib/mu_w": 0.979042553191489,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3611646586345382,
"calib/std_conf": 0.0217298396792282,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8437311557788943,
"calib/step_q_c_n": 1592.0,
"calib/step_q_gap": 0.010224547849378984,
"calib/step_q_w": 0.8335066079295154,
"calib/step_q_w_n": 1135.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2505.0,
"completions/max_terminated_length": 2505.0,
"completions/mean_length": 826.0078125,
"completions/mean_terminated_length": 839.1190795898438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 471.0,
"epoch": 0.09813333333333334,
"grad_norm": 0.015464968979358673,
"kl": 0.1487884521484375,
"learning_rate": 3e-06,
"loss": -0.0939,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01752406731247902,
"mask/share_reasoning": 0.833724856376648,
"mask/share_step_conf": 0.13312605023384094,
"num_tokens": 30201707.0,
"reward": 0.844549298286438,
"reward_std": 0.23641173541545868,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/asymmetric_l2_reward": 0.753333330154419,
"rewards/final_brier_reward_step": 0.6201401948928833,
"rewards/format_reward_step": 0.97265625,
"step": 92
},
{
"adv/mean_abs_final_conf": 0.5908430218696594,
"adv/mean_abs_reasoning": 0.513970673084259,
"adv/mean_abs_step_conf": 0.7497128844261169,
"adv/ratio_final_to_reasoning": 1.149565632459341,
"adv/ratio_step_to_reasoning": 1.458668604430687,
"adv/std_final_conf": 0.8375146985054016,
"adv/std_reasoning": 0.79283207654953,
"adv/std_step_conf": 0.9285351634025574,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6064102564102565,
"calib/avg_num_step_conf": 11.14453125,
"calib/ece": 0.34967479674796753,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9959349593495935,
"calib/gap": 0.008128205128205113,
"calib/mean_conf": 0.9838211382113821,
"calib/mu_c": 0.9867948717948718,
"calib/mu_w": 0.9786666666666667,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34967479674796753,
"calib/std_conf": 0.0176699793735646,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8417461263408821,
"calib/step_q_c_n": 1678.0,
"calib/step_q_gap": 0.014946126340882104,
"calib/step_q_w": 0.8268,
"calib/step_q_w_n": 1175.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2862.0,
"completions/max_terminated_length": 2862.0,
"completions/mean_length": 850.3984375,
"completions/mean_terminated_length": 877.8306274414062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 438.0,
"epoch": 0.0992,
"grad_norm": 0.02208016626536846,
"kl": 0.137359619140625,
"learning_rate": 2.9722222222222225e-06,
"loss": -0.1543,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.016855405643582344,
"mask/share_reasoning": 0.8227949738502502,
"mask/share_step_conf": 0.12909963726997375,
"num_tokens": 30525185.0,
"reward": 0.8347985744476318,
"reward_std": 0.2563871741294861,
"rewards/accuracy_reward_step": 0.609375,
"rewards/asymmetric_l2_reward": 0.7325701117515564,
"rewards/final_brier_reward_step": 0.6237456798553467,
"rewards/format_reward_step": 0.95703125,
"step": 93
},
{
"adv/mean_abs_final_conf": 0.5753787755966187,
"adv/mean_abs_reasoning": 0.5310263633728027,
"adv/mean_abs_step_conf": 0.7260737419128418,
"adv/ratio_final_to_reasoning": 1.0835220532971517,
"adv/ratio_step_to_reasoning": 1.3673026275027096,
"adv/std_final_conf": 0.8271228075027466,
"adv/std_reasoning": 0.7928115725517273,
"adv/std_step_conf": 0.931684136390686,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5608802961744138,
"calib/avg_num_step_conf": 10.5078125,
"calib/ece": 0.4028163265306123,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00356026326614578,
"calib/mean_conf": 0.9864897959183674,
"calib/mu_c": 0.9879720279720278,
"calib/mu_w": 0.984411764705882,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4028163265306123,
"calib/std_conf": 0.009472026374109198,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8471638655462185,
"calib/step_q_c_n": 1428.0,
"calib/step_q_gap": 0.01842377045905519,
"calib/step_q_w": 0.8287400950871633,
"calib/step_q_w_n": 1262.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2749.0,
"completions/max_terminated_length": 2749.0,
"completions/mean_length": 820.51953125,
"completions/mean_terminated_length": 843.5863037109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 470.0,
"epoch": 0.10026666666666667,
"grad_norm": 0.023399489000439644,
"kl": 0.1503753662109375,
"learning_rate": 2.944444444444445e-06,
"loss": -0.0828,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.0174710713326931,
"mask/share_reasoning": 0.8291428685188293,
"mask/share_step_conf": 0.12604230642318726,
"num_tokens": 30843918.0,
"reward": 0.7747670412063599,
"reward_std": 0.24693280458450317,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/asymmetric_l2_reward": 0.6803410053253174,
"rewards/final_brier_reward_step": 0.5668492317199707,
"rewards/format_reward_step": 0.953125,
"step": 94
},
{
"adv/mean_abs_final_conf": 0.4851650297641754,
"adv/mean_abs_reasoning": 0.40888628363609314,
"adv/mean_abs_step_conf": 0.7229888439178467,
"adv/ratio_final_to_reasoning": 1.1865524699184333,
"adv/ratio_step_to_reasoning": 1.7681905039429089,
"adv/std_final_conf": 0.7402144074440002,
"adv/std_reasoning": 0.7013313174247742,
"adv/std_step_conf": 0.92943274974823,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5654044117647059,
"calib/avg_num_step_conf": 10.4921875,
"calib/ece": 0.30435999999999996,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0036544117647059338,
"calib/mean_conf": 0.98436,
"calib/mu_c": 0.9855294117647059,
"calib/mu_w": 0.9818749999999999,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.30435999999999996,
"calib/std_conf": 0.011618536913054079,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8448533640023004,
"calib/step_q_c_n": 1739.0,
"calib/step_q_gap": 0.004853364002300409,
"calib/step_q_w": 0.84,
"calib/step_q_w_n": 947.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2896.0,
"completions/max_terminated_length": 2896.0,
"completions/mean_length": 849.15234375,
"completions/mean_terminated_length": 866.0677490234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 387.0,
"epoch": 0.10133333333333333,
"grad_norm": 0.013517041690647602,
"kl": 0.136505126953125,
"learning_rate": 2.916666666666667e-06,
"loss": -0.0705,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017546728253364563,
"mask/share_reasoning": 0.8325512409210205,
"mask/share_step_conf": 0.13037078082561493,
"num_tokens": 31167429.0,
"reward": 0.871839165687561,
"reward_std": 0.1962461769580841,
"rewards/accuracy_reward_step": 0.6640625,
"rewards/asymmetric_l2_reward": 0.7405334711074829,
"rewards/final_brier_reward_step": 0.6750198602676392,
"rewards/format_reward_step": 0.9765625,
"step": 95
},
{
"adv/mean_abs_final_conf": 0.5261593461036682,
"adv/mean_abs_reasoning": 0.468733549118042,
"adv/mean_abs_step_conf": 0.7362979650497437,
"adv/ratio_final_to_reasoning": 1.1225126665110217,
"adv/ratio_step_to_reasoning": 1.570824120516111,
"adv/std_final_conf": 0.7908560633659363,
"adv/std_reasoning": 0.7574588060379028,
"adv/std_step_conf": 0.9261574149131775,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6119724025974026,
"calib/avg_num_step_conf": 10.1875,
"calib/ece": 0.27158536585365856,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.9959349593495935,
"calib/gap": 0.006435064935065293,
"calib/mean_conf": 0.9870325203252033,
"calib/mu_c": 0.9888636363636365,
"calib/mu_w": 0.9824285714285712,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.27158536585365856,
"calib/std_conf": 0.010346386949860982,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.852012910798122,
"calib/step_q_c_n": 1704.0,
"calib/step_q_gap": 0.012189901948564508,
"calib/step_q_w": 0.8398230088495575,
"calib/step_q_w_n": 904.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2270.0,
"completions/max_terminated_length": 2270.0,
"completions/mean_length": 757.72265625,
"completions/mean_terminated_length": 782.165283203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 462.0,
"epoch": 0.1024,
"grad_norm": 0.015436707064509392,
"kl": 0.1490325927734375,
"learning_rate": 2.888888888888889e-06,
"loss": -0.1416,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.018313266336917877,
"mask/share_reasoning": 0.8223634958267212,
"mask/share_step_conf": 0.12807327508926392,
"num_tokens": 31467222.0,
"reward": 0.9001051187515259,
"reward_std": 0.2061901092529297,
"rewards/accuracy_reward_step": 0.6875,
"rewards/asymmetric_l2_reward": 0.7736778259277344,
"rewards/final_brier_reward_step": 0.6968449354171753,
"rewards/format_reward_step": 0.9609375,
"step": 96
},
{
"adv/mean_abs_final_conf": 0.6138774156570435,
"adv/mean_abs_reasoning": 0.5735644102096558,
"adv/mean_abs_step_conf": 0.7456457614898682,
"adv/ratio_final_to_reasoning": 1.0702850538314468,
"adv/ratio_step_to_reasoning": 1.3000209709966337,
"adv/std_final_conf": 0.8496063351631165,
"adv/std_reasoning": 0.8264009356498718,
"adv/std_step_conf": 0.9341704845428467,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5817634746206175,
"calib/avg_num_step_conf": 9.87109375,
"calib/ece": 0.3716929133858268,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.004733124018838186,
"calib/mean_conf": 0.9858661417322835,
"calib/mu_c": 0.9876923076923075,
"calib/mu_w": 0.9829591836734693,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3716929133858268,
"calib/std_conf": 0.010302405654948981,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8512392880685564,
"calib/step_q_c_n": 1517.0,
"calib/step_q_gap": 0.008981862325982082,
"calib/step_q_w": 0.8422574257425743,
"calib/step_q_w_n": 1010.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2489.0,
"completions/max_terminated_length": 2489.0,
"completions/mean_length": 841.16796875,
"completions/mean_terminated_length": 844.4667358398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 430.0,
"epoch": 0.10346666666666667,
"grad_norm": 0.014179172925651073,
"kl": 0.142333984375,
"learning_rate": 2.861111111111111e-06,
"loss": 0.023,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.017768729478120804,
"mask/share_reasoning": 0.8533096313476562,
"mask/share_step_conf": 0.12501543760299683,
"num_tokens": 31787633.0,
"reward": 0.8328101634979248,
"reward_std": 0.26197588443756104,
"rewards/accuracy_reward_step": 0.609375,
"rewards/asymmetric_l2_reward": 0.7231894731521606,
"rewards/final_brier_reward_step": 0.622118353843689,
"rewards/format_reward_step": 0.9921875,
"step": 97
},
{
"adv/mean_abs_final_conf": 0.6074684858322144,
"adv/mean_abs_reasoning": 0.4974590241909027,
"adv/mean_abs_step_conf": 0.7617883682250977,
"adv/ratio_final_to_reasoning": 1.2211427600901152,
"adv/ratio_step_to_reasoning": 1.5313590289453811,
"adv/std_final_conf": 0.8122832179069519,
"adv/std_reasoning": 0.7577930092811584,
"adv/std_step_conf": 0.9296541810035706,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5782942547648431,
"calib/avg_num_step_conf": 10.1015625,
"calib/ece": 0.4033061224489797,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.9959183673469387,
"calib/gap": 0.0021644042232277227,
"calib/mean_conf": 0.9837142857142857,
"calib/mu_c": 0.9846153846153843,
"calib/mu_w": 0.9824509803921566,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40167346938775517,
"calib/std_conf": 0.02682673050126972,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8541364902506964,
"calib/step_q_c_n": 1436.0,
"calib/step_q_gap": 0.012719098946348595,
"calib/step_q_w": 0.8414173913043478,
"calib/step_q_w_n": 1150.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2472.0,
"completions/max_terminated_length": 2472.0,
"completions/mean_length": 837.97265625,
"completions/mean_terminated_length": 858.0840454101562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 285.0,
"epoch": 0.10453333333333334,
"grad_norm": 0.018236543983221054,
"kl": 0.1473541259765625,
"learning_rate": 2.8333333333333335e-06,
"loss": -0.1734,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.0168771892786026,
"mask/share_reasoning": 0.8379881381988525,
"mask/share_step_conf": 0.12169711291790009,
"num_tokens": 32108338.0,
"reward": 0.7924503087997437,
"reward_std": 0.26125431060791016,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/asymmetric_l2_reward": 0.7101399898529053,
"rewards/final_brier_reward_step": 0.5716354846954346,
"rewards/format_reward_step": 0.95703125,
"step": 98
},
{
"adv/mean_abs_final_conf": 0.6645784378051758,
"adv/mean_abs_reasoning": 0.5807794332504272,
"adv/mean_abs_step_conf": 0.7467468976974487,
"adv/ratio_final_to_reasoning": 1.1442871419977008,
"adv/ratio_step_to_reasoning": 1.2857667729694857,
"adv/std_final_conf": 0.8672755360603333,
"adv/std_reasoning": 0.8265784382820129,
"adv/std_step_conf": 0.9330686330795288,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.609092894583576,
"calib/avg_num_step_conf": 10.4765625,
"calib/ece": 0.5567932489451476,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0064130751310423895,
"calib/mean_conf": 0.9829535864978902,
"calib/mu_c": 0.9866336633663365,
"calib/mu_w": 0.9802205882352941,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.5567932489451476,
"calib/std_conf": 0.012849746660799886,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8542406311637082,
"calib/step_q_c_n": 1014.0,
"calib/step_q_gap": 0.01385693811814459,
"calib/step_q_w": 0.8403836930455636,
"calib/step_q_w_n": 1668.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2788.0,
"completions/max_terminated_length": 2788.0,
"completions/mean_length": 880.34375,
"completions/mean_terminated_length": 935.136962890625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 392.0,
"epoch": 0.1056,
"grad_norm": 0.02690640278160572,
"kl": 0.131011962890625,
"learning_rate": 2.805555555555556e-06,
"loss": -0.2087,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.015637118369340897,
"mask/share_reasoning": 0.8111535310745239,
"mask/share_step_conf": 0.11461564898490906,
"num_tokens": 32439506.0,
"reward": 0.6306707262992859,
"reward_std": 0.251910537481308,
"rewards/accuracy_reward_step": 0.39453125,
"rewards/asymmetric_l2_reward": 0.5821539163589478,
"rewards/final_brier_reward_step": 0.4151250123977661,
"rewards/format_reward_step": 0.92578125,
"step": 99
},
{
"adv/mean_abs_final_conf": 0.5684931874275208,
"adv/mean_abs_reasoning": 0.4415876865386963,
"adv/mean_abs_step_conf": 0.7473946809768677,
"adv/ratio_final_to_reasoning": 1.2873846005162641,
"adv/ratio_step_to_reasoning": 1.6925170328801131,
"adv/std_final_conf": 0.7858427166938782,
"adv/std_reasoning": 0.7207019925117493,
"adv/std_step_conf": 0.9307321310043335,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5973707664884135,
"calib/avg_num_step_conf": 9.8359375,
"calib/ece": 0.3480497925311204,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.995850622406639,
"calib/gap": 0.007260101010101105,
"calib/mean_conf": 0.9829045643153527,
"calib/mu_c": 0.9855555555555557,
"calib/mu_w": 0.9782954545454546,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3480497925311204,
"calib/std_conf": 0.016867411307882017,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8569694915254237,
"calib/step_q_c_n": 1475.0,
"calib/step_q_gap": 0.016451754229162785,
"calib/step_q_w": 0.8405177372962609,
"calib/step_q_w_n": 1043.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2486.0,
"completions/max_terminated_length": 2486.0,
"completions/mean_length": 877.6484375,
"completions/mean_terminated_length": 913.3251953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 479.0,
"epoch": 0.10666666666666667,
"grad_norm": 0.020915953442454338,
"kl": 0.13641357421875,
"learning_rate": 2.7777777777777783e-06,
"loss": -0.1353,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016081880778074265,
"mask/share_reasoning": 0.8332370519638062,
"mask/share_step_conf": 0.11161854863166809,
"num_tokens": 32771592.0,
"reward": 0.8156791925430298,
"reward_std": 0.21008773148059845,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/asymmetric_l2_reward": 0.7115105986595154,
"rewards/final_brier_reward_step": 0.6120351552963257,
"rewards/format_reward_step": 0.94140625,
"step": 100
},
{
"adv/mean_abs_final_conf": 0.6653465032577515,
"adv/mean_abs_reasoning": 0.47705355286598206,
"adv/mean_abs_step_conf": 0.7555596828460693,
"adv/ratio_final_to_reasoning": 1.3946998177889396,
"adv/ratio_step_to_reasoning": 1.5838047496070689,
"adv/std_final_conf": 0.8447847962379456,
"adv/std_reasoning": 0.7393490672111511,
"adv/std_step_conf": 0.9337489604949951,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.5492322198275863,
"calib/avg_num_step_conf": 9.953125,
"calib/ece": 0.456844262295082,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0027343750000000666,
"calib/mean_conf": 0.9814344262295083,
"calib/mu_c": 0.982734375,
"calib/mu_w": 0.98,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.456844262295082,
"calib/std_conf": 0.013519559422032574,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8568569131832798,
"calib/step_q_c_n": 1244.0,
"calib/step_q_gap": 0.010031759809046692,
"calib/step_q_w": 0.8468251533742331,
"calib/step_q_w_n": 1304.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2727.0,
"completions/max_terminated_length": 2727.0,
"completions/mean_length": 938.17578125,
"completions/mean_terminated_length": 956.8645629882812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 353.0,
"epoch": 0.10773333333333333,
"grad_norm": 0.014655107632279396,
"kl": 0.13433837890625,
"learning_rate": 2.7500000000000004e-06,
"loss": -0.0587,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.015749523416161537,
"mask/share_reasoning": 0.8516314029693604,
"mask/share_step_conf": 0.11308782547712326,
"num_tokens": 33118757.0,
"reward": 0.7222472429275513,
"reward_std": 0.224051371216774,
"rewards/accuracy_reward_step": 0.5,
"rewards/asymmetric_l2_reward": 0.6362471580505371,
"rewards/final_brier_reward_step": 0.5176222324371338,
"rewards/format_reward_step": 0.953125,
"step": 101
},
{
"adv/mean_abs_final_conf": 0.5480314493179321,
"adv/mean_abs_reasoning": 0.4374270439147949,
"adv/mean_abs_step_conf": 0.7339659929275513,
"adv/ratio_final_to_reasoning": 1.2528522343138013,
"adv/ratio_step_to_reasoning": 1.677916359168955,
"adv/std_final_conf": 0.782140851020813,
"adv/std_reasoning": 0.7205737829208374,
"adv/std_step_conf": 0.9318551421165466,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5557624890446977,
"calib/avg_num_step_conf": 10.27734375,
"calib/ece": 0.32514170040485824,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.003519573473561355,
"calib/mean_conf": 0.9850607287449392,
"calib/mu_c": 0.9862576687116564,
"calib/mu_w": 0.9827380952380951,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.32514170040485824,
"calib/std_conf": 0.010830697314911472,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8625914445133293,
"calib/step_q_c_n": 1613.0,
"calib/step_q_gap": 0.014035452371875423,
"calib/step_q_w": 0.8485559921414538,
"calib/step_q_w_n": 1018.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2785.0,
"completions/max_terminated_length": 2785.0,
"completions/mean_length": 792.71875,
"completions/mean_terminated_length": 811.7440185546875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 454.0,
"epoch": 0.1088,
"grad_norm": 0.0197947658598423,
"kl": 0.147491455078125,
"learning_rate": 2.7222222222222224e-06,
"loss": -0.1023,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.018000053241848946,
"mask/share_reasoning": 0.8290439248085022,
"mask/share_step_conf": 0.12951849400997162,
"num_tokens": 33428389.0,
"reward": 0.8536818027496338,
"reward_std": 0.2171253263950348,
"rewards/accuracy_reward_step": 0.640625,
"rewards/asymmetric_l2_reward": 0.7385513782501221,
"rewards/final_brier_reward_step": 0.6477183103561401,
"rewards/format_reward_step": 0.96484375,
"step": 102
},
{
"adv/mean_abs_final_conf": 0.5411339998245239,
"adv/mean_abs_reasoning": 0.37070244550704956,
"adv/mean_abs_step_conf": 0.7217574715614319,
"adv/ratio_final_to_reasoning": 1.4597529808155347,
"adv/ratio_step_to_reasoning": 1.9469994878890173,
"adv/std_final_conf": 0.790514349937439,
"adv/std_reasoning": 0.6816906929016113,
"adv/std_step_conf": 0.9258291125297546,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6147619751626257,
"calib/avg_num_step_conf": 9.95703125,
"calib/ece": 0.3512448132780084,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006833234772324115,
"calib/mean_conf": 0.9819502074688797,
"calib/mu_c": 0.9844736842105264,
"calib/mu_w": 0.9776404494382023,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3512448132780084,
"calib/std_conf": 0.013292652712665354,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8647315436241612,
"calib/step_q_c_n": 1490.0,
"calib/step_q_gap": 0.012096982717645477,
"calib/step_q_w": 0.8526345609065157,
"calib/step_q_w_n": 1059.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2746.0,
"completions/max_terminated_length": 2746.0,
"completions/mean_length": 918.1015625,
"completions/mean_terminated_length": 951.5546875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 389.0,
"epoch": 0.10986666666666667,
"grad_norm": 0.01717713288962841,
"kl": 0.1207427978515625,
"learning_rate": 2.6944444444444444e-06,
"loss": -0.163,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01587964966893196,
"mask/share_reasoning": 0.838042140007019,
"mask/share_step_conf": 0.1109219565987587,
"num_tokens": 33767975.0,
"reward": 0.8112150430679321,
"reward_std": 0.18265941739082336,
"rewards/accuracy_reward_step": 0.59375,
"rewards/asymmetric_l2_reward": 0.7065750360488892,
"rewards/final_brier_reward_step": 0.6088237762451172,
"rewards/format_reward_step": 0.94140625,
"step": 103
},
{
"adv/mean_abs_final_conf": 0.5487852096557617,
"adv/mean_abs_reasoning": 0.40357086062431335,
"adv/mean_abs_step_conf": 0.7396882772445679,
"adv/ratio_final_to_reasoning": 1.3598236721224264,
"adv/ratio_step_to_reasoning": 1.8328584876031184,
"adv/std_final_conf": 0.7952535152435303,
"adv/std_reasoning": 0.7012667059898376,
"adv/std_step_conf": 0.9316074252128601,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6060175619834711,
"calib/avg_num_step_conf": 10.015625,
"calib/ece": 0.46823293172690783,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006382360537190213,
"calib/mean_conf": 0.9822891566265062,
"calib/mu_c": 0.9853906250000002,
"calib/mu_w": 0.97900826446281,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.46823293172690783,
"calib/std_conf": 0.013048439369190952,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8659394409937887,
"calib/step_q_c_n": 1288.0,
"calib/step_q_gap": 0.005853234097237148,
"calib/step_q_w": 0.8600862068965516,
"calib/step_q_w_n": 1276.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2665.0,
"completions/max_terminated_length": 2665.0,
"completions/mean_length": 884.9296875,
"completions/mean_terminated_length": 895.4229736328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 400.0,
"epoch": 0.11093333333333333,
"grad_norm": 0.0153735326603055,
"kl": 0.131439208984375,
"learning_rate": 2.666666666666667e-06,
"loss": -0.023,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.01663128286600113,
"mask/share_reasoning": 0.8520343899726868,
"mask/share_step_conf": 0.11961561441421509,
"num_tokens": 34101197.0,
"reward": 0.7398614883422852,
"reward_std": 0.18474718928337097,
"rewards/accuracy_reward_step": 0.5,
"rewards/asymmetric_l2_reward": 0.6658186316490173,
"rewards/final_brier_reward_step": 0.5193730592727661,
"rewards/format_reward_step": 0.97265625,
"step": 104
},
{
"adv/mean_abs_final_conf": 0.7172216176986694,
"adv/mean_abs_reasoning": 0.5644861459732056,
"adv/mean_abs_step_conf": 0.7502240538597107,
"adv/ratio_final_to_reasoning": 1.270574349459258,
"adv/ratio_step_to_reasoning": 1.3290389130211204,
"adv/std_final_conf": 0.8941024541854858,
"adv/std_reasoning": 0.8100602030754089,
"adv/std_step_conf": 0.9334349632263184,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6729426263179515,
"calib/avg_num_step_conf": 10.48046875,
"calib/ece": 0.42987654320987667,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009874709023688921,
"calib/mean_conf": 0.9813168724279837,
"calib/mu_c": 0.9857462686567162,
"calib/mu_w": 0.9758715596330273,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.42987654320987667,
"calib/std_conf": 0.013391631934015873,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8656762917933131,
"calib/step_q_c_n": 1316.0,
"calib/step_q_gap": 0.018456101595800245,
"calib/step_q_w": 0.8472201901975128,
"calib/step_q_w_n": 1367.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2609.0,
"completions/max_terminated_length": 2609.0,
"completions/mean_length": 896.37109375,
"completions/mean_terminated_length": 925.2862548828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 530.0,
"epoch": 0.112,
"grad_norm": 0.016667574644088745,
"kl": 0.126190185546875,
"learning_rate": 2.6388888888888893e-06,
"loss": -0.151,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.016032157465815544,
"mask/share_reasoning": 0.8382542133331299,
"mask/share_step_conf": 0.11446364969015121,
"num_tokens": 34436428.0,
"reward": 0.7515039443969727,
"reward_std": 0.2816525399684906,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/asymmetric_l2_reward": 0.6649938821792603,
"rewards/final_brier_reward_step": 0.543482780456543,
"rewards/format_reward_step": 0.94921875,
"step": 105
},
{
"adv/mean_abs_final_conf": 0.5788710117340088,
"adv/mean_abs_reasoning": 0.46618354320526123,
