zhaohq/PureRL-1.5B-v7-s2-l2-kl-w0-b1
1
0
Fork 0
Code Issues Pull Requests Actions Projects Releases Wiki Activity
Files
main
PureRL-1.5B-v7-s2-l2-kl-w0-b1/trainer_state.json
ModelHub XC 65de069662 初始化项目，由ModelHub XC社区提供模型 
Model: zhaohq/PureRL-1.5B-v7-s2-l2-kl-w0-b1
Source: Original Platform
2026-06-04 12:31:12 +08:00

12243 lines
503 KiB
JSON
Raw Permalink Blame History

{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.21333333333333335,
"eval_steps": 500,
"global_step": 200,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"adv/mean_abs_final_conf": 0.773959219455719,
"adv/mean_abs_reasoning": 0.47714588046073914,
"adv/mean_abs_step_conf": 0.7490277290344238,
"adv/ratio_final_to_reasoning": 1.622059942565935,
"adv/ratio_step_to_reasoning": 1.5698086470140988,
"adv/std_final_conf": 0.9294352531433105,
"adv/std_reasoning": 0.7393431663513184,
"adv/std_step_conf": 0.9343300461769104,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.38076182006817844,
"calib/avg_num_step_conf": 5.23046875,
"calib/ece": 0.2003187250996017,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.2948207171314741,
"calib/gap": -0.026059730250481805,
"calib/mean_conf": 0.8737051792828686,
"calib/mu_c": 0.865606936416185,
"calib/mu_w": 0.8916666666666668,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.19239043824701207,
"calib/std_conf": 0.09027744273295583,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7959393232205367,
"calib/step_q_c_n": 857.0,
"calib/step_q_gap": -0.006446568895645877,
"calib/step_q_w": 0.8023858921161826,
"calib/step_q_w_n": 482.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2492.0,
"completions/max_terminated_length": 2492.0,
"completions/mean_length": 474.94921875,
"completions/mean_terminated_length": 478.68896484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 138.0,
"epoch": 0.0010666666666666667,
"grad_norm": 0.04298506677150726,
"kl": 0.000291675329208374,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.0135,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03466901555657387,
"mask/share_reasoning": 0.8340686559677124,
"mask/share_step_conf": 0.12344987690448761,
"num_tokens": 229171.0,
"reward": 0.8933746814727783,
"reward_std": 0.19672557711601257,
"rewards/accuracy_reward_step": 0.67578125,
"rewards/final_brier_reward_step": 0.7142800688743591,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7420004606246948,
"step": 1
},
{
"adv/mean_abs_final_conf": 0.7672724723815918,
"adv/mean_abs_reasoning": 0.5104547739028931,
"adv/mean_abs_step_conf": 0.7698483467102051,
"adv/ratio_final_to_reasoning": 1.503115479781084,
"adv/ratio_step_to_reasoning": 1.5081617139634353,
"adv/std_final_conf": 0.9330522418022156,
"adv/std_reasoning": 0.7575037479400635,
"adv/std_step_conf": 0.9345317482948303,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.44343065693430656,
"calib/avg_num_step_conf": 5.05859375,
"calib/ece": 0.3349411764705883,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.2823529411764706,
"calib/gap": 0.002352468143016151,
"calib/mean_conf": 0.8721960784313726,
"calib/mu_c": 0.8732846715328467,
"calib/mu_w": 0.8709322033898306,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3349411764705883,
"calib/std_conf": 0.07627016470309335,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7954391371340525,
"calib/step_q_c_n": 649.0,
"calib/step_q_gap": 0.011011892552009073,
"calib/step_q_w": 0.7844272445820434,
"calib/step_q_w_n": 646.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1966.0,
"completions/max_terminated_length": 1966.0,
"completions/mean_length": 492.9765625,
"completions/mean_terminated_length": 494.9098205566406,
"completions/min_length": 0.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.0021333333333333334,
"grad_norm": 0.04043004661798477,
"kl": 0.00037539005279541016,
"learning_rate": 5.000000000000001e-07,
"loss": -0.0158,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03364308178424835,
"mask/share_reasoning": 0.8523939251899719,
"mask/share_step_conf": 0.11005672812461853,
"num_tokens": 458661.0,
"reward": 0.8337589502334595,
"reward_std": 0.1928534209728241,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.6320762038230896,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7291916012763977,
"step": 2
},
{
"adv/mean_abs_final_conf": 0.7821295261383057,
"adv/mean_abs_reasoning": 0.46012091636657715,
"adv/mean_abs_step_conf": 0.7513811588287354,
"adv/ratio_final_to_reasoning": 1.6998347571645387,
"adv/ratio_step_to_reasoning": 1.633008046585111,
"adv/std_final_conf": 0.9285802841186523,
"adv/std_reasoning": 0.7206122875213623,
"adv/std_step_conf": 0.9327278137207031,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.509598183569561,
"calib/avg_num_step_conf": 4.88671875,
"calib/ece": 0.22188235294117642,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.2627450980392157,
"calib/gap": 0.0024157148754645474,
"calib/mean_conf": 0.8810196078431374,
"calib/mu_c": 0.8818343195266273,
"calib/mu_w": 0.8794186046511627,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.220078431372549,
"calib/std_conf": 0.04377907558034709,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7875064599483204,
"calib/step_q_c_n": 774.0,
"calib/step_q_gap": 0.019602896007020698,
"calib/step_q_w": 0.7679035639412997,
"calib/step_q_w_n": 477.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2506.0,
"completions/max_terminated_length": 2506.0,
"completions/mean_length": 495.42578125,
"completions/mean_terminated_length": 495.42578125,
"completions/min_length": 201.0,
"completions/min_terminated_length": 201.0,
"epoch": 0.0032,
"grad_norm": 0.04484294354915619,
"kl": 0.0013908743858337402,
"learning_rate": 7.5e-07,
"loss": 0.0599,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.033238910138607025,
"mask/share_reasoning": 0.8569374084472656,
"mask/share_step_conf": 0.10982376337051392,
"num_tokens": 690746.0,
"reward": 0.9005277156829834,
"reward_std": 0.18030250072479248,
"rewards/accuracy_reward_step": 0.66015625,
"rewards/final_brier_reward_step": 0.7222340106964111,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7491339445114136,
"step": 3
},
{
"adv/mean_abs_final_conf": 0.7650107145309448,
"adv/mean_abs_reasoning": 0.44886037707328796,
"adv/mean_abs_step_conf": 0.7658126354217529,
"adv/ratio_final_to_reasoning": 1.7043400433761995,
"adv/ratio_step_to_reasoning": 1.706126614282807,
"adv/std_final_conf": 0.9289597868919373,
"adv/std_reasoning": 0.7205116152763367,
"adv/std_step_conf": 0.9340852499008179,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.5186476242046838,
"calib/avg_num_step_conf": 4.87109375,
"calib/ece": 0.222549019607843,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.22745098039215686,
"calib/gap": 0.0038682821172331128,
"calib/mean_conf": 0.873529411764706,
"calib/mu_c": 0.8748795180722891,
"calib/mu_w": 0.871011235955056,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.222549019607843,
"calib/std_conf": 0.05047319340595715,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7892537313432837,
"calib/step_q_c_n": 804.0,
"calib/step_q_gap": 0.0006307065125839006,
"calib/step_q_w": 0.7886230248306998,
"calib/step_q_w_n": 443.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1498.0,
"completions/max_terminated_length": 1498.0,
"completions/mean_length": 498.46484375,
"completions/mean_terminated_length": 500.4196472167969,
"completions/min_length": 0.0,
"completions/min_terminated_length": 135.0,
"epoch": 0.004266666666666667,
"grad_norm": 0.04768529161810875,
"kl": 0.00032705068588256836,
"learning_rate": 1.0000000000000002e-06,
"loss": 0.0365,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03280118107795715,
"mask/share_reasoning": 0.8532472252845764,
"mask/share_step_conf": 0.11004534363746643,
"num_tokens": 924521.0,
"reward": 0.8905909657478333,
"reward_std": 0.18536730110645294,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/final_brier_reward_step": 0.7146550416946411,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7391830086708069,
"step": 4
},
{
"adv/mean_abs_final_conf": 0.762508749961853,
"adv/mean_abs_reasoning": 0.3974190950393677,
"adv/mean_abs_step_conf": 0.7614568471908569,
"adv/ratio_final_to_reasoning": 1.918651517955674,
"adv/ratio_step_to_reasoning": 1.9160046829542206,
"adv/std_final_conf": 0.9297828674316406,
"adv/std_reasoning": 0.6815359592437744,
"adv/std_step_conf": 0.9332932233810425,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.4592454587796926,
"calib/avg_num_step_conf": 4.765625,
"calib/ece": 0.33922764227642277,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.2967479674796748,
"calib/gap": -0.005463437354447986,
"calib/mean_conf": 0.8798780487804878,
"calib/mu_c": 0.8773684210526316,
"calib/mu_w": 0.8828318584070796,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.9765625,
"calib/nonempty_step_conf_rate": 0.96484375,
"calib/pce": 0.33922764227642277,
"calib/std_conf": 0.045989409414631144,
"calib/step_conf_rate": 0.96484375,
"calib/step_q_c": 0.8018296529968455,
"calib/step_q_c_n": 634.0,
"calib/step_q_gap": 0.010174362894456501,
"calib/step_q_w": 0.791655290102389,
"calib/step_q_w_n": 586.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2953.0,
"completions/max_terminated_length": 2953.0,
"completions/mean_length": 525.87109375,
"completions/mean_terminated_length": 527.933349609375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 133.0,
"epoch": 0.005333333333333333,
"grad_norm": 0.04267508536577225,
"kl": 0.0002726316452026367,
"learning_rate": 1.25e-06,
"loss": 0.0323,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03395113721489906,
"mask/share_reasoning": 0.851130485534668,
"mask/share_step_conf": 0.11101210117340088,
"num_tokens": 1165832.0,
"reward": 0.7854695320129395,
"reward_std": 0.16234935820102692,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.6045964956283569,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.6725925207138062,
"step": 5
},
{
"adv/mean_abs_final_conf": 0.7647924423217773,
"adv/mean_abs_reasoning": 0.4423978328704834,
"adv/mean_abs_step_conf": 0.75323486328125,
"adv/ratio_final_to_reasoning": 1.7287436454185308,
"adv/ratio_step_to_reasoning": 1.7026187908605046,
"adv/std_final_conf": 0.9318432211875916,
"adv/std_reasoning": 0.720503568649292,
"adv/std_step_conf": 0.9338743090629578,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5081408557364634,
"calib/avg_num_step_conf": 5.06640625,
"calib/ece": 0.33411067193675875,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.308300395256917,
"calib/gap": 1.640792629054033e-05,
"calib/mean_conf": 0.8835177865612648,
"calib/mu_c": 0.8835251798561151,
"calib/mu_w": 0.8835087719298246,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.33411067193675875,
"calib/std_conf": 0.04051882741810088,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.795828488372093,
"calib/step_q_c_n": 688.0,
"calib/step_q_gap": -0.008621429526100788,
"calib/step_q_w": 0.8044499178981938,
"calib/step_q_w_n": 609.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2746.0,
"completions/max_terminated_length": 2746.0,
"completions/mean_length": 433.66796875,
"completions/mean_terminated_length": 437.0826721191406,
"completions/min_length": 0.0,
"completions/min_terminated_length": 164.0,
"epoch": 0.0064,
"grad_norm": 0.039841946214437485,
"kl": 0.0005662441253662109,
"learning_rate": 1.5e-06,
"loss": -0.0496,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03768611699342728,
"mask/share_reasoning": 0.8277445435523987,
"mask/share_step_conf": 0.12675684690475464,
"num_tokens": 1382803.0,
"reward": 0.8113628625869751,
"reward_std": 0.18264266848564148,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.6277871131896973,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.6894698143005371,
"step": 6
},
{
"adv/mean_abs_final_conf": 0.8003997206687927,
"adv/mean_abs_reasoning": 0.5040899515151978,
"adv/mean_abs_step_conf": 0.7537604570388794,
"adv/ratio_final_to_reasoning": 1.587811298882163,
"adv/ratio_step_to_reasoning": 1.4952895902273393,
"adv/std_final_conf": 0.929995596408844,
"adv/std_reasoning": 0.7393463253974915,
"adv/std_step_conf": 0.9334385395050049,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.4976365378473331,
"calib/avg_num_step_conf": 5.20703125,
"calib/ece": 0.277109375,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.34375,
"calib/gap": 0.004912168636218572,
"calib/mean_conf": 0.882578125,
"calib/mu_c": 0.8845161290322582,
"calib/mu_w": 0.8796039603960396,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.277109375,
"calib/std_conf": 0.05675806789773922,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7917788461538462,
"calib/step_q_c_n": 832.0,
"calib/step_q_gap": 0.001998407032089644,
"calib/step_q_w": 0.7897804391217566,
"calib/step_q_w_n": 501.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1335.0,
"completions/max_terminated_length": 1335.0,
"completions/mean_length": 525.24609375,
"completions/mean_terminated_length": 527.305908203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 178.0,
"epoch": 0.007466666666666667,
"grad_norm": 0.04400845989584923,
"kl": 0.0003158748149871826,
"learning_rate": 1.75e-06,
"loss": 0.0579,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.030677923932671547,
"mask/share_reasoning": 0.8573629856109619,
"mask/share_step_conf": 0.10805287212133408,
"num_tokens": 1624690.0,
"reward": 0.876061737537384,
"reward_std": 0.19940611720085144,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.6821796894073486,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7496312260627747,
"step": 7
},
{
"adv/mean_abs_final_conf": 0.7505788207054138,
"adv/mean_abs_reasoning": 0.4054693579673767,
"adv/mean_abs_step_conf": 0.7549181580543518,
"adv/ratio_final_to_reasoning": 1.851135741719363,
"adv/ratio_step_to_reasoning": 1.86183775227496,
"adv/std_final_conf": 0.9302457571029663,
"adv/std_reasoning": 0.7013283967971802,
"adv/std_step_conf": 0.934516966342926,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.4533621140763998,
"calib/avg_num_step_conf": 4.47265625,
"calib/ece": 0.2893625498007968,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.2749003984063745,
"calib/gap": 0.0003623757195183597,
"calib/mean_conf": 0.8750199203187251,
"calib/mu_c": 0.8751700680272108,
"calib/mu_w": 0.8748076923076924,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.2893625498007968,
"calib/std_conf": 0.047481382402938206,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.7935275590551181,
"calib/step_q_c_n": 635.0,
"calib/step_q_gap": 0.02386089238845146,
"calib/step_q_w": 0.7696666666666666,
"calib/step_q_w_n": 510.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2555.0,
"completions/max_terminated_length": 2555.0,
"completions/mean_length": 545.45703125,
"completions/mean_terminated_length": 545.45703125,
"completions/min_length": 168.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.008533333333333334,
"grad_norm": 0.03767187148332596,
"kl": 0.00035335123538970947,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.0414,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.032488659024238586,
"mask/share_reasoning": 0.8679938316345215,
"mask/share_step_conf": 0.09951749444007874,
"num_tokens": 1870839.0,
"reward": 0.8519612550735474,
"reward_std": 0.1759214848279953,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.6545824408531189,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7391838431358337,
"step": 8
},
{
"adv/mean_abs_final_conf": 0.7705340385437012,
"adv/mean_abs_reasoning": 0.47810250520706177,
"adv/mean_abs_step_conf": 0.74006587266922,
"adv/ratio_final_to_reasoning": 1.6116502845137572,
"adv/ratio_step_to_reasoning": 1.5479230177819383,
"adv/std_final_conf": 0.9310494661331177,
"adv/std_reasoning": 0.7393581867218018,
"adv/std_step_conf": 0.934630811214447,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5215696465696467,
"calib/avg_num_step_conf": 4.5625,
"calib/ece": 0.2912301587301586,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.28174603174603174,
"calib/gap": 0.00011954261954272294,
"calib/mean_conf": 0.8785317460317461,
"calib/mu_c": 0.8785810810810811,
"calib/mu_w": 0.8784615384615384,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.2912301587301586,
"calib/std_conf": 0.04584120045173485,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.7793759512937595,
"calib/step_q_c_n": 657.0,
"calib/step_q_gap": 0.0028593172428788005,
"calib/step_q_w": 0.7765166340508807,
"calib/step_q_w_n": 511.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2343.0,
"completions/max_terminated_length": 2343.0,
"completions/mean_length": 486.640625,
"completions/mean_terminated_length": 488.5490417480469,
"completions/min_length": 0.0,
"completions/min_terminated_length": 182.0,
"epoch": 0.0096,
"grad_norm": 0.04332495480775833,
"kl": 0.00040915608406066895,
"learning_rate": 2.25e-06,
"loss": 0.0087,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.034009773284196854,
"mask/share_reasoning": 0.8562586307525635,
"mask/share_step_conf": 0.10582535713911057,
"num_tokens": 2102955.0,
"reward": 0.8214284181594849,
"reward_std": 0.21525982022285461,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.6493328213691711,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.6857114434242249,
"step": 9
},
{
"adv/mean_abs_final_conf": 0.7739413976669312,
"adv/mean_abs_reasoning": 0.4154004752635956,
"adv/mean_abs_step_conf": 0.7806507349014282,
"adv/ratio_final_to_reasoning": 1.863121117460977,
"adv/ratio_step_to_reasoning": 1.8792726089349328,
"adv/std_final_conf": 0.9283263683319092,
"adv/std_reasoning": 0.6815720200538635,
"adv/std_step_conf": 0.9333322048187256,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.5619734132184984,
"calib/avg_num_step_conf": 5.04296875,
"calib/ece": 0.3107031250000001,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 1.0,
"calib/frac_conf_gt_0.9": 0.35546875,
"calib/gap": 0.012565686825188549,
"calib/mean_conf": 0.884921875,
"calib/mu_c": 0.8902721088435375,
"calib/mu_w": 0.8777064220183489,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3107031250000001,
"calib/std_conf": 0.043499211446696086,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7838218390804598,
"calib/step_q_c_n": 696.0,
"calib/step_q_gap": 0.0015865449628127415,
"calib/step_q_w": 0.782235294117647,
"calib/step_q_w_n": 595.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1314.0,
"completions/max_terminated_length": 1314.0,
"completions/mean_length": 506.203125,
"completions/mean_terminated_length": 508.1882629394531,
"completions/min_length": 0.0,
"completions/min_terminated_length": 162.0,
"epoch": 0.010666666666666666,
"grad_norm": 0.04865778982639313,
"kl": 0.0004032254219055176,
"learning_rate": 2.5e-06,
"loss": 0.078,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03207051753997803,
"mask/share_reasoning": 0.8554325103759766,
"mask/share_step_conf": 0.1085907593369484,
"num_tokens": 2339343.0,
"reward": 0.8535362482070923,
"reward_std": 0.16450417041778564,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.6632242202758789,
"rewards/format_reward_step": 1.0,
"rewards/step_l2_reward": 0.7290045022964478,
"step": 10
},
{
"adv/mean_abs_final_conf": 0.7742735743522644,
"adv/mean_abs_reasoning": 0.3947698175907135,