"adv/mean_abs_step_conf": 0.7207705974578857,
"adv/ratio_final_to_reasoning": 1.2417233945110138,
"adv/ratio_step_to_reasoning": 1.5461090550348524,
"adv/std_final_conf": 0.8257135152816772,
"adv/std_reasoning": 0.757561445236206,
"adv/std_step_conf": 0.9304320812225342,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6588401697312588,
"calib/avg_num_step_conf": 10.21484375,
"calib/ece": 0.4016597510373445,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.995850622406639,
"calib/gap": 0.009882602545968955,
"calib/mean_conf": 0.9825726141078839,
"calib/mu_c": 0.9867142857142857,
"calib/mu_w": 0.9768316831683167,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4016597510373445,
"calib/std_conf": 0.016994626836127357,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8627635542168675,
"calib/step_q_c_n": 1328.0,
"calib/step_q_gap": 0.011737913191226412,
"calib/step_q_w": 0.851025641025641,
"calib/step_q_w_n": 1287.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3047.0,
"completions/max_terminated_length": 3047.0,
"completions/mean_length": 841.17578125,
"completions/mean_terminated_length": 875.369873046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 518.0,
"epoch": 0.11306666666666666,
"grad_norm": 0.03570663556456566,
"kl": 0.138427734375,
"learning_rate": 2.6111111111111113e-06,
"loss": -0.1133,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016544461250305176,
"mask/share_reasoning": 0.8300613164901733,
"mask/share_step_conf": 0.11433173716068268,
"num_tokens": 34756353.0,
"reward": 0.7651780843734741,
"reward_std": 0.2139662802219391,
"rewards/accuracy_reward_step": 0.546875,
"rewards/asymmetric_l2_reward": 0.6681015491485596,
"rewards/final_brier_reward_step": 0.5645984411239624,
"rewards/format_reward_step": 0.94140625,
"step": 106
},
{
"adv/mean_abs_final_conf": 0.6399675011634827,
"adv/mean_abs_reasoning": 0.5369926691055298,
"adv/mean_abs_step_conf": 0.7458997368812561,
"adv/ratio_final_to_reasoning": 1.1917620816490444,
"adv/ratio_step_to_reasoning": 1.3890315078671436,
"adv/std_final_conf": 0.8324251770973206,
"adv/std_reasoning": 0.7928897142410278,
"adv/std_step_conf": 0.9346669912338257,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5739526823387583,
"calib/avg_num_step_conf": 11.05859375,
"calib/ece": 0.3011336032388664,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.004510247136830126,
"calib/mean_conf": 0.9812955465587044,
"calib/mu_c": 0.9827380952380954,
"calib/mu_w": 0.9782278481012653,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3011336032388664,
"calib/std_conf": 0.013525418430851373,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8619087136929461,
"calib/step_q_c_n": 1687.0,
"calib/step_q_gap": -0.00026785973362730697,
"calib/step_q_w": 0.8621765734265734,
"calib/step_q_w_n": 1144.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2013.0,
"completions/max_terminated_length": 2013.0,
"completions/mean_length": 829.03125,
"completions/mean_terminated_length": 852.3373413085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 446.0,
"epoch": 0.11413333333333334,
"grad_norm": 0.01998072676360607,
"kl": 0.144256591796875,
"learning_rate": 2.5833333333333337e-06,
"loss": -0.1303,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.017042508348822594,
"mask/share_reasoning": 0.8300793170928955,
"mask/share_step_conf": 0.12553441524505615,
"num_tokens": 35073201.0,
"reward": 0.8738058805465698,
"reward_std": 0.2717360854148865,
"rewards/accuracy_reward_step": 0.65625,
"rewards/asymmetric_l2_reward": 0.7542195320129395,
"rewards/final_brier_reward_step": 0.6691734194755554,
"rewards/format_reward_step": 0.96484375,
"step": 107
},
{
"adv/mean_abs_final_conf": 0.5755144357681274,
"adv/mean_abs_reasoning": 0.48384425044059753,
"adv/mean_abs_step_conf": 0.7267765998840332,
"adv/ratio_final_to_reasoning": 1.1894621776409524,
"adv/ratio_step_to_reasoning": 1.5020879120134567,
"adv/std_final_conf": 0.7753496766090393,
"adv/std_reasoning": 0.7394925355911255,
"adv/std_step_conf": 0.9305362701416016,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6644811397118342,
"calib/avg_num_step_conf": 9.90625,
"calib/ece": 0.2712653061224489,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009406669904484688,
"calib/mean_conf": 0.981469387755102,
"calib/mu_c": 0.9841954022988506,
"calib/mu_w": 0.9747887323943659,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2712653061224489,
"calib/std_conf": 0.013381531654525757,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8594519740718917,
"calib/step_q_c_n": 1697.0,
"calib/step_q_gap": 0.0017642505915579365,
"calib/step_q_w": 0.8576877234803337,
"calib/step_q_w_n": 839.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2823.0,
"completions/max_terminated_length": 2823.0,
"completions/mean_length": 869.64453125,
"completions/mean_terminated_length": 890.5160522460938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 288.0,
"epoch": 0.1152,
"grad_norm": 0.03150971978902817,
"kl": 0.1255035400390625,
"learning_rate": 2.5555555555555557e-06,
"loss": -0.1344,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01689765229821205,
"mask/share_reasoning": 0.8402672410011292,
"mask/share_step_conf": 0.11939756572246552,
"num_tokens": 35399062.0,
"reward": 0.8926793336868286,
"reward_std": 0.23910677433013916,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/asymmetric_l2_reward": 0.7657020092010498,
"rewards/final_brier_reward_step": 0.6930941343307495,
"rewards/format_reward_step": 0.953125,
"step": 108
},
{
"adv/mean_abs_final_conf": 0.6754946112632751,
"adv/mean_abs_reasoning": 0.5071865320205688,
"adv/mean_abs_step_conf": 0.7378177642822266,
"adv/ratio_final_to_reasoning": 1.3318465073829693,
"adv/ratio_step_to_reasoning": 1.4547266492721154,
"adv/std_final_conf": 0.8642860651016235,
"adv/std_reasoning": 0.7753763794898987,
"adv/std_step_conf": 0.9282304048538208,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.6707766439909297,
"calib/avg_num_step_conf": 10.73046875,
"calib/ece": 0.44756302521008395,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009970238095237782,
"calib/mean_conf": 0.9769747899159663,
"calib/mu_c": 0.9816666666666665,
"calib/mu_w": 0.9716964285714287,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.44756302521008395,
"calib/std_conf": 0.014755977102516214,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8645178849144635,
"calib/step_q_c_n": 1286.0,
"calib/step_q_gap": 0.017775927351150522,
"calib/step_q_w": 0.8467419575633129,
"calib/step_q_w_n": 1461.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2871.0,
"completions/max_terminated_length": 2871.0,
"completions/mean_length": 915.2734375,
"completions/mean_terminated_length": 956.3673095703125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 417.0,
"epoch": 0.11626666666666667,
"grad_norm": 0.01444100122898817,
"kl": 0.1240234375,
"learning_rate": 2.5277777777777778e-06,
"loss": -0.2686,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.015557501465082169,
"mask/share_reasoning": 0.8258721828460693,
"mask/share_step_conf": 0.11560158431529999,
"num_tokens": 35737972.0,
"reward": 0.7213988304138184,
"reward_std": 0.21585328876972198,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/asymmetric_l2_reward": 0.6432522535324097,
"rewards/final_brier_reward_step": 0.5159515142440796,
"rewards/format_reward_step": 0.92578125,
"step": 109
},
{
"adv/mean_abs_final_conf": 0.6226210594177246,
"adv/mean_abs_reasoning": 0.4714178144931793,
"adv/mean_abs_step_conf": 0.7386190891265869,
"adv/ratio_final_to_reasoning": 1.3207414744966812,
"adv/ratio_step_to_reasoning": 1.5668035157319529,
"adv/std_final_conf": 0.8406306505203247,
"adv/std_reasoning": 0.7575274109840393,
"adv/std_step_conf": 0.928333580493927,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5543624161073825,
"calib/avg_num_step_conf": 10.32421875,
"calib/ece": 0.38096385542168676,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.9959839357429718,
"calib/gap": 0.0036053691275168553,
"calib/mean_conf": 0.9793574297188755,
"calib/mu_c": 0.9808053691275167,
"calib/mu_w": 0.9771999999999998,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.38096385542168676,
"calib/std_conf": 0.014848253687567507,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8625860904794058,
"calib/step_q_c_n": 1481.0,
"calib/step_q_gap": 0.004290049171316146,
"calib/step_q_w": 0.8582960413080897,
"calib/step_q_w_n": 1162.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2907.0,
"completions/max_terminated_length": 2907.0,
"completions/mean_length": 815.44140625,
"completions/mean_terminated_length": 831.685302734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 307.0,
"epoch": 0.11733333333333333,
"grad_norm": 0.01637006178498268,
"kl": 0.1345367431640625,
"learning_rate": 2.5e-06,
"loss": -0.0706,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.017808744683861732,
"mask/share_reasoning": 0.8331002593040466,
"mask/share_step_conf": 0.129559725522995,
"num_tokens": 36051645.0,
"reward": 0.8081911206245422,
"reward_std": 0.22839653491973877,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/asymmetric_l2_reward": 0.7062299251556396,
"rewards/final_brier_reward_step": 0.5992147922515869,
"rewards/format_reward_step": 0.97265625,
"step": 110
},
{
"adv/mean_abs_final_conf": 0.6984493136405945,
"adv/mean_abs_reasoning": 0.542428731918335,
"adv/mean_abs_step_conf": 0.7579163312911987,
"adv/ratio_final_to_reasoning": 1.2876333286595687,
"adv/ratio_step_to_reasoning": 1.3972643532557312,
"adv/std_final_conf": 0.8653133511543274,
"adv/std_reasoning": 0.7929964661598206,
"adv/std_step_conf": 0.9306600689888,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6711769759450171,
"calib/avg_num_step_conf": 10.12109375,
"calib/ece": 0.3814107883817428,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.010099513172966867,
"calib/mean_conf": 0.9789211618257262,
"calib/mu_c": 0.9829861111111112,
"calib/mu_w": 0.9728865979381444,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3814107883817428,
"calib/std_conf": 0.014478414620961383,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8633705357142857,
"calib/step_q_c_n": 1344.0,
"calib/step_q_gap": 0.015615924647725876,
"calib/step_q_w": 0.8477546110665598,
"calib/step_q_w_n": 1247.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 3053.0,
"completions/max_terminated_length": 3053.0,
"completions/mean_length": 873.1015625,
"completions/mean_terminated_length": 904.9149780273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 411.0,
"epoch": 0.1184,
"grad_norm": 0.029058046638965607,
"kl": 0.1292877197265625,
"learning_rate": 2.4722222222222226e-06,
"loss": -0.1011,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016590723767876625,
"mask/share_reasoning": 0.8334500789642334,
"mask/share_step_conf": 0.11480294167995453,
"num_tokens": 36382567.0,
"reward": 0.7858656644821167,
"reward_std": 0.26266515254974365,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.6896809339523315,
"rewards/final_brier_reward_step": 0.5820503830909729,
"rewards/format_reward_step": 0.9375,
"step": 111
},
{
"adv/mean_abs_final_conf": 0.6206170916557312,
"adv/mean_abs_reasoning": 0.44810646772384644,
"adv/mean_abs_step_conf": 0.7446877956390381,
"adv/ratio_final_to_reasoning": 1.3849768667882685,
"adv/ratio_step_to_reasoning": 1.6618546021477314,
"adv/std_final_conf": 0.8321802020072937,
"adv/std_reasoning": 0.7207315564155579,
"adv/std_step_conf": 0.9297779202461243,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.758238510418206,
"calib/avg_num_step_conf": 10.796875,
"calib/ece": 0.40736170212765965,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.015019949756169715,
"calib/mean_conf": 0.9775744680851065,
"calib/mu_c": 0.9840298507462686,
"calib/mu_w": 0.9690099009900989,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40736170212765965,
"calib/std_conf": 0.014603620479184115,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8601647564469914,
"calib/step_q_c_n": 1396.0,
"calib/step_q_gap": 0.0030083236984532746,
"calib/step_q_w": 0.8571564327485381,
"calib/step_q_w_n": 1368.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2912.0,
"completions/max_terminated_length": 2912.0,
"completions/mean_length": 902.65234375,
"completions/mean_terminated_length": 950.9423828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 476.0,
"epoch": 0.11946666666666667,
"grad_norm": 0.03636123239994049,
"kl": 0.1224365234375,
"learning_rate": 2.4444444444444447e-06,
"loss": -0.1538,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.015544177033007145,
"mask/share_reasoning": 0.8192787170410156,
"mask/share_step_conf": 0.11439579725265503,
"num_tokens": 36721566.0,
"reward": 0.7470136880874634,
"reward_std": 0.20874658226966858,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/asymmetric_l2_reward": 0.6588191986083984,
"rewards/final_brier_reward_step": 0.5469269156455994,
"rewards/format_reward_step": 0.9140625,
"step": 112
},
{
"adv/mean_abs_final_conf": 0.6355844736099243,
"adv/mean_abs_reasoning": 0.5024422407150269,
"adv/mean_abs_step_conf": 0.7353067994117737,
"adv/ratio_final_to_reasoning": 1.2649901264380605,
"adv/ratio_step_to_reasoning": 1.4634653295976003,
"adv/std_final_conf": 0.8575465083122253,
"adv/std_reasoning": 0.7753759026527405,
"adv/std_step_conf": 0.9332528710365295,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.658067940552017,
"calib/avg_num_step_conf": 9.56640625,
"calib/ece": 0.34550607287449386,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009125265392781134,
"calib/mean_conf": 0.9811336032388663,
"calib/mu_c": 0.9844585987261145,
"calib/mu_w": 0.9753333333333334,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.34550607287449386,
"calib/std_conf": 0.013237566329213,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8643951093951093,
"calib/step_q_c_n": 1554.0,
"calib/step_q_gap": 0.008707958557120477,
"calib/step_q_w": 0.8556871508379889,
"calib/step_q_w_n": 895.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2820.0,
"completions/max_terminated_length": 2820.0,
"completions/mean_length": 800.5546875,
"completions/mean_terminated_length": 810.0474853515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 472.0,
"epoch": 0.12053333333333334,
"grad_norm": 0.02966303937137127,
"kl": 0.1484527587890625,
"learning_rate": 2.4166666666666667e-06,
"loss": -0.0225,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.018074627965688705,
"mask/share_reasoning": 0.8422780632972717,
"mask/share_step_conf": 0.12792859971523285,
"num_tokens": 37031708.0,
"reward": 0.829046368598938,
"reward_std": 0.25141140818595886,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/asymmetric_l2_reward": 0.7154799699783325,
"rewards/final_brier_reward_step": 0.626206636428833,
"rewards/format_reward_step": 0.9609375,
"step": 113
},
{
"adv/mean_abs_final_conf": 0.5766456127166748,
"adv/mean_abs_reasoning": 0.3910996913909912,
"adv/mean_abs_step_conf": 0.7413148283958435,
"adv/ratio_final_to_reasoning": 1.4744210374233948,
"adv/ratio_step_to_reasoning": 1.8954625756907957,
"adv/std_final_conf": 0.7828994393348694,
"adv/std_reasoning": 0.681779682636261,
"adv/std_step_conf": 0.9287707805633545,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6706652929055049,
"calib/avg_num_step_conf": 11.3125,
"calib/ece": 0.3340163934426229,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00994701206947346,
"calib/mean_conf": 0.9815573770491803,
"calib/mu_c": 0.9850632911392404,
"calib/mu_w": 0.975116279069767,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3340163934426229,
"calib/std_conf": 0.013338007556571024,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.86858125,
"calib/step_q_c_n": 1600.0,
"calib/step_q_gap": 0.016721682098765234,
"calib/step_q_w": 0.8518595679012347,
"calib/step_q_w_n": 1296.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 3020.0,
"completions/max_terminated_length": 3020.0,
"completions/mean_length": 825.6171875,
"completions/mean_terminated_length": 855.700439453125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 372.0,
"epoch": 0.1216,
"grad_norm": 0.02331947349011898,
"kl": 0.1299896240234375,
"learning_rate": 2.388888888888889e-06,
"loss": -0.1173,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01743999309837818,
"mask/share_reasoning": 0.8186057209968567,
"mask/share_step_conf": 0.12879803776741028,
"num_tokens": 37348090.0,
"reward": 0.8373966217041016,
"reward_std": 0.20491275191307068,
"rewards/accuracy_reward_step": 0.6171875,
"rewards/asymmetric_l2_reward": 0.727318286895752,
"rewards/final_brier_reward_step": 0.6334124803543091,
"rewards/format_reward_step": 0.953125,
"step": 114
},
{
"adv/mean_abs_final_conf": 0.59438157081604,
"adv/mean_abs_reasoning": 0.5020387768745422,
"adv/mean_abs_step_conf": 0.7483762502670288,
"adv/ratio_final_to_reasoning": 1.183935580666459,
"adv/ratio_step_to_reasoning": 1.4906741963759613,
"adv/std_final_conf": 0.8215389251708984,
"adv/std_reasoning": 0.7576356530189514,
"adv/std_step_conf": 0.9327325820922852,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5949225252913305,
"calib/avg_num_step_conf": 10.0546875,
"calib/ece": 0.43605577689243025,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.9960159362549801,
"calib/gap": 0.005359200922013141,
"calib/mean_conf": 0.9818725099601593,
"calib/mu_c": 0.9843065693430658,
"calib/mu_w": 0.9789473684210527,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43605577689243025,
"calib/std_conf": 0.014144941280175828,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8640315315315314,
"calib/step_q_c_n": 1332.0,
"calib/step_q_gap": -0.0008074378726554698,
"calib/step_q_w": 0.8648389694041869,
"calib/step_q_w_n": 1242.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2514.0,
"completions/max_terminated_length": 2514.0,
"completions/mean_length": 865.6328125,
"completions/mean_terminated_length": 879.373046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.12266666666666666,
"grad_norm": 0.02993854694068432,
"kl": 0.129150390625,
"learning_rate": 2.361111111111111e-06,
"loss": -0.073,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.016907256096601486,
"mask/share_reasoning": 0.8484659194946289,
"mask/share_step_conf": 0.1190018355846405,
"num_tokens": 37674956.0,
"reward": 0.7629421949386597,
"reward_std": 0.25496959686279297,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/asymmetric_l2_reward": 0.6740585565567017,
"rewards/final_brier_reward_step": 0.549481987953186,
"rewards/format_reward_step": 0.9765625,
"step": 115
},
{
"adv/mean_abs_final_conf": 0.6308901309967041,
"adv/mean_abs_reasoning": 0.4708402454853058,
"adv/mean_abs_step_conf": 0.7694852352142334,
"adv/ratio_final_to_reasoning": 1.3399239700642651,
"adv/ratio_step_to_reasoning": 1.6342809319987237,
"adv/std_final_conf": 0.8313889503479004,
"adv/std_reasoning": 0.7394453883171082,
"adv/std_step_conf": 0.9321046471595764,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5725085910652922,
"calib/avg_num_step_conf": 10.48046875,
"calib/ece": 0.3777732793522268,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0042591065292093555,
"calib/mean_conf": 0.9850607287449393,
"calib/mu_c": 0.9867333333333331,
"calib/mu_w": 0.9824742268041238,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3777732793522268,
"calib/std_conf": 0.011198266397935598,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8641393168117886,
"calib/step_q_c_n": 1493.0,
"calib/step_q_gap": 0.019441837820191932,
"calib/step_q_w": 0.8446974789915966,
"calib/step_q_w_n": 1190.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2885.0,
"completions/max_terminated_length": 2885.0,
"completions/mean_length": 854.0390625,
"completions/mean_terminated_length": 881.5886840820312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 394.0,
"epoch": 0.12373333333333333,
"grad_norm": 0.028440121561288834,
"kl": 0.1218414306640625,
"learning_rate": 2.3333333333333336e-06,
"loss": -0.1257,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.016650334000587463,
"mask/share_reasoning": 0.8338371515274048,
"mask/share_step_conf": 0.11826249212026596,
"num_tokens": 37998110.0,
"reward": 0.8119787573814392,
"reward_std": 0.24483713507652283,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/asymmetric_l2_reward": 0.714918851852417,
"rewards/final_brier_reward_step": 0.5988823771476746,
"rewards/format_reward_step": 0.96484375,
"step": 116
},
{
"adv/mean_abs_final_conf": 0.6085239052772522,
"adv/mean_abs_reasoning": 0.5667084455490112,
"adv/mean_abs_step_conf": 0.741919994354248,
"adv/ratio_final_to_reasoning": 1.0737865476624957,
"adv/ratio_step_to_reasoning": 1.309174056221267,
"adv/std_final_conf": 0.837278425693512,
"adv/std_reasoning": 0.8099818229675293,
"adv/std_step_conf": 0.9349244236946106,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5788172334372037,
"calib/avg_num_step_conf": 9.9375,
"calib/ece": 0.48271604938271606,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.9958847736625515,
"calib/gap": 0.004240617802465763,
"calib/mean_conf": 0.9847736625514404,
"calib/mu_c": 0.9868852459016394,
"calib/mu_w": 0.9826446280991736,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.48271604938271606,
"calib/std_conf": 0.012316563747766475,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.8623914893617022,
"calib/step_q_c_n": 1175.0,
"calib/step_q_gap": 0.001441891114806526,
"calib/step_q_w": 0.8609495982468957,
"calib/step_q_w_n": 1369.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2696.0,
"completions/max_terminated_length": 2696.0,
"completions/mean_length": 851.1015625,
"completions/mean_terminated_length": 882.1134033203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 447.0,
"epoch": 0.1248,
"grad_norm": 0.01481497474014759,
"kl": 0.1201629638671875,
"learning_rate": 2.305555555555556e-06,
"loss": -0.1201,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.016754884272813797,
"mask/share_reasoning": 0.831017017364502,
"mask/share_step_conf": 0.11707185953855515,
"num_tokens": 38322592.0,
"reward": 0.7045153379440308,
"reward_std": 0.2681896686553955,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/asymmetric_l2_reward": 0.6370803117752075,
"rewards/final_brier_reward_step": 0.4883566200733185,
"rewards/format_reward_step": 0.94140625,
"step": 117
},
{
"adv/mean_abs_final_conf": 0.44347405433654785,
"adv/mean_abs_reasoning": 0.3869093060493469,
"adv/mean_abs_step_conf": 0.7286380529403687,
"adv/ratio_final_to_reasoning": 1.1461964015928492,
"adv/ratio_step_to_reasoning": 1.8832270031970675,
"adv/std_final_conf": 0.7195388674736023,
"adv/std_reasoning": 0.6816406846046448,
"adv/std_step_conf": 0.9258517026901245,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6229621921609434,
"calib/avg_num_step_conf": 10.890625,
"calib/ece": 0.3604435483870969,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00698924731182804,
"calib/mean_conf": 0.9854435483870969,
"calib/mu_c": 0.9880645161290321,
"calib/mu_w": 0.9810752688172041,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3604435483870969,
"calib/std_conf": 0.01069202477833652,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8498784194528877,
"calib/step_q_c_n": 1645.0,
"calib/step_q_gap": 0.003736687169422992,
"calib/step_q_w": 0.8461417322834647,
"calib/step_q_w_n": 1143.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2624.0,
"completions/max_terminated_length": 2624.0,
"completions/mean_length": 849.3671875,
"completions/mean_terminated_length": 866.286865234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 408.0,
"epoch": 0.12586666666666665,
"grad_norm": 0.025349227711558342,
"kl": 0.115478515625,
"learning_rate": 2.277777777777778e-06,
"loss": -0.1366,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.017265360802412033,
"mask/share_reasoning": 0.8359560966491699,
"mask/share_step_conf": 0.12724733352661133,
"num_tokens": 38644038.0,
"reward": 0.8097965717315674,
"reward_std": 0.17407071590423584,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/asymmetric_l2_reward": 0.6867055892944336,
"rewards/final_brier_reward_step": 0.6188250184059143,
"rewards/format_reward_step": 0.96484375,
"step": 118
},
{
"adv/mean_abs_final_conf": 0.6102824211120605,
"adv/mean_abs_reasoning": 0.5631183385848999,
"adv/mean_abs_step_conf": 0.7842050790786743,
"adv/ratio_final_to_reasoning": 1.0837551883777798,
"adv/ratio_step_to_reasoning": 1.3926115087094464,
"adv/std_final_conf": 0.8209497332572937,
"adv/std_reasoning": 0.7930397987365723,
"adv/std_step_conf": 0.932422935962677,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.5962677317173943,
"calib/avg_num_step_conf": 10.0546875,
"calib/ece": 0.4205349794238683,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005656245696185058,
"calib/mean_conf": 0.984320987654321,
"calib/mu_c": 0.9867883211678832,
"calib/mu_w": 0.9811320754716981,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4205349794238683,
"calib/std_conf": 0.011646890286489645,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8619156804733727,
"calib/step_q_c_n": 1352.0,
"calib/step_q_gap": 0.01773401107893724,
"calib/step_q_w": 0.8441816693944355,
"calib/step_q_w_n": 1222.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 3039.0,
"completions/max_terminated_length": 3039.0,
"completions/mean_length": 907.3125,
"completions/mean_terminated_length": 932.8192138671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 466.0,
"epoch": 0.12693333333333334,
"grad_norm": 0.017647327855229378,
"kl": 0.115753173828125,