"adv/mean_abs_step_conf": 0.75969398021698,
"adv/ratio_final_to_reasoning": 1.961329209709264,
"adv/ratio_step_to_reasoning": 1.924397323111996,
"adv/std_final_conf": 0.92948317527771,
"adv/std_reasoning": 0.6816731095314026,
"adv/std_step_conf": 0.9346326589584351,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.3907295216458412,
"calib/avg_num_step_conf": 5.3203125,
"calib/ece": 0.35083003952569175,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.383399209486166,
"calib/gap": -0.028455761706424165,
"calib/mean_conf": 0.8849802371541501,
"calib/mu_c": 0.8721582733812951,
"calib/mu_w": 0.9006140350877193,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.34320158102766807,
"calib/std_conf": 0.0881905254513794,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7702435530085959,
"calib/step_q_c_n": 698.0,
"calib/step_q_gap": 0.023481504815824894,
"calib/step_q_w": 0.746762048192771,
"calib/step_q_w_n": 664.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2673.0,
"completions/max_terminated_length": 2673.0,
"completions/mean_length": 519.19921875,
"completions/mean_terminated_length": 521.2353515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 158.0,
"epoch": 0.011733333333333333,
"grad_norm": 0.05495906248688698,
"kl": 0.0007111132144927979,
"learning_rate": 2.7500000000000004e-06,
"loss": 0.0205,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03233877941966057,
"mask/share_reasoning": 0.8515212535858154,
"mask/share_step_conf": 0.1122337281703949,
"num_tokens": 2576738.0,
"reward": 0.8109133243560791,
"reward_std": 0.17835165560245514,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.6096394062042236,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.706718385219574,
"step": 11
},
{
"adv/mean_abs_final_conf": 0.7674424648284912,
"adv/mean_abs_reasoning": 0.4356115460395813,
"adv/mean_abs_step_conf": 0.7697429060935974,
"adv/ratio_final_to_reasoning": 1.7617587775296446,
"adv/ratio_step_to_reasoning": 1.7670397240197477,
"adv/std_final_conf": 0.9264889359474182,
"adv/std_reasoning": 0.7014715075492859,
"adv/std_step_conf": 0.9340823292732239,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.4732954545454545,
"calib/avg_num_step_conf": 5.6875,
"calib/ece": 0.1991633466135458,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.4063745019920319,
"calib/gap": -0.006133333333333546,
"calib/mean_conf": 0.8918326693227091,
"calib/mu_c": 0.8899999999999999,
"calib/mu_w": 0.8961333333333334,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.1949003984063745,
"calib/std_conf": 0.04934458769820447,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7774606741573034,
"calib/step_q_c_n": 890.0,
"calib/step_q_gap": 0.01652427839758608,
"calib/step_q_w": 0.7609363957597173,
"calib/step_q_w_n": 566.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2948.0,
"completions/max_terminated_length": 2948.0,
"completions/mean_length": 477.0859375,
"completions/mean_terminated_length": 482.74310302734375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 140.0,
"epoch": 0.0128,
"grad_norm": 0.0514008104801178,
"kl": 0.00114363431930542,
"learning_rate": 3e-06,
"loss": 0.015,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.036076635122299194,
"mask/share_reasoning": 0.8272448778152466,
"mask/share_step_conf": 0.12495977431535721,
"num_tokens": 2803048.0,
"reward": 0.9234218597412109,
"reward_std": 0.18709824979305267,
"rewards/accuracy_reward_step": 0.6875,
"rewards/final_brier_reward_step": 0.7339004278182983,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7801306843757629,
"step": 12
},
{
"adv/mean_abs_final_conf": 0.737118124961853,
"adv/mean_abs_reasoning": 0.469147264957428,
"adv/mean_abs_step_conf": 0.7225381135940552,
"adv/ratio_final_to_reasoning": 1.5711870877660166,
"adv/ratio_step_to_reasoning": 1.5401094018093033,
"adv/std_final_conf": 0.9289529323577881,
"adv/std_reasoning": 0.7573885321617126,
"adv/std_step_conf": 0.9343307614326477,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.4623179850452578,
"calib/avg_num_step_conf": 4.80859375,
"calib/ece": 0.290395256916996,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.4980237154150198,
"calib/gap": -0.0016594516594514719,
"calib/mean_conf": 0.899090909090909,
"calib/mu_c": 0.8984415584415585,
"calib/mu_w": 0.90010101010101,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.290395256916996,
"calib/std_conf": 0.042755436012976406,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.7684196185286103,
"calib/step_q_c_n": 734.0,
"calib/step_q_gap": -0.0028479871051925176,
"calib/step_q_w": 0.7712676056338028,
"calib/step_q_w_n": 497.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1238.0,
"completions/max_terminated_length": 1238.0,
"completions/mean_length": 478.9921875,
"completions/mean_terminated_length": 480.87060546875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 141.0,
"epoch": 0.013866666666666666,
"grad_norm": 0.03835929557681084,
"kl": 0.0017654895782470703,
"learning_rate": 3.2500000000000002e-06,
"loss": -0.0057,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03364657610654831,
"mask/share_reasoning": 0.8520339727401733,
"mask/share_step_conf": 0.11041321605443954,
"num_tokens": 3030262.0,
"reward": 0.8755779266357422,
"reward_std": 0.18408486247062683,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.666958212852478,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7654476165771484,
"step": 13
},
{
"adv/mean_abs_final_conf": 0.7786264419555664,
"adv/mean_abs_reasoning": 0.5486913919448853,
"adv/mean_abs_step_conf": 0.7712372541427612,
"adv/ratio_final_to_reasoning": 1.4190607933462487,
"adv/ratio_step_to_reasoning": 1.4055938647206445,
"adv/std_final_conf": 0.9291496276855469,
"adv/std_reasoning": 0.792828381061554,
"adv/std_step_conf": 0.9350869059562683,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.4334833953154564,
"calib/avg_num_step_conf": 5.3515625,
"calib/ece": 0.3871370967741936,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.6451612903225806,
"calib/gap": -0.007322372284204293,
"calib/mean_conf": 0.9153629032258064,
"calib/mu_c": 0.9119083969465649,
"calib/mu_w": 0.9192307692307692,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.3871370967741936,
"calib/std_conf": 0.03840401412936802,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.7520642201834862,
"calib/step_q_c_n": 654.0,
"calib/step_q_gap": 0.06305584029521794,
"calib/step_q_w": 0.6890083798882682,
"calib/step_q_w_n": 716.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2834.0,
"completions/max_terminated_length": 2834.0,
"completions/mean_length": 537.93359375,
"completions/mean_terminated_length": 546.4722290039062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 172.0,
"epoch": 0.014933333333333333,
"grad_norm": 0.04159555211663246,
"kl": 0.003778219223022461,
"learning_rate": 3.5e-06,
"loss": -0.051,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.0319044254720211,
"mask/share_reasoning": 0.8418941497802734,
"mask/share_step_conf": 0.11057643592357635,
"num_tokens": 3273373.0,
"reward": 0.8043519258499146,
"reward_std": 0.2247830182313919,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.5733602046966553,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.7400311827659607,
"step": 14
},
{
"adv/mean_abs_final_conf": 0.7623423933982849,
"adv/mean_abs_reasoning": 0.49396762251853943,
"adv/mean_abs_step_conf": 0.7478142976760864,
"adv/ratio_final_to_reasoning": 1.5433043759253127,
"adv/ratio_step_to_reasoning": 1.5138933476313414,
"adv/std_final_conf": 0.9219062924385071,
"adv/std_reasoning": 0.739289402961731,
"adv/std_step_conf": 0.9342023134231567,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.4932983682983683,
"calib/avg_num_step_conf": 4.90234375,
"calib/ece": 0.3167450980392157,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.792156862745098,
"calib/gap": -0.0031429681429681056,
"calib/mean_conf": 0.9254509803921569,
"calib/mu_c": 0.9242307692307693,
"calib/mu_w": 0.9273737373737374,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3152156862745098,
"calib/std_conf": 0.04695943753441511,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.7143354838709679,
"calib/step_q_c_n": 775.0,
"calib/step_q_gap": 0.0030854838709679644,
"calib/step_q_w": 0.7112499999999999,
"calib/step_q_w_n": 480.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1329.0,
"completions/max_terminated_length": 1329.0,
"completions/mean_length": 467.5234375,
"completions/mean_terminated_length": 469.3569030761719,
"completions/min_length": 0.0,
"completions/min_terminated_length": 150.0,
"epoch": 0.016,
"grad_norm": 0.040848296135663986,
"kl": 0.008719921112060547,
"learning_rate": 3.7500000000000005e-06,
"loss": -0.0177,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03405185788869858,
"mask/share_reasoning": 0.8486453294754028,
"mask/share_step_conf": 0.11339649558067322,
"num_tokens": 3500939.0,
"reward": 0.8766745328903198,
"reward_std": 0.20056243240833282,
"rewards/accuracy_reward_step": 0.61328125,
"rewards/final_brier_reward_step": 0.6578144431114197,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7736595869064331,
"step": 15
},
{
"adv/mean_abs_final_conf": 0.7344765067100525,
"adv/mean_abs_reasoning": 0.37689459323883057,
"adv/mean_abs_step_conf": 0.7676898241043091,
"adv/ratio_final_to_reasoning": 1.9487584058936855,
"adv/ratio_step_to_reasoning": 2.0368820298195134,
"adv/std_final_conf": 0.9199317097663879,
"adv/std_reasoning": 0.6814736127853394,
"adv/std_step_conf": 0.9343593716621399,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.47638728694092014,
"calib/avg_num_step_conf": 6.484375,
"calib/ece": 0.34105882352941186,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.8941176470588236,
"calib/gap": -0.001993464052287397,
"calib/mean_conf": 0.9410588235294116,
"calib/mu_c": 0.9402614379084968,
"calib/mu_w": 0.9422549019607842,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.34105882352941186,
"calib/std_conf": 0.03228634857797129,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6677367896311067,
"calib/step_q_c_n": 1003.0,
"calib/step_q_gap": 0.019365404547392884,
"calib/step_q_w": 0.6483713850837138,
"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": 0.00390625,
"completions/max_length": 2503.0,
"completions/max_terminated_length": 2503.0,
"completions/mean_length": 615.65234375,
"completions/mean_terminated_length": 618.0667114257812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 184.0,
"epoch": 0.017066666666666667,
"grad_norm": 0.03159378096461296,
"kl": 0.009853363037109375,
"learning_rate": 4.000000000000001e-06,
"loss": -0.0392,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.026258613914251328,
"mask/share_reasoning": 0.8569262623786926,
"mask/share_step_conf": 0.11290884017944336,
"num_tokens": 3767394.0,
"reward": 0.8711050748825073,
"reward_std": 0.15706929564476013,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.6391730308532715,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.7842870950698853,
"step": 16
},
{
"adv/mean_abs_final_conf": 0.7703932523727417,
"adv/mean_abs_reasoning": 0.4900427758693695,
"adv/mean_abs_step_conf": 0.7754462957382202,
"adv/ratio_final_to_reasoning": 1.5720938871224277,
"adv/ratio_step_to_reasoning": 1.5824053203570347,
"adv/std_final_conf": 0.9156169295310974,
"adv/std_reasoning": 0.739263117313385,
"adv/std_step_conf": 0.9346990585327148,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.5078077158603796,
"calib/avg_num_step_conf": 5.88671875,
"calib/ece": 0.22843137254901952,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.8784313725490196,
"calib/gap": -0.004872167789344961,
"calib/mean_conf": 0.9390196078431372,
"calib/mu_c": 0.9376630434782609,
"calib/mu_w": 0.9425352112676059,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.22294117647058817,
"calib/std_conf": 0.04889584376101013,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6399820788530465,
"calib/step_q_c_n": 1116.0,
"calib/step_q_gap": 0.030442436909312476,
"calib/step_q_w": 0.609539641943734,
"calib/step_q_w_n": 391.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2383.0,
"completions/max_terminated_length": 2383.0,
"completions/mean_length": 524.06640625,
"completions/mean_terminated_length": 524.06640625,
"completions/min_length": 152.0,
"completions/min_terminated_length": 152.0,
"epoch": 0.018133333333333335,
"grad_norm": 0.03880058601498604,
"kl": 0.014672279357910156,
"learning_rate": 4.25e-06,
"loss": 0.0472,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03221951052546501,
"mask/share_reasoning": 0.8422893285751343,
"mask/share_step_conf": 0.12549114227294922,
"num_tokens": 4005083.0,
"reward": 0.9624072313308716,
"reward_std": 0.1909254938364029,
"rewards/accuracy_reward_step": 0.71875,
"rewards/final_brier_reward_step": 0.7406648397445679,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8419621586799622,
"step": 17
},
{
"adv/mean_abs_final_conf": 0.7564767003059387,
"adv/mean_abs_reasoning": 0.43438613414764404,
"adv/mean_abs_step_conf": 0.7423365712165833,
"adv/ratio_final_to_reasoning": 1.7414844555070879,
"adv/ratio_step_to_reasoning": 1.7089324747282784,
"adv/std_final_conf": 0.9087929129600525,
"adv/std_reasoning": 0.7205876708030701,
"adv/std_step_conf": 0.934881865978241,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5161087339541908,
"calib/avg_num_step_conf": 4.5859375,
"calib/ece": 0.41592885375494065,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.9209486166007905,
"calib/gap": -0.00337528316133906,
"calib/mean_conf": 0.9416205533596839,
"calib/mu_c": 0.9400729927007299,
"calib/mu_w": 0.9434482758620689,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.40802371541501975,
"calib/std_conf": 0.09777031188382188,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.6413122923588038,
"calib/step_q_c_n": 602.0,
"calib/step_q_gap": 0.020525579072090627,
"calib/step_q_w": 0.6207867132867132,
"calib/step_q_w_n": 572.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2096.0,
"completions/max_terminated_length": 2096.0,
"completions/mean_length": 504.70703125,
"completions/mean_terminated_length": 506.6863098144531,
"completions/min_length": 0.0,
"completions/min_terminated_length": 170.0,
"epoch": 0.0192,
"grad_norm": 0.02734338492155075,
"kl": 0.014324188232421875,
"learning_rate": 4.5e-06,
"loss": -0.0854,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03250627592206001,
"mask/share_reasoning": 0.8629031777381897,
"mask/share_step_conf": 0.1006842851638794,
"num_tokens": 4245008.0,
"reward": 0.8083161115646362,
"reward_std": 0.19949162006378174,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.5616640448570251,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7541868686676025,
"step": 18
},
{
"adv/mean_abs_final_conf": 0.7312251329421997,
"adv/mean_abs_reasoning": 0.3974907696247101,
"adv/mean_abs_step_conf": 0.7509687542915344,
"adv/ratio_final_to_reasoning": 1.8396028004186968,
"adv/ratio_step_to_reasoning": 1.889273441495408,
"adv/std_final_conf": 0.9088385701179504,
"adv/std_reasoning": 0.6815569400787354,
"adv/std_step_conf": 0.9349246621131897,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.472909825713564,
"calib/avg_num_step_conf": 4.671875,
"calib/ece": 0.3792549019607844,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.984313725490196,
"calib/gap": -0.0028624652690071306,
"calib/mean_conf": 0.9596470588235294,
"calib/mu_c": 0.9584459459459461,
"calib/mu_w": 0.9613084112149533,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3792549019607844,
"calib/std_conf": 0.021353076216285564,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6354014598540146,
"calib/step_q_c_n": 685.0,
"calib/step_q_gap": 0.02362063793620639,
"calib/step_q_w": 0.6117808219178082,
"calib/step_q_w_n": 511.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1457.0,
"completions/max_terminated_length": 1457.0,
"completions/mean_length": 499.2421875,
"completions/mean_terminated_length": 501.2000427246094,
"completions/min_length": 0.0,
"completions/min_terminated_length": 183.0,
"epoch": 0.020266666666666665,
"grad_norm": 0.02526911534368992,
"kl": 0.023448944091796875,
"learning_rate": 4.75e-06,
"loss": -0.0057,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03129266947507858,
"mask/share_reasoning": 0.8586671948432922,
"mask/share_step_conf": 0.1061338409781456,
"num_tokens": 4477574.0,
"reward": 0.8613945841789246,
"reward_std": 0.18759620189666748,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.60454261302948,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.804184079170227,
"step": 19
},
{
"adv/mean_abs_final_conf": 0.7194894552230835,
"adv/mean_abs_reasoning": 0.4517136514186859,
"adv/mean_abs_step_conf": 0.7461144924163818,
"adv/ratio_final_to_reasoning": 1.5927998920630375,
"adv/ratio_step_to_reasoning": 1.6517421824048897,
"adv/std_final_conf": 0.8810336589813232,
"adv/std_reasoning": 0.739335298538208,
"adv/std_step_conf": 0.9348995685577393,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.49452873254564983,
"calib/avg_num_step_conf": 5.44140625,
"calib/ece": 0.35896,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.984,
"calib/gap": 0.0010418904403868279,
"calib/mean_conf": 0.96696,
"calib/mu_c": 0.9673684210526318,
"calib/mu_w": 0.9663265306122449,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.35896,
"calib/std_conf": 0.019543756036135944,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6032474226804124,
"calib/step_q_c_n": 776.0,
"calib/step_q_gap": 0.044171247639893885,
"calib/step_q_w": 0.5590761750405185,
"calib/step_q_w_n": 617.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2581.0,
"completions/max_terminated_length": 2581.0,
"completions/mean_length": 513.9375,
"completions/mean_terminated_length": 515.9529418945312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 144.0,
"epoch": 0.021333333333333333,
"grad_norm": 0.03067217580974102,
"kl": 0.025417327880859375,
"learning_rate": 5e-06,
"loss": 0.0229,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03458629548549652,
"mask/share_reasoning": 0.8356574773788452,
"mask/share_step_conf": 0.12584996223449707,
"num_tokens": 4714014.0,
"reward": 0.8612836599349976,
"reward_std": 0.20020237565040588,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.6180921792984009,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7904125452041626,
"step": 20
},
{
"adv/mean_abs_final_conf": 0.7106478214263916,
"adv/mean_abs_reasoning": 0.43507465720176697,
"adv/mean_abs_step_conf": 0.7549261450767517,
"adv/ratio_final_to_reasoning": 1.6333928204345556,
"adv/ratio_step_to_reasoning": 1.735164603546772,
"adv/std_final_conf": 0.8982265591621399,
"adv/std_reasoning": 0.7205180525779724,
"adv/std_step_conf": 0.9351524710655212,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.579504724666015,
"calib/avg_num_step_conf": 5.8671875,
"calib/ece": 0.36047244094488184,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.9881889763779528,
"calib/gap": 0.002816552623004176,
"calib/mean_conf": 0.9707086614173228,
"calib/mu_c": 0.9718064516129032,
"calib/mu_w": 0.9689898989898991,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.36047244094488184,
"calib/std_conf": 0.021284762297996477,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5655878928987195,
"calib/step_q_c_n": 859.0,
"calib/step_q_gap": 0.008558343909605859,
"calib/step_q_w": 0.5570295489891136,
"calib/step_q_w_n": 643.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1916.0,
"completions/max_terminated_length": 1916.0,
"completions/mean_length": 500.91796875,
"completions/mean_terminated_length": 502.88238525390625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.0224,
"grad_norm": 0.027707798406481743,
"kl": 0.031322479248046875,
"learning_rate": 4.9722222222222224e-06,
"loss": -0.0308,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03268470615148544,
"mask/share_reasoning": 0.836051344871521,
"mask/share_step_conf": 0.12735772132873535,
"num_tokens": 4945209.0,
"reward": 0.8745825886726379,
"reward_std": 0.1900649070739746,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.6281523704528809,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8014814853668213,
"step": 21
},
{
"adv/mean_abs_final_conf": 0.7099519968032837,
"adv/mean_abs_reasoning": 0.3854062557220459,
"adv/mean_abs_step_conf": 0.7544239163398743,
"adv/ratio_final_to_reasoning": 1.842087372124285,
"adv/ratio_step_to_reasoning": 1.9574770910931012,
"adv/std_final_conf": 0.8757226467132568,
"adv/std_reasoning": 0.6815925240516663,
"adv/std_step_conf": 0.9346233606338501,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5026154891304349,
"calib/avg_num_step_conf": 6.0546875,
"calib/ece": 0.3418253968253968,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0003532608695651085,
"calib/mean_conf": 0.9767460317460317,
"calib/mu_c": 0.976875,
"calib/mu_w": 0.9765217391304349,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3418253968253968,
"calib/std_conf": 0.013934618594137112,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5992184154175589,
"calib/step_q_c_n": 934.0,
"calib/step_q_gap": 0.06386127256041607,
"calib/step_q_w": 0.5353571428571429,
"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": 2731.0,
"completions/max_terminated_length": 2731.0,
"completions/mean_length": 505.08984375,
"completions/mean_terminated_length": 511.0790710449219,
"completions/min_length": 0.0,
"completions/min_terminated_length": 199.0,
"epoch": 0.023466666666666667,
"grad_norm": 0.02160457894206047,
"kl": 0.03806304931640625,
"learning_rate": 4.944444444444445e-06,
"loss": -0.0644,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.031118055805563927,
"mask/share_reasoning": 0.8294792175292969,
"mask/share_step_conf": 0.12768399715423584,
"num_tokens": 5176328.0,
"reward": 0.8897135257720947,
"reward_std": 0.18438906967639923,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.6411515474319458,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.816400408744812,
"step": 22
},
{
"adv/mean_abs_final_conf": 0.725068211555481,
"adv/mean_abs_reasoning": 0.4642287790775299,
"adv/mean_abs_step_conf": 0.771478533744812,
"adv/ratio_final_to_reasoning": 1.5618769111993998,
"adv/ratio_step_to_reasoning": 1.6618498647968762,
"adv/std_final_conf": 0.8838828206062317,
"adv/std_reasoning": 0.7392085194587708,
"adv/std_step_conf": 0.9351414442062378,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5082431412031211,
"calib/avg_num_step_conf": 5.7265625,
"calib/ece": 0.44043478260869573,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00023345079285186365,
"calib/mean_conf": 0.9819367588932807,
"calib/mu_c": 0.9820437956204381,
"calib/mu_w": 0.9818103448275862,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.44043478260869573,
"calib/std_conf": 0.011820627134753209,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.6099600000000001,
"calib/step_q_c_n": 750.0,
"calib/step_q_gap": 0.024080111731843656,
"calib/step_q_w": 0.5858798882681564,
"calib/step_q_w_n": 716.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2293.0,
"completions/max_terminated_length": 2293.0,
"completions/mean_length": 528.421875,
"completions/mean_terminated_length": 530.494140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 166.0,
"epoch": 0.024533333333333334,
"grad_norm": 0.021991299465298653,
"kl": 0.032794952392578125,
"learning_rate": 4.9166666666666665e-06,
"loss": 0.005,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03355323523283005,
"mask/share_reasoning": 0.8336979150772095,
"mask/share_step_conf": 0.12884265184402466,
"num_tokens": 5415540.0,
"reward": 0.8157504200935364,
"reward_std": 0.20109497010707855,
"rewards/accuracy_reward_step": 0.53515625,
"rewards/final_brier_reward_step": 0.5472742319107056,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7803203463554382,
"step": 23
},
{
"adv/mean_abs_final_conf": 0.7268941402435303,
"adv/mean_abs_reasoning": 0.634642481803894,
"adv/mean_abs_step_conf": 0.7647470235824585,
"adv/ratio_final_to_reasoning": 1.1453600429921145,
"adv/ratio_step_to_reasoning": 1.2050044639444213,
"adv/std_final_conf": 0.8899421095848083,
"adv/std_reasoning": 0.843029260635376,
"adv/std_step_conf": 0.9352526068687439,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6084887334887334,
"calib/avg_num_step_conf": 6.765625,
"calib/ece": 0.45120481927710854,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0050796425796426625,
"calib/mean_conf": 0.9813253012048194,
"calib/mu_c": 0.9837121212121211,
"calib/mu_w": 0.9786324786324785,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.45120481927710854,
"calib/std_conf": 0.013089535751676469,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.6250829383886255,
"calib/step_q_c_n": 844.0,
"calib/step_q_gap": 0.005251857307544405,
"calib/step_q_w": 0.6198310810810811,
"calib/step_q_w_n": 888.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2925.0,
"completions/max_terminated_length": 2925.0,
"completions/mean_length": 599.82421875,
"completions/mean_terminated_length": 604.5472412109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 173.0,
"epoch": 0.0256,
"grad_norm": 0.027556290850043297,
"kl": 0.04826927185058594,
"learning_rate": 4.888888888888889e-06,
"loss": 0.0361,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.030089421197772026,
"mask/share_reasoning": 0.8356856107711792,
"mask/share_step_conf": 0.12641246616840363,
"num_tokens": 5673607.0,
"reward": 0.7990022897720337,
"reward_std": 0.2679530680179596,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.5345726609230042,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7665569186210632,
"step": 24
},
{
"adv/mean_abs_final_conf": 0.6522500514984131,
"adv/mean_abs_reasoning": 0.40848758816719055,
"adv/mean_abs_step_conf": 0.759333610534668,
"adv/ratio_final_to_reasoning": 1.5967438678490584,
"adv/ratio_step_to_reasoning": 1.8588902882989902,
"adv/std_final_conf": 0.8359845876693726,
"adv/std_reasoning": 0.6815344095230103,
"adv/std_step_conf": 0.9345144629478455,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.534307953950811,
"calib/avg_num_step_conf": 5.859375,
"calib/ece": 0.3708661417322835,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0013932496075353962,
"calib/mean_conf": 0.9850393700787402,
"calib/mu_c": 0.9855769230769231,
"calib/mu_w": 0.9841836734693877,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3708661417322835,
"calib/std_conf": 0.010712713452924015,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6474572405929304,
"calib/step_q_c_n": 877.0,
"calib/step_q_gap": 0.0247124572863493,
"calib/step_q_w": 0.6227447833065811,
"calib/step_q_w_n": 623.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2779.0,
"completions/max_terminated_length": 2779.0,
"completions/mean_length": 493.07421875,
"completions/mean_terminated_length": 495.00787353515625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 172.0,
"epoch": 0.02666666666666667,
"grad_norm": 0.02888176217675209,
"kl": 0.034931182861328125,
"learning_rate": 4.861111111111111e-06,
"loss": 0.0065,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03211010992527008,
"mask/share_reasoning": 0.8384443521499634,
"mask/share_step_conf": 0.12553933262825012,
"num_tokens": 5903058.0,
"reward": 0.8710969686508179,
"reward_std": 0.1723031848669052,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.6211484670639038,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8007329702377319,
"step": 25
},
{
"adv/mean_abs_final_conf": 0.6101415753364563,
"adv/mean_abs_reasoning": 0.4364333152770996,
"adv/mean_abs_step_conf": 0.7746877670288086,
"adv/ratio_final_to_reasoning": 1.3980178734729867,
"adv/ratio_step_to_reasoning": 1.775042692460233,
"adv/std_final_conf": 0.8025437593460083,
"adv/std_reasoning": 0.7013906240463257,
"adv/std_step_conf": 0.9353902339935303,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5800822047031872,
"calib/avg_num_step_conf": 5.51171875,
"calib/ece": 0.37240000000000006,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.996,
"calib/gap": 0.003988949531702479,
"calib/mean_conf": 0.9844,
"calib/mu_c": 0.9859477124183006,
"calib/mu_w": 0.9819587628865981,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.37240000000000006,
"calib/std_conf": 0.015357083056361975,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.649262453066333,
"calib/step_q_c_n": 799.0,
"calib/step_q_gap": 0.008036962870254682,
"calib/step_q_w": 0.6412254901960783,
"calib/step_q_w_n": 612.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2910.0,
"completions/max_terminated_length": 2910.0,
"completions/mean_length": 528.96484375,
"completions/mean_terminated_length": 531.0392456054688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 223.0,
"epoch": 0.027733333333333332,
"grad_norm": 0.02937263622879982,
"kl": 0.042026519775390625,
"learning_rate": 4.833333333333333e-06,
"loss": 0.0452,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.02984796278178692,
"mask/share_reasoning": 0.8507357239723206,
"mask/share_step_conf": 0.11551006883382797,
"num_tokens": 6143713.0,
"reward": 0.848415732383728,
"reward_std": 0.206464484333992,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.6108601093292236,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.7703462839126587,
"step": 26
},
{
"adv/mean_abs_final_conf": 0.6748393774032593,
"adv/mean_abs_reasoning": 0.5690269470214844,
"adv/mean_abs_step_conf": 0.7577905654907227,
"adv/ratio_final_to_reasoning": 1.18595328557925,
"adv/ratio_step_to_reasoning": 1.3317305436188969,
"adv/std_final_conf": 0.8738255500793457,
"adv/std_reasoning": 0.8098766207695007,
"adv/std_step_conf": 0.9352293014526367,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.5000635404752828,
"calib/avg_num_step_conf": 6.3203125,
"calib/ece": 0.501593625498008,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": -0.000529927563858279,
"calib/mean_conf": 0.9876494023904383,
"calib/mu_c": 0.9873770491803279,
"calib/mu_w": 0.9879069767441861,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.501593625498008,
"calib/std_conf": 0.0074451908086178315,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.6452361111111111,
"calib/step_q_c_n": 720.0,
"calib/step_q_gap": 0.030247246968572172,
"calib/step_q_w": 0.614988864142539,
"calib/step_q_w_n": 898.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 1888.0,
"completions/max_terminated_length": 1888.0,
"completions/mean_length": 539.953125,
"completions/mean_terminated_length": 544.2047119140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 188.0,
"epoch": 0.0288,
"grad_norm": 0.02530396170914173,
"kl": 0.03910064697265625,
"learning_rate": 4.805555555555556e-06,
"loss": 0.0118,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.030356114730238914,
"mask/share_reasoning": 0.8339431285858154,
"mask/share_step_conf": 0.12788823246955872,
"num_tokens": 6387157.0,
"reward": 0.7715263366699219,
"reward_std": 0.2471705675125122,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/final_brier_reward_step": 0.48854607343673706,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.763100266456604,
"step": 27
},
{
"adv/mean_abs_final_conf": 0.5680310130119324,
"adv/mean_abs_reasoning": 0.3548010289669037,
"adv/mean_abs_step_conf": 0.7654111981391907,
"adv/ratio_final_to_reasoning": 1.600984683347463,
"adv/ratio_step_to_reasoning": 2.157297007756393,
"adv/std_final_conf": 0.7786206603050232,
"adv/std_reasoning": 0.6401953101158142,
"adv/std_step_conf": 0.934259831905365,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5256724548859382,
"calib/avg_num_step_conf": 5.7734375,
"calib/ece": 0.3370866141732284,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.9960629921259843,
"calib/gap": 0.0014817841334693327,
"calib/mean_conf": 0.9866929133858268,
"calib/mu_c": 0.9872121212121211,
"calib/mu_w": 0.9857303370786518,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.3370866141732284,
"calib/std_conf": 0.010615329734419499,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.6254288702928871,
"calib/step_q_c_n": 956.0,
"calib/step_q_gap": -0.008191819362285302,
"calib/step_q_w": 0.6336206896551724,
"calib/step_q_w_n": 522.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1989.0,
"completions/max_terminated_length": 1989.0,
"completions/mean_length": 553.17578125,
"completions/mean_terminated_length": 555.3451538085938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 204.0,
"epoch": 0.029866666666666666,
"grad_norm": 0.025527317076921463,
"kl": 0.041477203369140625,
"learning_rate": 4.777777777777778e-06,
"loss": -0.0162,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.02951807528734207,
"mask/share_reasoning": 0.8487849235534668,
"mask/share_step_conf": 0.11779077351093292,
"num_tokens": 6635714.0,
"reward": 0.8864841461181641,
"reward_std": 0.16156134009361267,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.6463539004325867,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.800832986831665,
"step": 28
},
{
"adv/mean_abs_final_conf": 0.606778621673584,
"adv/mean_abs_reasoning": 0.4561738073825836,
"adv/mean_abs_step_conf": 0.7378709316253662,
"adv/ratio_final_to_reasoning": 1.330147877527504,
"adv/ratio_step_to_reasoning": 1.617521479058812,
"adv/std_final_conf": 0.8212588429450989,
"adv/std_reasoning": 0.7205647230148315,
"adv/std_step_conf": 0.9352527260780334,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5364016997875265,
"calib/avg_num_step_conf": 6.4140625,
"calib/ece": 0.4859683794466403,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.0014391951006123005,
"calib/mean_conf": 0.9879446640316205,
"calib/mu_c": 0.9886614173228344,
"calib/mu_w": 0.9872222222222221,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4859683794466403,
"calib/std_conf": 0.00837161241651468,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5999868073878627,
"calib/step_q_c_n": 758.0,
"calib/step_q_gap": 0.024421196528134126,
"calib/step_q_w": 0.5755656108597286,
"calib/step_q_w_n": 884.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2556.0,
"completions/max_terminated_length": 2556.0,
"completions/mean_length": 584.58984375,
"completions/mean_terminated_length": 586.8823852539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 187.0,
"epoch": 0.030933333333333334,
"grad_norm": 0.027113767340779305,
"kl": 0.0341796875,
"learning_rate": 4.75e-06,
"loss": -0.0488,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.028182100504636765,
"mask/share_reasoning": 0.849419355392456,
"mask/share_step_conf": 0.11849230527877808,
"num_tokens": 6892497.0,
"reward": 0.8000890016555786,
"reward_std": 0.19791510701179504,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.5045530796051025,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.7995311617851257,
"step": 29
},
{
"adv/mean_abs_final_conf": 0.695144534111023,
"adv/mean_abs_reasoning": 0.5578685402870178,
"adv/mean_abs_step_conf": 0.747866153717041,
"adv/ratio_final_to_reasoning": 1.246072298239614,
"adv/ratio_step_to_reasoning": 1.3405777521210844,
"adv/std_final_conf": 0.8613123297691345,
"adv/std_reasoning": 0.7755047082901001,
"adv/std_step_conf": 0.9354647397994995,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5202317880794703,
"calib/avg_num_step_conf": 6.5078125,
"calib/ece": 0.3870517928286853,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 1.0,
"calib/gap": 0.00024172185430437754,
"calib/mean_conf": 0.9886454183266933,
"calib/mu_c": 0.9887417218543043,
"calib/mu_w": 0.9884999999999999,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3870517928286853,
"calib/std_conf": 0.007179941690125934,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5561439312567132,
"calib/step_q_c_n": 931.0,
"calib/step_q_gap": 0.0007017543859649811,
"calib/step_q_w": 0.5554421768707483,
"calib/step_q_w_n": 735.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2962.0,
"completions/max_terminated_length": 2962.0,
"completions/mean_length": 625.05859375,
"completions/mean_terminated_length": 627.5098266601562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 194.0,
"epoch": 0.032,
"grad_norm": 0.031422603875398636,
"kl": 0.03484344482421875,
"learning_rate": 4.722222222222222e-06,
"loss": 0.0001,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.02694585919380188,
"mask/share_reasoning": 0.8529764413833618,
"mask/share_step_conf": 0.11617143452167511,
"num_tokens": 7159496.0,
"reward": 0.8489872813224792,
"reward_std": 0.24480488896369934,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.5945906639099121,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.7908838987350464,
"step": 30
},
{
"adv/mean_abs_final_conf": 0.7133569717407227,
"adv/mean_abs_reasoning": 0.4811326265335083,
"adv/mean_abs_step_conf": 0.7683807015419006,
"adv/ratio_final_to_reasoning": 1.4826618117344432,
"adv/ratio_step_to_reasoning": 1.5970247270029754,
"adv/std_final_conf": 0.8834726214408875,
"adv/std_reasoning": 0.720693826675415,
"adv/std_step_conf": 0.9351906776428223,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.4464917825537295,
"calib/avg_num_step_conf": 7.140625,
"calib/ece": 0.5399209486166008,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.9920948616600791,
"calib/gap": -0.0009823008849556558,
"calib/mean_conf": 0.9865612648221344,
"calib/mu_c": 0.9860176991150444,
"calib/mu_w": 0.9870000000000001,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.5399209486166008,
"calib/std_conf": 0.015336579589303448,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5300267379679144,
"calib/step_q_c_n": 748.0,
"calib/step_q_gap": 0.005850812041988562,
"calib/step_q_w": 0.5241759259259259,
"calib/step_q_w_n": 1080.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2759.0,
"completions/max_terminated_length": 2759.0,
"completions/mean_length": 620.8359375,
"completions/mean_terminated_length": 623.2706298828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 202.0,
"epoch": 0.03306666666666667,
"grad_norm": 0.09976796060800552,
"kl": 0.056781768798828125,
"learning_rate": 4.694444444444445e-06,
"loss": 0.0319,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.02804284356534481,
"mask/share_reasoning": 0.84264075756073,
"mask/share_step_conf": 0.12541010975837708,
"num_tokens": 7424342.0,
"reward": 0.7682323455810547,
"reward_std": 0.21610067784786224,
"rewards/accuracy_reward_step": 0.44140625,
"rewards/final_brier_reward_step": 0.4552140533924103,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7953131198883057,
"step": 31
},
{