"learning_rate": 2.25e-06,
"loss": -0.1018,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.016389839351177216,
"mask/share_reasoning": 0.8423788547515869,
"mask/share_step_conf": 0.11388753354549408,
"num_tokens": 38981374.0,
"reward": 0.7642459273338318,
"reward_std": 0.28795790672302246,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/asymmetric_l2_reward": 0.680416464805603,
"rewards/final_brier_reward_step": 0.5504191517829895,
"rewards/format_reward_step": 0.94921875,
"step": 119
},
{
"adv/mean_abs_final_conf": 0.5616532564163208,
"adv/mean_abs_reasoning": 0.5108141899108887,
"adv/mean_abs_step_conf": 0.7378266453742981,
"adv/ratio_final_to_reasoning": 1.099525556473482,
"adv/ratio_step_to_reasoning": 1.4444129782358075,
"adv/std_final_conf": 0.8124690055847168,
"adv/std_reasoning": 0.792822539806366,
"adv/std_step_conf": 0.9319746494293213,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6360431813174707,
"calib/avg_num_step_conf": 9.78515625,
"calib/ece": 0.35239669421487607,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007853418520966327,
"calib/mean_conf": 0.9846280991735538,
"calib/mu_c": 0.9875163398692809,
"calib/mu_w": 0.9796629213483146,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.35239669421487607,
"calib/std_conf": 0.011430083715451561,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8604258241758244,
"calib/step_q_c_n": 1456.0,
"calib/step_q_gap": 0.009243746006138975,
"calib/step_q_w": 0.8511820781696854,
"calib/step_q_w_n": 1049.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2772.0,
"completions/max_terminated_length": 2772.0,
"completions/mean_length": 829.47265625,
"completions/mean_terminated_length": 866.7142333984375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 481.0,
"epoch": 0.128,
"grad_norm": 0.013983098790049553,
"kl": 0.118865966796875,
"learning_rate": 2.222222222222222e-06,
"loss": -0.1756,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.01693129353225231,
"mask/share_reasoning": 0.8256010413169861,
"mask/share_step_conf": 0.11449891328811646,
"num_tokens": 39300407.0,
"reward": 0.8072173595428467,
"reward_std": 0.2549954056739807,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/asymmetric_l2_reward": 0.699937105178833,
"rewards/final_brier_reward_step": 0.6074663996696472,
"rewards/format_reward_step": 0.9375,
"step": 120
},
{
"adv/mean_abs_final_conf": 0.6494415998458862,
"adv/mean_abs_reasoning": 0.5642135143280029,
"adv/mean_abs_step_conf": 0.763487696647644,
"adv/ratio_final_to_reasoning": 1.1510564411407138,
"adv/ratio_step_to_reasoning": 1.3531893108887747,
"adv/std_final_conf": 0.8462623357772827,
"adv/std_reasoning": 0.8100326657295227,
"adv/std_step_conf": 0.9326791763305664,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5847723704866562,
"calib/avg_num_step_conf": 9.84765625,
"calib/ece": 0.3912448132780084,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00501070358213207,
"calib/mean_conf": 0.9846058091286308,
"calib/mu_c": 0.9866433566433565,
"calib/mu_w": 0.9816326530612244,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3912448132780084,
"calib/std_conf": 0.011412221713071413,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8529175946547883,
"calib/step_q_c_n": 1347.0,
"calib/step_q_gap": 0.007457628726338594,
"calib/step_q_w": 0.8454599659284497,
"calib/step_q_w_n": 1174.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2620.0,
"completions/max_terminated_length": 2620.0,
"completions/mean_length": 899.78515625,
"completions/mean_terminated_length": 932.5708618164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 443.0,
"epoch": 0.12906666666666666,
"grad_norm": 0.02037104405462742,
"kl": 0.1131134033203125,
"learning_rate": 2.1944444444444445e-06,
"loss": -0.0718,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.015997782349586487,
"mask/share_reasoning": 0.8389323949813843,
"mask/share_step_conf": 0.10991354286670685,
"num_tokens": 39635808.0,
"reward": 0.7773492932319641,
"reward_std": 0.271402508020401,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/asymmetric_l2_reward": 0.6823880672454834,
"rewards/final_brier_reward_step": 0.5723105072975159,
"rewards/format_reward_step": 0.94140625,
"step": 121
},
{
"adv/mean_abs_final_conf": 0.5551102161407471,
"adv/mean_abs_reasoning": 0.4751325845718384,
"adv/mean_abs_step_conf": 0.7243285775184631,
"adv/ratio_final_to_reasoning": 1.1683269768605322,
"adv/ratio_step_to_reasoning": 1.5244767482558275,
"adv/std_final_conf": 0.7889186143875122,
"adv/std_reasoning": 0.7576159238815308,
"adv/std_step_conf": 0.931524932384491,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6446941612604262,
"calib/avg_num_step_conf": 9.8359375,
"calib/ece": 0.30397540983606564,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008775100401606672,
"calib/mean_conf": 0.9843032786885246,
"calib/mu_c": 0.9871084337349396,
"calib/mu_w": 0.9783333333333329,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.30397540983606564,
"calib/std_conf": 0.011870468484313796,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8596106304079111,
"calib/step_q_c_n": 1618.0,
"calib/step_q_gap": 0.013721741519022279,
"calib/step_q_w": 0.8458888888888888,
"calib/step_q_w_n": 900.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2888.0,
"completions/max_terminated_length": 2888.0,
"completions/mean_length": 868.0546875,
"completions/mean_terminated_length": 896.056396484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 461.0,
"epoch": 0.13013333333333332,
"grad_norm": 0.043400537222623825,
"kl": 0.1139068603515625,
"learning_rate": 2.166666666666667e-06,
"loss": -0.0989,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01688627153635025,
"mask/share_reasoning": 0.8362627029418945,
"mask/share_step_conf": 0.11560102552175522,
"num_tokens": 39965374.0,
"reward": 0.8607096672058105,
"reward_std": 0.22507581114768982,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/asymmetric_l2_reward": 0.739835262298584,
"rewards/final_brier_reward_step": 0.6612714529037476,
"rewards/format_reward_step": 0.953125,
"step": 122
},
{
"adv/mean_abs_final_conf": 0.6799473762512207,
"adv/mean_abs_reasoning": 0.5501300096511841,
"adv/mean_abs_step_conf": 0.7506182193756104,
"adv/ratio_final_to_reasoning": 1.2359757953985253,
"adv/ratio_step_to_reasoning": 1.3644378714252436,
"adv/std_final_conf": 0.8617650270462036,
"adv/std_reasoning": 0.7928981184959412,
"adv/std_step_conf": 0.9324036240577698,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.5671607378129118,
"calib/avg_num_step_conf": 9.46484375,
"calib/ece": 0.44789473684210523,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0038300395256917197,
"calib/mean_conf": 0.9823076923076923,
"calib/mu_c": 0.9840909090909091,
"calib/mu_w": 0.9802608695652174,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44789473684210523,
"calib/std_conf": 0.01294407393907125,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8543739837398375,
"calib/step_q_c_n": 1230.0,
"calib/step_q_gap": 0.011959901593986744,
"calib/step_q_w": 0.8424140821458508,
"calib/step_q_w_n": 1193.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2405.0,
"completions/max_terminated_length": 2405.0,
"completions/mean_length": 937.48046875,
"completions/mean_terminated_length": 967.7217407226562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 509.0,
"epoch": 0.1312,
"grad_norm": 0.021335626021027565,
"kl": 0.1084747314453125,
"learning_rate": 2.138888888888889e-06,
"loss": -0.1323,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.015504155308008194,
"mask/share_reasoning": 0.8473955392837524,
"mask/share_step_conf": 0.10585027933120728,
"num_tokens": 40310657.0,
"reward": 0.7545461654663086,
"reward_std": 0.26759061217308044,
"rewards/accuracy_reward_step": 0.515625,
"rewards/asymmetric_l2_reward": 0.6801028251647949,
"rewards/final_brier_reward_step": 0.5328956842422485,
"rewards/format_reward_step": 0.96484375,
"step": 123
},
{
"adv/mean_abs_final_conf": 0.6220813989639282,
"adv/mean_abs_reasoning": 0.49169105291366577,
"adv/mean_abs_step_conf": 0.7388211488723755,
"adv/ratio_final_to_reasoning": 1.2651875507548787,
"adv/ratio_step_to_reasoning": 1.5026125541521749,
"adv/std_final_conf": 0.838358461856842,
"adv/std_reasoning": 0.757723331451416,
"adv/std_step_conf": 0.9337208867073059,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6167235009946007,
"calib/avg_num_step_conf": 10.2421875,
"calib/ece": 0.35746938775510195,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006618357487922943,
"calib/mean_conf": 0.9819591836734693,
"calib/mu_c": 0.9844444444444445,
"calib/mu_w": 0.9778260869565215,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35746938775510195,
"calib/std_conf": 0.013133473643551283,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8522455902306649,
"calib/step_q_c_n": 1474.0,
"calib/step_q_gap": 0.009092280126135277,
"calib/step_q_w": 0.8431533101045297,
"calib/step_q_w_n": 1148.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2648.0,
"completions/max_terminated_length": 2648.0,
"completions/mean_length": 884.62109375,
"completions/mean_terminated_length": 916.854248046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 427.0,
"epoch": 0.13226666666666667,
"grad_norm": 0.02123633213341236,
"kl": 0.1136016845703125,
"learning_rate": 2.1111111111111114e-06,
"loss": -0.1747,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016168804839253426,
"mask/share_reasoning": 0.8370611667633057,
"mask/share_step_conf": 0.11161378026008606,
"num_tokens": 40643936.0,
"reward": 0.8192969560623169,
"reward_std": 0.2530980706214905,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/asymmetric_l2_reward": 0.7145392298698425,
"rewards/final_brier_reward_step": 0.6131171584129333,
"rewards/format_reward_step": 0.95703125,
"step": 124
},
{
"adv/mean_abs_final_conf": 0.5496960282325745,
"adv/mean_abs_reasoning": 0.509536862373352,
"adv/mean_abs_step_conf": 0.7545234560966492,
"adv/ratio_final_to_reasoning": 1.078815035426812,
"adv/ratio_step_to_reasoning": 1.4808024930368797,
"adv/std_final_conf": 0.7946847677230835,
"adv/std_reasoning": 0.7754969596862793,
"adv/std_step_conf": 0.9305669069290161,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5919524336283186,
"calib/avg_num_step_conf": 10.23828125,
"calib/ece": 0.4519502074688797,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005281388274336329,
"calib/mean_conf": 0.9830705394190872,
"calib/mu_c": 0.985546875,
"calib/mu_w": 0.9802654867256637,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4519502074688797,
"calib/std_conf": 0.012544989714182483,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8530210016155089,
"calib/step_q_c_n": 1238.0,
"calib/step_q_gap": 0.0013941035677863534,
"calib/step_q_w": 0.8516268980477225,
"calib/step_q_w_n": 1383.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2821.0,
"completions/max_terminated_length": 2821.0,
"completions/mean_length": 891.6171875,
"completions/mean_terminated_length": 931.64892578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 397.0,
"epoch": 0.13333333333333333,
"grad_norm": 0.027528828009963036,
"kl": 0.1133575439453125,
"learning_rate": 2.0833333333333334e-06,
"loss": -0.1541,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01594950258731842,
"mask/share_reasoning": 0.8337675333023071,
"mask/share_step_conf": 0.10731419920921326,
"num_tokens": 40976998.0,
"reward": 0.7393513321876526,
"reward_std": 0.24845367670059204,
"rewards/accuracy_reward_step": 0.5,
"rewards/asymmetric_l2_reward": 0.6734174489974976,
"rewards/final_brier_reward_step": 0.5170038938522339,
"rewards/format_reward_step": 0.94140625,
"step": 125
},
{
"adv/mean_abs_final_conf": 0.6131834983825684,
"adv/mean_abs_reasoning": 0.505977988243103,
"adv/mean_abs_step_conf": 0.7746909856796265,
"adv/ratio_final_to_reasoning": 1.211877814115418,
"adv/ratio_step_to_reasoning": 1.5310764572379325,
"adv/std_final_conf": 0.8193503618240356,
"adv/std_reasoning": 0.7578449845314026,
"adv/std_step_conf": 0.9281861186027527,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.6112836812829588,
"calib/avg_num_step_conf": 10.14453125,
"calib/ece": 0.4433474576271186,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006677020876977524,
"calib/mean_conf": 0.9814830508474576,
"calib/mu_c": 0.9845669291338581,
"calib/mu_w": 0.9778899082568806,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4433474576271186,
"calib/std_conf": 0.013526642292501095,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8479055118110236,
"calib/step_q_c_n": 1270.0,
"calib/step_q_gap": 0.015275519346818589,
"calib/step_q_w": 0.832629992464205,
"calib/step_q_w_n": 1327.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 3045.0,
"completions/max_terminated_length": 3045.0,
"completions/mean_length": 887.1015625,
"completions/mean_terminated_length": 934.5596313476562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 425.0,
"epoch": 0.1344,
"grad_norm": 0.027923274785280228,
"kl": 0.1035614013671875,
"learning_rate": 2.0555555555555555e-06,
"loss": -0.2056,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.016019921749830246,
"mask/share_reasoning": 0.8219481706619263,
"mask/share_step_conf": 0.1112506240606308,
"num_tokens": 41309560.0,
"reward": 0.731331467628479,
"reward_std": 0.26831701397895813,
"rewards/accuracy_reward_step": 0.5,
"rewards/asymmetric_l2_reward": 0.6638506650924683,
"rewards/final_brier_reward_step": 0.5144370794296265,
"rewards/format_reward_step": 0.921875,
"step": 126
},
{
"adv/mean_abs_final_conf": 0.5383285284042358,
"adv/mean_abs_reasoning": 0.45178020000457764,
"adv/mean_abs_step_conf": 0.7675524950027466,
"adv/ratio_final_to_reasoning": 1.1915717607783194,
"adv/ratio_step_to_reasoning": 1.6989511603097467,
"adv/std_final_conf": 0.7692522406578064,
"adv/std_reasoning": 0.7207222580909729,
"adv/std_step_conf": 0.9278998374938965,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5820910973084885,
"calib/avg_num_step_conf": 9.23046875,
"calib/ece": 0.46024896265560167,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.004875086266390438,
"calib/mean_conf": 0.9830705394190872,
"calib/mu_c": 0.9853968253968253,
"calib/mu_w": 0.9805217391304348,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.46024896265560167,
"calib/std_conf": 0.012610968161338316,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.850793388429752,
"calib/step_q_c_n": 1210.0,
"calib/step_q_gap": 0.016578297362969674,
"calib/step_q_w": 0.8342150910667824,
"calib/step_q_w_n": 1153.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 3035.0,
"completions/max_terminated_length": 3035.0,
"completions/mean_length": 838.109375,
"completions/mean_terminated_length": 868.6477661132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 488.0,
"epoch": 0.13546666666666668,
"grad_norm": 0.06415653973817825,
"kl": 0.127960205078125,
"learning_rate": 2.027777777777778e-06,
"loss": -0.0919,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016996802762150764,
"mask/share_reasoning": 0.8339184522628784,
"mask/share_step_conf": 0.11392848193645477,
"num_tokens": 41627788.0,
"reward": 0.7279948592185974,
"reward_std": 0.21570391952991486,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/asymmetric_l2_reward": 0.6600029468536377,
"rewards/final_brier_reward_step": 0.5092679262161255,
"rewards/format_reward_step": 0.94140625,
"step": 127
},
{
"adv/mean_abs_final_conf": 0.6560871005058289,
"adv/mean_abs_reasoning": 0.5164062976837158,
"adv/mean_abs_step_conf": 0.7905282974243164,
"adv/ratio_final_to_reasoning": 1.2704862497778127,
"adv/ratio_step_to_reasoning": 1.5308262137199817,
"adv/std_final_conf": 0.8383203148841858,
"adv/std_reasoning": 0.7577394843101501,
"adv/std_step_conf": 0.92868971824646,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6346334586466165,
"calib/avg_num_step_conf": 9.8359375,
"calib/ece": 0.4346530612244899,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008078007518796837,
"calib/mean_conf": 0.9775102040816327,
"calib/mu_c": 0.981203007518797,
"calib/mu_w": 0.9731250000000001,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4346530612244899,
"calib/std_conf": 0.01478847103214376,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8522678018575852,
"calib/step_q_c_n": 1292.0,
"calib/step_q_gap": 0.0211095636846651,
"calib/step_q_w": 0.8311582381729201,
"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": 0.01953125,
"completions/max_length": 2984.0,
"completions/max_terminated_length": 2984.0,
"completions/mean_length": 950.8515625,
"completions/mean_terminated_length": 969.7928466796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 499.0,
"epoch": 0.13653333333333334,
"grad_norm": 0.04046202450990677,
"kl": 0.11383056640625,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.0728,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.015950171276926994,
"mask/share_reasoning": 0.8544788360595703,
"mask/share_step_conf": 0.11003974080085754,
"num_tokens": 41977870.0,
"reward": 0.7633705139160156,
"reward_std": 0.2553948760032654,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/asymmetric_l2_reward": 0.6937562823295593,
"rewards/final_brier_reward_step": 0.5384535193443298,
"rewards/format_reward_step": 0.953125,
"step": 128
},
{
"adv/mean_abs_final_conf": 0.5023715496063232,
"adv/mean_abs_reasoning": 0.3759290277957916,
"adv/mean_abs_step_conf": 0.763197124004364,
"adv/ratio_final_to_reasoning": 1.3363467901159696,
"adv/ratio_step_to_reasoning": 2.030162790245983,
"adv/std_final_conf": 0.7428443431854248,
"adv/std_reasoning": 0.6614742875099182,
"adv/std_step_conf": 0.9317966103553772,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.62,
"calib/avg_num_step_conf": 9.74609375,
"calib/ece": 0.3892244897959183,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007200000000000317,
"calib/mean_conf": 0.9810612244897958,
"calib/mu_c": 0.9840000000000001,
"calib/mu_w": 0.9767999999999998,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3892244897959183,
"calib/std_conf": 0.013720843912977747,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.847810650887574,
"calib/step_q_c_n": 1352.0,
"calib/step_q_gap": 0.004477317554240634,
"calib/step_q_w": 0.8433333333333334,
"calib/step_q_w_n": 1143.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2828.0,
"completions/max_terminated_length": 2828.0,
"completions/mean_length": 838.765625,
"completions/mean_terminated_length": 862.3453369140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 492.0,
"epoch": 0.1376,
"grad_norm": 0.03307565301656723,
"kl": 0.1181793212890625,
"learning_rate": 1.9722222222222224e-06,
"loss": -0.1346,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01715748943388462,
"mask/share_reasoning": 0.8388442993164062,
"mask/share_step_conf": 0.1166544258594513,
"num_tokens": 42294978.0,
"reward": 0.8008184432983398,
"reward_std": 0.20455628633499146,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/asymmetric_l2_reward": 0.7129415273666382,
"rewards/final_brier_reward_step": 0.5840077996253967,
"rewards/format_reward_step": 0.95703125,
"step": 129
},
{
"adv/mean_abs_final_conf": 0.49385571479797363,
"adv/mean_abs_reasoning": 0.31653401255607605,
"adv/mean_abs_step_conf": 0.7410389184951782,
"adv/ratio_final_to_reasoning": 1.5601979414786709,
"adv/ratio_step_to_reasoning": 2.34110360687984,
"adv/std_final_conf": 0.7291826009750366,
"adv/std_reasoning": 0.6186844110488892,
"adv/std_step_conf": 0.9263778924942017,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.6813129697745083,
"calib/avg_num_step_conf": 9.4375,
"calib/ece": 0.28510288065843614,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.010738045738046265,
"calib/mean_conf": 0.9805761316872428,
"calib/mu_c": 0.983846153846154,
"calib/mu_w": 0.9731081081081078,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.28510288065843614,
"calib/std_conf": 0.013746580804025008,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8522900262467191,
"calib/step_q_c_n": 1524.0,
"calib/step_q_gap": 0.007346080058378224,
"calib/step_q_w": 0.8449439461883409,
"calib/step_q_w_n": 892.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2392.0,
"completions/max_terminated_length": 2392.0,
"completions/mean_length": 793.53125,
"completions/mean_terminated_length": 825.78857421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 359.0,
"epoch": 0.13866666666666666,
"grad_norm": 0.03069177456200123,
"kl": 0.114990234375,
"learning_rate": 1.944444444444445e-06,
"loss": -0.1845,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01768389716744423,
"mask/share_reasoning": 0.8277462124824524,
"mask/share_step_conf": 0.11550737917423248,
"num_tokens": 42603410.0,
"reward": 0.8770005106925964,
"reward_std": 0.16171851754188538,
"rewards/accuracy_reward_step": 0.66015625,
"rewards/asymmetric_l2_reward": 0.7616474628448486,
"rewards/final_brier_reward_step": 0.6712597608566284,
"rewards/format_reward_step": 0.9453125,
"step": 130
},
{
"adv/mean_abs_final_conf": 0.5914018154144287,
"adv/mean_abs_reasoning": 0.5124457478523254,
"adv/mean_abs_step_conf": 0.7483264803886414,
"adv/ratio_final_to_reasoning": 1.1540769298857692,
"adv/ratio_step_to_reasoning": 1.4603038146474991,
"adv/std_final_conf": 0.8130688667297363,
"adv/std_reasoning": 0.7754942178726196,
"adv/std_step_conf": 0.9339196085929871,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.6212319790301443,
"calib/avg_num_step_conf": 10.33984375,
"calib/ece": 0.5120425531914893,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.9957446808510638,
"calib/gap": 0.007698412698412649,
"calib/mean_conf": 0.9758723404255318,
"calib/mu_c": 0.9799999999999998,
"calib/mu_w": 0.9723015873015871,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.5120425531914893,
"calib/std_conf": 0.01575020029764783,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8409661835748792,
"calib/step_q_c_n": 1035.0,
"calib/step_q_gap": 0.018280079356516876,
"calib/step_q_w": 0.8226861042183623,
"calib/step_q_w_n": 1612.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.06640625,
"completions/max_length": 2675.0,
"completions/max_terminated_length": 2675.0,
"completions/mean_length": 808.34375,
"completions/mean_terminated_length": 865.8410034179688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 464.0,
"epoch": 0.13973333333333332,
"grad_norm": 0.033274900168180466,
"kl": 0.125213623046875,
"learning_rate": 1.916666666666667e-06,
"loss": -0.2252,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.01633717119693756,
"mask/share_reasoning": 0.8041282892227173,
"mask/share_step_conf": 0.11312822997570038,
"num_tokens": 42916554.0,
"reward": 0.6535139083862305,
"reward_std": 0.24080276489257812,
"rewards/accuracy_reward_step": 0.4296875,
"rewards/asymmetric_l2_reward": 0.5873867273330688,
"rewards/final_brier_reward_step": 0.4516722857952118,
"rewards/format_reward_step": 0.91015625,
"step": 131
},
{
"adv/mean_abs_final_conf": 0.5862385034561157,
"adv/mean_abs_reasoning": 0.5381482243537903,
"adv/mean_abs_step_conf": 0.7538403272628784,
"adv/ratio_final_to_reasoning": 1.0893625156156046,
"adv/ratio_step_to_reasoning": 1.4008042638588871,
"adv/std_final_conf": 0.7897200584411621,
"adv/std_reasoning": 0.7754783630371094,
"adv/std_step_conf": 0.9292224645614624,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7671005261700359,
"calib/avg_num_step_conf": 9.6875,
"calib/ece": 0.34542168674698803,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.015805178620880822,
"calib/mean_conf": 0.9759437751004016,
"calib/mu_c": 0.9817834394904458,
"calib/mu_w": 0.965978260869565,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.34542168674698803,
"calib/std_conf": 0.014916529853053997,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.846875420875421,
"calib/step_q_c_n": 1485.0,
"calib/step_q_gap": 0.02651361183019485,
"calib/step_q_w": 0.8203618090452262,
"calib/step_q_w_n": 995.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 3067.0,
"completions/max_terminated_length": 3067.0,
"completions/mean_length": 872.87109375,
"completions/mean_terminated_length": 883.2213745117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 469.0,
"epoch": 0.1408,
"grad_norm": 0.04500338062644005,
"kl": 0.3972930908203125,
"learning_rate": 1.888888888888889e-06,
"loss": -0.0107,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.017480727285146713,
"mask/share_reasoning": 0.8506555557250977,
"mask/share_step_conf": 0.12014497816562653,
"num_tokens": 43245601.0,
"reward": 0.8479660749435425,
"reward_std": 0.24857370555400848,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/asymmetric_l2_reward": 0.7464767694473267,
"rewards/final_brier_reward_step": 0.6330491900444031,
"rewards/format_reward_step": 0.96875,
"step": 132
},
{
"adv/mean_abs_final_conf": 0.6809636950492859,
"adv/mean_abs_reasoning": 0.560234010219574,
"adv/mean_abs_step_conf": 0.7415635585784912,
"adv/ratio_final_to_reasoning": 1.2154986713184264,
"adv/ratio_step_to_reasoning": 1.3236675122380526,
"adv/std_final_conf": 0.894681453704834,
"adv/std_reasoning": 0.8429984450340271,
"adv/std_step_conf": 0.9343292713165283,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.6300944669365721,
"calib/avg_num_step_conf": 9.78125,
"calib/ece": 0.5029508196721313,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007924426450742206,
"calib/mean_conf": 0.9701639344262296,
"calib/mu_c": 0.9743859649122807,
"calib/mu_w": 0.9664615384615385,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.5029508196721313,
"calib/std_conf": 0.014256641049997975,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.833780260707635,
"calib/step_q_c_n": 1074.0,
"calib/step_q_gap": 0.026066973994348097,
"calib/step_q_w": 0.8077132867132869,
"calib/step_q_w_n": 1430.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 3056.0,
"completions/max_terminated_length": 3056.0,
"completions/mean_length": 946.0859375,
"completions/mean_terminated_length": 976.6047973632812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 505.0,
"epoch": 0.14186666666666667,
"grad_norm": 0.032102812081575394,
"kl": 0.1101837158203125,
"learning_rate": 1.8611111111111113e-06,
"loss": -0.0748,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.015190929174423218,
"mask/share_reasoning": 0.8482433557510376,
"mask/share_step_conf": 0.10531570762395859,
"num_tokens": 43594143.0,
"reward": 0.6962805986404419,
"reward_std": 0.29506897926330566,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/asymmetric_l2_reward": 0.6364873647689819,
"rewards/final_brier_reward_step": 0.4779488146305084,
"rewards/format_reward_step": 0.9453125,
"step": 133
},
{
"adv/mean_abs_final_conf": 0.6505323648452759,
"adv/mean_abs_reasoning": 0.5010393261909485,
"adv/mean_abs_step_conf": 0.7333165407180786,
"adv/ratio_final_to_reasoning": 1.2983658783649148,
"adv/ratio_step_to_reasoning": 1.4635907849648675,
"adv/std_final_conf": 0.8285301923751831,
"adv/std_reasoning": 0.7577097415924072,
"adv/std_step_conf": 0.9305702447891235,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6528799628819513,
"calib/avg_num_step_conf": 9.2890625,
"calib/ece": 0.40048387096774196,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00881818784383892,
"calib/mean_conf": 0.9690322580645162,
"calib/mu_c": 0.972836879432624,
"calib/mu_w": 0.9640186915887851,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40048387096774196,
"calib/std_conf": 0.013703034443715784,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.827359375,
"calib/step_q_c_n": 1280.0,
"calib/step_q_gap": 0.019189976092896188,
"calib/step_q_w": 0.8081693989071038,
"calib/step_q_w_n": 1098.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2677.0,
"completions/max_terminated_length": 2677.0,
"completions/mean_length": 957.24609375,
"completions/mean_terminated_length": 972.4405517578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 494.0,
"epoch": 0.14293333333333333,
"grad_norm": 0.05993098393082619,
"kl": 0.1142578125,
"learning_rate": 1.8333333333333333e-06,
"loss": -0.0637,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.015441222116351128,
"mask/share_reasoning": 0.8652579188346863,
"mask/share_step_conf": 0.10367587953805923,
"num_tokens": 43948150.0,
"reward": 0.7980170845985413,
"reward_std": 0.24089252948760986,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/asymmetric_l2_reward": 0.7133153676986694,
"rewards/final_brier_reward_step": 0.5795937776565552,
"rewards/format_reward_step": 0.96484375,
"step": 134
},
{
"adv/mean_abs_final_conf": 0.6484160423278809,
"adv/mean_abs_reasoning": 0.49199366569519043,
"adv/mean_abs_step_conf": 0.7641174793243408,
"adv/ratio_final_to_reasoning": 1.3179357531192288,
"adv/ratio_step_to_reasoning": 1.553104303171541,
"adv/std_final_conf": 0.84324711561203,
"adv/std_reasoning": 0.7575876712799072,
"adv/std_step_conf": 0.9321054220199585,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.604211882799783,
"calib/avg_num_step_conf": 9.78515625,
"calib/ece": 0.36265060240963853,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006247965274009837,
"calib/mean_conf": 0.9730923694779117,
"calib/mu_c": 0.9755263157894737,
"calib/mu_w": 0.9692783505154638,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.36265060240963853,
"calib/std_conf": 0.014851186246889227,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8211809392265192,
"calib/step_q_c_n": 1448.0,
"calib/step_q_gap": 0.0074723299549960265,
"calib/step_q_w": 0.8137086092715232,
"calib/step_q_w_n": 1057.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.015625,
"completions/max_length": 2738.0,
"completions/max_terminated_length": 2738.0,
"completions/mean_length": 891.28125,
"completions/mean_terminated_length": 905.4286499023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 480.0,
"epoch": 0.144,
"grad_norm": 0.02977120503783226,
"kl": 0.1153411865234375,
"learning_rate": 1.8055555555555557e-06,
"loss": -0.0757,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.016633452847599983,
"mask/share_reasoning": 0.8540046215057373,
"mask/share_step_conf": 0.11373695731163025,
"num_tokens": 44282198.0,
"reward": 0.8326214551925659,
"reward_std": 0.22683849930763245,
"rewards/accuracy_reward_step": 0.59375,
"rewards/asymmetric_l2_reward": 0.7369366884231567,
"rewards/final_brier_reward_step": 0.6158062219619751,
"rewards/format_reward_step": 0.96875,
"step": 135
},
{
"adv/mean_abs_final_conf": 0.5686486959457397,
"adv/mean_abs_reasoning": 0.4569851756095886,
"adv/mean_abs_step_conf": 0.7533724904060364,
"adv/ratio_final_to_reasoning": 1.2443482333692755,
"adv/ratio_step_to_reasoning": 1.6485709616314934,
"adv/std_final_conf": 0.7785744667053223,
"adv/std_reasoning": 0.7207177877426147,
"adv/std_step_conf": 0.9300398826599121,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6370212468549064,
"calib/avg_num_step_conf": 10.1875,
"calib/ece": 0.37372950819672135,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008268101761252145,
"calib/mean_conf": 0.9720901639344263,
"calib/mu_c": 0.9754109589041093,
"calib/mu_w": 0.9671428571428572,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.37372950819672135,
"calib/std_conf": 0.014687177466768042,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8182135661517799,
"calib/step_q_c_n": 1489.0,
"calib/step_q_gap": 0.022136711817552945,
"calib/step_q_w": 0.7960768543342269,
"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": 0.03125,
"completions/max_length": 2614.0,
"completions/max_terminated_length": 2614.0,
"completions/mean_length": 882.46875,
"completions/mean_terminated_length": 910.9354858398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 463.0,
"epoch": 0.14506666666666668,
"grad_norm": 0.024910585954785347,
"kl": 0.11865234375,
"learning_rate": 1.777777777777778e-06,
"loss": -0.2104,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01636427454650402,
"mask/share_reasoning": 0.8356093168258667,
"mask/share_step_conf": 0.11677637696266174,
"num_tokens": 44616598.0,
"reward": 0.7921877503395081,
"reward_std": 0.21893033385276794,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/asymmetric_l2_reward": 0.6898484230041504,
"rewards/final_brier_reward_step": 0.590620756149292,
"rewards/format_reward_step": 0.94921875,
"step": 136
},
{
"adv/mean_abs_final_conf": 0.5874014496803284,
"adv/mean_abs_reasoning": 0.35178643465042114,
"adv/mean_abs_step_conf": 0.7439720034599304,
"adv/ratio_final_to_reasoning": 1.669767199136157,
"adv/ratio_step_to_reasoning": 2.1148399431582225,
"adv/std_final_conf": 0.8005043268203735,
"adv/std_reasoning": 0.6611693501472473,
"adv/std_step_conf": 0.9287843704223633,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.6661753144654087,
"calib/avg_num_step_conf": 10.35546875,
"calib/ece": 0.39668000000000014,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009730083857442229,
"calib/mean_conf": 0.9726800000000001,
"calib/mu_c": 0.9768055555555555,
"calib/mu_w": 0.9670754716981133,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.39668000000000014,
"calib/std_conf": 0.01468392318149344,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8195276162790698,
"calib/step_q_c_n": 1376.0,
"calib/step_q_gap": 0.02033545941632464,
"calib/step_q_w": 0.7991921568627451,
"calib/step_q_w_n": 1275.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2593.0,
"completions/max_terminated_length": 2593.0,
"completions/mean_length": 914.8203125,
"completions/mean_terminated_length": 925.6680297851562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 489.0,
"epoch": 0.14613333333333334,
"grad_norm": 0.032471757382154465,
"kl": 0.114349365234375,
"learning_rate": 1.75e-06,
"loss": -0.0626,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.016321057453751564,
"mask/share_reasoning": 0.8556821346282959,
"mask/share_step_conf": 0.11627800762653351,
"num_tokens": 44957776.0,
"reward": 0.8128588199615479,
"reward_std": 0.1617228239774704,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.7290788888931274,
"rewards/final_brier_reward_step": 0.5888261795043945,
"rewards/format_reward_step": 0.9765625,
"step": 137
},
{
"adv/mean_abs_final_conf": 0.5582438111305237,
"adv/mean_abs_reasoning": 0.46346306800842285,
"adv/mean_abs_step_conf": 0.7503811120986938,
"adv/ratio_final_to_reasoning": 1.2045054928095766,
"adv/ratio_step_to_reasoning": 1.6190742345947975,
"adv/std_final_conf": 0.7708762288093567,
"adv/std_reasoning": 0.7208356261253357,
"adv/std_step_conf": 0.9279402494430542,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6776187883972316,
"calib/avg_num_step_conf": 9.95703125,
"calib/ece": 0.2922131147540983,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.010649350649350797,
"calib/mean_conf": 0.976639344262295,
"calib/mu_c": 0.9799999999999998,
"calib/mu_w": 0.969350649350649,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2922131147540983,
"calib/std_conf": 0.014855073413732612,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8179331306990881,
"calib/step_q_c_n": 1645.0,
"calib/step_q_gap": 0.0156101218495307,
"calib/step_q_w": 0.8023230088495574,
"calib/step_q_w_n": 904.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2815.0,
"completions/max_terminated_length": 2815.0,
"completions/mean_length": 896.2890625,
"completions/mean_terminated_length": 925.2015991210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 446.0,
"epoch": 0.1472,
"grad_norm": 0.049276694655418396,
"kl": 0.121612548828125,
"learning_rate": 1.7222222222222224e-06,
"loss": -0.0866,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.016306929290294647,
"mask/share_reasoning": 0.8377681970596313,
"mask/share_step_conf": 0.11467483639717102,
"num_tokens": 45291562.0,
"reward": 0.8790077567100525,
"reward_std": 0.24240699410438538,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/asymmetric_l2_reward": 0.7668709754943848,
"rewards/final_brier_reward_step": 0.6700507402420044,
"rewards/format_reward_step": 0.953125,
"step": 138
},
{
"adv/mean_abs_final_conf": 0.5593092441558838,
"adv/mean_abs_reasoning": 0.3993344306945801,
"adv/mean_abs_step_conf": 0.7353893518447876,
"adv/ratio_final_to_reasoning": 1.4006036073149326,
"adv/ratio_step_to_reasoning": 1.8415375567934182,
"adv/std_final_conf": 0.7827629446983337,
"adv/std_reasoning": 0.6816564798355103,
"adv/std_step_conf": 0.9297609925270081,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6579830053667263,
"calib/avg_num_step_conf": 9.88671875,
"calib/ece": 0.2951600000000001,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009251639833035341,
"calib/mean_conf": 0.98316,
"calib/mu_c": 0.9860465116279068,
"calib/mu_w": 0.9767948717948715,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2951600000000001,
"calib/std_conf": 0.01277553912756719,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8246811070998796,
"calib/step_q_c_n": 1662.0,
"calib/step_q_gap": 0.0137835236706505,
"calib/step_q_w": 0.8108975834292291,
"calib/step_q_w_n": 869.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1964.0,
"completions/max_terminated_length": 1964.0,
"completions/mean_length": 792.83203125,
"completions/mean_terminated_length": 811.8600463867188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 524.0,
"epoch": 0.14826666666666666,
"grad_norm": 0.02060694247484207,
"kl": 0.12457275390625,
"learning_rate": 1.6944444444444446e-06,
"loss": -0.0802,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017769603058695793,
"mask/share_reasoning": 0.8357161283493042,
"mask/share_step_conf": 0.12307681888341904,
"num_tokens": 45597623.0,
"reward": 0.9000256061553955,
"reward_std": 0.20213943719863892,
"rewards/accuracy_reward_step": 0.671875,
"rewards/asymmetric_l2_reward": 0.7916465997695923,
"rewards/final_brier_reward_step": 0.6810609102249146,
"rewards/format_reward_step": 0.96484375,
"step": 139
},
{
"adv/mean_abs_final_conf": 0.48904263973236084,
"adv/mean_abs_reasoning": 0.42571309208869934,
"adv/mean_abs_step_conf": 0.7449794411659241,
"adv/ratio_final_to_reasoning": 1.1487610994835613,
"adv/ratio_step_to_reasoning": 1.749956613997448,
"adv/std_final_conf": 0.7239442467689514,
"adv/std_reasoning": 0.7015735507011414,
"adv/std_step_conf": 0.9283146858215332,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6101296958855098,
"calib/avg_num_step_conf": 10.92578125,
"calib/ece": 0.3378925619834712,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0064355992844369325,
"calib/mean_conf": 0.9825206611570249,
"calib/mu_c": 0.9848076923076923,
"calib/mu_w": 0.9783720930232553,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3378925619834712,
"calib/std_conf": 0.012946552030568872,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8200129032258064,
"calib/step_q_c_n": 1550.0,
"calib/step_q_gap": 0.039202959360529754,
"calib/step_q_w": 0.7808099438652767,
"calib/step_q_w_n": 1247.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2956.0,
"completions/max_terminated_length": 2956.0,
"completions/mean_length": 814.66015625,
"completions/mean_terminated_length": 851.2366943359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 471.0,
"epoch": 0.14933333333333335,
"grad_norm": 0.02468651905655861,
"kl": 0.1251983642578125,
"learning_rate": 1.6666666666666667e-06,
"loss": -0.1502,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016852054744958878,
"mask/share_reasoning": 0.8168226480484009,
"mask/share_step_conf": 0.12335656583309174,
"num_tokens": 45911192.0,
"reward": 0.8422061204910278,
"reward_std": 0.21512775123119354,
"rewards/accuracy_reward_step": 0.609375,
"rewards/asymmetric_l2_reward": 0.7500156760215759,
"rewards/final_brier_reward_step": 0.6234589219093323,
"rewards/format_reward_step": 0.9453125,
"step": 140
},
{
"adv/mean_abs_final_conf": 0.4663376808166504,
"adv/mean_abs_reasoning": 0.4102482199668884,
"adv/mean_abs_step_conf": 0.749131441116333,
"adv/ratio_final_to_reasoning": 1.1367207902920067,
"adv/ratio_step_to_reasoning": 1.8260443425611845,
"adv/std_final_conf": 0.7231074571609497,
"adv/std_reasoning": 0.7014703154563904,
"adv/std_step_conf": 0.9257126450538635,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6944444444444445,
"calib/avg_num_step_conf": 10.15625,
"calib/ece": 0.3129795918367346,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.011467178560674585,
"calib/mean_conf": 0.9823673469387755,
"calib/mu_c": 0.9861585365853658,
"calib/mu_w": 0.9746913580246912,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3129795918367346,
"calib/std_conf": 0.013034794687800158,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8134662576687115,
"calib/step_q_c_n": 1630.0,
"calib/step_q_gap": 0.02948687622541246,
"calib/step_q_w": 0.783979381443299,
"calib/step_q_w_n": 970.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.02734375,
"completions/max_length": 2181.0,
"completions/max_terminated_length": 2181.0,
"completions/mean_length": 861.04296875,
"completions/mean_terminated_length": 885.2489624023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 459.0,
"epoch": 0.1504,
"grad_norm": 0.02938557043671608,
"kl": 0.1165771484375,
"learning_rate": 1.638888888888889e-06,
"loss": -0.1667,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.01663772389292717,
"mask/share_reasoning": 0.835965633392334,
"mask/share_step_conf": 0.12005292624235153,
"num_tokens": 46238715.0,
"reward": 0.8724101781845093,
"reward_std": 0.194613516330719,
"rewards/accuracy_reward_step": 0.640625,
"rewards/asymmetric_l2_reward": 0.7691086530685425,
"rewards/final_brier_reward_step": 0.6561804413795471,
"rewards/format_reward_step": 0.95703125,
"step": 141
},
{
"adv/mean_abs_final_conf": 0.5036216974258423,
"adv/mean_abs_reasoning": 0.4271680414676666,
"adv/mean_abs_step_conf": 0.750043511390686,
"adv/ratio_final_to_reasoning": 1.1789779396780145,
"adv/ratio_step_to_reasoning": 1.7558511840297832,
"adv/std_final_conf": 0.758115828037262,
"adv/std_reasoning": 0.7014580965042114,
"adv/std_step_conf": 0.9286013245582581,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.6115348588251037,
"calib/avg_num_step_conf": 10.296875,
"calib/ece": 0.4116599190283402,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006710825638966633,
"calib/mean_conf": 0.98251012145749,
"calib/mu_c": 0.9853900709219856,
"calib/mu_w": 0.978679245283019,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4116599190283402,
"calib/std_conf": 0.013108659050052051,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.815952380952381,
"calib/step_q_c_n": 1428.0,
"calib/step_q_gap": 0.025207347839798278,
"calib/step_q_w": 0.7907450331125827,
"calib/step_q_w_n": 1208.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2588.0,
"completions/max_terminated_length": 2588.0,
"completions/mean_length": 893.21875,
"completions/mean_terminated_length": 914.6560668945312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 413.0,
"epoch": 0.15146666666666667,
"grad_norm": 0.022861601784825325,
"kl": 0.1132965087890625,
"learning_rate": 1.6111111111111113e-06,
"loss": -0.1392,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.016534026712179184,
"mask/share_reasoning": 0.8400761485099792,
"mask/share_step_conf": 0.11995226889848709,
"num_tokens": 46572539.0,
"reward": 0.7890394926071167,
"reward_std": 0.20801588892936707,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/asymmetric_l2_reward": 0.7069774866104126,
"rewards/final_brier_reward_step": 0.5679764747619629,
"rewards/format_reward_step": 0.96484375,
"step": 142
},
{
"adv/mean_abs_final_conf": 0.5079202651977539,
"adv/mean_abs_reasoning": 0.4386137127876282,
"adv/mean_abs_step_conf": 0.723576545715332,
"adv/ratio_final_to_reasoning": 1.158012735100426,
"adv/ratio_step_to_reasoning": 1.6496897489059574,
"adv/std_final_conf": 0.7746670246124268,
"adv/std_reasoning": 0.7205743789672852,
"adv/std_step_conf": 0.9297850131988525,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.658318615330272,
"calib/avg_num_step_conf": 10.03125,
"calib/ece": 0.3740573770491803,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00978664782762284,
"calib/mean_conf": 0.9847131147540983,
"calib/mu_c": 0.9885234899328856,
"calib/mu_w": 0.9787368421052628,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3740573770491803,
"calib/std_conf": 0.012024551052044528,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8181613756613758,
"calib/step_q_c_n": 1512.0,
"calib/step_q_gap": 0.020670845358345402,
"calib/step_q_w": 0.7974905303030304,
"calib/step_q_w_n": 1056.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2830.0,
"completions/max_terminated_length": 2830.0,
"completions/mean_length": 829.546875,
"completions/mean_terminated_length": 856.306396484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 380.0,
"epoch": 0.15253333333333333,
"grad_norm": 0.027216384187340736,
"kl": 0.1242828369140625,
"learning_rate": 1.5833333333333333e-06,
"loss": -0.1348,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01705748960375786,
"mask/share_reasoning": 0.8287699818611145,
"mask/share_step_conf": 0.12292252480983734,
"num_tokens": 46892239.0,
"reward": 0.804224967956543,
"reward_std": 0.20171010494232178,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/asymmetric_l2_reward": 0.7086536884307861,
"rewards/final_brier_reward_step": 0.5935460329055786,
"rewards/format_reward_step": 0.94921875,
"step": 143
},
{
"adv/mean_abs_final_conf": 0.5042335987091064,
"adv/mean_abs_reasoning": 0.4892060458660126,
"adv/mean_abs_step_conf": 0.736696720123291,
"adv/ratio_final_to_reasoning": 1.0307182484151263,
"adv/ratio_step_to_reasoning": 1.5059027302476615,
"adv/std_final_conf": 0.7585073709487915,
"adv/std_reasoning": 0.7576208710670471,
"adv/std_step_conf": 0.9292923212051392,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6311817970745298,
"calib/avg_num_step_conf": 10.15625,
"calib/ece": 0.2748400000000001,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007880968965250479,
"calib/mean_conf": 0.98284,
"calib/mu_c": 0.9851412429378531,
"calib/mu_w": 0.9772602739726026,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2748400000000001,
"calib/std_conf": 0.012693872537567102,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.8190409356725146,
"calib/step_q_c_n": 1710.0,
"calib/step_q_gap": 0.026872396346671934,
"calib/step_q_w": 0.7921685393258426,
"calib/step_q_w_n": 890.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2123.0,
"completions/max_terminated_length": 2123.0,
"completions/mean_length": 852.80078125,
"completions/mean_terminated_length": 869.7888793945312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 423.0,
"epoch": 0.1536,
"grad_norm": 0.02283472567796707,
"kl": 0.125030517578125,
"learning_rate": 1.5555555555555558e-06,
"loss": -0.0987,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.01706160604953766,
"mask/share_reasoning": 0.8418867588043213,
"mask/share_step_conf": 0.12152038514614105,
"num_tokens": 47214684.0,
"reward": 0.9079037308692932,
"reward_std": 0.23519548773765564,
"rewards/accuracy_reward_step": 0.69140625,
"rewards/asymmetric_l2_reward": 0.7887454032897949,
"rewards/final_brier_reward_step": 0.6958121061325073,
"rewards/format_reward_step": 0.96484375,
"step": 144
},
{
"adv/mean_abs_final_conf": 0.6386491656303406,
"adv/mean_abs_reasoning": 0.5374621152877808,
"adv/mean_abs_step_conf": 0.7449643611907959,
"adv/ratio_final_to_reasoning": 1.1882682471272972,
"adv/ratio_step_to_reasoning": 1.3860779020525182,
"adv/std_final_conf": 0.8432608842849731,
"adv/std_reasoning": 0.7929659485816956,
"adv/std_step_conf": 0.9301500916481018,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.5916508538899431,
"calib/avg_num_step_conf": 11.1796875,
"calib/ece": 0.3404166666666667,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0053055028462999365,
"calib/mean_conf": 0.9862500000000001,
"calib/mu_c": 0.9881290322580644,
"calib/mu_w": 0.9828235294117644,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3404166666666667,
"calib/std_conf": 0.009837216747298675,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8118131868131868,
"calib/step_q_c_n": 1638.0,
"calib/step_q_gap": 0.02967266393736978,
"calib/step_q_w": 0.782140522875817,
"calib/step_q_w_n": 1224.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2888.0,
"completions/max_terminated_length": 2888.0,
"completions/mean_length": 894.8984375,
"completions/mean_terminated_length": 916.3760375976562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 463.0,
"epoch": 0.15466666666666667,
"grad_norm": 0.0357648991048336,
"kl": 0.1115264892578125,
"learning_rate": 1.527777777777778e-06,
"loss": -0.0304,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.016461670398712158,
"mask/share_reasoning": 0.8312902450561523,
"mask/share_step_conf": 0.12881061434745789,
"num_tokens": 47546482.0,
"reward": 0.8309597373008728,
"reward_std": 0.2730830907821655,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/asymmetric_l2_reward": 0.7351561188697815,
"rewards/final_brier_reward_step": 0.6166070103645325,
"rewards/format_reward_step": 0.9375,
"step": 145
},
{
"adv/mean_abs_final_conf": 0.6367698907852173,
"adv/mean_abs_reasoning": 0.4902743697166443,
"adv/mean_abs_step_conf": 0.7329921722412109,
"adv/ratio_final_to_reasoning": 1.2988031398688873,
"adv/ratio_step_to_reasoning": 1.4950652481891848,
"adv/std_final_conf": 0.8359977602958679,
"adv/std_reasoning": 0.7575299739837646,
"adv/std_step_conf": 0.930633008480072,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5765886287625417,
"calib/avg_num_step_conf": 10.46484375,
"calib/ece": 0.514,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.004595317725752279,
"calib/mean_conf": 0.9833877551020408,
"calib/mu_c": 0.9858260869565216,
"calib/mu_w": 0.9812307692307694,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.514,
"calib/std_conf": 0.0124356569729222,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8136788445199661,
"calib/step_q_c_n": 1177.0,
"calib/step_q_gap": 0.015303345185745165,
"calib/step_q_w": 0.7983754993342209,
"calib/step_q_w_n": 1502.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2712.0,
"completions/max_terminated_length": 2712.0,
"completions/mean_length": 875.3671875,
"completions/mean_terminated_length": 907.26318359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 492.0,
"epoch": 0.15573333333333333,
"grad_norm": 0.03923913091421127,
"kl": 0.11285400390625,
"learning_rate": 1.5e-06,
"loss": -0.1544,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016228225082159042,
"mask/share_reasoning": 0.8297832012176514,
"mask/share_step_conf": 0.11883234977722168,
"num_tokens": 47877792.0,
"reward": 0.7035650014877319,
"reward_std": 0.21938246488571167,
"rewards/accuracy_reward_step": 0.44921875,
"rewards/asymmetric_l2_reward": 0.6580109596252441,
"rewards/final_brier_reward_step": 0.4678691327571869,
"rewards/format_reward_step": 0.95703125,
"step": 146
},
{
"adv/mean_abs_final_conf": 0.4933481216430664,
"adv/mean_abs_reasoning": 0.43553024530410767,
"adv/mean_abs_step_conf": 0.7302189469337463,
"adv/ratio_final_to_reasoning": 1.1327528385510577,
"adv/ratio_step_to_reasoning": 1.6766205213230903,
"adv/std_final_conf": 0.7655453681945801,