"adv/mean_abs_final_conf": 0.6915620565414429,
"adv/mean_abs_reasoning": 0.4747565984725952,
"adv/mean_abs_step_conf": 0.7673255205154419,
"adv/ratio_final_to_reasoning": 1.4566665503257086,
"adv/ratio_step_to_reasoning": 1.6162503543586555,
"adv/std_final_conf": 0.8616219162940979,
"adv/std_reasoning": 0.7393125295639038,
"adv/std_step_conf": 0.93488609790802,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5823407071280016,
"calib/avg_num_step_conf": 6.13671875,
"calib/ece": 0.4732142857142858,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.996031746031746,
"calib/gap": 0.0052353942144072185,
"calib/mean_conf": 0.9851190476190477,
"calib/mu_c": 0.9876744186046511,
"calib/mu_w": 0.9824390243902439,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4732142857142858,
"calib/std_conf": 0.013728725037543098,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4811369680851063,
"calib/step_q_c_n": 752.0,
"calib/step_q_gap": 0.035080802028940206,
"calib/step_q_w": 0.4460561660561661,
"calib/step_q_w_n": 819.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2550.0,
"completions/max_terminated_length": 2550.0,
"completions/mean_length": 552.30859375,
"completions/mean_terminated_length": 558.8577270507812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 199.0,
"epoch": 0.034133333333333335,
"grad_norm": 0.03148404136300087,
"kl": 0.046142578125,
"learning_rate": 4.666666666666667e-06,
"loss": -0.0457,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.029168495908379555,
"mask/share_reasoning": 0.838029146194458,
"mask/share_step_conf": 0.12108359485864639,
"num_tokens": 7672437.0,
"reward": 0.8188750147819519,
"reward_std": 0.19034592807292938,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.5203777551651001,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8197159767150879,
"step": 32
},
{
"adv/mean_abs_final_conf": 0.6719990372657776,
"adv/mean_abs_reasoning": 0.44133204221725464,
"adv/mean_abs_step_conf": 0.7471027374267578,
"adv/ratio_final_to_reasoning": 1.5226608833785344,
"adv/ratio_step_to_reasoning": 1.692835928416413,
"adv/std_final_conf": 0.8424936532974243,
"adv/std_reasoning": 0.701448380947113,
"adv/std_step_conf": 0.9346904754638672,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.540872713414634,
"calib/avg_num_step_conf": 6.61328125,
"calib/ece": 0.4918725099601594,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9840637450199203,
"calib/gap": 0.006772103658536421,
"calib/mean_conf": 0.9819123505976096,
"calib/mu_c": 0.9853658536585366,
"calib/mu_w": 0.9785937500000002,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.4918725099601594,
"calib/std_conf": 0.02858426222007358,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4880940594059406,
"calib/step_q_c_n": 808.0,
"calib/step_q_gap": 0.07595846618560159,
"calib/step_q_w": 0.412135593220339,
"calib/step_q_w_n": 885.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2719.0,
"completions/max_terminated_length": 2719.0,
"completions/mean_length": 576.78515625,
"completions/mean_terminated_length": 579.047119140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 181.0,
"epoch": 0.0352,
"grad_norm": 0.028174152597784996,
"kl": 0.0565185546875,
"learning_rate": 4.638888888888889e-06,
"loss": 0.0181,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.028950443491339684,
"mask/share_reasoning": 0.8444273471832275,
"mask/share_step_conf": 0.12271599471569061,
"num_tokens": 7926966.0,
"reward": 0.8060372471809387,
"reward_std": 0.18721714615821838,
"rewards/accuracy_reward_step": 0.48046875,
"rewards/final_brier_reward_step": 0.5007531046867371,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8191338777542114,
"step": 33
},
{
"adv/mean_abs_final_conf": 0.720805287361145,
"adv/mean_abs_reasoning": 0.5097736120223999,
"adv/mean_abs_step_conf": 0.7887823581695557,
"adv/ratio_final_to_reasoning": 1.4139713597601287,
"adv/ratio_step_to_reasoning": 1.5473189266118699,
"adv/std_final_conf": 0.9024256467819214,
"adv/std_reasoning": 0.7575446963310242,
"adv/std_step_conf": 0.9344650506973267,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5065359477124183,
"calib/avg_num_step_conf": 6.8125,
"calib/ece": 0.44379446640316206,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.9723320158102767,
"calib/gap": 0.00667106586224242,
"calib/mean_conf": 0.9747826086956521,
"calib/mu_c": 0.9778676470588236,
"calib/mu_w": 0.9711965811965811,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.44051383399209487,
"calib/std_conf": 0.061158546097693,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.39326460481099657,
"calib/step_q_c_n": 873.0,
"calib/step_q_gap": 0.01340237748608264,
"calib/step_q_w": 0.3798622273249139,
"calib/step_q_w_n": 871.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 1466.0,
"completions/max_terminated_length": 1466.0,
"completions/mean_length": 489.14453125,
"completions/mean_terminated_length": 494.9446716308594,
"completions/min_length": 0.0,
"completions/min_terminated_length": 176.0,
"epoch": 0.03626666666666667,
"grad_norm": 0.028920266777276993,
"kl": 0.06124114990234375,
"learning_rate": 4.611111111111112e-06,
"loss": -0.0353,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03213128075003624,
"mask/share_reasoning": 0.8149935007095337,
"mask/share_step_conf": 0.14115644991397858,
"num_tokens": 8157299.0,
"reward": 0.8410252332687378,
"reward_std": 0.19692179560661316,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.553253173828125,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8248911499977112,
"step": 34
},
{
"adv/mean_abs_final_conf": 0.7265925407409668,
"adv/mean_abs_reasoning": 0.492765873670578,
"adv/mean_abs_step_conf": 0.7652642726898193,
"adv/ratio_final_to_reasoning": 1.4745187919135523,
"adv/ratio_step_to_reasoning": 1.5529977086063613,
"adv/std_final_conf": 0.8988257050514221,
"adv/std_reasoning": 0.7393568754196167,
"adv/std_step_conf": 0.9353464841842651,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5056186868686869,
"calib/avg_num_step_conf": 6.48046875,
"calib/ece": 0.4386111111111111,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9087301587301587,
"calib/gap": 0.010030303030302723,
"calib/mean_conf": 0.9624206349206349,
"calib/mu_c": 0.9671969696969696,
"calib/mu_w": 0.9571666666666668,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4386111111111111,
"calib/std_conf": 0.0845159751194443,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.3924677002583979,
"calib/step_q_c_n": 774.0,
"calib/step_q_gap": 0.041778434721674695,
"calib/step_q_w": 0.3506892655367232,
"calib/step_q_w_n": 885.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2814.0,
"completions/max_terminated_length": 2814.0,
"completions/mean_length": 592.49609375,
"completions/mean_terminated_length": 594.8196411132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 218.0,
"epoch": 0.037333333333333336,
"grad_norm": 0.04211479797959328,
"kl": 0.05974578857421875,
"learning_rate": 4.583333333333333e-06,
"loss": 0.0351,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.028332440182566643,
"mask/share_reasoning": 0.8531514406204224,
"mask/share_step_conf": 0.11460986733436584,
"num_tokens": 8418234.0,
"reward": 0.8261188268661499,
"reward_std": 0.1979362666606903,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.5472050905227661,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8058139085769653,
"step": 35
},
{
"adv/mean_abs_final_conf": 0.7033538818359375,
"adv/mean_abs_reasoning": 0.40417081117630005,
"adv/mean_abs_step_conf": 0.7413933277130127,
"adv/ratio_final_to_reasoning": 1.7402391820153813,
"adv/ratio_step_to_reasoning": 1.834356433497162,
"adv/std_final_conf": 0.9182852506637573,
"adv/std_reasoning": 0.7012822031974792,
"adv/std_step_conf": 0.9347103834152222,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.517733490964364,
"calib/avg_num_step_conf": 6.6171875,
"calib/ece": 0.24321818181818186,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.8932806324110671,
"calib/gap": -0.03352055396047959,
"calib/mean_conf": 0.9403391304347826,
"calib/mu_c": 0.9321246073298429,
"calib/mu_w": 0.9656451612903225,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.21430830039525697,
"calib/std_conf": 0.15752448260193302,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.36025121555915723,
"calib/step_q_c_n": 1234.0,
"calib/step_q_gap": 0.007251215559157254,
"calib/step_q_w": 0.353,
"calib/step_q_w_n": 460.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2335.0,
"completions/max_terminated_length": 2335.0,
"completions/mean_length": 527.109375,
"completions/mean_terminated_length": 527.109375,
"completions/min_length": 179.0,
"completions/min_terminated_length": 179.0,
"epoch": 0.0384,
"grad_norm": 0.05301982909440994,
"kl": 0.07215118408203125,
"learning_rate": 4.555555555555556e-06,
"loss": 0.0231,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.0321599505841732,
"mask/share_reasoning": 0.8275481462478638,
"mask/share_step_conf": 0.14029183983802795,
"num_tokens": 8655886.0,
"reward": 0.9387648105621338,
"reward_std": 0.18246082961559296,
"rewards/accuracy_reward_step": 0.75,
"rewards/final_brier_reward_step": 0.7346935272216797,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.7951799035072327,
"step": 36
},
{
"adv/mean_abs_final_conf": 0.7183753252029419,
"adv/mean_abs_reasoning": 0.5007466077804565,
"adv/mean_abs_step_conf": 0.7447097301483154,
"adv/ratio_final_to_reasoning": 1.4346084707135964,
"adv/ratio_step_to_reasoning": 1.4871987519780068,
"adv/std_final_conf": 0.9124584197998047,
"adv/std_reasoning": 0.7575705647468567,
"adv/std_step_conf": 0.9348015785217285,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.4699532085561497,
"calib/avg_num_step_conf": 6.80859375,
"calib/ece": 0.5294707317073172,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.8617886178861789,
"calib/gap": -0.0421933689839572,
"calib/mean_conf": 0.9406918699186994,
"calib/mu_c": 0.9173654545454546,
"calib/mu_w": 0.9595588235294118,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.5115040650406505,
"calib/std_conf": 0.12652019157822678,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.40833070866141724,
"calib/step_q_c_n": 635.0,
"calib/step_q_gap": 0.07433251371556882,
"calib/step_q_w": 0.3339981949458484,
"calib/step_q_w_n": 1108.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3027.0,
"completions/max_terminated_length": 3027.0,
"completions/mean_length": 575.2734375,
"completions/mean_terminated_length": 582.0949096679688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 203.0,
"epoch": 0.039466666666666664,
"grad_norm": 0.030364058911800385,
"kl": 0.06581878662109375,
"learning_rate": 4.527777777777778e-06,
"loss": -0.0514,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.029105212539434433,
"mask/share_reasoning": 0.8325843811035156,
"mask/share_step_conf": 0.12659165263175964,
"num_tokens": 8910252.0,
"reward": 0.7674344182014465,
"reward_std": 0.20056301355361938,
"rewards/accuracy_reward_step": 0.4296875,
"rewards/final_brier_reward_step": 0.4537394344806671,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8037855625152588,
"step": 37
},
{
"adv/mean_abs_final_conf": 0.7179021835327148,
"adv/mean_abs_reasoning": 0.4177252948284149,
"adv/mean_abs_step_conf": 0.7406455278396606,
"adv/ratio_final_to_reasoning": 1.7185987835081924,
"adv/ratio_step_to_reasoning": 1.77304447925254,
"adv/std_final_conf": 0.90470290184021,
"adv/std_reasoning": 0.7013022899627686,
"adv/std_step_conf": 0.9349610209465027,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.5494932984635502,
"calib/avg_num_step_conf": 6.11328125,
"calib/ece": 0.4177016129032257,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.8991935483870968,
"calib/gap": 0.012976135992154303,
"calib/mean_conf": 0.9505241935483871,
"calib/mu_c": 0.9565413533834587,
"calib/mu_w": 0.9435652173913044,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.41596774193548375,
"calib/std_conf": 0.09957230827877064,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.39140328697850824,
"calib/step_q_c_n": 791.0,
"calib/step_q_gap": 0.010447214627087031,
"calib/step_q_w": 0.3809560723514212,
"calib/step_q_w_n": 774.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2462.0,
"completions/max_terminated_length": 2462.0,
"completions/mean_length": 551.609375,
"completions/mean_terminated_length": 558.1502075195312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.04053333333333333,
"grad_norm": 0.04029100760817528,
"kl": 0.11200714111328125,
"learning_rate": 4.5e-06,
"loss": -0.0309,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.03054647706449032,
"mask/share_reasoning": 0.8312405943870544,
"mask/share_step_conf": 0.12649419903755188,
"num_tokens": 9158352.0,
"reward": 0.8352245092391968,
"reward_std": 0.17706511914730072,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5582581758499146,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8145345449447632,
"step": 38
},
{
"adv/mean_abs_final_conf": 0.717568576335907,
"adv/mean_abs_reasoning": 0.4351946711540222,
"adv/mean_abs_step_conf": 0.7223755121231079,
"adv/ratio_final_to_reasoning": 1.648845043146101,
"adv/ratio_step_to_reasoning": 1.659890527169272,
"adv/std_final_conf": 0.9146440625190735,
"adv/std_reasoning": 0.7205963730812073,
"adv/std_step_conf": 0.9348421692848206,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5192158385093167,
"calib/avg_num_step_conf": 7.0234375,
"calib/ece": 0.3912872,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.868,
"calib/gap": 0.029644901656314726,
"calib/mean_conf": 0.9404872000000001,
"calib/mu_c": 0.953768115942029,
"calib/mu_w": 0.9241232142857143,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3898872,
"calib/std_conf": 0.13081278070647379,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4256962040332147,
"calib/step_q_c_n": 843.0,
"calib/step_q_gap": 0.03835588989708899,
"calib/step_q_w": 0.3873403141361257,
"calib/step_q_w_n": 955.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2845.0,
"completions/max_terminated_length": 2845.0,
"completions/mean_length": 569.48046875,
"completions/mean_terminated_length": 573.9645385742188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 165.0,
"epoch": 0.0416,
"grad_norm": 0.022142568603157997,
"kl": 0.05435943603515625,
"learning_rate": 4.472222222222223e-06,
"loss": -0.012,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03002547100186348,
"mask/share_reasoning": 0.8311752080917358,
"mask/share_step_conf": 0.13098683953285217,
"num_tokens": 9410227.0,
"reward": 0.8570247888565063,
"reward_std": 0.18772698938846588,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.585284948348999,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8256394863128662,
"step": 39
},
{
"adv/mean_abs_final_conf": 0.7763040661811829,
"adv/mean_abs_reasoning": 0.5241175293922424,
"adv/mean_abs_step_conf": 0.752269983291626,
"adv/ratio_final_to_reasoning": 1.4811640951627234,
"adv/ratio_step_to_reasoning": 1.4353078099943446,
"adv/std_final_conf": 0.9272984266281128,
"adv/std_reasoning": 0.7576014995574951,
"adv/std_step_conf": 0.9347425103187561,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5090937499999999,
"calib/avg_num_step_conf": 6.1796875,
"calib/ece": 0.4414229249011857,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.8656126482213439,
"calib/gap": 0.01803937500000008,
"calib/mean_conf": 0.9394466403162055,
"calib/mu_c": 0.9483593750000001,
"calib/mu_w": 0.93032,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.4374703557312252,
"calib/std_conf": 0.11467986941563385,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4363092269326683,
"calib/step_q_c_n": 802.0,
"calib/step_q_gap": 0.012219483342924775,
"calib/step_q_w": 0.42408974358974355,
"calib/step_q_w_n": 780.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2608.0,
"completions/max_terminated_length": 2608.0,
"completions/mean_length": 597.90625,
"completions/mean_terminated_length": 597.90625,
"completions/min_length": 179.0,
"completions/min_terminated_length": 179.0,
"epoch": 0.042666666666666665,
"grad_norm": 0.020455822348594666,
"kl": 0.05359649658203125,
"learning_rate": 4.444444444444444e-06,
"loss": -0.009,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.030211972072720528,
"mask/share_reasoning": 0.8481245040893555,
"mask/share_step_conf": 0.12166358530521393,
"num_tokens": 9670051.0,
"reward": 0.8377189040184021,
"reward_std": 0.20407214760780334,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.5514258146286011,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8263556957244873,
"step": 40
},
{
"adv/mean_abs_final_conf": 0.6707231998443604,
"adv/mean_abs_reasoning": 0.3246734142303467,
"adv/mean_abs_step_conf": 0.7762365937232971,
"adv/ratio_final_to_reasoning": 2.0658396112731947,
"adv/ratio_step_to_reasoning": 2.3908227766766856,
"adv/std_final_conf": 0.8706308007240295,
"adv/std_reasoning": 0.6184402704238892,
"adv/std_step_conf": 0.9344105124473572,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.5263590755818734,
"calib/avg_num_step_conf": 6.06640625,
"calib/ece": 0.20425781250000002,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 1.0,
"calib/frac_conf_gt_0.9": 0.890625,
"calib/gap": 0.039305041533020724,
"calib/mean_conf": 0.9556640625000001,
"calib/mu_c": 0.9653367875647668,
"calib/mu_w": 0.9260317460317461,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.20300781250000002,
"calib/std_conf": 0.08102582317382585,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4902378929481733,
"calib/step_q_c_n": 1177.0,
"calib/step_q_gap": 0.008642148267322292,
"calib/step_q_w": 0.48159574468085103,
"calib/step_q_w_n": 376.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1262.0,
"completions/max_terminated_length": 1262.0,
"completions/mean_length": 498.671875,
"completions/mean_terminated_length": 500.6274719238281,
"completions/min_length": 0.0,
"completions/min_terminated_length": 176.0,
"epoch": 0.04373333333333333,
"grad_norm": 0.025205468758940697,
"kl": 0.0605010986328125,
"learning_rate": 4.416666666666667e-06,
"loss": -0.0149,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.033144064247608185,
"mask/share_reasoning": 0.8285462856292725,
"mask/share_step_conf": 0.13440342247486115,
"num_tokens": 9904959.0,
"reward": 0.9856371879577637,
"reward_std": 0.14432279765605927,
"rewards/accuracy_reward_step": 0.75390625,
"rewards/final_brier_reward_step": 0.7817816138267517,
"rewards/format_reward_step": 1.0,
"rewards/step_l2_reward": 0.8387115001678467,
"step": 41
},
{
"adv/mean_abs_final_conf": 0.6925047636032104,
"adv/mean_abs_reasoning": 0.3465518355369568,
"adv/mean_abs_step_conf": 0.7726593017578125,
"adv/ratio_final_to_reasoning": 1.9982718098440448,
"adv/ratio_step_to_reasoning": 2.229563437633026,
"adv/std_final_conf": 0.8931399583816528,
"adv/std_reasoning": 0.6401625275611877,
"adv/std_step_conf": 0.9346557855606079,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5946341153283744,
"calib/avg_num_step_conf": 6.3515625,
"calib/ece": 0.3798031496062992,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.8937007874015748,
"calib/gap": 0.024753639773666625,
"calib/mean_conf": 0.9546062992125984,
"calib/mu_c": 0.9650340136054423,
"calib/mu_w": 0.9402803738317757,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.37783464566929126,