"adv/std_reasoning": 0.7393311262130737,
"adv/std_step_conf": 0.9292852878570557,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6033057851239669,
"calib/avg_num_step_conf": 11.26171875,
"calib/ece": 0.48541322314049595,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006198347107438051,
"calib/mean_conf": 0.985413223140496,
"calib/mu_c": 0.9885123966942149,
"calib/mu_w": 0.9823140495867768,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.48541322314049595,
"calib/std_conf": 0.010834723351670466,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8105288461538462,
"calib/step_q_c_n": 1248.0,
"calib/step_q_gap": 0.023158815572806524,
"calib/step_q_w": 0.7873700305810397,
"calib/step_q_w_n": 1635.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 3001.0,
"completions/max_terminated_length": 3001.0,
"completions/mean_length": 849.50390625,
"completions/mean_terminated_length": 887.6448364257812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 355.0,
"epoch": 0.1568,
"grad_norm": 0.05791052430868149,
"kl": 0.151153564453125,
"learning_rate": 1.4722222222222225e-06,
"loss": -0.1968,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016502130776643753,
"mask/share_reasoning": 0.8159295916557312,
"mask/share_step_conf": 0.12459948658943176,
"num_tokens": 48198945.0,
"reward": 0.7110254168510437,
"reward_std": 0.19756188988685608,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/asymmetric_l2_reward": 0.649394154548645,
"rewards/final_brier_reward_step": 0.4890628457069397,
"rewards/format_reward_step": 0.9453125,
"step": 147
},
{
"adv/mean_abs_final_conf": 0.4634344279766083,
"adv/mean_abs_reasoning": 0.41140034794807434,
"adv/mean_abs_step_conf": 0.7451192140579224,
"adv/ratio_final_to_reasoning": 1.1264803986872212,
"adv/ratio_step_to_reasoning": 1.8111778897959732,
"adv/std_final_conf": 0.7121722102165222,
"adv/std_reasoning": 0.6816737651824951,
"adv/std_step_conf": 0.9280598163604736,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6173570019723866,
"calib/avg_num_step_conf": 10.6171875,
"calib/ece": 0.30992,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006931112572138187,
"calib/mean_conf": 0.98592,
"calib/mu_c": 0.9881656804733728,
"calib/mu_w": 0.9812345679012346,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.30992,
"calib/std_conf": 0.010322480322093142,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8252042529378847,
"calib/step_q_c_n": 1787.0,
"calib/step_q_gap": 0.027298774957218663,
"calib/step_q_w": 0.797905477980666,
"calib/step_q_w_n": 931.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2847.0,
"completions/max_terminated_length": 2847.0,
"completions/mean_length": 833.15234375,
"completions/mean_terminated_length": 843.0316772460938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 417.0,
"epoch": 0.15786666666666666,
"grad_norm": 0.020612573251128197,
"kl": 0.1186370849609375,
"learning_rate": 1.4444444444444445e-06,
"loss": -0.0327,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.017827773466706276,
"mask/share_reasoning": 0.8354048728942871,
"mask/share_step_conf": 0.13504862785339355,
"num_tokens": 48517344.0,
"reward": 0.8763802647590637,
"reward_std": 0.20133787393569946,
"rewards/accuracy_reward_step": 0.66015625,
"rewards/asymmetric_l2_reward": 0.7545421123504639,
"rewards/final_brier_reward_step": 0.6716558933258057,
"rewards/format_reward_step": 0.97265625,
"step": 148
},
{
"adv/mean_abs_final_conf": 0.6034526824951172,
"adv/mean_abs_reasoning": 0.5531489849090576,
"adv/mean_abs_step_conf": 0.7516754269599915,
"adv/ratio_final_to_reasoning": 1.0909405945928472,
"adv/ratio_step_to_reasoning": 1.358902298417077,
"adv/std_final_conf": 0.8103293776512146,
"adv/std_reasoning": 0.7930197715759277,
"adv/std_step_conf": 0.9299349188804626,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.6671369737644904,
"calib/avg_num_step_conf": 10.43359375,
"calib/ece": 0.3564135021097047,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00992830994508831,
"calib/mean_conf": 0.9851054852320675,
"calib/mu_c": 0.9887919463087247,
"calib/mu_w": 0.9788636363636364,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3564135021097047,
"calib/std_conf": 0.01108508765073446,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8246044624746451,
"calib/step_q_c_n": 1479.0,
"calib/step_q_gap": 0.05184439536055108,
"calib/step_q_w": 0.7727600671140941,
"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": 0.05078125,
"completions/max_length": 3056.0,
"completions/max_terminated_length": 3056.0,
"completions/mean_length": 881.70703125,
"completions/mean_terminated_length": 928.8765258789062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 409.0,
"epoch": 0.15893333333333334,
"grad_norm": 0.01654718443751335,
"kl": 0.10540771484375,
"learning_rate": 1.4166666666666667e-06,
"loss": -0.2621,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.016193099319934845,
"mask/share_reasoning": 0.8126864433288574,
"mask/share_step_conf": 0.12033917754888535,
"num_tokens": 48847517.0,
"reward": 0.8093559145927429,
"reward_std": 0.2620517313480377,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/asymmetric_l2_reward": 0.7209059596061707,
"rewards/final_brier_reward_step": 0.5962433815002441,
"rewards/format_reward_step": 0.92578125,
"step": 149
},
{
"adv/mean_abs_final_conf": 0.5678569078445435,
"adv/mean_abs_reasoning": 0.5295990705490112,
"adv/mean_abs_step_conf": 0.7428033351898193,
"adv/ratio_final_to_reasoning": 1.0722392455407297,
"adv/ratio_step_to_reasoning": 1.402576734924759,
"adv/std_final_conf": 0.8097455501556396,
"adv/std_reasoning": 0.7755541205406189,
"adv/std_step_conf": 0.932079553604126,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.551062271062271,
"calib/avg_num_step_conf": 12.6640625,
"calib/ece": 0.36302904564315364,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.995850622406639,
"calib/gap": 0.002730402930402831,
"calib/mean_conf": 0.985435684647303,
"calib/mu_c": 0.9864666666666665,
"calib/mu_w": 0.9837362637362637,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36302904564315364,
"calib/std_conf": 0.011733424454417551,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8237485101311085,
"calib/step_q_c_n": 1678.0,
"calib/step_q_gap": 0.010666668698883441,
"calib/step_q_w": 0.8130818414322251,
"calib/step_q_w_n": 1564.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2614.0,
"completions/max_terminated_length": 2614.0,
"completions/mean_length": 816.42578125,
"completions/mean_terminated_length": 860.1028442382812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 463.0,
"epoch": 0.16,
"grad_norm": 0.025528794154524803,
"kl": 0.11785888671875,
"learning_rate": 1.3888888888888892e-06,
"loss": -0.2231,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.0170246921479702,
"mask/share_reasoning": 0.7955192923545837,
"mask/share_step_conf": 0.13667477667331696,
"num_tokens": 49161482.0,
"reward": 0.8199504613876343,
"reward_std": 0.24832351505756378,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/asymmetric_l2_reward": 0.7381203174591064,
"rewards/final_brier_reward_step": 0.5970929861068726,
"rewards/format_reward_step": 0.9375,
"step": 150
},
{
"adv/mean_abs_final_conf": 0.6715290546417236,
"adv/mean_abs_reasoning": 0.6039446592330933,
"adv/mean_abs_step_conf": 0.7560272812843323,
"adv/ratio_final_to_reasoning": 1.1119049475401455,
"adv/ratio_step_to_reasoning": 1.2518154929035352,
"adv/std_final_conf": 0.859036386013031,
"adv/std_reasoning": 0.826852023601532,
"adv/std_step_conf": 0.9320445656776428,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.6452629291612342,
"calib/avg_num_step_conf": 12.24609375,
"calib/ece": 0.4825106382978723,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008428943937418554,
"calib/mean_conf": 0.9803829787234042,
"calib/mu_c": 0.9846153846153844,
"calib/mu_w": 0.9761864406779659,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4825106382978723,
"calib/std_conf": 0.01393992153506622,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8167237354085604,
"calib/step_q_c_n": 1285.0,
"calib/step_q_gap": 0.034350762435587434,
"calib/step_q_w": 0.7823729729729729,
"calib/step_q_w_n": 1850.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0703125,
"completions/max_length": 2835.0,
"completions/max_terminated_length": 2835.0,
"completions/mean_length": 897.87890625,
"completions/mean_terminated_length": 965.7857666015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 541.0,
"epoch": 0.16106666666666666,
"grad_norm": 0.025068072602152824,
"kl": 0.107177734375,
"learning_rate": 1.3611111111111112e-06,
"loss": -0.2567,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.015033187344670296,
"mask/share_reasoning": 0.7970659732818604,
"mask/share_step_conf": 0.11758832633495331,
"num_tokens": 49498363.0,
"reward": 0.7052844762802124,
"reward_std": 0.29788458347320557,
"rewards/accuracy_reward_step": 0.45703125,
"rewards/asymmetric_l2_reward": 0.657116174697876,
"rewards/final_brier_reward_step": 0.4784526824951172,
"rewards/format_reward_step": 0.91796875,
"step": 151
},
{
"adv/mean_abs_final_conf": 0.628909707069397,
"adv/mean_abs_reasoning": 0.5587596297264099,
"adv/mean_abs_step_conf": 0.7436010241508484,
"adv/ratio_final_to_reasoning": 1.1255460731429994,
"adv/ratio_step_to_reasoning": 1.3308066377575343,
"adv/std_final_conf": 0.8389648199081421,
"adv/std_reasoning": 0.810052752494812,
"adv/std_step_conf": 0.9334518313407898,
"calib/answer_extract_rate": 0.91015625,
"calib/auroc": 0.60296992481203,
"calib/avg_num_step_conf": 12.109375,
"calib/ece": 0.41248927038626615,
"calib/final_conf_rate": 0.91015625,
"calib/format_rate": 0.90625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006315037593985107,
"calib/mean_conf": 0.983304721030043,
"calib/mu_c": 0.9860150375939849,
"calib/mu_w": 0.9796999999999998,
"calib/nonempty_final_conf_rate": 0.91015625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.41248927038626615,
"calib/std_conf": 0.012593915829054145,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8207415107415107,
"calib/step_q_c_n": 1443.0,
"calib/step_q_gap": 0.020783755762633138,
"calib/step_q_w": 0.7999577549788776,
"calib/step_q_w_n": 1657.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0703125,
"completions/max_length": 2641.0,
"completions/max_terminated_length": 2641.0,
"completions/mean_length": 852.71875,
"completions/mean_terminated_length": 917.2101440429688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 487.0,
"epoch": 0.16213333333333332,
"grad_norm": 0.020201267674565315,
"kl": 0.1133880615234375,
"learning_rate": 1.3333333333333334e-06,
"loss": -0.3514,
"mask/has_final_conf_rate": 0.91015625,
"mask/share_final_conf": 0.01569315977394581,
"mask/share_reasoning": 0.7901881337165833,
"mask/share_step_conf": 0.12380620837211609,
"num_tokens": 49822051.0,
"reward": 0.7507580518722534,
"reward_std": 0.28133276104927063,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/asymmetric_l2_reward": 0.6852730512619019,
"rewards/final_brier_reward_step": 0.5310866832733154,
"rewards/format_reward_step": 0.90625,
"step": 152
},
{
"adv/mean_abs_final_conf": 0.5813193917274475,
"adv/mean_abs_reasoning": 0.5272849798202515,
"adv/mean_abs_step_conf": 0.7468937635421753,
"adv/ratio_final_to_reasoning": 1.1024766757543825,
"adv/ratio_step_to_reasoning": 1.4164897391857951,
"adv/std_final_conf": 0.7978730797767639,
"adv/std_reasoning": 0.7755938768386841,
"adv/std_step_conf": 0.9315870404243469,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.5637419871794871,
"calib/avg_num_step_conf": 10.6875,
"calib/ece": 0.3234322033898306,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0037051282051281387,
"calib/mean_conf": 0.984449152542373,
"calib/mu_c": 0.9857051282051281,
"calib/mu_w": 0.982,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3234322033898306,
"calib/std_conf": 0.011613184688269178,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8166891064871481,
"calib/step_q_c_n": 1634.0,
"calib/step_q_gap": 0.02629890685012448,
"calib/step_q_w": 0.7903901996370236,
"calib/step_q_w_n": 1102.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2913.0,
"completions/max_terminated_length": 2913.0,
"completions/mean_length": 850.31640625,
"completions/mean_terminated_length": 903.24072265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 473.0,
"epoch": 0.1632,
"grad_norm": 0.019481275230646133,
"kl": 0.1083831787109375,
"learning_rate": 1.3055555555555556e-06,
"loss": -0.2613,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.01606718823313713,
"mask/share_reasoning": 0.8041765689849854,
"mask/share_step_conf": 0.12116247415542603,
"num_tokens": 50147052.0,
"reward": 0.8226182460784912,
"reward_std": 0.2654617428779602,
"rewards/accuracy_reward_step": 0.609375,
"rewards/asymmetric_l2_reward": 0.7187087535858154,
"rewards/final_brier_reward_step": 0.6202777624130249,
"rewards/format_reward_step": 0.921875,
"step": 153
},
{
"adv/mean_abs_final_conf": 0.5330643653869629,
"adv/mean_abs_reasoning": 0.41719967126846313,
"adv/mean_abs_step_conf": 0.7645653486251831,
"adv/ratio_final_to_reasoning": 1.2777200033888383,
"adv/ratio_step_to_reasoning": 1.8326125385012448,
"adv/std_final_conf": 0.7777272462844849,
"adv/std_reasoning": 0.7014922499656677,
"adv/std_step_conf": 0.9263600707054138,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.5758741258741258,
"calib/avg_num_step_conf": 11.234375,
"calib/ece": 0.4437083333333335,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.004552447552447769,
"calib/mean_conf": 0.9853750000000001,
"calib/mu_c": 0.9874615384615385,
"calib/mu_w": 0.9829090909090907,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4437083333333335,
"calib/std_conf": 0.01083325320483188,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8275321725965178,
"calib/step_q_c_n": 1321.0,
"calib/step_q_gap": 0.028303876776582082,
"calib/step_q_w": 0.7992282958199357,
"calib/step_q_w_n": 1555.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2928.0,
"completions/max_terminated_length": 2928.0,
"completions/mean_length": 813.80859375,
"completions/mean_terminated_length": 860.8883666992188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 420.0,
"epoch": 0.16426666666666667,
"grad_norm": 0.049363043159246445,
"kl": 0.12261962890625,
"learning_rate": 1.2777777777777779e-06,
"loss": -0.1861,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.016554001718759537,
"mask/share_reasoning": 0.802830696105957,
"mask/share_step_conf": 0.12592774629592896,
"num_tokens": 50459827.0,
"reward": 0.7538589239120483,
"reward_std": 0.19900038838386536,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/asymmetric_l2_reward": 0.6964658498764038,
"rewards/final_brier_reward_step": 0.5221894383430481,
"rewards/format_reward_step": 0.9375,
"step": 154
},
{
"adv/mean_abs_final_conf": 0.597705602645874,
"adv/mean_abs_reasoning": 0.48494482040405273,
"adv/mean_abs_step_conf": 0.7656376957893372,
"adv/ratio_final_to_reasoning": 1.2325229129117614,
"adv/ratio_step_to_reasoning": 1.5788140497127343,
"adv/std_final_conf": 0.7969999313354492,
"adv/std_reasoning": 0.7393460869789124,
"adv/std_step_conf": 0.9281103014945984,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.600145137880987,
"calib/avg_num_step_conf": 12.203125,
"calib/ece": 0.43076271186440684,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005834542815675081,
"calib/mean_conf": 0.9816101694915255,
"calib/mu_c": 0.9842307692307694,
"calib/mu_w": 0.9783962264150943,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43076271186440684,
"calib/std_conf": 0.013401893300540396,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8178236529041288,
"calib/step_q_c_n": 1429.0,
"calib/step_q_gap": 0.010714508361356012,
"calib/step_q_w": 0.8071091445427728,
"calib/step_q_w_n": 1695.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0703125,
"completions/max_length": 2104.0,
"completions/max_terminated_length": 2104.0,
"completions/mean_length": 804.8515625,
"completions/mean_terminated_length": 865.7227172851562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 514.0,
"epoch": 0.16533333333333333,
"grad_norm": 0.020702464506030083,
"kl": 0.127410888671875,
"learning_rate": 1.25e-06,
"loss": -0.2912,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.015847964212298393,
"mask/share_reasoning": 0.784995436668396,
"mask/share_step_conf": 0.12884411215782166,
"num_tokens": 50773085.0,
"reward": 0.7318023443222046,
"reward_std": 0.19392001628875732,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/asymmetric_l2_reward": 0.6524415016174316,
"rewards/final_brier_reward_step": 0.5252257585525513,
"rewards/format_reward_step": 0.921875,
"step": 155
},
{
"adv/mean_abs_final_conf": 0.6328127384185791,
"adv/mean_abs_reasoning": 0.5709925889968872,
"adv/mean_abs_step_conf": 0.7423877120018005,
"adv/ratio_final_to_reasoning": 1.108267866541485,
"adv/ratio_step_to_reasoning": 1.3001704861108936,
"adv/std_final_conf": 0.8355315327644348,
"adv/std_reasoning": 0.8098950386047363,
"adv/std_step_conf": 0.9297937750816345,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5610687022900763,
"calib/avg_num_step_conf": 11.86328125,
"calib/ece": 0.43983402489626555,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.003583622484386262,
"calib/mean_conf": 0.983402489626556,
"calib/mu_c": 0.9850381679389315,
"calib/mu_w": 0.9814545454545452,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.43983402489626555,
"calib/std_conf": 0.012325118316657337,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8224016282225237,
"calib/step_q_c_n": 1474.0,
"calib/step_q_gap": 0.011454731229561421,
"calib/step_q_w": 0.8109468969929623,
"calib/step_q_w_n": 1563.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 3059.0,
"completions/max_terminated_length": 3059.0,
"completions/mean_length": 858.23828125,
"completions/mean_terminated_length": 893.1259765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 485.0,
"epoch": 0.1664,
"grad_norm": 0.02893920987844467,
"kl": 0.11334228515625,
"learning_rate": 1.2222222222222223e-06,
"loss": -0.1956,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.0159677155315876,
"mask/share_reasoning": 0.8142074346542358,
"mask/share_step_conf": 0.13076233863830566,
"num_tokens": 51097554.0,
"reward": 0.764479398727417,
"reward_std": 0.25560951232910156,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/asymmetric_l2_reward": 0.7110797762870789,
"rewards/final_brier_reward_step": 0.5272538661956787,
"rewards/format_reward_step": 0.94140625,
"step": 156
},
{
"adv/mean_abs_final_conf": 0.5657306909561157,
"adv/mean_abs_reasoning": 0.46107548475265503,
"adv/mean_abs_step_conf": 0.726364254951477,
"adv/ratio_final_to_reasoning": 1.2269806347643992,
"adv/ratio_step_to_reasoning": 1.5753694979925832,
"adv/std_final_conf": 0.7988712787628174,
"adv/std_reasoning": 0.7394241690635681,
"adv/std_step_conf": 0.9278371930122375,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6322088353413655,
"calib/avg_num_step_conf": 13.40625,
"calib/ece": 0.2919502074688797,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007859437751004172,
"calib/mean_conf": 0.980746887966805,
"calib/mu_c": 0.9831927710843372,
"calib/mu_w": 0.975333333333333,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2919502074688797,
"calib/std_conf": 0.013915208576745691,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8242296511627907,
"calib/step_q_c_n": 2064.0,
"calib/step_q_gap": 0.06155421256629934,
"calib/step_q_w": 0.7626754385964913,
"calib/step_q_w_n": 1368.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2931.0,
"completions/max_terminated_length": 2931.0,
"completions/mean_length": 860.27734375,
"completions/mean_terminated_length": 913.8216552734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 480.0,
"epoch": 0.16746666666666668,
"grad_norm": 0.020823344588279724,
"kl": 0.1176910400390625,
"learning_rate": 1.1944444444444446e-06,
"loss": -0.2683,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.015636589378118515,
"mask/share_reasoning": 0.7919765710830688,
"mask/share_step_conf": 0.13379308581352234,
"num_tokens": 51421513.0,
"reward": 0.8676108121871948,
"reward_std": 0.2390524446964264,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/asymmetric_l2_reward": 0.7548935413360596,
"rewards/final_brier_reward_step": 0.662359356880188,
"rewards/format_reward_step": 0.94140625,
"step": 157
},
{
"adv/mean_abs_final_conf": 0.5855216979980469,
"adv/mean_abs_reasoning": 0.5074455738067627,
"adv/mean_abs_step_conf": 0.7488413453102112,
"adv/ratio_final_to_reasoning": 1.1538610803234948,
"adv/ratio_step_to_reasoning": 1.4757077092870121,
"adv/std_final_conf": 0.8118085861206055,
"adv/std_reasoning": 0.7754456400871277,
"adv/std_step_conf": 0.9327535033226013,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6261183261183261,
"calib/avg_num_step_conf": 11.73828125,
"calib/ece": 0.35143442622950816,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00743578643578624,
"calib/mean_conf": 0.9825819672131147,
"calib/mu_c": 0.9853246753246752,
"calib/mu_w": 0.9778888888888889,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35143442622950816,
"calib/std_conf": 0.012847363806859777,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8337210648148149,
"calib/step_q_c_n": 1728.0,
"calib/step_q_gap": 0.029805638033295834,
"calib/step_q_w": 0.8039154267815191,
"calib/step_q_w_n": 1277.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2647.0,
"completions/max_terminated_length": 2647.0,
"completions/mean_length": 843.34765625,
"completions/mean_terminated_length": 870.5523681640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 497.0,
"epoch": 0.16853333333333334,
"grad_norm": 0.041330933570861816,
"kl": 0.1220550537109375,
"learning_rate": 1.1666666666666668e-06,
"loss": -0.1525,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01676858961582184,
"mask/share_reasoning": 0.8167569637298584,
"mask/share_step_conf": 0.13522447645664215,
"num_tokens": 51742650.0,
"reward": 0.8231799602508545,
"reward_std": 0.2605646252632141,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/asymmetric_l2_reward": 0.7187595367431641,
"rewards/final_brier_reward_step": 0.616662859916687,
"rewards/format_reward_step": 0.953125,
"step": 158
},
{
"adv/mean_abs_final_conf": 0.6384714841842651,
"adv/mean_abs_reasoning": 0.5273313522338867,
"adv/mean_abs_step_conf": 0.7569658756256104,
"adv/ratio_final_to_reasoning": 1.2107595755108536,
"adv/ratio_step_to_reasoning": 1.4354653339289298,
"adv/std_final_conf": 0.8230856657028198,
"adv/std_reasoning": 0.775598406791687,
"adv/std_step_conf": 0.9249182343482971,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6248127978216473,
"calib/avg_num_step_conf": 12.60546875,
"calib/ece": 0.44563786008230455,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007439754935330223,
"calib/mean_conf": 0.9806172839506173,
"calib/mu_c": 0.9840769230769232,
"calib/mu_w": 0.976637168141593,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44563786008230455,
"calib/std_conf": 0.013908490968402642,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8364338983050849,
"calib/step_q_c_n": 1475.0,
"calib/step_q_gap": 0.006091432551660114,
"calib/step_q_w": 0.8303424657534247,
"calib/step_q_w_n": 1752.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2692.0,
"completions/max_terminated_length": 2692.0,
"completions/mean_length": 860.046875,
"completions/mean_terminated_length": 895.0081176757812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 490.0,
"epoch": 0.1696,
"grad_norm": 0.026793481782078743,
"kl": 0.1242218017578125,
"learning_rate": 1.138888888888889e-06,
"loss": -0.1202,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01638055592775345,
"mask/share_reasoning": 0.8136330246925354,
"mask/share_step_conf": 0.13092389702796936,
"num_tokens": 52067606.0,
"reward": 0.7726041078567505,
"reward_std": 0.25494319200515747,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/asymmetric_l2_reward": 0.7267637848854065,
"rewards/final_brier_reward_step": 0.5278195142745972,
"rewards/format_reward_step": 0.9453125,
"step": 159
},
{
"adv/mean_abs_final_conf": 0.5665127038955688,
"adv/mean_abs_reasoning": 0.4910869002342224,
"adv/mean_abs_step_conf": 0.7735710144042969,
"adv/ratio_final_to_reasoning": 1.1535895248384194,
"adv/ratio_step_to_reasoning": 1.5752222550333648,
"adv/std_final_conf": 0.7866808176040649,
"adv/std_reasoning": 0.7395008206367493,
"adv/std_step_conf": 0.9210416674613953,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.5971209641422407,
"calib/avg_num_step_conf": 12.1171875,
"calib/ece": 0.37843881856540085,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 0.9957805907172996,
"calib/gap": 0.00635694093140915,
"calib/mean_conf": 0.9818143459915611,
"calib/mu_c": 0.9843356643356642,
"calib/mu_w": 0.9779787234042551,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.37843881856540085,
"calib/std_conf": 0.01418976739886446,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8283772198407838,
"calib/step_q_c_n": 1633.0,
"calib/step_q_gap": 0.019091991794493723,
"calib/step_q_w": 0.8092852280462901,
"calib/step_q_w_n": 1469.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 3024.0,
"completions/max_terminated_length": 3024.0,
"completions/mean_length": 889.9140625,
"completions/mean_terminated_length": 929.8693237304688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 557.0,
"epoch": 0.17066666666666666,
"grad_norm": 0.0199897438287735,
"kl": 0.1261749267578125,
"learning_rate": 1.111111111111111e-06,
"loss": -0.1778,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.01571989618241787,
"mask/share_reasoning": 0.8111594319343567,
"mask/share_step_conf": 0.1301519125699997,
"num_tokens": 52400264.0,
"reward": 0.7833555936813354,
"reward_std": 0.2323831170797348,
"rewards/accuracy_reward_step": 0.55859375,
"rewards/asymmetric_l2_reward": 0.6977380514144897,
"rewards/final_brier_reward_step": 0.5736604928970337,
"rewards/format_reward_step": 0.91796875,
"step": 160
},
{
"adv/mean_abs_final_conf": 0.5213401317596436,
"adv/mean_abs_reasoning": 0.38532155752182007,
"adv/mean_abs_step_conf": 0.7431133985519409,
"adv/ratio_final_to_reasoning": 1.3530001672177945,
"adv/ratio_step_to_reasoning": 1.9285539156730407,
"adv/std_final_conf": 0.7485088109970093,
"adv/std_reasoning": 0.6817570328712463,
"adv/std_step_conf": 0.9272958040237427,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.6418430335097002,
"calib/avg_num_step_conf": 13.37890625,
"calib/ece": 0.24082304526748965,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00853968253968218,
"calib/mean_conf": 0.9815637860082304,
"calib/mu_c": 0.9837777777777776,
"calib/mu_w": 0.9752380952380955,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.24082304526748965,
"calib/std_conf": 0.013487650397239634,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.830957345971564,
"calib/step_q_c_n": 2110.0,
"calib/step_q_gap": 0.004637954336582872,
"calib/step_q_w": 0.8263193916349811,
"calib/step_q_w_n": 1315.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2860.0,
"completions/max_terminated_length": 2860.0,
"completions/mean_length": 834.53125,
"completions/mean_terminated_length": 875.57373046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 494.0,
"epoch": 0.17173333333333332,
"grad_norm": 0.04618752375245094,
"kl": 0.12261962890625,
"learning_rate": 1.0833333333333335e-06,
"loss": -0.2442,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01637279987335205,
"mask/share_reasoning": 0.7970735430717468,
"mask/share_step_conf": 0.13967867195606232,
"num_tokens": 52717824.0,
"reward": 0.9100635051727295,
"reward_std": 0.20310011506080627,
"rewards/accuracy_reward_step": 0.703125,
"rewards/asymmetric_l2_reward": 0.7748411893844604,
"rewards/final_brier_reward_step": 0.7148171663284302,
"rewards/format_reward_step": 0.94921875,
"step": 161
},
{
"adv/mean_abs_final_conf": 0.4991297721862793,
"adv/mean_abs_reasoning": 0.36160850524902344,
"adv/mean_abs_step_conf": 0.7166227102279663,
"adv/ratio_final_to_reasoning": 1.3803042930158713,
"adv/ratio_step_to_reasoning": 1.981763979070295,
"adv/std_final_conf": 0.7472751140594482,
"adv/std_reasoning": 0.6614494919776917,
"adv/std_step_conf": 0.9230762720108032,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6346153846153846,
"calib/avg_num_step_conf": 12.59765625,
"calib/ece": 0.25020746887966816,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.995850622406639,
"calib/gap": 0.008897727272727418,
"calib/mean_conf": 0.9804979253112034,
"calib/mu_c": 0.9828977272727275,
"calib/mu_w": 0.9740000000000001,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.25020746887966816,
"calib/std_conf": 0.014737009746776298,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8359530791788857,
"calib/step_q_c_n": 2046.0,
"calib/step_q_gap": 0.03344247697362701,
"calib/step_q_w": 0.8025106022052587,
"calib/step_q_w_n": 1179.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2886.0,
"completions/max_terminated_length": 2886.0,
"completions/mean_length": 840.6953125,
"completions/mean_terminated_length": 882.0409545898438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 500.0,
"epoch": 0.1728,
"grad_norm": 0.03857972472906113,
"kl": 0.1229400634765625,
"learning_rate": 1.0555555555555557e-06,
"loss": -0.1944,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.016243983060121536,
"mask/share_reasoning": 0.8029369711875916,
"mask/share_step_conf": 0.1339440643787384,
"num_tokens": 53037186.0,
"reward": 0.9082566499710083,
"reward_std": 0.2010715901851654,
"rewards/accuracy_reward_step": 0.69140625,
"rewards/asymmetric_l2_reward": 0.789810061454773,
"rewards/final_brier_reward_step": 0.7001405954360962,
"rewards/format_reward_step": 0.94140625,
"step": 162
},
{
"adv/mean_abs_final_conf": 0.5536430478096008,
"adv/mean_abs_reasoning": 0.4788112938404083,
"adv/mean_abs_step_conf": 0.7330542206764221,
"adv/ratio_final_to_reasoning": 1.156286526512331,
"adv/ratio_step_to_reasoning": 1.530987740069379,
"adv/std_final_conf": 0.7959635257720947,
"adv/std_reasoning": 0.7577126622200012,
"adv/std_step_conf": 0.9270132184028625,
"calib/answer_extract_rate": 0.90234375,
"calib/auroc": 0.6336287313432836,
"calib/avg_num_step_conf": 12.34765625,
"calib/ece": 0.398,
"calib/final_conf_rate": 0.8984375,
"calib/format_rate": 0.8984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008017723880596739,
"calib/mean_conf": 0.9806086956521739,
"calib/mu_c": 0.9839552238805968,
"calib/mu_w": 0.9759375,
"calib/nonempty_final_conf_rate": 0.8984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.398,
"calib/std_conf": 0.013911956479276587,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.838726322664925,
"calib/step_q_c_n": 1531.0,
"calib/step_q_gap": 0.020057611008483023,
"calib/step_q_w": 0.818668711656442,
"calib/step_q_w_n": 1630.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 3050.0,
"completions/max_terminated_length": 3050.0,
"completions/mean_length": 901.875,
"completions/mean_terminated_length": 958.0083618164062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 461.0,
"epoch": 0.17386666666666667,
"grad_norm": 0.023081474006175995,
"kl": 0.1114044189453125,
"learning_rate": 1.0277777777777777e-06,
"loss": -0.2066,
"mask/has_final_conf_rate": 0.8984375,
"mask/share_final_conf": 0.015411942265927792,
"mask/share_reasoning": 0.7985349893569946,
"mask/share_step_conf": 0.12745928764343262,
"num_tokens": 53372898.0,
"reward": 0.745888888835907,
"reward_std": 0.23261308670043945,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/asymmetric_l2_reward": 0.6664301156997681,
"rewards/final_brier_reward_step": 0.5409726500511169,
"rewards/format_reward_step": 0.8984375,
"step": 163
},
{
"adv/mean_abs_final_conf": 0.6406596302986145,
"adv/mean_abs_reasoning": 0.5433213710784912,
"adv/mean_abs_step_conf": 0.7142074704170227,
"adv/ratio_final_to_reasoning": 1.1791541146760105,
"adv/ratio_step_to_reasoning": 1.3145212178923187,
"adv/std_final_conf": 0.8660091161727905,
"adv/std_reasoning": 0.8098387718200684,
"adv/std_step_conf": 0.9305307269096375,
"calib/answer_extract_rate": 0.90625,
"calib/auroc": 0.6472543123912011,
"calib/avg_num_step_conf": 12.48046875,
"calib/ece": 0.36359307359307347,
"calib/final_conf_rate": 0.90234375,
"calib/format_rate": 0.8984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008951574616236524,
"calib/mean_conf": 0.9783116883116881,
"calib/mu_c": 0.9817605633802815,
"calib/mu_w": 0.972808988764045,
"calib/nonempty_final_conf_rate": 0.90234375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.36359307359307347,
"calib/std_conf": 0.014747743013404825,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8337029702970297,
"calib/step_q_c_n": 1515.0,
"calib/step_q_gap": 0.028054160773219983,
"calib/step_q_w": 0.8056488095238097,
"calib/step_q_w_n": 1680.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.07421875,
"completions/max_length": 3060.0,
"completions/max_terminated_length": 3060.0,
"completions/mean_length": 896.9609375,
"completions/mean_terminated_length": 968.869140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 565.0,
"epoch": 0.17493333333333333,
"grad_norm": 0.0218285471200943,
"kl": 0.1168670654296875,
"learning_rate": 1.0000000000000002e-06,
"loss": -0.3059,
"mask/has_final_conf_rate": 0.90234375,
"mask/share_final_conf": 0.014433878473937511,
"mask/share_reasoning": 0.7988092303276062,
"mask/share_step_conf": 0.1125381588935852,
"num_tokens": 53708656.0,
"reward": 0.7782033681869507,
"reward_std": 0.24336914718151093,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/asymmetric_l2_reward": 0.6931148171424866,
"rewards/final_brier_reward_step": 0.5726667642593384,
"rewards/format_reward_step": 0.8984375,
"step": 164
},
{
"adv/mean_abs_final_conf": 0.5370346307754517,
"adv/mean_abs_reasoning": 0.49270907044410706,
"adv/mean_abs_step_conf": 0.7504063844680786,
"adv/ratio_final_to_reasoning": 1.0899629476912034,
"adv/ratio_step_to_reasoning": 1.5230212502312859,
"adv/std_final_conf": 0.7757647633552551,
"adv/std_reasoning": 0.7753999829292297,
"adv/std_step_conf": 0.927631139755249,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6621621621621621,
"calib/avg_num_step_conf": 11.73046875,
"calib/ece": 0.433155737704918,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009493327914380734,
"calib/mean_conf": 0.9782377049180327,
"calib/mu_c": 0.9825563909774436,
"calib/mu_w": 0.9730630630630629,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.433155737704918,
"calib/std_conf": 0.014590336617139838,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8378965053763441,
"calib/step_q_c_n": 1488.0,
"calib/step_q_gap": 0.023467462472053557,
"calib/step_q_w": 0.8144290429042905,
"calib/step_q_w_n": 1515.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 2974.0,
"completions/max_terminated_length": 2974.0,
"completions/mean_length": 913.09375,
"completions/mean_terminated_length": 946.3643798828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 522.0,
"epoch": 0.176,
"grad_norm": 0.05142681300640106,
"kl": 0.124298095703125,
"learning_rate": 9.722222222222224e-07,
"loss": -0.1896,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.015462111681699753,
"mask/share_reasoning": 0.8250299692153931,
"mask/share_step_conf": 0.12435169517993927,
"num_tokens": 54047984.0,
"reward": 0.7622162699699402,
"reward_std": 0.23104625940322876,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/asymmetric_l2_reward": 0.6923520565032959,
"rewards/final_brier_reward_step": 0.5383304357528687,
"rewards/format_reward_step": 0.94921875,
"step": 165
},
{
"adv/mean_abs_final_conf": 0.5480644106864929,
"adv/mean_abs_reasoning": 0.49569880962371826,
"adv/mean_abs_step_conf": 0.7524327039718628,
"adv/ratio_final_to_reasoning": 1.1056399572605895,
"adv/ratio_step_to_reasoning": 1.5179231609271557,
"adv/std_final_conf": 0.8023747801780701,
"adv/std_reasoning": 0.775547981262207,
"adv/std_step_conf": 0.9304532408714294,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.8121466768525591,
"calib/avg_num_step_conf": 12.640625,
"calib/ece": 0.33221757322175716,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.018593582887700855,
"calib/mean_conf": 0.9765690376569036,
"calib/mu_c": 0.9831818181818182,
"calib/mu_w": 0.9645882352941173,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33221757322175716,
"calib/std_conf": 0.014861510039826178,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8385502283105023,
"calib/step_q_c_n": 1752.0,
"calib/step_q_gap": 0.027559662272766294,
"calib/step_q_w": 0.810990566037736,
"calib/step_q_w_n": 1484.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2825.0,
"completions/max_terminated_length": 2825.0,
"completions/mean_length": 921.578125,
"completions/mean_terminated_length": 966.901611328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 519.0,
"epoch": 0.17706666666666668,
"grad_norm": 0.016354931518435478,
"kl": 0.12017822265625,
"learning_rate": 9.444444444444445e-07,
"loss": -0.1826,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.015144339762628078,
"mask/share_reasoning": 0.8095753192901611,
"mask/share_step_conf": 0.12840533256530762,
"num_tokens": 54390092.0,
"reward": 0.8301776647567749,
"reward_std": 0.25855541229248047,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/asymmetric_l2_reward": 0.728964626789093,
"rewards/final_brier_reward_step": 0.6243593692779541,
"rewards/format_reward_step": 0.93359375,
"step": 166
},
{
"adv/mean_abs_final_conf": 0.5595499277114868,
"adv/mean_abs_reasoning": 0.46411240100860596,
"adv/mean_abs_step_conf": 0.7970150709152222,
"adv/ratio_final_to_reasoning": 1.2056345111560836,
"adv/ratio_step_to_reasoning": 1.7172888920510514,
"adv/std_final_conf": 0.7713521122932434,
"adv/std_reasoning": 0.7207490801811218,
"adv/std_step_conf": 0.9261698126792908,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5107586512866017,
"calib/avg_num_step_conf": 12.05859375,
"calib/ece": 0.3246530612244899,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.000559006211180324,
"calib/mean_conf": 0.981795918367347,
"calib/mu_c": 0.9819875776397515,
"calib/mu_w": 0.9814285714285712,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3246530612244899,
"calib/std_conf": 0.013249535366265032,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.830916890080429,
"calib/step_q_c_n": 1865.0,
"calib/step_q_gap": -0.014811424158523545,
"calib/step_q_w": 0.8457283142389526,
"calib/step_q_w_n": 1222.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0390625,
"completions/max_length": 2656.0,
"completions/max_terminated_length": 2656.0,
"completions/mean_length": 838.125,
"completions/mean_terminated_length": 872.195068359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 522.0,
"epoch": 0.17813333333333334,
"grad_norm": 0.033019956201314926,
"kl": 0.1278839111328125,
"learning_rate": 9.166666666666666e-07,
"loss": -0.1369,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016273878514766693,
"mask/share_reasoning": 0.8090370893478394,
"mask/share_step_conf": 0.13562653958797455,
"num_tokens": 54710260.0,
"reward": 0.8514053821563721,
"reward_std": 0.24220512807369232,
"rewards/accuracy_reward_step": 0.62890625,
"rewards/asymmetric_l2_reward": 0.7450146675109863,
"rewards/final_brier_reward_step": 0.640608549118042,
"rewards/format_reward_step": 0.95703125,
"step": 167
},
{
"adv/mean_abs_final_conf": 0.6799944043159485,
"adv/mean_abs_reasoning": 0.6203447580337524,
"adv/mean_abs_step_conf": 0.7689940929412842,
"adv/ratio_final_to_reasoning": 1.096155638473132,
"adv/ratio_step_to_reasoning": 1.239623746283746,
"adv/std_final_conf": 0.8511113524436951,
"adv/std_reasoning": 0.8269323110580444,
"adv/std_step_conf": 0.9306549429893494,
"calib/answer_extract_rate": 0.9140625,
"calib/auroc": 0.5853057199211045,
"calib/avg_num_step_conf": 12.35546875,
"calib/ece": 0.3105982905982907,
"calib/final_conf_rate": 0.9140625,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005128205128205332,
"calib/mean_conf": 0.9772649572649573,
"calib/mu_c": 0.9789743589743589,
"calib/mu_w": 0.9738461538461536,
"calib/nonempty_final_conf_rate": 0.9140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3105982905982907,
"calib/std_conf": 0.014741297976223637,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8346118721461188,
"calib/step_q_c_n": 1752.0,
"calib/step_q_gap": 0.04859486293279491,
"calib/step_q_w": 0.7860170092133238,
"calib/step_q_w_n": 1411.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2964.0,
"completions/max_terminated_length": 2964.0,
"completions/mean_length": 932.375,
"completions/mean_terminated_length": 982.255126953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 532.0,
"epoch": 0.1792,
"grad_norm": 0.03438134863972664,
"kl": 0.1192779541015625,
"learning_rate": 8.88888888888889e-07,
"loss": -0.1497,
"mask/has_final_conf_rate": 0.9140625,
"mask/share_final_conf": 0.014473002403974533,
"mask/share_reasoning": 0.8152675628662109,
"mask/share_step_conf": 0.11947821080684662,
"num_tokens": 55053620.0,
"reward": 0.8218560218811035,
"reward_std": 0.31332412362098694,
"rewards/accuracy_reward_step": 0.609375,
"rewards/asymmetric_l2_reward": 0.714383065700531,
"rewards/final_brier_reward_step": 0.6246414184570312,
"rewards/format_reward_step": 0.9140625,
"step": 168
},
{
"adv/mean_abs_final_conf": 0.5080597400665283,
"adv/mean_abs_reasoning": 0.4036991000175476,
"adv/mean_abs_step_conf": 0.7368338108062744,
"adv/ratio_final_to_reasoning": 1.2585109554230973,
"adv/ratio_step_to_reasoning": 1.8252054829308424,
"adv/std_final_conf": 0.759669303894043,
"adv/std_reasoning": 0.6817628741264343,
"adv/std_step_conf": 0.9267999529838562,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.6489621064363333,
"calib/avg_num_step_conf": 11.41015625,
"calib/ece": 0.3732653061224489,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00891961549178033,
"calib/mean_conf": 0.9773469387755102,
"calib/mu_c": 0.9808783783783783,
"calib/mu_w": 0.971958762886598,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3732653061224489,
"calib/std_conf": 0.014760055058257152,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8386280487804878,
"calib/step_q_c_n": 1640.0,
"calib/step_q_gap": 0.016090968374554926,
"calib/step_q_w": 0.8225370804059329,
"calib/step_q_w_n": 1281.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2865.0,
"completions/max_terminated_length": 2865.0,
"completions/mean_length": 910.59375,
"completions/mean_terminated_length": 936.1927490234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 516.0,
"epoch": 0.18026666666666666,
"grad_norm": 11.406038284301758,
"kl": 19.868133544921875,
"learning_rate": 8.611111111111112e-07,
"loss": 0.1358,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.015651900321245193,
"mask/share_reasoning": 0.8308191299438477,
"mask/share_step_conf": 0.12618522346019745,
"num_tokens": 55390916.0,
"reward": 0.8216732740402222,
"reward_std": 0.18951186537742615,
"rewards/accuracy_reward_step": 0.578125,
"rewards/asymmetric_l2_reward": 0.738727331161499,
"rewards/final_brier_reward_step": 0.5983691215515137,
"rewards/format_reward_step": 0.953125,
"step": 169
},
{
"adv/mean_abs_final_conf": 0.5418843626976013,
"adv/mean_abs_reasoning": 0.4897979497909546,
"adv/mean_abs_step_conf": 0.7558741569519043,
"adv/ratio_final_to_reasoning": 1.1063426519626658,
"adv/ratio_step_to_reasoning": 1.5432366698850226,
"adv/std_final_conf": 0.785415768623352,
"adv/std_reasoning": 0.7576139569282532,
"adv/std_step_conf": 0.9276391267776489,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.696646458445246,
"calib/avg_num_step_conf": 12.20703125,
"calib/ece": 0.3245,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.9958333333333333,
"calib/gap": 0.011222469495817733,
"calib/mean_conf": 0.9786666666666667,
"calib/mu_c": 0.9825477707006369,
"calib/mu_w": 0.9713253012048192,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3245,
"calib/std_conf": 0.015244306769705509,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.839802036199095,
"calib/step_q_c_n": 1768.0,
"calib/step_q_gap": 0.04934514600012663,
"calib/step_q_w": 0.7904568901989684,
"calib/step_q_w_n": 1357.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03515625,
"completions/max_length": 3046.0,
"completions/max_terminated_length": 3046.0,
"completions/mean_length": 904.96875,
"completions/mean_terminated_length": 937.943359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 545.0,
"epoch": 0.18133333333333335,
"grad_norm": 0.01993340626358986,
"kl": 0.12060546875,
"learning_rate": 8.333333333333333e-07,
"loss": -0.1043,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.015275785699486732,
"mask/share_reasoning": 0.8242030143737793,
"mask/share_step_conf": 0.12536495923995972,
"num_tokens": 55726740.0,
"reward": 0.8266474008560181,
"reward_std": 0.2241785079240799,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/asymmetric_l2_reward": 0.7176440954208374,
"rewards/final_brier_reward_step": 0.6270570158958435,
"rewards/format_reward_step": 0.9296875,
"step": 170
},
{
"adv/mean_abs_final_conf": 0.5801984071731567,
"adv/mean_abs_reasoning": 0.5012568831443787,
"adv/mean_abs_step_conf": 0.7429395914077759,
"adv/ratio_final_to_reasoning": 1.1574871621384604,
"adv/ratio_step_to_reasoning": 1.4821533955749882,
"adv/std_final_conf": 0.8193222880363464,
"adv/std_reasoning": 0.7755275368690491,
"adv/std_step_conf": 0.9323210120201111,
"calib/answer_extract_rate": 0.91796875,
"calib/auroc": 0.6916470933646506,
"calib/avg_num_step_conf": 12.890625,
"calib/ece": 0.422468085106383,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.010886670581327085,
"calib/mean_conf": 0.9799148936170213,
"calib/mu_c": 0.9847328244274809,
"calib/mu_w": 0.9738461538461538,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.422468085106383,
"calib/std_conf": 0.01402100216048866,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8441294196130753,
"calib/step_q_c_n": 1499.0,
"calib/step_q_gap": 0.035167731661936985,
"calib/step_q_w": 0.8089616879511383,
"calib/step_q_w_n": 1801.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0703125,
"completions/max_length": 2960.0,
"completions/max_terminated_length": 2960.0,
"completions/mean_length": 834.046875,
"completions/mean_terminated_length": 897.1260986328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 570.0,
"epoch": 0.1824,
"grad_norm": 0.019295768812298775,
"kl": 0.1244049072265625,
"learning_rate": 8.055555555555557e-07,
"loss": -0.2481,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.015632905066013336,
"mask/share_reasoning": 0.7890222668647766,
"mask/share_step_conf": 0.12503235042095184,
"num_tokens": 56047152.0,
"reward": 0.7329943180084229,
"reward_std": 0.2551029324531555,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/asymmetric_l2_reward": 0.6523202061653137,
"rewards/final_brier_reward_step": 0.5285121202468872,
"rewards/format_reward_step": 0.9140625,
"step": 171
},
{
"adv/mean_abs_final_conf": 0.5102229714393616,
"adv/mean_abs_reasoning": 0.43718355894088745,
"adv/mean_abs_step_conf": 0.7247519493103027,
"adv/ratio_final_to_reasoning": 1.1670680678738652,
"adv/ratio_step_to_reasoning": 1.6577749425574764,
"adv/std_final_conf": 0.7555689811706543,
"adv/std_reasoning": 0.7207115888595581,
"adv/std_step_conf": 0.9293190240859985,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.6016486060942854,
"calib/avg_num_step_conf": 12.12890625,
"calib/ece": 0.2252066115702479,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0061165138464384095,
"calib/mean_conf": 0.981404958677686,
"calib/mu_c": 0.9828961748633879,
"calib/mu_w": 0.9767796610169495,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2252066115702479,
"calib/std_conf": 0.01362458795337016,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8448167539267016,
"calib/step_q_c_n": 2101.0,
"calib/step_q_gap": 0.019657391376900812,
"calib/step_q_w": 0.8251593625498008,
"calib/step_q_w_n": 1004.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2751.0,
"completions/max_terminated_length": 2751.0,
"completions/mean_length": 829.60546875,
"completions/mean_terminated_length": 870.4057006835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 521.0,
"epoch": 0.18346666666666667,
"grad_norm": 0.018698470667004585,
"kl": 0.1277313232421875,
"learning_rate": 7.777777777777779e-07,
"loss": -0.1593,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.016454197466373444,
"mask/share_reasoning": 0.8024311065673828,
"mask/share_step_conf": 0.13423970341682434,
"num_tokens": 56362883.0,
"reward": 0.91644287109375,
"reward_std": 0.2156335860490799,
"rewards/accuracy_reward_step": 0.71484375,
"rewards/asymmetric_l2_reward": 0.776668906211853,
"rewards/final_brier_reward_step": 0.7249667644500732,
"rewards/format_reward_step": 0.94140625,
"step": 172
},
{
"adv/mean_abs_final_conf": 0.6531821489334106,
"adv/mean_abs_reasoning": 0.4964807629585266,
"adv/mean_abs_step_conf": 0.7733575701713562,
"adv/ratio_final_to_reasoning": 1.3156242853018134,
"adv/ratio_step_to_reasoning": 1.5576788223634728,
"adv/std_final_conf": 0.8502808213233948,
"adv/std_reasoning": 0.7576800584793091,
"adv/std_step_conf": 0.9313940405845642,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6137486645299145,
"calib/avg_num_step_conf": 12.765625,
"calib/ece": 0.4017741935483873,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.006816239316239203,
"calib/mean_conf": 0.9824193548387099,
"calib/mu_c": 0.9852777777777777,
"calib/mu_w": 0.9784615384615385,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4017741935483873,
"calib/std_conf": 0.013036608983064408,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8396873190503764,
"calib/step_q_c_n": 1727.0,
"calib/step_q_gap": 0.0056444897187734044,
"calib/step_q_w": 0.834042829331603,
"calib/step_q_w_n": 1541.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0234375,
"completions/max_length": 2124.0,
"completions/max_terminated_length": 2124.0,
"completions/mean_length": 893.078125,
"completions/mean_terminated_length": 914.5120239257812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 397.0,
"epoch": 0.18453333333333333,
"grad_norm": 0.017594095319509506,
"kl": 0.1277618408203125,
"learning_rate": 7.5e-07,
"loss": -0.152,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.015863744542002678,
"mask/share_reasoning": 0.8241561055183411,
"mask/share_step_conf": 0.13654263317584991,
"num_tokens": 56694671.0,
"reward": 0.800077497959137,
"reward_std": 0.2479168325662613,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.7143690586090088,
"rewards/final_brier_reward_step": 0.579535961151123,
"rewards/format_reward_step": 0.96875,
"step": 173
},
{
"adv/mean_abs_final_conf": 0.7132623195648193,
"adv/mean_abs_reasoning": 0.6211596727371216,
"adv/mean_abs_step_conf": 0.7582811117172241,
"adv/ratio_final_to_reasoning": 1.1482753161064854,
"adv/ratio_step_to_reasoning": 1.220750710322003,
"adv/std_final_conf": 0.8836770057678223,
"adv/std_reasoning": 0.8431268930435181,