"calib/std_conf": 0.08252679426309702,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5342200854700855,
"calib/step_q_c_n": 936.0,
"calib/step_q_gap": 0.042683853586027554,
"calib/step_q_w": 0.49153623188405793,
"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": 0.00390625,
"completions/max_length": 1674.0,
"completions/max_terminated_length": 1674.0,
"completions/mean_length": 444.89453125,
"completions/mean_terminated_length": 446.6392517089844,
"completions/min_length": 0.0,
"completions/min_terminated_length": 193.0,
"epoch": 0.0448,
"grad_norm": 0.01810487173497677,
"kl": 0.059814453125,
"learning_rate": 4.388888888888889e-06,
"loss": -0.0085,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03483423590660095,
"mask/share_reasoning": 0.8130151033401489,
"mask/share_step_conf": 0.1482444405555725,
"num_tokens": 10123220.0,
"reward": 0.8757699131965637,
"reward_std": 0.14750750362873077,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.6153386831283569,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8229199051856995,
"step": 42
},
{
"adv/mean_abs_final_conf": 0.7330894470214844,
"adv/mean_abs_reasoning": 0.6000100374221802,
"adv/mean_abs_step_conf": 0.7539304494857788,
"adv/ratio_final_to_reasoning": 1.2217953055769741,
"adv/ratio_step_to_reasoning": 1.2565297286106847,
"adv/std_final_conf": 0.9077485799789429,
"adv/std_reasoning": 0.8099002242088318,
"adv/std_step_conf": 0.934234619140625,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.6020503552287174,
"calib/avg_num_step_conf": 6.51171875,
"calib/ece": 0.38486274509803925,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.99609375,
"calib/frac_conf_gt_0.9": 0.8823529411764706,
"calib/gap": 0.04346379159915226,
"calib/mean_conf": 0.9417254901960785,
"calib/mu_c": 0.9609859154929576,
"calib/mu_w": 0.9175221238938054,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.38486274509803925,
"calib/std_conf": 0.11158467784889671,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5562444444444444,
"calib/step_q_c_n": 900.0,
"calib/step_q_gap": 0.03515904679125015,
"calib/step_q_w": 0.5210853976531943,
"calib/step_q_w_n": 767.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2905.0,
"completions/max_terminated_length": 2905.0,
"completions/mean_length": 547.1015625,
"completions/mean_terminated_length": 547.1015625,
"completions/min_length": 152.0,
"completions/min_terminated_length": 152.0,
"epoch": 0.04586666666666667,
"grad_norm": 0.022294873371720314,
"kl": 0.0514373779296875,
"learning_rate": 4.361111111111112e-06,
"loss": -0.0052,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.031736839562654495,
"mask/share_reasoning": 0.8334815502166748,
"mask/share_step_conf": 0.1347815990447998,
"num_tokens": 10368502.0,
"reward": 0.8724399209022522,
"reward_std": 0.22675156593322754,
"rewards/accuracy_reward_step": 0.5546875,
"rewards/final_brier_reward_step": 0.6117148399353027,
"rewards/format_reward_step": 0.99609375,
"rewards/step_l2_reward": 0.82300865650177,
"step": 43
},
{
"adv/mean_abs_final_conf": 0.7103047370910645,
"adv/mean_abs_reasoning": 0.3640681505203247,
"adv/mean_abs_step_conf": 0.7505558729171753,
"adv/ratio_final_to_reasoning": 1.9510213570615826,
"adv/ratio_step_to_reasoning": 2.061580700878349,
"adv/std_final_conf": 0.9105544686317444,
"adv/std_reasoning": 0.6612716913223267,
"adv/std_step_conf": 0.9344580173492432,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5427115987460815,
"calib/avg_num_step_conf": 6.7578125,
"calib/ece": 0.49834677419354845,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.8991935483870968,
"calib/gap": -0.015864681295715832,
"calib/mean_conf": 0.9456854838709678,
"calib/mu_c": 0.9372413793103448,
"calib/mu_w": 0.9531060606060606,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.48814516129032265,
"calib/std_conf": 0.10945398701350703,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5724705882352941,
"calib/step_q_c_n": 765.0,
"calib/step_q_gap": 0.03525815300213331,
"calib/step_q_w": 0.5372124352331608,
"calib/step_q_w_n": 965.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2901.0,
"completions/max_terminated_length": 2901.0,
"completions/mean_length": 614.7890625,
"completions/mean_terminated_length": 614.7890625,
"completions/min_length": 184.0,
"completions/min_terminated_length": 184.0,
"epoch": 0.046933333333333334,
"grad_norm": 0.025560567155480385,
"kl": 0.042316436767578125,
"learning_rate": 4.333333333333334e-06,
"loss": 0.009,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.02792154811322689,
"mask/share_reasoning": 0.8449289798736572,
"mask/share_step_conf": 0.12714949250221252,
"num_tokens": 10632208.0,
"reward": 0.7812137603759766,
"reward_std": 0.17581892013549805,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.48702070116996765,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7910317778587341,
"step": 44
},
{
"adv/mean_abs_final_conf": 0.7185328006744385,
"adv/mean_abs_reasoning": 0.49002134799957275,
"adv/mean_abs_step_conf": 0.7524739503860474,
"adv/ratio_final_to_reasoning": 1.4663295866756094,
"adv/ratio_step_to_reasoning": 1.5355942214719662,
"adv/std_final_conf": 0.8993702530860901,
"adv/std_reasoning": 0.7574601769447327,
"adv/std_step_conf": 0.933860719203949,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.6072868217054264,
"calib/avg_num_step_conf": 6.75390625,
"calib/ece": 0.44074803149606306,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.84251968503937,
"calib/gap": 0.0285748837209302,
"calib/mean_conf": 0.9354724409448818,
"calib/mu_c": 0.9495348837209302,
"calib/mu_w": 0.92096,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.4341732283464568,
"calib/std_conf": 0.1341361000438982,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5701189060642093,
"calib/step_q_c_n": 841.0,
"calib/step_q_gap": 0.04057949164979491,
"calib/step_q_w": 0.5295394144144144,
"calib/step_q_w_n": 888.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1842.0,
"completions/max_terminated_length": 1842.0,
"completions/mean_length": 525.51171875,
"completions/mean_terminated_length": 527.5725708007812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.048,
"grad_norm": 0.028553619980812073,
"kl": 0.05028533935546875,
"learning_rate": 4.305555555555556e-06,
"loss": 0.0151,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03267733007669449,
"mask/share_reasoning": 0.8204450607299805,
"mask/share_step_conf": 0.14297136664390564,
"num_tokens": 10871787.0,
"reward": 0.8383032083511353,
"reward_std": 0.19443970918655396,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.5568546652793884,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8213140964508057,
"step": 45
},
{
"adv/mean_abs_final_conf": 0.7056925296783447,
"adv/mean_abs_reasoning": 0.3723521828651428,
"adv/mean_abs_step_conf": 0.760047435760498,
"adv/ratio_final_to_reasoning": 1.895228662950876,
"adv/ratio_step_to_reasoning": 2.0412058012179544,
"adv/std_final_conf": 0.899570107460022,
"adv/std_reasoning": 0.640304684638977,
"adv/std_step_conf": 0.9345961809158325,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.5852539062500001,
"calib/avg_num_step_conf": 7.61328125,
"calib/ece": 0.4351701612903226,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.8870967741935484,
"calib/gap": 0.03033583333333334,
"calib/mean_conf": 0.950008870967742,
"calib/mu_c": 0.9646875000000001,
"calib/mu_w": 0.9343516666666668,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.434525,
"calib/std_conf": 0.10489374078938077,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5282490974729241,
"calib/step_q_c_n": 831.0,
"calib/step_q_gap": 0.07193424953016919,
"calib/step_q_w": 0.45631484794275495,
"calib/step_q_w_n": 1118.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2389.0,
"completions/max_terminated_length": 2389.0,
"completions/mean_length": 563.71484375,
"completions/mean_terminated_length": 572.6627197265625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 164.0,
"epoch": 0.04906666666666667,
"grad_norm": 0.02243691496551037,
"kl": 0.04651641845703125,
"learning_rate": 4.277777777777778e-06,
"loss": -0.0769,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.032937683165073395,
"mask/share_reasoning": 0.810023844242096,
"mask/share_step_conf": 0.14141345024108887,
"num_tokens": 11120866.0,
"reward": 0.8275113105773926,
"reward_std": 0.17046914994716644,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.5484654307365417,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8128072023391724,
"step": 46
},
{
"adv/mean_abs_final_conf": 0.7315204739570618,
"adv/mean_abs_reasoning": 0.43789246678352356,
"adv/mean_abs_step_conf": 0.7421537637710571,
"adv/ratio_final_to_reasoning": 1.6705482040610122,
"adv/ratio_step_to_reasoning": 1.6948310831251365,
"adv/std_final_conf": 0.9107234477996826,
"adv/std_reasoning": 0.7205823063850403,
"adv/std_step_conf": 0.93376624584198,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6430202217873451,
"calib/avg_num_step_conf": 7.2890625,
"calib/ece": 0.3410450199203188,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.8007968127490039,
"calib/gap": 0.0683270841487279,
"calib/mean_conf": 0.9227183266932272,
"calib/mu_c": 0.9513013698630136,
"calib/mu_w": 0.8829742857142857,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3410450199203188,
"calib/std_conf": 0.13605958057016465,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.49314814814814817,
"calib/step_q_c_n": 1026.0,
"calib/step_q_gap": 0.051064814814814785,
"calib/step_q_w": 0.4420833333333334,
"calib/step_q_w_n": 840.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2595.0,
"completions/max_terminated_length": 2595.0,
"completions/mean_length": 601.453125,
"completions/mean_terminated_length": 601.453125,
"completions/min_length": 173.0,
"completions/min_terminated_length": 173.0,
"epoch": 0.050133333333333335,
"grad_norm": 0.02189200557768345,
"kl": 0.07508087158203125,
"learning_rate": 4.25e-06,
"loss": 0.0298,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.028607338666915894,
"mask/share_reasoning": 0.8348825573921204,
"mask/share_step_conf": 0.13651013374328613,
"num_tokens": 11380814.0,
"reward": 0.9010224938392639,
"reward_std": 0.19029831886291504,
"rewards/accuracy_reward_step": 0.57421875,
"rewards/final_brier_reward_step": 0.6423037052154541,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8488037586212158,
"step": 47
},
{
"adv/mean_abs_final_conf": 0.7594094276428223,
"adv/mean_abs_reasoning": 0.5068789124488831,
"adv/mean_abs_step_conf": 0.759912371635437,
"adv/ratio_final_to_reasoning": 1.498206788627069,
"adv/ratio_step_to_reasoning": 1.4991990255899066,
"adv/std_final_conf": 0.9114439487457275,
"adv/std_reasoning": 0.7393609881401062,
"adv/std_step_conf": 0.9339820146560669,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5656954887218045,
"calib/avg_num_step_conf": 6.6171875,
"calib/ece": 0.3942640316205534,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.6996047430830039,
"calib/gap": 0.03819072681704261,
"calib/mean_conf": 0.8779098814229249,
"calib/mu_c": 0.896024060150376,
"calib/mu_w": 0.8578333333333333,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3732411067193676,
"calib/std_conf": 0.20810508600068744,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.48002675159235675,
"calib/step_q_c_n": 785.0,
"calib/step_q_gap": 0.03750749966716421,
"calib/step_q_w": 0.44251925192519254,
"calib/step_q_w_n": 909.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 3040.0,
"completions/max_terminated_length": 3040.0,
"completions/mean_length": 528.59765625,
"completions/mean_terminated_length": 528.59765625,
"completions/min_length": 141.0,
"completions/min_terminated_length": 141.0,
"epoch": 0.0512,
"grad_norm": 0.04436640441417694,
"kl": 0.05141448974609375,
"learning_rate": 4.222222222222223e-06,
"loss": 0.0771,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03390214592218399,
"mask/share_reasoning": 0.8238903284072876,
"mask/share_step_conf": 0.14220750331878662,
"num_tokens": 11619823.0,
"reward": 0.8720307946205139,
"reward_std": 0.21540537476539612,
"rewards/accuracy_reward_step": 0.51953125,
"rewards/final_brier_reward_step": 0.5952809453010559,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8472180962562561,
"step": 48
},
{
"adv/mean_abs_final_conf": 0.7109767198562622,
"adv/mean_abs_reasoning": 0.4105602502822876,
"adv/mean_abs_step_conf": 0.7638152837753296,
"adv/ratio_final_to_reasoning": 1.7317232230042197,
"adv/ratio_step_to_reasoning": 1.8604219070164625,
"adv/std_final_conf": 0.9296323657035828,
"adv/std_reasoning": 0.7013821005821228,
"adv/std_step_conf": 0.9330959320068359,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6210515157939369,
"calib/avg_num_step_conf": 6.69921875,
"calib/ece": 0.262055241935484,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.6411290322580645,
"calib/gap": 0.05743836477987441,
"calib/mean_conf": 0.8805254032258064,
"calib/mu_c": 0.9011383647798743,
"calib/mu_w": 0.8436999999999999,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.2507258064516131,
"calib/std_conf": 0.17448653971054612,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.44392989289191825,
"calib/step_q_c_n": 1027.0,
"calib/step_q_gap": 0.038566520798895,
"calib/step_q_w": 0.40536337209302326,
"calib/step_q_w_n": 688.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2943.0,
"completions/max_terminated_length": 2943.0,
"completions/mean_length": 540.40625,
"completions/mean_terminated_length": 542.5255126953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 182.0,
"epoch": 0.05226666666666667,
"grad_norm": 0.026353314518928528,
"kl": 0.05274200439453125,
"learning_rate": 4.194444444444445e-06,
"loss": 0.0058,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.03257454186677933,
"mask/share_reasoning": 0.8204771280288696,
"mask/share_step_conf": 0.14304211735725403,
"num_tokens": 11862703.0,
"reward": 0.9203234314918518,
"reward_std": 0.17258694767951965,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.686448872089386,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8362292647361755,
"step": 49
},
{
"adv/mean_abs_final_conf": 0.7396889925003052,
"adv/mean_abs_reasoning": 0.4333023130893707,
"adv/mean_abs_step_conf": 0.7512565851211548,
"adv/ratio_final_to_reasoning": 1.707096800906624,
"adv/ratio_step_to_reasoning": 1.7337931564796527,
"adv/std_final_conf": 0.9285359382629395,
"adv/std_reasoning": 0.701409637928009,
"adv/std_step_conf": 0.9330706000328064,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6411962365591398,
"calib/avg_num_step_conf": 6.4921875,
"calib/ece": 0.2160474308300395,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.4980237154150198,
"calib/gap": 0.08282526881720431,
"calib/mean_conf": 0.8113043478260871,
"calib/mu_c": 0.84175,
"calib/mu_w": 0.7589247311827957,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1974703557312253,
"calib/std_conf": 0.21444725864414702,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4211234817813765,
"calib/step_q_c_n": 988.0,
"calib/step_q_gap": 0.02628668652915095,
"calib/step_q_w": 0.39483679525222554,
"calib/step_q_w_n": 674.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2081.0,
"completions/max_terminated_length": 2081.0,
"completions/mean_length": 524.390625,
"completions/mean_terminated_length": 526.4470825195312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 156.0,
"epoch": 0.05333333333333334,
"grad_norm": 0.027222493663430214,
"kl": 0.05865478515625,
"learning_rate": 4.166666666666667e-06,
"loss": -0.0132,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.0337475948035717,
"mask/share_reasoning": 0.8215821385383606,
"mask/share_step_conf": 0.1407640427350998,
"num_tokens": 12102307.0,
"reward": 0.9507993459701538,
"reward_std": 0.17239651083946228,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.7195187211036682,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8594236373901367,
"step": 50
},
{
"adv/mean_abs_final_conf": 0.7422425746917725,
"adv/mean_abs_reasoning": 0.32978689670562744,
"adv/mean_abs_step_conf": 0.747180700302124,
"adv/ratio_final_to_reasoning": 2.2506733351334725,
"adv/ratio_step_to_reasoning": 2.265647021655528,
"adv/std_final_conf": 0.9353103637695312,
"adv/std_reasoning": 0.6185762286186218,
"adv/std_step_conf": 0.9331775307655334,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6924078525641026,
"calib/avg_num_step_conf": 6.62890625,
"calib/ece": 0.19300833333333328,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.4007936507936508,
"calib/gap": 0.14890641025641027,
"calib/mean_conf": 0.7490551587301587,
"calib/mu_c": 0.8057814102564103,
"calib/mu_w": 0.656875,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1615079365079365,
"calib/std_conf": 0.26039730848654774,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.41321656050955413,
"calib/step_q_c_n": 942.0,
"calib/step_q_gap": 0.04111060024465346,
"calib/step_q_w": 0.37210596026490067,
"calib/step_q_w_n": 755.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2370.0,
"completions/max_terminated_length": 2370.0,
"completions/mean_length": 511.4140625,
"completions/mean_terminated_length": 517.478271484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 160.0,
"epoch": 0.0544,
"grad_norm": 0.04199579730629921,
"kl": 0.055328369140625,
"learning_rate": 4.138888888888889e-06,
"loss": -0.0797,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.032312776893377304,
"mask/share_reasoning": 0.8194234371185303,
"mask/share_step_conf": 0.1365450620651245,
"num_tokens": 12342525.0,
"reward": 0.9573413133621216,
"reward_std": 0.16113626956939697,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.7379820942878723,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8579504489898682,
"step": 51
},
{
"adv/mean_abs_final_conf": 0.731773316860199,
"adv/mean_abs_reasoning": 0.32766151428222656,
"adv/mean_abs_step_conf": 0.7813364267349243,
"adv/ratio_final_to_reasoning": 2.2333209271257184,
"adv/ratio_step_to_reasoning": 2.384584068246512,
"adv/std_final_conf": 0.9118570685386658,
"adv/std_reasoning": 0.6184503436088562,
"adv/std_step_conf": 0.9324238896369934,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.6821321321321321,
"calib/avg_num_step_conf": 6.40234375,
"calib/ece": 0.11342519685039368,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.4015748031496063,
"calib/gap": 0.1586441441441443,
"calib/mean_conf": 0.7516141732283463,
"calib/mu_c": 0.7978333333333334,
"calib/mu_w": 0.6391891891891891,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.07818897637795273,
"calib/std_conf": 0.24566347724293106,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4085675917636526,
"calib/step_q_c_n": 1117.0,
"calib/step_q_gap": 0.04006184463721579,
"calib/step_q_w": 0.3685057471264368,
"calib/step_q_w_n": 522.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2640.0,
"completions/max_terminated_length": 2640.0,
"completions/mean_length": 500.421875,
"completions/mean_terminated_length": 500.421875,
"completions/min_length": 186.0,
"completions/min_terminated_length": 186.0,
"epoch": 0.055466666666666664,
"grad_norm": 0.5362410545349121,
"kl": 1.5911026000976562,
"learning_rate": 4.111111111111111e-06,
"loss": 0.0556,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03464528173208237,
"mask/share_reasoning": 0.8255054354667664,
"mask/share_step_conf": 0.13984927535057068,
"num_tokens": 12578585.0,
"reward": 0.998847246170044,
"reward_std": 0.11867427080869675,
"rewards/accuracy_reward_step": 0.703125,
"rewards/final_brier_reward_step": 0.7906261682510376,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8680056929588318,
"step": 52
},
{
"adv/mean_abs_final_conf": 0.7077825665473938,
"adv/mean_abs_reasoning": 0.42978763580322266,
"adv/mean_abs_step_conf": 0.758415937423706,
"adv/ratio_final_to_reasoning": 1.6468192837251616,
"adv/ratio_step_to_reasoning": 1.7646294919729733,
"adv/std_final_conf": 0.8943660259246826,
"adv/std_reasoning": 0.7013234496116638,
"adv/std_step_conf": 0.9318793416023254,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.6782853359814631,
"calib/avg_num_step_conf": 7.140625,
"calib/ece": 0.2157874015748033,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.4448818897637795,
"calib/gap": 0.1158788480635552,
"calib/mean_conf": 0.7804330708661417,
"calib/mu_c": 0.8237735849056604,
"calib/mu_w": 0.7078947368421052,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.1851181102362206,
"calib/std_conf": 0.23070290028011436,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.40601916376306624,
"calib/step_q_c_n": 1148.0,
"calib/step_q_gap": 0.033298575527772156,
"calib/step_q_w": 0.3727205882352941,
"calib/step_q_w_n": 680.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2708.0,
"completions/max_terminated_length": 2708.0,
"completions/mean_length": 524.7890625,
"completions/mean_terminated_length": 526.8471069335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 171.0,
"epoch": 0.05653333333333333,
"grad_norm": 0.03012523055076599,
"kl": 0.06369400024414062,
"learning_rate": 4.083333333333334e-06,
"loss": -0.0806,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03163750469684601,
"mask/share_reasoning": 0.8229259252548218,
"mask/share_step_conf": 0.14153030514717102,
"num_tokens": 12818755.0,
"reward": 0.9539192318916321,
"reward_std": 0.14013896882534027,
"rewards/accuracy_reward_step": 0.62109375,
"rewards/final_brier_reward_step": 0.7372496128082275,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8479325771331787,
"step": 53
},
{
"adv/mean_abs_final_conf": 0.6720455884933472,
"adv/mean_abs_reasoning": 0.2754554748535156,
"adv/mean_abs_step_conf": 0.762405276298523,
"adv/ratio_final_to_reasoning": 2.4397612312869588,
"adv/ratio_step_to_reasoning": 2.7677985950504786,
"adv/std_final_conf": 0.870083749294281,
"adv/std_reasoning": 0.5726470351219177,
"adv/std_step_conf": 0.9328941702842712,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.724006359300477,
"calib/avg_num_step_conf": 6.20703125,
"calib/ece": 0.16122529644268782,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.6363636363636364,
"calib/gap": 0.12781796502384746,
"calib/mean_conf": 0.8553754940711463,
"calib/mu_c": 0.8897297297297297,
"calib/mu_w": 0.7619117647058823,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.14268774703557321,
"calib/std_conf": 0.2091568744618195,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4453250222617987,
"calib/step_q_c_n": 1123.0,
"calib/step_q_gap": 0.03588510809870865,
"calib/step_q_w": 0.40943991416309006,
"calib/step_q_w_n": 466.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2739.0,
"completions/max_terminated_length": 2739.0,
"completions/mean_length": 482.38671875,
"completions/mean_terminated_length": 484.2784729003906,
"completions/min_length": 0.0,
"completions/min_terminated_length": 148.0,
"epoch": 0.0576,
"grad_norm": 0.042764466255903244,
"kl": 0.05710601806640625,
"learning_rate": 4.055555555555556e-06,
"loss": 0.01,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.036248303949832916,
"mask/share_reasoning": 0.8180440664291382,
"mask/share_step_conf": 0.14180132746696472,
"num_tokens": 13048478.0,
"reward": 0.9917140007019043,
"reward_std": 0.13152074813842773,
"rewards/accuracy_reward_step": 0.72265625,
"rewards/final_brier_reward_step": 0.7852355241775513,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8560049533843994,
"step": 54
},
{
"adv/mean_abs_final_conf": 0.7059681415557861,
"adv/mean_abs_reasoning": 0.44087427854537964,
"adv/mean_abs_step_conf": 0.7620362639427185,
"adv/ratio_final_to_reasoning": 1.601291288493983,
"adv/ratio_step_to_reasoning": 1.728466143357196,
"adv/std_final_conf": 0.8803775310516357,
"adv/std_reasoning": 0.7205796241760254,
"adv/std_step_conf": 0.9334477782249451,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.8048804012345679,
"calib/avg_num_step_conf": 6.890625,
"calib/ece": 0.26468253968253963,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.5833333333333334,
"calib/gap": 0.22023148148148142,
"calib/mean_conf": 0.8249206349206349,
"calib/mu_c": 0.9193055555555555,
"calib/mu_w": 0.6990740740740741,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.25908730158730153,
"calib/std_conf": 0.22471762185952407,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4419675925925926,
"calib/step_q_c_n": 864.0,
"calib/step_q_gap": 0.07078981481481483,
"calib/step_q_w": 0.3711777777777778,
"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": 0.01171875,
"completions/max_length": 2216.0,
"completions/max_terminated_length": 2216.0,
"completions/mean_length": 498.51171875,
"completions/mean_terminated_length": 504.4229431152344,
"completions/min_length": 0.0,
"completions/min_terminated_length": 131.0,
"epoch": 0.058666666666666666,
"grad_norm": 0.0345403328537941,
"kl": 0.060577392578125,
"learning_rate": 4.027777777777779e-06,
"loss": -0.0871,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.034203819930553436,
"mask/share_reasoning": 0.8095616102218628,
"mask/share_step_conf": 0.14451584219932556,
"num_tokens": 13283921.0,
"reward": 0.9504541754722595,
"reward_std": 0.18151721358299255,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.7330425977706909,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8592719435691833,
"step": 55
},
{
"adv/mean_abs_final_conf": 0.7152382135391235,
"adv/mean_abs_reasoning": 0.5132660865783691,
"adv/mean_abs_step_conf": 0.7421176433563232,
"adv/ratio_final_to_reasoning": 1.3935037444363,
"adv/ratio_step_to_reasoning": 1.4458731304527976,
"adv/std_final_conf": 0.888146162033081,
"adv/std_reasoning": 0.7575376629829407,
"adv/std_step_conf": 0.9327020645141602,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.675488342785129,
"calib/avg_num_step_conf": 6.74609375,
"calib/ece": 0.36122529644268786,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.7154150197628458,
"calib/gap": 0.08611594202898532,
"calib/mean_conf": 0.8881027667984189,
"calib/mu_c": 0.9272463768115942,
"calib/mu_w": 0.8411304347826088,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.3519367588932808,
"calib/std_conf": 0.19404259065574303,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4460133630289532,
"calib/step_q_c_n": 898.0,
"calib/step_q_gap": 0.05667681296863958,
"calib/step_q_w": 0.38933655006031365,
"calib/step_q_w_n": 829.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2208.0,
"completions/max_terminated_length": 2208.0,
"completions/mean_length": 516.9765625,
"completions/mean_terminated_length": 519.0039672851562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 186.0,
"epoch": 0.05973333333333333,
"grad_norm": 0.030760738998651505,
"kl": 0.05310821533203125,
"learning_rate": 4.000000000000001e-06,
"loss": -0.0016,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.032516516745090485,
"mask/share_reasoning": 0.8238645792007446,
"mask/share_step_conf": 0.1397126466035843,
"num_tokens": 13523107.0,
"reward": 0.8982587456703186,
"reward_std": 0.1786113828420639,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.6322113275527954,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.858837366104126,
"step": 56
},
{
"adv/mean_abs_final_conf": 0.6410858035087585,
"adv/mean_abs_reasoning": 0.45202022790908813,
"adv/mean_abs_step_conf": 0.7799255847930908,
"adv/ratio_final_to_reasoning": 1.41826795334853,
"adv/ratio_step_to_reasoning": 1.7254218653019044,
"adv/std_final_conf": 0.8326322436332703,
"adv/std_reasoning": 0.7205796241760254,
"adv/std_step_conf": 0.9327417612075806,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.749898097826087,
"calib/avg_num_step_conf": 6.65234375,
"calib/ece": 0.2768650793650795,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.7579365079365079,
"calib/gap": 0.128470108695652,
"calib/mean_conf": 0.9117857142857142,
"calib/mu_c": 0.9586874999999999,
"calib/mu_w": 0.8302173913043479,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2768650793650795,
"calib/std_conf": 0.1650408370408865,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4636482412060301,
"calib/step_q_c_n": 995.0,
"calib/step_q_gap": 0.06506067058456111,
"calib/step_q_w": 0.398587570621469,
"calib/step_q_w_n": 708.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2541.0,
"completions/max_terminated_length": 2541.0,
"completions/mean_length": 530.0,
"completions/mean_terminated_length": 532.0784912109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 188.0,
"epoch": 0.0608,
"grad_norm": 0.03361690789461136,
"kl": 0.0550384521484375,
"learning_rate": 3.972222222222223e-06,
"loss": 0.0228,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.032974641770124435,
"mask/share_reasoning": 0.8260525465011597,
"mask/share_step_conf": 0.1370665282011032,
"num_tokens": 13765579.0,
"reward": 0.9451895356178284,
"reward_std": 0.18180344998836517,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.7125582098960876,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.855945885181427,
"step": 57
},
{
"adv/mean_abs_final_conf": 0.7059772610664368,
"adv/mean_abs_reasoning": 0.6071633696556091,
"adv/mean_abs_step_conf": 0.7528630495071411,
"adv/ratio_final_to_reasoning": 1.1627467932837847,
"adv/ratio_step_to_reasoning": 1.2399678358959216,
"adv/std_final_conf": 0.8767675757408142,
"adv/std_reasoning": 0.826632022857666,
"adv/std_step_conf": 0.9340095520019531,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5451073232323232,
"calib/avg_num_step_conf": 7.3515625,
"calib/ece": 0.39047619047619053,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.6865079365079365,
"calib/gap": 0.034446969696969476,
"calib/mean_conf": 0.8624603174603176,
"calib/mu_c": 0.8788636363636363,
"calib/mu_w": 0.8444166666666668,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.3645634920634921,
"calib/std_conf": 0.22655174435378,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4163152507676561,
"calib/step_q_c_n": 977.0,
"calib/step_q_gap": 0.020679891651633975,
"calib/step_q_w": 0.3956353591160221,
"calib/step_q_w_n": 905.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2937.0,
"completions/max_terminated_length": 2937.0,
"completions/mean_length": 581.890625,
"completions/mean_terminated_length": 584.172607421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 182.0,
"epoch": 0.06186666666666667,
"grad_norm": 0.027973540127277374,
"kl": 0.06337738037109375,
"learning_rate": 3.944444444444445e-06,
"loss": 0.0905,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.030530782416462898,
"mask/share_reasoning": 0.8286088705062866,
"mask/share_step_conf": 0.13695411384105682,
"num_tokens": 14020863.0,
"reward": 0.8600574731826782,
"reward_std": 0.22369416058063507,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.589409351348877,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8314868807792664,
"step": 58
},
{
"adv/mean_abs_final_conf": 0.59950852394104,
"adv/mean_abs_reasoning": 0.46238672733306885,
"adv/mean_abs_step_conf": 0.7642239332199097,
"adv/ratio_final_to_reasoning": 1.2965521899792745,
"adv/ratio_step_to_reasoning": 1.652780860791058,
"adv/std_final_conf": 0.8129385113716125,
"adv/std_reasoning": 0.7206915616989136,
"adv/std_step_conf": 0.9333196878433228,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.5510875402792696,
"calib/avg_num_step_conf": 6.86328125,
"calib/ece": 0.35996,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.816,
"calib/gap": 0.0012755102040816757,
"calib/mean_conf": 0.927,
"calib/mu_c": 0.9275000000000001,
"calib/mu_w": 0.9262244897959184,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.33948,
"calib/std_conf": 0.15880868993855468,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4582951146560319,
"calib/step_q_c_n": 1003.0,
"calib/step_q_gap": 0.032300419695819815,
"calib/step_q_w": 0.4259946949602121,
"calib/step_q_w_n": 754.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2506.0,
"completions/max_terminated_length": 2506.0,
"completions/mean_length": 536.48046875,
"completions/mean_terminated_length": 540.7047119140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 135.0,
"epoch": 0.06293333333333333,
"grad_norm": 0.04841792583465576,
"kl": 0.07193756103515625,
"learning_rate": 3.916666666666667e-06,
"loss": 0.0297,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03387182950973511,
"mask/share_reasoning": 0.8236122727394104,
"mask/share_step_conf": 0.1347033530473709,
"num_tokens": 14264450.0,
"reward": 0.8812941312789917,
"reward_std": 0.1919928938150406,
"rewards/accuracy_reward_step": 0.59375,
"rewards/final_brier_reward_step": 0.62040114402771,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8281245231628418,
"step": 59
},
{
"adv/mean_abs_final_conf": 0.6542816162109375,
"adv/mean_abs_reasoning": 0.4994843602180481,
"adv/mean_abs_step_conf": 0.7410473227500916,
"adv/ratio_final_to_reasoning": 1.3099141200843871,
"adv/ratio_step_to_reasoning": 1.4836246773103967,
"adv/std_final_conf": 0.8408299684524536,
"adv/std_reasoning": 0.7753770351409912,
"adv/std_step_conf": 0.9341368079185486,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6835099508917032,
"calib/avg_num_step_conf": 6.91015625,
"calib/ece": 0.3258734126984126,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.7182539682539683,
"calib/gap": 0.10675041354355141,
"calib/mean_conf": 0.8592059523809524,
"calib/mu_c": 0.9041089041095892,
"calib/mu_w": 0.7973584905660378,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.30285714285714277,
"calib/std_conf": 0.24918923389658307,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.47209534368070943,
"calib/step_q_c_n": 902.0,
"calib/step_q_gap": 0.06223375198520775,
"calib/step_q_w": 0.4098615916955017,
"calib/step_q_w_n": 867.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2882.0,
"completions/max_terminated_length": 2882.0,
"completions/mean_length": 524.90625,
"completions/mean_terminated_length": 526.9647216796875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 180.0,
"epoch": 0.064,
"grad_norm": 0.06499022990465164,
"kl": 0.062458038330078125,
"learning_rate": 3.88888888888889e-06,
"loss": 0.0025,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03385772183537483,
"mask/share_reasoning": 0.8223803043365479,
"mask/share_step_conf": 0.13985571265220642,
"num_tokens": 14507682.0,
"reward": 0.908770740032196,
"reward_std": 0.21273939311504364,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.6553382873535156,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8528280854225159,
"step": 60
},
{
"adv/mean_abs_final_conf": 0.624822735786438,
"adv/mean_abs_reasoning": 0.48236283659935,
"adv/mean_abs_step_conf": 0.7643167972564697,
"adv/ratio_final_to_reasoning": 1.2953376346142833,
"adv/ratio_step_to_reasoning": 1.5845267074157092,
"adv/std_final_conf": 0.837339460849762,
"adv/std_reasoning": 0.7392899990081787,
"adv/std_step_conf": 0.9319630861282349,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.5754061889031215,
"calib/avg_num_step_conf": 6.05078125,
"calib/ece": 0.29094488188976375,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.7677165354330708,
"calib/gap": 0.07063574462347466,
"calib/mean_conf": 0.9023622047244094,
"calib/mu_c": 0.9276687116564416,
"calib/mu_w": 0.8570329670329669,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.27578740157480314,
"calib/std_conf": 0.19318285855435954,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5004536082474226,
"calib/step_q_c_n": 970.0,
"calib/step_q_gap": 0.0710926410626212,
"calib/step_q_w": 0.4293609671848014,
"calib/step_q_w_n": 579.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1334.0,
"completions/max_terminated_length": 1334.0,
"completions/mean_length": 427.23046875,
"completions/mean_terminated_length": 428.9059143066406,
"completions/min_length": 0.0,
"completions/min_terminated_length": 124.0,
"epoch": 0.06506666666666666,
"grad_norm": 0.03414275497198105,
"kl": 0.07006072998046875,
"learning_rate": 3.861111111111112e-06,
"loss": -0.0542,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.04184265434741974,
"mask/share_reasoning": 0.8065690994262695,
"mask/share_step_conf": 0.14768198132514954,
"num_tokens": 14721117.0,
"reward": 0.9312187433242798,
"reward_std": 0.1829134076833725,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.6918726563453674,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8440022468566895,
"step": 61
},
{
"adv/mean_abs_final_conf": 0.7021551132202148,
"adv/mean_abs_reasoning": 0.6482760906219482,
"adv/mean_abs_step_conf": 0.7477187514305115,
"adv/ratio_final_to_reasoning": 1.0831112289619933,
"adv/ratio_step_to_reasoning": 1.1533955397200584,
"adv/std_final_conf": 0.9061055183410645,
"adv/std_reasoning": 0.8589903712272644,
"adv/std_step_conf": 0.9343938827514648,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.5849002849002849,
"calib/avg_num_step_conf": 6.49609375,
"calib/ece": 0.2740079365079365,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.5317460317460317,
"calib/gap": 0.07605128205128187,
"calib/mean_conf": 0.7631349206349206,
"calib/mu_c": 0.7984444444444444,
"calib/mu_w": 0.7223931623931625,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.25071428571428567,
"calib/std_conf": 0.2770466654054787,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.48075178997613366,
"calib/step_q_c_n": 838.0,
"calib/step_q_gap": 0.04790330512764879,
"calib/step_q_w": 0.43284848484848487,
"calib/step_q_w_n": 825.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3048.0,
"completions/max_terminated_length": 3048.0,
"completions/mean_length": 534.63671875,
"completions/mean_terminated_length": 536.7333374023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.06613333333333334,
"grad_norm": 0.0399576835334301,
"kl": 0.06037139892578125,
"learning_rate": 3.833333333333334e-06,
"loss": 0.021,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.033392585813999176,
"mask/share_reasoning": 0.8314791321754456,
"mask/share_step_conf": 0.13122203946113586,
"num_tokens": 14965064.0,
"reward": 0.8811693787574768,
"reward_std": 0.1998165249824524,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.6458152532577515,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8165234327316284,
"step": 62
},
{
"adv/mean_abs_final_conf": 0.7243236899375916,
"adv/mean_abs_reasoning": 0.5651422142982483,
"adv/mean_abs_step_conf": 0.777940571308136,
"adv/ratio_final_to_reasoning": 1.2816662277423445,
"adv/ratio_step_to_reasoning": 1.3765394826753916,
"adv/std_final_conf": 0.9065544009208679,
"adv/std_reasoning": 0.77553391456604,