"adv/std_step_conf": 0.9293094277381897,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.5396544245228455,
"calib/avg_num_step_conf": 12.86328125,
"calib/ece": 0.41666666666666663,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.002333718912666205,
"calib/mean_conf": 0.9778481012658228,
"calib/mu_c": 0.9788721804511278,
"calib/mu_w": 0.9765384615384616,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.41666666666666663,
"calib/std_conf": 0.01472712868075848,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8338524077548467,
"calib/step_q_c_n": 1599.0,
"calib/step_q_gap": 0.007913800907148882,
"calib/step_q_w": 0.8259386068476978,
"calib/step_q_w_n": 1694.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 3044.0,
"completions/max_terminated_length": 3044.0,
"completions/mean_length": 937.640625,
"completions/mean_terminated_length": 991.88427734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 579.0,
"epoch": 0.1856,
"grad_norm": 0.03041265159845352,
"kl": 0.1407012939453125,
"learning_rate": 7.222222222222222e-07,
"loss": -0.2731,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.014524579979479313,
"mask/share_reasoning": 0.8076785206794739,
"mask/share_step_conf": 0.12310938537120819,
"num_tokens": 57038939.0,
"reward": 0.7551385164260864,
"reward_std": 0.3065335750579834,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/asymmetric_l2_reward": 0.6824946403503418,
"rewards/final_brier_reward_step": 0.53793865442276,
"rewards/format_reward_step": 0.92578125,
"step": 174
},
{
"adv/mean_abs_final_conf": 0.6171407699584961,
"adv/mean_abs_reasoning": 0.5069169998168945,
"adv/mean_abs_step_conf": 0.7473276257514954,
"adv/ratio_final_to_reasoning": 1.217439482561082,
"adv/ratio_step_to_reasoning": 1.474260334574379,
"adv/std_final_conf": 0.8295073509216309,
"adv/std_reasoning": 0.7755605578422546,
"adv/std_step_conf": 0.929857075214386,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.6424601433377212,
"calib/avg_num_step_conf": 12.9609375,
"calib/ece": 0.5252340425531915,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008462776071376243,
"calib/mean_conf": 0.9762978723404255,
"calib/mu_c": 0.980943396226415,
"calib/mu_w": 0.9724806201550388,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.5252340425531915,
"calib/std_conf": 0.014915242658373733,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8433551020408164,
"calib/step_q_c_n": 1225.0,
"calib/step_q_gap": 0.01862027165381208,
"calib/step_q_w": 0.8247348303870043,
"calib/step_q_w_n": 2093.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2957.0,
"completions/max_terminated_length": 2957.0,
"completions/mean_length": 930.4140625,
"completions/mean_terminated_length": 988.32373046875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 501.0,
"epoch": 0.18666666666666668,
"grad_norm": 0.018979249522089958,
"kl": 0.115966796875,
"learning_rate": 6.944444444444446e-07,
"loss": -0.2135,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.014532214030623436,
"mask/share_reasoning": 0.8057027459144592,
"mask/share_step_conf": 0.12117130309343338,
"num_tokens": 57382949.0,
"reward": 0.6719663143157959,
"reward_std": 0.24395278096199036,
"rewards/accuracy_reward_step": 0.4140625,
"rewards/asymmetric_l2_reward": 0.63644939661026,
"rewards/final_brier_reward_step": 0.44107693433761597,
"rewards/format_reward_step": 0.91796875,
"step": 175
},
{
"adv/mean_abs_final_conf": 0.5018015503883362,
"adv/mean_abs_reasoning": 0.41027840971946716,
"adv/mean_abs_step_conf": 0.7303222417831421,
"adv/ratio_final_to_reasoning": 1.2230756932382796,
"adv/ratio_step_to_reasoning": 1.7800650106899576,
"adv/std_final_conf": 0.7439115047454834,
"adv/std_reasoning": 0.6816770434379578,
"adv/std_step_conf": 0.9248234033584595,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.629388908512893,
"calib/avg_num_step_conf": 11.5703125,
"calib/ece": 0.36909836065573776,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00769904627340201,
"calib/mean_conf": 0.9797540983606557,
"calib/mu_c": 0.9827516778523488,
"calib/mu_w": 0.9750526315789468,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36909836065573776,
"calib/std_conf": 0.014139997609043935,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.840911664779162,
"calib/step_q_c_n": 1766.0,
"calib/step_q_gap": 0.011555477488192079,
"calib/step_q_w": 0.8293561872909699,
"calib/step_q_w_n": 1196.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2610.0,
"completions/max_terminated_length": 2610.0,
"completions/mean_length": 879.80078125,
"completions/mean_terminated_length": 908.181396484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 429.0,
"epoch": 0.18773333333333334,
"grad_norm": 0.01840066723525524,
"kl": 0.121307373046875,
"learning_rate": 6.666666666666667e-07,
"loss": -0.1031,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.01617918536067009,
"mask/share_reasoning": 0.818038821220398,
"mask/share_step_conf": 0.13453197479248047,
"num_tokens": 57712242.0,
"reward": 0.829666256904602,
"reward_std": 0.19717296957969666,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/asymmetric_l2_reward": 0.7515543699264526,
"rewards/final_brier_reward_step": 0.5999656319618225,
"rewards/format_reward_step": 0.953125,
"step": 176
},
{
"adv/mean_abs_final_conf": 0.5797469019889832,
"adv/mean_abs_reasoning": 0.4305408000946045,
"adv/mean_abs_step_conf": 0.7243553400039673,
"adv/ratio_final_to_reasoning": 1.346555081101705,
"adv/ratio_step_to_reasoning": 1.6824313510933266,
"adv/std_final_conf": 0.8125232458114624,
"adv/std_reasoning": 0.7207408547401428,
"adv/std_step_conf": 0.9289398193359375,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.7121413782904467,
"calib/avg_num_step_conf": 13.609375,
"calib/ece": 0.3640585774058577,
"calib/final_conf_rate": 0.93359375,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.012709257616089986,
"calib/mean_conf": 0.9791213389121338,
"calib/mu_c": 0.9840136054421768,
"calib/mu_w": 0.9713043478260868,
"calib/nonempty_final_conf_rate": 0.93359375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3640585774058577,
"calib/std_conf": 0.014480614825990485,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8476618911174786,
"calib/step_q_c_n": 1745.0,
"calib/step_q_gap": 0.02403911940960035,
"calib/step_q_w": 0.8236227717078782,
"calib/step_q_w_n": 1739.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3055.0,
"completions/max_terminated_length": 3055.0,
"completions/mean_length": 909.4765625,
"completions/mean_terminated_length": 954.2048950195312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 329.0,
"epoch": 0.1888,
"grad_norm": 0.04756280779838562,
"kl": 0.1220703125,
"learning_rate": 6.388888888888889e-07,
"loss": -0.2128,
"mask/has_final_conf_rate": 0.93359375,
"mask/share_final_conf": 0.015417687594890594,
"mask/share_reasoning": 0.8047745227813721,
"mask/share_step_conf": 0.13293278217315674,
"num_tokens": 58048900.0,
"reward": 0.795788049697876,
"reward_std": 0.21256600320339203,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/asymmetric_l2_reward": 0.6957724690437317,
"rewards/final_brier_reward_step": 0.5942409634590149,
"rewards/format_reward_step": 0.93359375,
"step": 177
},
{
"adv/mean_abs_final_conf": 0.6708763837814331,
"adv/mean_abs_reasoning": 0.6096030473709106,
"adv/mean_abs_step_conf": 0.7516494989395142,
"adv/ratio_final_to_reasoning": 1.1005135008343239,
"adv/ratio_step_to_reasoning": 1.2330146677927873,
"adv/std_final_conf": 0.8560056090354919,
"adv/std_reasoning": 0.8266636729240417,
"adv/std_step_conf": 0.9325826168060303,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7006932654216185,
"calib/avg_num_step_conf": 11.84765625,
"calib/ece": 0.35873469387755097,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.011963780418789205,
"calib/mean_conf": 0.9791428571428571,
"calib/mu_c": 0.9836842105263158,
"calib/mu_w": 0.9717204301075266,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35873469387755097,
"calib/std_conf": 0.014444828182588057,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8468819729888433,
"calib/step_q_c_n": 1703.0,
"calib/step_q_gap": 0.027904529379820686,
"calib/step_q_w": 0.8189774436090226,
"calib/step_q_w_n": 1330.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2968.0,
"completions/max_terminated_length": 2968.0,
"completions/mean_length": 843.47265625,
"completions/mean_terminated_length": 877.7601318359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 503.0,
"epoch": 0.18986666666666666,
"grad_norm": 0.02961582876741886,
"kl": 0.1240234375,
"learning_rate": 6.111111111111112e-07,
"loss": -0.1428,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.016271937638521194,
"mask/share_reasoning": 0.8166067600250244,
"mask/share_step_conf": 0.12805885076522827,
"num_tokens": 58370901.0,
"reward": 0.8209068775177002,
"reward_std": 0.28923019766807556,
"rewards/accuracy_reward_step": 0.59375,
"rewards/asymmetric_l2_reward": 0.7179765701293945,
"rewards/final_brier_reward_step": 0.6136808395385742,
"rewards/format_reward_step": 0.95703125,
"step": 178
},
{
"adv/mean_abs_final_conf": 0.6773865222930908,
"adv/mean_abs_reasoning": 0.6010725498199463,
"adv/mean_abs_step_conf": 0.7447128891944885,
"adv/ratio_final_to_reasoning": 1.1269629972222233,
"adv/ratio_step_to_reasoning": 1.238973380863209,
"adv/std_final_conf": 0.8693088889122009,
"adv/std_reasoning": 0.8430590629577637,
"adv/std_step_conf": 0.9291117787361145,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.6356553266990665,
"calib/avg_num_step_conf": 12.76953125,
"calib/ece": 0.3469915254237288,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0081238910745971,
"calib/mean_conf": 0.9783474576271186,
"calib/mu_c": 0.9813422818791945,
"calib/mu_w": 0.9732183908045974,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3469915254237288,
"calib/std_conf": 0.014592706048143606,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8434924199887703,
"calib/step_q_c_n": 1781.0,
"calib/step_q_gap": 0.018707473752211046,
"calib/step_q_w": 0.8247849462365593,
"calib/step_q_w_n": 1488.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2946.0,
"completions/max_terminated_length": 2946.0,
"completions/mean_length": 896.1953125,
"completions/mean_terminated_length": 951.9751586914062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 502.0,
"epoch": 0.19093333333333334,
"grad_norm": 0.02130270004272461,
"kl": 0.1211700439453125,
"learning_rate": 5.833333333333334e-07,
"loss": -0.2417,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.015163568779826164,
"mask/share_reasoning": 0.7996172904968262,
"mask/share_step_conf": 0.1266254037618637,
"num_tokens": 58706591.0,
"reward": 0.8096022605895996,
"reward_std": 0.2978968620300293,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/asymmetric_l2_reward": 0.7180361747741699,
"rewards/final_brier_reward_step": 0.5996058583259583,
"rewards/format_reward_step": 0.921875,
"step": 179
},
{
"adv/mean_abs_final_conf": 0.5014601349830627,
"adv/mean_abs_reasoning": 0.4010167121887207,
"adv/mean_abs_step_conf": 0.7332862615585327,
"adv/ratio_final_to_reasoning": 1.2504719123702575,
"adv/ratio_step_to_reasoning": 1.828567835879728,
"adv/std_final_conf": 0.7621612548828125,
"adv/std_reasoning": 0.701465368270874,
"adv/std_step_conf": 0.9304397702217102,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.5817487755690004,
"calib/avg_num_step_conf": 13.08984375,
"calib/ece": 0.34236734693877546,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.004942379717660539,
"calib/mean_conf": 0.9791020408163265,
"calib/mu_c": 0.9808974358974358,
"calib/mu_w": 0.9759550561797753,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34236734693877546,
"calib/std_conf": 0.014314606352210409,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.832957992998833,
"calib/step_q_c_n": 1714.0,
"calib/step_q_gap": -0.008196558009108568,
"calib/step_q_w": 0.8411545510079416,
"calib/step_q_w_n": 1637.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3020.0,
"completions/max_terminated_length": 3020.0,
"completions/mean_length": 940.8671875,
"completions/mean_terminated_length": 971.2177124023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 523.0,
"epoch": 0.192,
"grad_norm": 0.014338175766170025,
"kl": 0.1213226318359375,
"learning_rate": 5.555555555555555e-07,
"loss": -0.1398,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.015053339302539825,
"mask/share_reasoning": 0.8310614228248596,
"mask/share_step_conf": 0.12263523042201996,
"num_tokens": 59051309.0,
"reward": 0.8462561368942261,
"reward_std": 0.19401851296424866,
"rewards/accuracy_reward_step": 0.609375,
"rewards/asymmetric_l2_reward": 0.7546102404594421,
"rewards/final_brier_reward_step": 0.6254019737243652,
"rewards/format_reward_step": 0.953125,
"step": 180
},
{
"adv/mean_abs_final_conf": 0.693315863609314,
"adv/mean_abs_reasoning": 0.5807839632034302,
"adv/mean_abs_step_conf": 0.7269697189331055,
"adv/ratio_final_to_reasoning": 1.1937586220273568,
"adv/ratio_step_to_reasoning": 1.2517041877729518,
"adv/std_final_conf": 0.9083855152130127,
"adv/std_reasoning": 0.8429928421974182,
"adv/std_step_conf": 0.9324417114257812,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6643928571428572,
"calib/avg_num_step_conf": 11.55078125,
"calib/ece": 0.39537500000000003,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009785714285714286,
"calib/mean_conf": 0.9787083333333334,
"calib/mu_c": 0.9827857142857142,
"calib/mu_w": 0.9729999999999999,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.39537500000000003,
"calib/std_conf": 0.014476933280989542,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8407907845579078,
"calib/step_q_c_n": 1606.0,
"calib/step_q_gap": 0.017674574343251814,
"calib/step_q_w": 0.823116210214656,
"calib/step_q_w_n": 1351.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2875.0,
"completions/max_terminated_length": 2875.0,
"completions/mean_length": 864.91796875,
"completions/mean_terminated_length": 911.1892700195312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 553.0,
"epoch": 0.19306666666666666,
"grad_norm": 0.031241774559020996,
"kl": 0.1353912353515625,
"learning_rate": 5.277777777777779e-07,
"loss": -0.1777,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.01560254767537117,
"mask/share_reasoning": 0.8068452477455139,
"mask/share_step_conf": 0.12677091360092163,
"num_tokens": 59378992.0,
"reward": 0.7764517664909363,
"reward_std": 0.28926050662994385,
"rewards/accuracy_reward_step": 0.546875,
"rewards/asymmetric_l2_reward": 0.6886812448501587,
"rewards/final_brier_reward_step": 0.5673472881317139,
"rewards/format_reward_step": 0.9375,
"step": 181
},
{
"adv/mean_abs_final_conf": 0.6113361120223999,
"adv/mean_abs_reasoning": 0.4344509243965149,
"adv/mean_abs_step_conf": 0.7469823360443115,
"adv/ratio_final_to_reasoning": 1.407146532998156,
"adv/ratio_step_to_reasoning": 1.7193710361692205,
"adv/std_final_conf": 0.8014238476753235,
"adv/std_reasoning": 0.7014479041099548,
"adv/std_step_conf": 0.9269827604293823,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.6751508295625943,
"calib/avg_num_step_conf": 11.79296875,
"calib/ece": 0.33224066390041485,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.010199849170437703,
"calib/mean_conf": 0.9795435684647302,
"calib/mu_c": 0.9831410256410257,
"calib/mu_w": 0.972941176470588,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33224066390041485,
"calib/std_conf": 0.014149438434543887,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8405216881594373,
"calib/step_q_c_n": 1706.0,
"calib/step_q_gap": -0.0003389363645535104,
"calib/step_q_w": 0.8408606245239908,
"calib/step_q_w_n": 1313.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2585.0,
"completions/max_terminated_length": 2585.0,
"completions/mean_length": 863.76953125,
"completions/mean_terminated_length": 906.2499389648438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 561.0,
"epoch": 0.19413333333333332,
"grad_norm": 0.022411292418837547,
"kl": 0.121917724609375,
"learning_rate": 5.000000000000001e-07,
"loss": -0.2028,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.01571817323565483,
"mask/share_reasoning": 0.8114913105964661,
"mask/share_step_conf": 0.1259155571460724,
"num_tokens": 59706277.0,
"reward": 0.8176075220108032,
"reward_std": 0.19320210814476013,
"rewards/accuracy_reward_step": 0.609375,
"rewards/asymmetric_l2_reward": 0.6993848085403442,
"rewards/final_brier_reward_step": 0.6264550685882568,
"rewards/format_reward_step": 0.9375,
"step": 182
},
{
"adv/mean_abs_final_conf": 0.6075547337532043,
"adv/mean_abs_reasoning": 0.5296026468276978,
"adv/mean_abs_step_conf": 0.7353948354721069,
"adv/ratio_final_to_reasoning": 1.147189760837558,
"adv/ratio_step_to_reasoning": 1.3885784745924092,
"adv/std_final_conf": 0.82039874792099,
"adv/std_reasoning": 0.7929397225379944,
"adv/std_step_conf": 0.932217538356781,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.7287499999999999,
"calib/avg_num_step_conf": 12.2890625,
"calib/ece": 0.39350000000000007,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.013600000000000168,
"calib/mean_conf": 0.9768333333333334,
"calib/mu_c": 0.9825,
"calib/mu_w": 0.9688999999999999,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.39350000000000007,
"calib/std_conf": 0.014887541398393801,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8456774193548389,
"calib/step_q_c_n": 1550.0,
"calib/step_q_gap": 0.03122879780095411,
"calib/step_q_w": 0.8144486215538848,
"calib/step_q_w_n": 1596.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 3001.0,
"completions/max_terminated_length": 3001.0,
"completions/mean_length": 941.71484375,
"completions/mean_terminated_length": 988.0286254882812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 546.0,
"epoch": 0.1952,
"grad_norm": 0.041146308183670044,
"kl": 0.1220550537109375,
"learning_rate": 4.7222222222222226e-07,
"loss": -0.2376,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.014763625338673592,
"mask/share_reasoning": 0.8222149610519409,
"mask/share_step_conf": 0.11614643037319183,
"num_tokens": 60054036.0,
"reward": 0.7858456373214722,
"reward_std": 0.25782373547554016,
"rewards/accuracy_reward_step": 0.546875,
"rewards/asymmetric_l2_reward": 0.7043553590774536,
"rewards/final_brier_reward_step": 0.5704609155654907,
"rewards/format_reward_step": 0.9375,
"step": 183
},
{
"adv/mean_abs_final_conf": 0.6665743589401245,
"adv/mean_abs_reasoning": 0.5123068690299988,
"adv/mean_abs_step_conf": 0.7528473138809204,
"adv/ratio_final_to_reasoning": 1.3011232119573481,
"adv/ratio_step_to_reasoning": 1.4695241453747452,
"adv/std_final_conf": 0.8536959290504456,
"adv/std_reasoning": 0.7754993438720703,
"adv/std_step_conf": 0.9273887276649475,
"calib/answer_extract_rate": 0.94140625,
"calib/auroc": 0.5960636952154178,
"calib/avg_num_step_conf": 11.93359375,
"calib/ece": 0.3506250000000002,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005740754520425551,
"calib/mean_conf": 0.9797916666666668,
"calib/mu_c": 0.9819205298013244,
"calib/mu_w": 0.9761797752808988,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3506250000000002,
"calib/std_conf": 0.014214075086179752,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8426749226006193,
"calib/step_q_c_n": 1615.0,
"calib/step_q_gap": 0.015758255933952636,
"calib/step_q_w": 0.8269166666666666,
"calib/step_q_w_n": 1440.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2817.0,
"completions/max_terminated_length": 2817.0,
"completions/mean_length": 904.36328125,
"completions/mean_terminated_length": 944.96728515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 442.0,
"epoch": 0.19626666666666667,
"grad_norm": 0.03529198095202446,
"kl": 0.120391845703125,
"learning_rate": 4.444444444444445e-07,
"loss": -0.1466,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.015524040907621384,
"mask/share_reasoning": 0.8197634816169739,
"mask/share_step_conf": 0.12174372375011444,
"num_tokens": 60390833.0,
"reward": 0.8256651759147644,
"reward_std": 0.25622591376304626,
"rewards/accuracy_reward_step": 0.59375,
"rewards/asymmetric_l2_reward": 0.7392463088035583,
"rewards/final_brier_reward_step": 0.6058340072631836,
"rewards/format_reward_step": 0.9375,
"step": 184
},
{
"adv/mean_abs_final_conf": 0.5692657232284546,
"adv/mean_abs_reasoning": 0.5020483136177063,
"adv/mean_abs_step_conf": 0.7326884865760803,
"adv/ratio_final_to_reasoning": 1.1338863368076806,
"adv/ratio_step_to_reasoning": 1.4593983620747686,
"adv/std_final_conf": 0.798140823841095,
"adv/std_reasoning": 0.7755592465400696,
"adv/std_step_conf": 0.9297898411750793,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.7315218222120309,
"calib/avg_num_step_conf": 12.76953125,
"calib/ece": 0.35533898305084755,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.013738439195903118,
"calib/mean_conf": 0.9782203389830509,
"calib/mu_c": 0.9834013605442177,
"calib/mu_w": 0.9696629213483146,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35533898305084755,
"calib/std_conf": 0.014592275407900104,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8490630524454921,
"calib/step_q_c_n": 1697.0,
"calib/step_q_gap": 0.049800965931497077,
"calib/step_q_w": 0.799262086513995,
"calib/step_q_w_n": 1572.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2965.0,
"completions/max_terminated_length": 2965.0,
"completions/mean_length": 917.31640625,
"completions/mean_terminated_length": 970.38427734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 508.0,
"epoch": 0.19733333333333333,
"grad_norm": 0.01986088417470455,
"kl": 0.1129608154296875,
"learning_rate": 4.1666666666666667e-07,
"loss": -0.1787,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.014929613098502159,
"mask/share_reasoning": 0.8069747686386108,
"mask/share_step_conf": 0.12340810149908066,
"num_tokens": 60732586.0,
"reward": 0.7960665225982666,
"reward_std": 0.26516661047935486,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/asymmetric_l2_reward": 0.698235273361206,
"rewards/final_brier_reward_step": 0.5946788787841797,
"rewards/format_reward_step": 0.921875,
"step": 185
},
{
"adv/mean_abs_final_conf": 0.617626965045929,
"adv/mean_abs_reasoning": 0.5716798305511475,
"adv/mean_abs_step_conf": 0.7495993375778198,
"adv/ratio_final_to_reasoning": 1.080372145455061,
"adv/ratio_step_to_reasoning": 1.3112222917767502,
"adv/std_final_conf": 0.8382736444473267,
"adv/std_reasoning": 0.8267985582351685,
"adv/std_step_conf": 0.9307035207748413,
"calib/answer_extract_rate": 0.9140625,
"calib/auroc": 0.7566045066045066,
"calib/avg_num_step_conf": 13.01171875,
"calib/ece": 0.3665665236051502,
"calib/final_conf_rate": 0.91015625,
"calib/format_rate": 0.90625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.015153846153846184,
"calib/mean_conf": 0.9803004291845493,
"calib/mu_c": 0.986153846153846,
"calib/mu_w": 0.9709999999999999,
"calib/nonempty_final_conf_rate": 0.91015625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3665665236051502,
"calib/std_conf": 0.013940241861726809,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8402585722315907,
"calib/step_q_c_n": 1779.0,
"calib/step_q_gap": 0.023061407283136992,
"calib/step_q_w": 0.8171971649484537,
"calib/step_q_w_n": 1552.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.07421875,
"completions/max_length": 2790.0,
"completions/max_terminated_length": 2790.0,
"completions/mean_length": 869.046875,
"completions/mean_terminated_length": 938.7172241210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 522.0,
"epoch": 0.1984,
"grad_norm": 0.05723670870065689,
"kl": 0.1216888427734375,
"learning_rate": 3.8888888888888895e-07,
"loss": -0.27,
"mask/has_final_conf_rate": 0.91015625,
"mask/share_final_conf": 0.01483369804918766,
"mask/share_reasoning": 0.7815354466438293,
"mask/share_step_conf": 0.12941214442253113,
"num_tokens": 61060102.0,
"reward": 0.7855908870697021,
"reward_std": 0.3000542223453522,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.702877402305603,
"rewards/final_brier_reward_step": 0.5745543241500854,
"rewards/format_reward_step": 0.90625,
"step": 186
},
{
"adv/mean_abs_final_conf": 0.6157701015472412,
"adv/mean_abs_reasoning": 0.5407514572143555,
"adv/mean_abs_step_conf": 0.7796497941017151,
"adv/ratio_final_to_reasoning": 1.138730360005573,
"adv/ratio_step_to_reasoning": 1.4417895387985236,
"adv/std_final_conf": 0.8065946102142334,
"adv/std_reasoning": 0.7578535676002502,
"adv/std_step_conf": 0.9338079690933228,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.6219548872180451,
"calib/avg_num_step_conf": 11.609375,
"calib/ece": 0.38578723404255333,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007165413533834819,
"calib/mean_conf": 0.9815319148936171,
"calib/mu_c": 0.9844285714285714,
"calib/mu_w": 0.9772631578947366,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.38578723404255333,
"calib/std_conf": 0.013471563531853206,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8369925280199253,
"calib/step_q_c_n": 1606.0,
"calib/step_q_gap": 0.013112586585078967,
"calib/step_q_w": 0.8238799414348463,
"calib/step_q_w_n": 1366.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 3036.0,
"completions/max_terminated_length": 3036.0,
"completions/mean_length": 948.04296875,
"completions/mean_terminated_length": 990.6080932617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 361.0,
"epoch": 0.19946666666666665,
"grad_norm": 0.036436568945646286,
"kl": 0.1116943359375,
"learning_rate": 3.611111111111111e-07,