"adv/std_step_conf": 0.9342755079269409,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.7336436170212766,
"calib/avg_num_step_conf": 5.91015625,
"calib/ece": 0.10771653543307089,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.41338582677165353,
"calib/gap": 0.2497606382978722,
"calib/mean_conf": 0.6888188976377954,
"calib/mu_c": 0.78125,
"calib/mu_w": 0.5314893617021278,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.953125,
"calib/pce": 0.08330708661417326,
"calib/std_conf": 0.28815354112855623,
"calib/step_conf_rate": 0.953125,
"calib/step_q_c": 0.48752183406113536,
"calib/step_q_c_n": 916.0,
"calib/step_q_gap": 0.060670912788103604,
"calib/step_q_w": 0.42685092127303176,
"calib/step_q_w_n": 597.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2383.0,
"completions/max_terminated_length": 2383.0,
"completions/mean_length": 542.08984375,
"completions/mean_terminated_length": 544.2156982421875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 137.0,
"epoch": 0.0672,
"grad_norm": 0.04744715243577957,
"kl": 0.0615386962890625,
"learning_rate": 3.8055555555555556e-06,
"loss": -0.077,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.032869912683963776,
"mask/share_reasoning": 0.8419344425201416,
"mask/share_step_conf": 0.12128940969705582,
"num_tokens": 15212479.0,
"reward": 0.9379910230636597,
"reward_std": 0.19364088773727417,
"rewards/accuracy_reward_step": 0.625,
"rewards/final_brier_reward_step": 0.7451125383377075,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.8175883293151855,
"step": 63
},
{
"adv/mean_abs_final_conf": 0.7251790761947632,
"adv/mean_abs_reasoning": 0.5222870111465454,
"adv/mean_abs_step_conf": 0.7692461609840393,
"adv/ratio_final_to_reasoning": 1.388468525385728,
"adv/ratio_step_to_reasoning": 1.4728418370875418,
"adv/std_final_conf": 0.9209416508674622,
"adv/std_reasoning": 0.7927463054656982,
"adv/std_step_conf": 0.9337198734283447,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6286424768842813,
"calib/avg_num_step_conf": 6.71875,
"calib/ece": 0.1541269841269841,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.38095238095238093,
"calib/gap": 0.1363519193051277,
"calib/mean_conf": 0.6988888888888889,
"calib/mu_c": 0.745421686746988,
"calib/mu_w": 0.6090697674418604,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.09714285714285711,
"calib/std_conf": 0.2715350900245463,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4557664233576642,
"calib/step_q_c_n": 1096.0,
"calib/step_q_gap": 0.04341065412689499,
"calib/step_q_w": 0.4123557692307692,
"calib/step_q_w_n": 624.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2289.0,
"completions/max_terminated_length": 2289.0,
"completions/mean_length": 512.3125,
"completions/mean_terminated_length": 512.3125,
"completions/min_length": 155.0,
"completions/min_terminated_length": 155.0,
"epoch": 0.06826666666666667,
"grad_norm": 0.12423020601272583,
"kl": 0.06578826904296875,
"learning_rate": 3.777777777777778e-06,
"loss": 0.0643,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.035044483840465546,
"mask/share_reasoning": 0.8245336413383484,
"mask/share_step_conf": 0.14042183756828308,
"num_tokens": 15447407.0,
"reward": 0.9516028165817261,
"reward_std": 0.15794092416763306,
"rewards/accuracy_reward_step": 0.6484375,
"rewards/final_brier_reward_step": 0.7492632865905762,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8273798823356628,
"step": 64
},
{
"adv/mean_abs_final_conf": 0.5918190479278564,
"adv/mean_abs_reasoning": 0.4216136932373047,
"adv/mean_abs_step_conf": 0.75426185131073,
"adv/ratio_final_to_reasoning": 1.4036997787800785,
"adv/ratio_step_to_reasoning": 1.7889880319570046,
"adv/std_final_conf": 0.8399807810783386,
"adv/std_reasoning": 0.7204056978225708,
"adv/std_step_conf": 0.9331092238426208,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.6555228758169934,
"calib/avg_num_step_conf": 5.9921875,
"calib/ece": 0.275494071146245,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.7114624505928854,
"calib/gap": 0.14319999999999977,
"calib/mean_conf": 0.863399209486166,
"calib/mu_c": 0.9199999999999999,
"calib/mu_w": 0.7768000000000002,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2670750988142292,
"calib/std_conf": 0.2268313845027238,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.5217622080679406,
"calib/step_q_c_n": 942.0,
"calib/step_q_gap": 0.04120477563550817,
"calib/step_q_w": 0.48055743243243243,
"calib/step_q_w_n": 592.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2184.0,
"completions/max_terminated_length": 2184.0,
"completions/mean_length": 453.61328125,
"completions/mean_terminated_length": 455.3921813964844,
"completions/min_length": 0.0,
"completions/min_terminated_length": 143.0,
"epoch": 0.06933333333333333,
"grad_norm": 0.04314102977514267,
"kl": 0.0760955810546875,
"learning_rate": 3.7500000000000005e-06,
"loss": -0.0123,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.038080520927906036,
"mask/share_reasoning": 0.8146101236343384,
"mask/share_step_conf": 0.1434030830860138,
"num_tokens": 15668556.0,
"reward": 0.9371784925460815,
"reward_std": 0.15816906094551086,
"rewards/accuracy_reward_step": 0.6015625,
"rewards/final_brier_reward_step": 0.7027406096458435,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8536475896835327,
"step": 65
},
{
"adv/mean_abs_final_conf": 0.6618846654891968,
"adv/mean_abs_reasoning": 0.5083737969398499,
"adv/mean_abs_step_conf": 0.7653838396072388,
"adv/ratio_final_to_reasoning": 1.301964557326526,
"adv/ratio_step_to_reasoning": 1.5055532842456827,
"adv/std_final_conf": 0.8603252172470093,
"adv/std_reasoning": 0.7575896382331848,
"adv/std_step_conf": 0.9338029026985168,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7378846275352247,
"calib/avg_num_step_conf": 7.546875,
"calib/ece": 0.28428571428571425,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.5277777777777778,
"calib/gap": 0.261530296329058,
"calib/mean_conf": 0.7565079365079366,
"calib/mu_c": 0.8945378151260505,
"calib/mu_w": 0.6330075187969925,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.28428571428571425,
"calib/std_conf": 0.2847076402438163,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.4879167733674776,
"calib/step_q_c_n": 781.0,
"calib/step_q_gap": 0.08008010959684331,
"calib/step_q_w": 0.40783666377063427,
"calib/step_q_w_n": 1151.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2759.0,
"completions/max_terminated_length": 2759.0,
"completions/mean_length": 596.5625,
"completions/mean_terminated_length": 598.9019775390625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 136.0,
"epoch": 0.0704,
"grad_norm": 0.05358945205807686,
"kl": 0.06067657470703125,
"learning_rate": 3.7222222222222225e-06,
"loss": 0.0073,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03181684762239456,
"mask/share_reasoning": 0.831584095954895,
"mask/share_step_conf": 0.13269281387329102,
"num_tokens": 15927628.0,
"reward": 0.9063929319381714,
"reward_std": 0.20524170994758606,
"rewards/accuracy_reward_step": 0.46484375,
"rewards/final_brier_reward_step": 0.6950390934944153,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8310280442237854,
"step": 66
},
{
"adv/mean_abs_final_conf": 0.5081011056900024,
"adv/mean_abs_reasoning": 0.35968446731567383,
"adv/mean_abs_step_conf": 0.7419739365577698,
"adv/ratio_final_to_reasoning": 1.4126301018277558,
"adv/ratio_step_to_reasoning": 2.0628467559222763,
"adv/std_final_conf": 0.7767390608787537,
"adv/std_reasoning": 0.6813791990280151,
"adv/std_step_conf": 0.9321821928024292,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7599078341013825,
"calib/avg_num_step_conf": 6.23046875,
"calib/ece": 0.23509881422924903,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.6719367588932806,
"calib/gap": 0.22916919025674787,
"calib/mean_conf": 0.8477470355731226,
"calib/mu_c": 0.936516129032258,
"calib/mu_w": 0.7073469387755101,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.23509881422924903,
"calib/std_conf": 0.23716887306077852,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.4979411764705883,
"calib/step_q_c_n": 918.0,
"calib/step_q_gap": 0.0674094187157877,
"calib/step_q_w": 0.43053175775480057,
"calib/step_q_w_n": 677.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2449.0,
"completions/max_terminated_length": 2449.0,
"completions/mean_length": 525.47265625,
"completions/mean_terminated_length": 527.5333862304688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 200.0,
"epoch": 0.07146666666666666,
"grad_norm": 0.07478371262550354,
"kl": 0.070709228515625,
"learning_rate": 3.694444444444445e-06,
"loss": -0.0358,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03388848155736923,
"mask/share_reasoning": 0.8310776352882385,
"mask/share_step_conf": 0.13112762570381165,
"num_tokens": 16167157.0,
"reward": 0.9643101096153259,
"reward_std": 0.14082594215869904,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.7473699450492859,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8632815480232239,
"step": 67
},
{
"adv/mean_abs_final_conf": 0.6216704845428467,
"adv/mean_abs_reasoning": 0.4290942847728729,
"adv/mean_abs_step_conf": 0.7697738409042358,
"adv/ratio_final_to_reasoning": 1.448796934855256,
"adv/ratio_step_to_reasoning": 1.7939503466276427,
"adv/std_final_conf": 0.8364220857620239,
"adv/std_reasoning": 0.7013837695121765,
"adv/std_step_conf": 0.9322218894958496,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7144844093673178,
"calib/avg_num_step_conf": 6.36328125,
"calib/ece": 0.3388353413654618,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.714859437751004,
"calib/gap": 0.19704942424634486,
"calib/mean_conf": 0.8649397590361446,
"calib/mu_c": 0.9583206106870228,
"calib/mu_w": 0.7612711864406779,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.3388353413654618,
"calib/std_conf": 0.23456380175533872,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.4838215712383489,
"calib/step_q_c_n": 751.0,
"calib/step_q_gap": 0.09090585369848553,
"calib/step_q_w": 0.3929157175398634,
"calib/step_q_w_n": 878.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2674.0,
"completions/max_terminated_length": 2674.0,
"completions/mean_length": 518.23828125,
"completions/mean_terminated_length": 518.23828125,
"completions/min_length": 141.0,
"completions/min_terminated_length": 141.0,
"epoch": 0.07253333333333334,
"grad_norm": 0.04946654289960861,
"kl": 0.06878662109375,
"learning_rate": 3.6666666666666666e-06,
"loss": 0.0202,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03565322607755661,
"mask/share_reasoning": 0.8229528665542603,
"mask/share_step_conf": 0.14139388501644135,
"num_tokens": 16403914.0,
"reward": 0.9006872177124023,
"reward_std": 0.1842219978570938,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.6572574377059937,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8480231761932373,
"step": 68
},
{
"adv/mean_abs_final_conf": 0.7201113700866699,
"adv/mean_abs_reasoning": 0.5343183279037476,
"adv/mean_abs_step_conf": 0.7530443668365479,
"adv/ratio_final_to_reasoning": 1.3477197626962016,
"adv/ratio_step_to_reasoning": 1.4093552990984088,
"adv/std_final_conf": 0.9012336134910583,
"adv/std_reasoning": 0.7576988339424133,
"adv/std_step_conf": 0.9333232045173645,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.714098237720285,
"calib/avg_num_step_conf": 6.6953125,
"calib/ece": 0.18699604743083,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.36363636363636365,
"calib/gap": 0.2355143107111609,
"calib/mean_conf": 0.662094861660079,
"calib/mu_c": 0.7803174603174603,
"calib/mu_w": 0.5448031496062994,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.17553359683794462,
"calib/std_conf": 0.2990141226542987,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.4238228155339806,
"calib/step_q_c_n": 824.0,
"calib/step_q_gap": 0.04873292789353112,
"calib/step_q_w": 0.3750898876404495,
"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": 0.0,
"completions/max_length": 2613.0,
"completions/max_terminated_length": 2613.0,
"completions/mean_length": 613.66796875,
"completions/mean_terminated_length": 613.66796875,
"completions/min_length": 171.0,
"completions/min_terminated_length": 171.0,
"epoch": 0.0736,
"grad_norm": 0.060156576335430145,
"kl": 0.06504058837890625,
"learning_rate": 3.638888888888889e-06,
"loss": -0.0109,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.030466951429843903,
"mask/share_reasoning": 0.8466631770133972,
"mask/share_step_conf": 0.12286990880966187,
"num_tokens": 16665509.0,
"reward": 0.934023916721344,
"reward_std": 0.16452187299728394,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/final_brier_reward_step": 0.7315406203269958,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8435383439064026,
"step": 69
},
{
"adv/mean_abs_final_conf": 0.6430877447128296,
"adv/mean_abs_reasoning": 0.41097357869148254,
"adv/mean_abs_step_conf": 0.7237237095832825,
"adv/ratio_final_to_reasoning": 1.564790969678357,
"adv/ratio_step_to_reasoning": 1.7609981446680327,
"adv/std_final_conf": 0.8517022728919983,
"adv/std_reasoning": 0.7013329863548279,
"adv/std_step_conf": 0.9332301616668701,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.8866443643849717,
"calib/avg_num_step_conf": 7.02734375,
"calib/ece": 0.13728,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.412,
"calib/gap": 0.5055249613998971,
"calib/mean_conf": 0.6012799999999999,
"calib/mu_c": 0.8722413793103448,
"calib/mu_w": 0.36671641791044773,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.13728,
"calib/std_conf": 0.3597743203732028,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.4723941605839415,
"calib/step_q_c_n": 685.0,
"calib/step_q_gap": 0.13617333473115872,
"calib/step_q_w": 0.3362208258527828,
"calib/step_q_w_n": 1114.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2651.0,
"completions/max_terminated_length": 2651.0,
"completions/mean_length": 593.4765625,
"completions/mean_terminated_length": 598.1495971679688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 131.0,
"epoch": 0.07466666666666667,
"grad_norm": 0.05660438910126686,
"kl": 0.10711669921875,
"learning_rate": 3.6111111111111115e-06,
"loss": -0.028,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03215644136071205,
"mask/share_reasoning": 0.8270474672317505,
"mask/share_step_conf": 0.13298359513282776,
"num_tokens": 16924431.0,
"reward": 0.9864169955253601,
"reward_std": 0.1601521074771881,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.8269773721694946,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8614815473556519,
"step": 70
},
{
"adv/mean_abs_final_conf": 0.6957278251647949,
"adv/mean_abs_reasoning": 0.547110915184021,
"adv/mean_abs_step_conf": 0.7761636972427368,
"adv/ratio_final_to_reasoning": 1.2716394534566844,
"adv/ratio_step_to_reasoning": 1.4186587686368382,
"adv/std_final_conf": 0.8768286108970642,
"adv/std_reasoning": 0.7576953768730164,
"adv/std_step_conf": 0.9334019422531128,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6453033268101761,
"calib/avg_num_step_conf": 7.08984375,
"calib/ece": 0.23900398406374493,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.3705179282868526,
"calib/gap": 0.17135746901500337,
"calib/mean_conf": 0.5838645418326693,
"calib/mu_c": 0.6555479452054795,
"calib/mu_w": 0.48419047619047617,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.12059760956175292,
"calib/std_conf": 0.3530129360880522,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4009497816593886,
"calib/step_q_c_n": 916.0,
"calib/step_q_gap": 0.07640028221556217,
"calib/step_q_w": 0.32454949944382644,
"calib/step_q_w_n": 899.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2621.0,
"completions/max_terminated_length": 2621.0,
"completions/mean_length": 533.2734375,
"completions/mean_terminated_length": 537.472412109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 172.0,
"epoch": 0.07573333333333333,
"grad_norm": 0.08311379700899124,
"kl": 0.07154083251953125,
"learning_rate": 3.5833333333333335e-06,
"loss": 0.0345,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03394794836640358,
"mask/share_reasoning": 0.8204426765441895,
"mask/share_step_conf": 0.13779692351818085,
"num_tokens": 17165357.0,
"reward": 0.9159102439880371,
"reward_std": 0.17106956243515015,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.6937718391418457,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8302360773086548,
"step": 71
},
{
"adv/mean_abs_final_conf": 0.7265099287033081,
"adv/mean_abs_reasoning": 0.38411909341812134,
"adv/mean_abs_step_conf": 0.7782511711120605,
"adv/ratio_final_to_reasoning": 1.8913663526548197,
"adv/ratio_step_to_reasoning": 2.0260673953661517,
"adv/std_final_conf": 0.8927708268165588,
"adv/std_reasoning": 0.6612346768379211,
"adv/std_step_conf": 0.9314785003662109,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.7946273291925465,
"calib/avg_num_step_conf": 6.46875,
"calib/ece": 0.13019607843137251,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.2627450980392157,
"calib/gap": 0.36564285714285705,
"calib/mean_conf": 0.4747450980392157,
"calib/mu_c": 0.6396428571428571,
"calib/mu_w": 0.274,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.02796078431372547,
"calib/std_conf": 0.3526024169468427,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.40148357870894674,
"calib/step_q_c_n": 883.0,
"calib/step_q_gap": 0.0988315735343025,
"calib/step_q_w": 0.30265200517464425,
"calib/step_q_w_n": 773.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 1670.0,
"completions/max_terminated_length": 1670.0,
"completions/mean_length": 487.4609375,
"completions/mean_terminated_length": 487.4609375,
"completions/min_length": 173.0,
"completions/min_terminated_length": 173.0,
"epoch": 0.0768,
"grad_norm": 0.06640713661909103,
"kl": 0.08196258544921875,
"learning_rate": 3.555555555555556e-06,
"loss": 0.0169,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.034188974648714066,
"mask/share_reasoning": 0.8254907727241516,
"mask/share_step_conf": 0.14032024145126343,
"num_tokens": 17394555.0,
"reward": 0.9930468797683716,
"reward_std": 0.1217828020453453,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.7926070690155029,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8872367143630981,
"step": 72
},
{
"adv/mean_abs_final_conf": 0.7764418721199036,
"adv/mean_abs_reasoning": 0.5932731628417969,
"adv/mean_abs_step_conf": 0.7526953220367432,
"adv/ratio_final_to_reasoning": 1.3087426176514085,
"adv/ratio_step_to_reasoning": 1.2687162831221106,
"adv/std_final_conf": 0.9236984252929688,
"adv/std_reasoning": 0.7929815053939819,
"adv/std_step_conf": 0.9335096478462219,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.7226810456624121,
"calib/avg_num_step_conf": 6.09375,
"calib/ece": 0.26289682539682546,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.15079365079365079,
"calib/gap": 0.22959866220735775,
"calib/mean_conf": 0.40361111111111114,
"calib/mu_c": 0.48652173913043467,
"calib/mu_w": 0.2569230769230769,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.013809523809523827,
"calib/std_conf": 0.30786395346775686,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.39301063829787236,
"calib/step_q_c_n": 940.0,
"calib/step_q_gap": 0.08380096087851752,