"loss": -0.0918,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.014702420681715012,
"mask/share_reasoning": 0.8228905200958252,
"mask/share_step_conf": 0.11943836510181427,
"num_tokens": 61404345.0,
"reward": 0.7795661091804504,
"reward_std": 0.2637999355792999,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/asymmetric_l2_reward": 0.703906774520874,
"rewards/final_brier_reward_step": 0.5630378723144531,
"rewards/format_reward_step": 0.91015625,
"step": 187
},
{
"adv/mean_abs_final_conf": 0.5346688628196716,
"adv/mean_abs_reasoning": 0.4271719455718994,
"adv/mean_abs_step_conf": 0.7676730155944824,
"adv/ratio_final_to_reasoning": 1.2516478864356482,
"adv/ratio_step_to_reasoning": 1.7971054128255517,
"adv/std_final_conf": 0.7731443643569946,
"adv/std_reasoning": 0.7208114266395569,
"adv/std_step_conf": 0.9298895001411438,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6706921413121846,
"calib/avg_num_step_conf": 11.53125,
"calib/ece": 0.374813278008299,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.010062725306417053,
"calib/mean_conf": 0.9806224066390044,
"calib/mu_c": 0.9845890410958904,
"calib/mu_w": 0.9745263157894734,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.374813278008299,
"calib/std_conf": 0.013756422717855155,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8384236760124611,
"calib/step_q_c_n": 1605.0,
"calib/step_q_gap": 0.019559533473485602,
"calib/step_q_w": 0.8188641425389755,
"calib/step_q_w_n": 1347.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2856.0,
"completions/max_terminated_length": 2856.0,
"completions/mean_length": 942.8203125,
"completions/mean_terminated_length": 985.1510009765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 474.0,
"epoch": 0.20053333333333334,
"grad_norm": 0.023577481508255005,
"kl": 0.1107635498046875,
"learning_rate": 3.3333333333333335e-07,
"loss": -0.1837,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.015136613510549068,
"mask/share_reasoning": 0.8198684453964233,
"mask/share_step_conf": 0.12202617526054382,
"num_tokens": 61749779.0,
"reward": 0.7894806861877441,
"reward_std": 0.23784969747066498,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/asymmetric_l2_reward": 0.6890180110931396,
"rewards/final_brier_reward_step": 0.5883808135986328,
"rewards/format_reward_step": 0.9375,
"step": 188
},
{
"adv/mean_abs_final_conf": 0.6499653458595276,
"adv/mean_abs_reasoning": 0.5608536601066589,
"adv/mean_abs_step_conf": 0.753533124923706,
"adv/ratio_final_to_reasoning": 1.158885805855171,
"adv/ratio_step_to_reasoning": 1.3435467725759422,
"adv/std_final_conf": 0.8551369905471802,
"adv/std_reasoning": 0.8101358413696289,
"adv/std_step_conf": 0.9269534945487976,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.6467896962706651,
"calib/avg_num_step_conf": 12.296875,
"calib/ece": 0.3350840336134453,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.008692810457516909,
"calib/mean_conf": 0.9779411764705882,
"calib/mu_c": 0.9810457516339871,
"calib/mu_w": 0.9723529411764702,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.3350840336134453,
"calib/std_conf": 0.014594113074491486,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.8391402714932126,
"calib/step_q_c_n": 1768.0,
"calib/step_q_gap": 0.018480851203357518,
"calib/step_q_w": 0.8206594202898551,
"calib/step_q_w_n": 1380.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2500.0,
"completions/max_terminated_length": 2500.0,
"completions/mean_length": 857.04296875,
"completions/mean_terminated_length": 906.6239013671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 552.0,
"epoch": 0.2016,
"grad_norm": 0.025642195716500282,
"kl": 0.130401611328125,
"learning_rate": 3.055555555555556e-07,
"loss": -0.2976,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.01532869040966034,
"mask/share_reasoning": 0.8045191764831543,
"mask/share_step_conf": 0.12546461820602417,
"num_tokens": 62076950.0,
"reward": 0.8125067949295044,
"reward_std": 0.29982349276542664,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/asymmetric_l2_reward": 0.7050384879112244,
"rewards/final_brier_reward_step": 0.6152874827384949,
"rewards/format_reward_step": 0.92578125,
"step": 189
},
{
"adv/mean_abs_final_conf": 0.6154840588569641,
"adv/mean_abs_reasoning": 0.5056318044662476,
"adv/mean_abs_step_conf": 0.741155743598938,
"adv/ratio_final_to_reasoning": 1.217257406318968,
"adv/ratio_step_to_reasoning": 1.4658012748650433,
"adv/std_final_conf": 0.830213725566864,
"adv/std_reasoning": 0.7755312323570251,
"adv/std_step_conf": 0.9308255314826965,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.6579521463757916,
"calib/avg_num_step_conf": 12.359375,
"calib/ece": 0.37893004115226336,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.009448275862068867,
"calib/mean_conf": 0.9756378600823045,
"calib/mu_c": 0.9794482758620688,
"calib/mu_w": 0.97,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.37893004115226336,
"calib/std_conf": 0.014958946796078698,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8391810620601406,
"calib/step_q_c_n": 1563.0,
"calib/step_q_gap": 0.0027101064074235293,
"calib/step_q_w": 0.8364709556527171,
"calib/step_q_w_n": 1601.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2540.0,
"completions/max_terminated_length": 2540.0,
"completions/mean_length": 917.296875,
"completions/mean_terminated_length": 958.4815673828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 524.0,
"epoch": 0.20266666666666666,
"grad_norm": 0.022940604016184807,
"kl": 0.11981201171875,
"learning_rate": 2.7777777777777776e-07,
"loss": -0.1979,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01512821763753891,
"mask/share_reasoning": 0.8211517333984375,
"mask/share_step_conf": 0.12075131386518478,
"num_tokens": 62417386.0,
"reward": 0.8009210824966431,
"reward_std": 0.24404773116111755,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/asymmetric_l2_reward": 0.7101178169250488,
"rewards/final_brier_reward_step": 0.5885992050170898,
"rewards/format_reward_step": 0.94921875,
"step": 190
},
{
"adv/mean_abs_final_conf": 0.6485990881919861,
"adv/mean_abs_reasoning": 0.517936646938324,
"adv/mean_abs_step_conf": 0.7152957320213318,
"adv/ratio_final_to_reasoning": 1.2522749491198322,
"adv/ratio_step_to_reasoning": 1.3810486982329895,
"adv/std_final_conf": 0.8772525191307068,
"adv/std_reasoning": 0.7929953336715698,
"adv/std_step_conf": 0.9332661628723145,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6184525847312564,
"calib/avg_num_step_conf": 12.9140625,
"calib/ece": 0.45429752066115714,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.007215337213282713,
"calib/mean_conf": 0.9790909090909092,
"calib/mu_c": 0.9825196850393698,
"calib/mu_w": 0.9753043478260871,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.45429752066115714,
"calib/std_conf": 0.014431370787625051,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8464498269896195,
"calib/step_q_c_n": 1445.0,
"calib/step_q_gap": 0.010404689966513558,
"calib/step_q_w": 0.8360451370231059,
"calib/step_q_w_n": 1861.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2616.0,
"completions/max_terminated_length": 2616.0,
"completions/mean_length": 893.30078125,
"completions/mean_terminated_length": 941.0905151367188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 548.0,
"epoch": 0.20373333333333332,
"grad_norm": 0.017885025590658188,
"kl": 0.120147705078125,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.2514,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.01538168266415596,
"mask/share_reasoning": 0.8069976568222046,
"mask/share_step_conf": 0.12683944404125214,
"num_tokens": 62750239.0,
"reward": 0.7270830869674683,
"reward_std": 0.26644009351730347,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/asymmetric_l2_reward": 0.6501602530479431,
"rewards/final_brier_reward_step": 0.5172871351242065,
"rewards/format_reward_step": 0.9375,
"step": 191
},
{
"adv/mean_abs_final_conf": 0.6051337718963623,
"adv/mean_abs_reasoning": 0.5304586887359619,
"adv/mean_abs_step_conf": 0.7536708116531372,
"adv/ratio_final_to_reasoning": 1.1407745499246036,
"adv/ratio_step_to_reasoning": 1.420790775336475,
"adv/std_final_conf": 0.8207889199256897,
"adv/std_reasoning": 0.7756054997444153,
"adv/std_step_conf": 0.9285756945610046,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.7128837072018891,
"calib/avg_num_step_conf": 12.7109375,
"calib/ece": 0.34314049586776874,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.012792207792207688,
"calib/mean_conf": 0.9795041322314051,
"calib/mu_c": 0.9841558441558441,
"calib/mu_w": 0.9713636363636364,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34314049586776874,
"calib/std_conf": 0.014307787838866616,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8471263736263737,
"calib/step_q_c_n": 1820.0,
"calib/step_q_gap": 0.02227281714101792,
"calib/step_q_w": 0.8248535564853557,
"calib/step_q_w_n": 1434.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.046875,
"completions/max_length": 2702.0,
"completions/max_terminated_length": 2702.0,
"completions/mean_length": 880.55859375,
"completions/mean_terminated_length": 923.8646850585938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 474.0,
"epoch": 0.2048,
"grad_norm": 0.031466081738471985,
"kl": 0.1272125244140625,
"learning_rate": 2.2222222222222224e-07,
"loss": -0.2225,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.015780529007315636,
"mask/share_reasoning": 0.806641161441803,
"mask/share_step_conf": 0.13070333003997803,
"num_tokens": 63080638.0,
"reward": 0.8223682641983032,
"reward_std": 0.27081209421157837,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/asymmetric_l2_reward": 0.7147022485733032,
"rewards/final_brier_reward_step": 0.620659351348877,
"rewards/format_reward_step": 0.9453125,
"step": 192
},
{
"adv/mean_abs_final_conf": 0.6623778343200684,
"adv/mean_abs_reasoning": 0.6136531233787537,
"adv/mean_abs_step_conf": 0.7547324895858765,
"adv/ratio_final_to_reasoning": 1.0794010640295255,
"adv/ratio_step_to_reasoning": 1.2299008362091348,
"adv/std_final_conf": 0.8701410889625549,
"adv/std_reasoning": 0.843064546585083,
"adv/std_step_conf": 0.9320592880249023,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.6003455608718766,
"calib/avg_num_step_conf": 11.41015625,
"calib/ece": 0.44345528455284544,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005797448165869135,
"calib/mean_conf": 0.980040650406504,
"calib/mu_c": 0.9827272727272727,
"calib/mu_w": 0.9769298245614035,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44345528455284544,
"calib/std_conf": 0.014070032950183454,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8425330549756439,
"calib/step_q_c_n": 1437.0,
"calib/step_q_gap": 0.029487232873218083,
"calib/step_q_w": 0.8130458221024258,
"calib/step_q_w_n": 1484.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2666.0,
"completions/max_terminated_length": 2666.0,
"completions/mean_length": 908.3984375,
"completions/mean_terminated_length": 933.9357299804688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 545.0,
"epoch": 0.20586666666666667,
"grad_norm": 0.01949903927743435,
"kl": 0.1169586181640625,
"learning_rate": 1.9444444444444447e-07,
"loss": -0.0695,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.015710387378931046,
"mask/share_reasoning": 0.8357419371604919,
"mask/share_step_conf": 0.12120389193296432,
"num_tokens": 63418900.0,
"reward": 0.7550156116485596,
"reward_std": 0.296281099319458,
"rewards/accuracy_reward_step": 0.515625,
"rewards/asymmetric_l2_reward": 0.6791197657585144,
"rewards/final_brier_reward_step": 0.5355988144874573,
"rewards/format_reward_step": 0.9609375,
"step": 193
},
{
"adv/mean_abs_final_conf": 0.5542274713516235,
"adv/mean_abs_reasoning": 0.5262541770935059,
"adv/mean_abs_step_conf": 0.736404299736023,
"adv/ratio_final_to_reasoning": 1.0531554816583382,
"adv/ratio_step_to_reasoning": 1.3993319802289705,
"adv/std_final_conf": 0.7933449149131775,
"adv/std_reasoning": 0.7755599021911621,
"adv/std_step_conf": 0.9261869788169861,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.6697740732616801,
"calib/avg_num_step_conf": 12.56640625,
"calib/ece": 0.3888235294117647,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.9957983193277311,
"calib/gap": 0.009396797543321012,
"calib/mean_conf": 0.9799999999999999,
"calib/mu_c": 0.9838297872340426,
"calib/mu_w": 0.9744329896907216,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.38819327731092435,
"calib/std_conf": 0.016218051981789737,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8422549647661757,
"calib/step_q_c_n": 1561.0,
"calib/step_q_gap": 0.04467646235071665,
"calib/step_q_w": 0.797578502415459,
"calib/step_q_w_n": 1656.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05859375,
"completions/max_length": 2677.0,
"completions/max_terminated_length": 2677.0,
"completions/mean_length": 857.75390625,
"completions/mean_terminated_length": 911.14111328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 24.0,
"epoch": 0.20693333333333333,
"grad_norm": 0.023266388103365898,
"kl": 0.12542724609375,
"learning_rate": 1.6666666666666668e-07,
"loss": -0.1916,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.01550736278295517,
"mask/share_reasoning": 0.8019325733184814,
"mask/share_step_conf": 0.12396633625030518,
"num_tokens": 63744429.0,
"reward": 0.7903183698654175,
"reward_std": 0.2568064332008362,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/asymmetric_l2_reward": 0.7150031328201294,
"rewards/final_brier_reward_step": 0.5695398449897766,
"rewards/format_reward_step": 0.9296875,
"step": 194
},
{
"adv/mean_abs_final_conf": 0.6420087218284607,
"adv/mean_abs_reasoning": 0.5460288524627686,
"adv/mean_abs_step_conf": 0.754043459892273,
"adv/ratio_final_to_reasoning": 1.1757780178332913,
"adv/ratio_step_to_reasoning": 1.3809590033407402,
"adv/std_final_conf": 0.8318362236022949,
"adv/std_reasoning": 0.7929378747940063,
"adv/std_step_conf": 0.9332217574119568,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.6601991758241758,
"calib/avg_num_step_conf": 11.66796875,
"calib/ece": 0.40479508196721314,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00973351648351617,
"calib/mean_conf": 0.9785655737704918,
"calib/mu_c": 0.9827142857142857,
"calib/mu_w": 0.9729807692307695,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.40479508196721314,
"calib/std_conf": 0.014654204177662597,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8443458838544992,
"calib/step_q_c_n": 1567.0,
"calib/step_q_gap": 0.0173599683615413,
"calib/step_q_w": 0.8269859154929579,
"calib/step_q_w_n": 1420.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.03125,
"completions/max_length": 2817.0,
"completions/max_terminated_length": 2817.0,
"completions/mean_length": 917.6171875,
"completions/mean_terminated_length": 947.2177124023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 466.0,
"epoch": 0.208,
"grad_norm": 0.03138196840882301,
"kl": 0.125946044921875,
"learning_rate": 1.3888888888888888e-07,
"loss": -0.0945,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.015517974272370338,
"mask/share_reasoning": 0.826676607131958,
"mask/share_step_conf": 0.1265554130077362,
"num_tokens": 64085323.0,
"reward": 0.7782267332077026,
"reward_std": 0.27050119638442993,
"rewards/accuracy_reward_step": 0.546875,
"rewards/asymmetric_l2_reward": 0.6882679462432861,
"rewards/final_brier_reward_step": 0.5681855082511902,
"rewards/format_reward_step": 0.953125,
"step": 195
},
{
"adv/mean_abs_final_conf": 0.5235567688941956,
"adv/mean_abs_reasoning": 0.41682004928588867,
"adv/mean_abs_step_conf": 0.74735426902771,
"adv/ratio_final_to_reasoning": 1.2560738615888851,
"adv/ratio_step_to_reasoning": 1.7929902131821744,
"adv/std_final_conf": 0.7714676260948181,
"adv/std_reasoning": 0.7014705538749695,
"adv/std_step_conf": 0.9283434748649597,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.5894736842105264,
"calib/avg_num_step_conf": 12.375,
"calib/ece": 0.37510288065843633,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.005441678520626048,
"calib/mean_conf": 0.9841563786008232,
"calib/mu_c": 0.9862837837837836,
"calib/mu_w": 0.9808421052631575,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.37510288065843633,
"calib/std_conf": 0.01184642079494479,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8543995243757431,
"calib/step_q_c_n": 1682.0,
"calib/step_q_gap": 0.0170307491402113,
"calib/step_q_w": 0.8373687752355318,
"calib/step_q_w_n": 1486.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3026.0,
"completions/max_terminated_length": 3026.0,
"completions/mean_length": 789.8828125,
"completions/mean_terminated_length": 828.7294921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 508.0,
"epoch": 0.20906666666666668,
"grad_norm": 0.021556729450821877,
"kl": 0.1281890869140625,
"learning_rate": 1.1111111111111112e-07,
"loss": -0.1975,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.01695971004664898,
"mask/share_reasoning": 0.7979605197906494,
"mask/share_step_conf": 0.13820475339889526,
"num_tokens": 64390077.0,
"reward": 0.7968078255653381,
"reward_std": 0.21186324954032898,
"rewards/accuracy_reward_step": 0.578125,
"rewards/asymmetric_l2_reward": 0.6961746215820312,
"rewards/final_brier_reward_step": 0.5919722318649292,
"rewards/format_reward_step": 0.94921875,
"step": 196
},
{
"adv/mean_abs_final_conf": 0.6259146928787231,
"adv/mean_abs_reasoning": 0.527590811252594,
"adv/mean_abs_step_conf": 0.7711992263793945,
"adv/ratio_final_to_reasoning": 1.1863639008281643,
"adv/ratio_step_to_reasoning": 1.4617374107567018,
"adv/std_final_conf": 0.8267181515693665,
"adv/std_reasoning": 0.7756460309028625,
"adv/std_step_conf": 0.9291678071022034,
"calib/answer_extract_rate": 0.9296875,
"calib/auroc": 0.583591903073286,
"calib/avg_num_step_conf": 12.50390625,
"calib/ece": 0.37638655462184867,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0048226950354613995,
"calib/mean_conf": 0.9814285714285713,
"calib/mu_c": 0.9833333333333335,
"calib/mu_w": 0.9785106382978721,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.37638655462184867,
"calib/std_conf": 0.013490470604054103,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.836875,
"calib/step_q_c_n": 1600.0,
"calib/step_q_gap": -0.0035060118675829255,
"calib/step_q_w": 0.840381011867583,
"calib/step_q_w_n": 1601.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.04296875,
"completions/max_length": 2836.0,
"completions/max_terminated_length": 2836.0,
"completions/mean_length": 873.21484375,
"completions/mean_terminated_length": 912.4203491210938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 572.0,
"epoch": 0.21013333333333334,
"grad_norm": 0.023413734510540962,
"kl": 0.128143310546875,
"learning_rate": 8.333333333333334e-08,
"loss": -0.0998,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.015757042914628983,
"mask/share_reasoning": 0.8129766583442688,
"mask/share_step_conf": 0.12829750776290894,
"num_tokens": 64718676.0,
"reward": 0.7949533462524414,
"reward_std": 0.271894633769989,
"rewards/accuracy_reward_step": 0.5625,
"rewards/asymmetric_l2_reward": 0.7136778235435486,
"rewards/final_brier_reward_step": 0.5777913928031921,
"rewards/format_reward_step": 0.9296875,
"step": 197
},
{
"adv/mean_abs_final_conf": 0.5809048414230347,
"adv/mean_abs_reasoning": 0.4991025924682617,
"adv/mean_abs_step_conf": 0.777890682220459,
"adv/ratio_final_to_reasoning": 1.1638986657036345,
"adv/ratio_step_to_reasoning": 1.5585787250141876,
"adv/std_final_conf": 0.7810680270195007,
"adv/std_reasoning": 0.7396623492240906,
"adv/std_step_conf": 0.9276652932167053,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.6636700798193985,
"calib/avg_num_step_conf": 13.34765625,
"calib/ece": 0.31288135593220345,
"calib/final_conf_rate": 0.921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 0.9957627118644068,
"calib/gap": 0.009945980811094435,
"calib/mean_conf": 0.978135593220339,
"calib/mu_c": 0.9814649681528661,
"calib/mu_w": 0.9715189873417717,
"calib/nonempty_final_conf_rate": 0.921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.31288135593220345,
"calib/std_conf": 0.015235393257114744,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.8487527839643654,
"calib/step_q_c_n": 1796.0,
"calib/step_q_gap": 0.010412253427659612,
"calib/step_q_w": 0.8383405305367058,
"calib/step_q_w_n": 1621.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2949.0,
"completions/max_terminated_length": 2949.0,
"completions/mean_length": 893.18359375,
"completions/mean_terminated_length": 944.8553466796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 462.0,
"epoch": 0.2112,
"grad_norm": 0.027668872848153114,
"kl": 0.126190185546875,
"learning_rate": 5.555555555555556e-08,
"loss": -0.2549,
"mask/has_final_conf_rate": 0.921875,
"mask/share_final_conf": 0.015574757941067219,
"mask/share_reasoning": 0.7983865141868591,
"mask/share_step_conf": 0.13135124742984772,
"num_tokens": 65052715.0,
"reward": 0.8287535905838013,
"reward_std": 0.24659347534179688,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/asymmetric_l2_reward": 0.7202712893486023,
"rewards/final_brier_reward_step": 0.6302046775817871,
"rewards/format_reward_step": 0.921875,
"step": 198
},
{
"adv/mean_abs_final_conf": 0.7182374000549316,
"adv/mean_abs_reasoning": 0.603756308555603,
"adv/mean_abs_step_conf": 0.7502772212028503,
"adv/ratio_final_to_reasoning": 1.1896147334231713,
"adv/ratio_step_to_reasoning": 1.242682205669663,
"adv/std_final_conf": 0.8792629837989807,
"adv/std_reasoning": 0.843132734298706,
"adv/std_step_conf": 0.9331202507019043,
"calib/answer_extract_rate": 0.921875,
"calib/auroc": 0.6757726269315673,
"calib/avg_num_step_conf": 12.4609375,
"calib/ece": 0.33880851063829787,
"calib/final_conf_rate": 0.91796875,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.010641753390097874,
"calib/mean_conf": 0.9813617021276596,
"calib/mu_c": 0.9851655629139072,
"calib/mu_w": 0.9745238095238093,
"calib/nonempty_final_conf_rate": 0.91796875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33880851063829787,
"calib/std_conf": 0.013646643312496537,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.841233993015134,
"calib/step_q_c_n": 1718.0,
"calib/step_q_gap": -0.0020812243761704208,
"calib/step_q_w": 0.8433152173913044,
"calib/step_q_w_n": 1472.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.05078125,
"completions/max_length": 2642.0,
"completions/max_terminated_length": 2642.0,
"completions/mean_length": 911.12109375,
"completions/mean_terminated_length": 959.8641967773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 494.0,
"epoch": 0.21226666666666666,
"grad_norm": 0.02316947840154171,
"kl": 0.1233367919921875,
"learning_rate": 2.777777777777778e-08,
"loss": -0.2208,
"mask/has_final_conf_rate": 0.91796875,
"mask/share_final_conf": 0.015151815488934517,
"mask/share_reasoning": 0.8085764646530151,
"mask/share_step_conf": 0.1254904866218567,
"num_tokens": 65390162.0,
"reward": 0.8104346394538879,
"reward_std": 0.31479138135910034,
"rewards/accuracy_reward_step": 0.59375,
"rewards/asymmetric_l2_reward": 0.7124529480934143,
"rewards/final_brier_reward_step": 0.6060726642608643,
"rewards/format_reward_step": 0.91796875,
"step": 199
},
{
"adv/mean_abs_final_conf": 0.5120427012443542,
"adv/mean_abs_reasoning": 0.47003117203712463,
"adv/mean_abs_step_conf": 0.7336593866348267,
"adv/ratio_final_to_reasoning": 1.0893803043426902,
"adv/ratio_step_to_reasoning": 1.560873895778299,
"adv/std_final_conf": 0.76967853307724,
"adv/std_reasoning": 0.7576777935028076,
"adv/std_step_conf": 0.926696240901947,
"calib/answer_extract_rate": 0.92578125,
"calib/auroc": 0.77136916679501,
"calib/avg_num_step_conf": 12.203125,
"calib/ece": 0.34371308016877633,
"calib/final_conf_rate": 0.92578125,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.016107346373016918,
"calib/mean_conf": 0.98084388185654,
"calib/mu_c": 0.986688741721854,
"calib/mu_w": 0.970581395348837,
"calib/nonempty_final_conf_rate": 0.92578125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.34371308016877633,
"calib/std_conf": 0.013814812515803535,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.8510152905198776,
"calib/step_q_c_n": 1635.0,
"calib/step_q_gap": 0.01500454505312121,
"calib/step_q_w": 0.8360107454667564,
"calib/step_q_w_n": 1489.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0546875,
"completions/max_length": 2971.0,
"completions/max_terminated_length": 2971.0,
"completions/mean_length": 876.6796875,
"completions/mean_terminated_length": 927.3966674804688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 444.0,
"epoch": 0.21333333333333335,
"grad_norm": 0.019465776160359383,
"kl": 0.119354248046875,
"learning_rate": 0.0,
"loss": -0.1049,
"mask/has_final_conf_rate": 0.92578125,
"mask/share_final_conf": 0.01544947735965252,
"mask/share_reasoning": 0.8081007599830627,
"mask/share_step_conf": 0.1217622458934784,
"num_tokens": 65722640.0,
"reward": 0.8022987842559814,
"reward_std": 0.23661327362060547,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/asymmetric_l2_reward": 0.6970667243003845,
"rewards/final_brier_reward_step": 0.6051871180534363,
"rewards/format_reward_step": 0.921875,
"step": 200
},
{
"epoch": 0.21333333333333335,
"step": 200,
"total_flos": 0.0,
"train_loss": 0.0124371408065781,
"train_runtime": 23828.9403,
"train_samples_per_second": 2.149,
"train_steps_per_second": 0.008
}
],
"logging_steps": 1,
"max_steps": 200,
"num_input_tokens_seen": 65722640,
"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