"calib/step_q_w": 0.30920967741935484,
"calib/step_q_w_n": 620.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2520.0,
"completions/max_terminated_length": 2520.0,
"completions/mean_length": 502.7265625,
"completions/mean_terminated_length": 504.69805908203125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 166.0,
"epoch": 0.07786666666666667,
"grad_norm": 0.10404765605926514,
"kl": 0.09230804443359375,
"learning_rate": 3.5277777777777784e-06,
"loss": -0.0109,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03293906897306442,
"mask/share_reasoning": 0.8376255631446838,
"mask/share_step_conf": 0.12552911043167114,
"num_tokens": 17630285.0,
"reward": 0.9415270090103149,
"reward_std": 0.15380805730819702,
"rewards/accuracy_reward_step": 0.6328125,
"rewards/final_brier_reward_step": 0.7067304849624634,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8544486165046692,
"step": 73
},
{
"adv/mean_abs_final_conf": 0.73908931016922,
"adv/mean_abs_reasoning": 0.44497135281562805,
"adv/mean_abs_step_conf": 0.7388650178909302,
"adv/ratio_final_to_reasoning": 1.660981781169761,
"adv/ratio_step_to_reasoning": 1.6604777211288828,
"adv/std_final_conf": 0.9192122220993042,
"adv/std_reasoning": 0.7205416560173035,
"adv/std_step_conf": 0.9333183765411377,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6947535256824946,
"calib/avg_num_step_conf": 6.47265625,
"calib/ece": 0.18489959839357437,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.15261044176706828,
"calib/gap": 0.2503215163669299,
"calib/mean_conf": 0.3292369477911647,
"calib/mu_c": 0.46093220338983065,
"calib/mu_w": 0.21061068702290076,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.02012048192771087,
"calib/std_conf": 0.3090207036128271,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.37147937411095305,
"calib/step_q_c_n": 703.0,
"calib/step_q_gap": 0.07340809528495723,
"calib/step_q_w": 0.2980712788259958,
"calib/step_q_w_n": 954.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2502.0,
"completions/max_terminated_length": 2502.0,
"completions/mean_length": 530.578125,
"completions/mean_terminated_length": 530.578125,
"completions/min_length": 125.0,
"completions/min_terminated_length": 125.0,
"epoch": 0.07893333333333333,
"grad_norm": 0.09156779944896698,
"kl": 0.091705322265625,
"learning_rate": 3.5e-06,
"loss": 0.0196,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03372114151716232,
"mask/share_reasoning": 0.8263232111930847,
"mask/share_step_conf": 0.13995562493801117,
"num_tokens": 17870041.0,
"reward": 0.9400572776794434,
"reward_std": 0.145626500248909,
"rewards/accuracy_reward_step": 0.4609375,
"rewards/final_brier_reward_step": 0.7383249998092651,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8550707697868347,
"step": 74
},
{
"adv/mean_abs_final_conf": 0.6418685913085938,
"adv/mean_abs_reasoning": 0.4402565062046051,
"adv/mean_abs_step_conf": 0.7393307685852051,
"adv/ratio_final_to_reasoning": 1.4579423183136135,
"adv/ratio_step_to_reasoning": 1.6793182114646774,
"adv/std_final_conf": 0.8484848737716675,
"adv/std_reasoning": 0.7205449342727661,
"adv/std_step_conf": 0.9325771927833557,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.8336708860759493,
"calib/avg_num_step_conf": 6.04296875,
"calib/ece": 0.137244094488189,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.42913385826771655,
"calib/gap": 0.41448245931283917,
"calib/mean_conf": 0.6235433070866142,
"calib/mu_c": 0.7524571428571429,
"calib/mu_w": 0.33797468354430377,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.035905511811023666,
"calib/std_conf": 0.3644130812605217,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.43970727101038715,
"calib/step_q_c_n": 1059.0,
"calib/step_q_gap": 0.08626464805956746,
"calib/step_q_w": 0.3534426229508197,
"calib/step_q_w_n": 488.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2525.0,
"completions/max_terminated_length": 2525.0,
"completions/mean_length": 474.8359375,
"completions/mean_terminated_length": 474.8359375,
"completions/min_length": 154.0,
"completions/min_terminated_length": 154.0,
"epoch": 0.08,
"grad_norm": 0.0735108032822609,
"kl": 0.07817840576171875,
"learning_rate": 3.4722222222222224e-06,
"loss": 0.0596,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03609136492013931,
"mask/share_reasoning": 0.822197437286377,
"mask/share_step_conf": 0.14171117544174194,
"num_tokens": 18096351.0,
"reward": 1.008620262145996,
"reward_std": 0.12839573621749878,
"rewards/accuracy_reward_step": 0.68359375,
"rewards/final_brier_reward_step": 0.8198156356811523,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8622686862945557,
"step": 75
},
{
"adv/mean_abs_final_conf": 0.6263427734375,
"adv/mean_abs_reasoning": 0.41424667835235596,
"adv/mean_abs_step_conf": 0.7314493060112,
"adv/ratio_final_to_reasoning": 1.5120043350226606,
"adv/ratio_step_to_reasoning": 1.7657336660380731,
"adv/std_final_conf": 0.8453238010406494,
"adv/std_reasoning": 0.6816251873970032,
"adv/std_step_conf": 0.9329613447189331,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7849274767378216,
"calib/avg_num_step_conf": 5.84375,
"calib/ece": 0.18490196078431378,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.42745098039215684,
"calib/gap": 0.33634236453201966,
"calib/mean_conf": 0.6125098039215686,
"calib/mu_c": 0.7272619047619048,
"calib/mu_w": 0.3909195402298851,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.0692941176470589,
"calib/std_conf": 0.36580823072832713,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.44701639344262295,
"calib/step_q_c_n": 915.0,
"calib/step_q_gap": 0.09937439688496369,
"calib/step_q_w": 0.34764199655765926,
"calib/step_q_w_n": 581.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1987.0,
"completions/max_terminated_length": 1987.0,
"completions/mean_length": 484.93359375,
"completions/mean_terminated_length": 486.8353271484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 154.0,
"epoch": 0.08106666666666666,
"grad_norm": 0.056952860206365585,
"kl": 0.0942840576171875,
"learning_rate": 3.444444444444445e-06,
"loss": -0.0246,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03601346164941788,
"mask/share_reasoning": 0.8301782608032227,
"mask/share_step_conf": 0.12990202009677887,
"num_tokens": 18323550.0,
"reward": 0.997848391532898,
"reward_std": 0.1068374365568161,
"rewards/accuracy_reward_step": 0.65625,
"rewards/final_brier_reward_step": 0.7835609316825867,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8824483156204224,
"step": 76
},
{
"adv/mean_abs_final_conf": 0.7339452505111694,
"adv/mean_abs_reasoning": 0.4712091088294983,
"adv/mean_abs_step_conf": 0.7212806940078735,
"adv/ratio_final_to_reasoning": 1.5575786561815792,
"adv/ratio_step_to_reasoning": 1.5307019335843544,
"adv/std_final_conf": 0.9045832753181458,
"adv/std_reasoning": 0.7392547726631165,
"adv/std_step_conf": 0.9330065846443176,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6528925619834711,
"calib/avg_num_step_conf": 6.3671875,
"calib/ece": 0.2300395256916996,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.47035573122529645,
"calib/gap": 0.15879545454545452,
"calib/mean_conf": 0.7045849802371542,
"calib/mu_c": 0.7598181818181818,
"calib/mu_w": 0.6010227272727273,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.14122529644268775,
"calib/std_conf": 0.32628633502282633,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4560182370820669,
"calib/step_q_c_n": 987.0,
"calib/step_q_gap": 0.08223907689544169,
"calib/step_q_w": 0.3737791601866252,
"calib/step_q_w_n": 643.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 3025.0,
"completions/max_terminated_length": 3025.0,
"completions/mean_length": 508.41015625,
"completions/mean_terminated_length": 508.41015625,
"completions/min_length": 139.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.08213333333333334,
"grad_norm": 0.09080254286527634,
"kl": 0.0932464599609375,
"learning_rate": 3.416666666666667e-06,
"loss": 0.0611,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03744114935398102,
"mask/share_reasoning": 0.8208218216896057,
"mask/share_step_conf": 0.14173701405525208,
"num_tokens": 18558367.0,
"reward": 0.9562437534332275,
"reward_std": 0.17072513699531555,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.7273656129837036,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8593405485153198,
"step": 77
},
{
"adv/mean_abs_final_conf": 0.680761456489563,
"adv/mean_abs_reasoning": 0.49786466360092163,
"adv/mean_abs_step_conf": 0.7462135553359985,
"adv/ratio_final_to_reasoning": 1.3673624706879133,
"adv/ratio_step_to_reasoning": 1.4988281151324057,
"adv/std_final_conf": 0.874717116355896,
"adv/std_reasoning": 0.7575218081474304,
"adv/std_step_conf": 0.9335688352584839,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6907335907335908,
"calib/avg_num_step_conf": 6.72265625,
"calib/ece": 0.23426877470355728,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.6284584980237155,
"calib/gap": 0.24018854568854564,
"calib/mean_conf": 0.8066007905138339,
"calib/mu_c": 0.9062837837837837,
"calib/mu_w": 0.6660952380952381,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.2279446640316205,
"calib/std_conf": 0.292426651590856,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5035102925243771,
"calib/step_q_c_n": 923.0,
"calib/step_q_gap": 0.09295891407826179,
"calib/step_q_w": 0.4105513784461153,
"calib/step_q_w_n": 798.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2771.0,
"completions/max_terminated_length": 2771.0,
"completions/mean_length": 577.2109375,
"completions/mean_terminated_length": 577.2109375,
"completions/min_length": 200.0,
"completions/min_terminated_length": 200.0,
"epoch": 0.0832,
"grad_norm": 0.06496492773294449,
"kl": 0.0711212158203125,
"learning_rate": 3.3888888888888893e-06,
"loss": 0.0643,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.029832325875759125,
"mask/share_reasoning": 0.8439598083496094,
"mask/share_step_conf": 0.1262078881263733,
"num_tokens": 18814157.0,
"reward": 0.9481760859489441,
"reward_std": 0.20212724804878235,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.7228991985321045,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.86173415184021,
"step": 78
},
{
"adv/mean_abs_final_conf": 0.6379029750823975,
"adv/mean_abs_reasoning": 0.48745274543762207,
"adv/mean_abs_step_conf": 0.7569434642791748,
"adv/ratio_final_to_reasoning": 1.308645773468165,
"adv/ratio_step_to_reasoning": 1.5528550641347012,
"adv/std_final_conf": 0.8597071170806885,
"adv/std_reasoning": 0.7575080394744873,
"adv/std_step_conf": 0.9334502816200256,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.5977476864452912,
"calib/avg_num_step_conf": 6.98828125,
"calib/ece": 0.2859215686274509,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.9921875,
"calib/frac_conf_gt_0.9": 0.7764705882352941,
"calib/gap": 0.08455906369080002,
"calib/mean_conf": 0.8921960784313725,
"calib/mu_c": 0.921377245508982,
"calib/mu_w": 0.836818181818182,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.2616078431372548,
"calib/std_conf": 0.2300858339077255,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.504735494880546,
"calib/step_q_c_n": 1172.0,
"calib/step_q_gap": 0.059743598608260895,
"calib/step_q_w": 0.44499189627228514,
"calib/step_q_w_n": 617.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 1947.0,
"completions/max_terminated_length": 1947.0,
"completions/mean_length": 558.203125,
"completions/mean_terminated_length": 558.203125,
"completions/min_length": 164.0,
"completions/min_terminated_length": 164.0,
"epoch": 0.08426666666666667,
"grad_norm": 0.03557606413960457,
"kl": 0.07753753662109375,
"learning_rate": 3.3611111111111117e-06,
"loss": -0.0357,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.030786633491516113,
"mask/share_reasoning": 0.8402998447418213,
"mask/share_step_conf": 0.1289135068655014,
"num_tokens": 19063433.0,
"reward": 0.9475151300430298,
"reward_std": 0.1952834278345108,
"rewards/accuracy_reward_step": 0.65234375,
"rewards/final_brier_reward_step": 0.6971789002418518,
"rewards/format_reward_step": 0.9921875,
"rewards/step_l2_reward": 0.8689451217651367,
"step": 79
},
{
"adv/mean_abs_final_conf": 0.5850554704666138,
"adv/mean_abs_reasoning": 0.4936892092227936,
"adv/mean_abs_step_conf": 0.7499476075172424,
"adv/ratio_final_to_reasoning": 1.1850683781151639,
"adv/ratio_step_to_reasoning": 1.5190682589515618,
"adv/std_final_conf": 0.7985376119613647,
"adv/std_reasoning": 0.7575132250785828,
"adv/std_step_conf": 0.9337923526763916,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.5993457300275482,
"calib/avg_num_step_conf": 6.6953125,
"calib/ece": 0.30507246376811586,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.9288537549407114,
"calib/gap": 0.09892929292929298,
"calib/mean_conf": 0.9572463768115942,
"calib/mu_c": 0.9916565656565657,
"calib/mu_w": 0.8927272727272727,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.30507246376811586,
"calib/std_conf": 0.15437612240392326,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.555724171539961,
"calib/step_q_c_n": 1026.0,
"calib/step_q_gap": 0.06543347386554244,
"calib/step_q_w": 0.4902906976744186,
"calib/step_q_w_n": 688.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2709.0,
"completions/max_terminated_length": 2709.0,
"completions/mean_length": 509.875,
"completions/mean_terminated_length": 509.875,
"completions/min_length": 175.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.08533333333333333,
"grad_norm": 0.024001065641641617,
"kl": 0.07498931884765625,
"learning_rate": 3.3333333333333333e-06,
"loss": -0.0313,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.034538500010967255,
"mask/share_reasoning": 0.8263256549835205,
"mask/share_step_conf": 0.13913580775260925,
"num_tokens": 19296121.0,
"reward": 0.9185687303543091,
"reward_std": 0.222752183675766,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.685109555721283,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8270277976989746,
"step": 80
},
{
"adv/mean_abs_final_conf": 0.563480794429779,
"adv/mean_abs_reasoning": 0.46398669481277466,
"adv/mean_abs_step_conf": 0.7496002316474915,
"adv/ratio_final_to_reasoning": 1.2144330876926368,
"adv/ratio_step_to_reasoning": 1.6155640668747322,
"adv/std_final_conf": 0.7881554365158081,
"adv/std_reasoning": 0.7393821477890015,
"adv/std_step_conf": 0.9339177012443542,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7111616997792494,
"calib/avg_num_step_conf": 7.3984375,
"calib/ece": 0.30008097165991915,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.8259109311740891,
"calib/gap": 0.15856719094922733,
"calib/mean_conf": 0.9114170040485831,
"calib/mu_c": 0.9730463576158941,
"calib/mu_w": 0.8144791666666668,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.30008097165991915,
"calib/std_conf": 0.21095314874616838,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.5747247706422018,
"calib/step_q_c_n": 872.0,
"calib/step_q_gap": 0.14297330293183003,
"calib/step_q_w": 0.4317514677103718,
"calib/step_q_w_n": 1022.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 3059.0,
"completions/max_terminated_length": 3059.0,
"completions/mean_length": 591.15625,
"completions/mean_terminated_length": 593.4745483398438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 155.0,
"epoch": 0.0864,
"grad_norm": 0.02519802935421467,
"kl": 0.06589508056640625,
"learning_rate": 3.3055555555555558e-06,
"loss": 0.1002,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.03203187882900238,
"mask/share_reasoning": 0.8315613269805908,
"mask/share_step_conf": 0.1325005143880844,
"num_tokens": 19553705.0,
"reward": 0.9012755751609802,
"reward_std": 0.200510174036026,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.6784765720367432,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8131371140480042,
"step": 81
},
{
"adv/mean_abs_final_conf": 0.5482626557350159,
"adv/mean_abs_reasoning": 0.422796368598938,
"adv/mean_abs_step_conf": 0.7653951644897461,
"adv/ratio_final_to_reasoning": 1.2967534644440015,
"adv/ratio_step_to_reasoning": 1.810316316164473,
"adv/std_final_conf": 0.7778072357177734,
"adv/std_reasoning": 0.6816251277923584,
"adv/std_step_conf": 0.933276891708374,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6104530744336569,
"calib/avg_num_step_conf": 6.21875,
"calib/ece": 0.3651383399209486,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.8893280632411067,
"calib/gap": 0.06926731391585794,
"calib/mean_conf": 0.9455335968379447,
"calib/mu_c": 0.9737333333333333,
"calib/mu_w": 0.9044660194174754,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.358893280632411,
"calib/std_conf": 0.167837359685009,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.6096951934349356,
"calib/step_q_c_n": 853.0,
"calib/step_q_gap": 0.12957340723737143,
"calib/step_q_w": 0.4801217861975642,
"calib/step_q_w_n": 739.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2608.0,
"completions/max_terminated_length": 2608.0,
"completions/mean_length": 508.69140625,
"completions/mean_terminated_length": 508.69140625,
"completions/min_length": 124.0,
"completions/min_terminated_length": 124.0,
"epoch": 0.08746666666666666,
"grad_norm": 0.05241883173584938,
"kl": 0.07489776611328125,
"learning_rate": 3.277777777777778e-06,
"loss": -0.0588,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03483765944838524,
"mask/share_reasoning": 0.8307796120643616,
"mask/share_step_conf": 0.13438266515731812,
"num_tokens": 19789482.0,
"reward": 0.8882876634597778,
"reward_std": 0.1930658370256424,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.6260604858398438,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8372336030006409,
"step": 82
},
{
"adv/mean_abs_final_conf": 0.5539191961288452,
"adv/mean_abs_reasoning": 0.41217041015625,
"adv/mean_abs_step_conf": 0.7351502776145935,
"adv/ratio_final_to_reasoning": 1.3439082051495632,
"adv/ratio_step_to_reasoning": 1.7836076037964608,
"adv/std_final_conf": 0.7939161062240601,
"adv/std_reasoning": 0.6817459464073181,
"adv/std_step_conf": 0.9330217242240906,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.6222155528878217,
"calib/avg_num_step_conf": 6.38671875,
"calib/ece": 0.3342629482071713,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.8366533864541833,
"calib/gap": 0.08308523409363733,
"calib/mean_conf": 0.9169721115537849,
"calib/mu_c": 0.9494117647058823,
"calib/mu_w": 0.866326530612245,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.3208366533864542,
"calib/std_conf": 0.20582840883510017,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.576103752759382,
"calib/step_q_c_n": 906.0,
"calib/step_q_gap": 0.1191215854068442,
"calib/step_q_w": 0.45698216735253777,
"calib/step_q_w_n": 729.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2486.0,
"completions/max_terminated_length": 2486.0,
"completions/mean_length": 602.44921875,
"completions/mean_terminated_length": 604.8118286132812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 143.0,
"epoch": 0.08853333333333334,
"grad_norm": 0.04311411827802658,
"kl": 0.06854248046875,
"learning_rate": 3.2500000000000002e-06,
"loss": -0.0854,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03103901445865631,
"mask/share_reasoning": 0.84500652551651,
"mask/share_step_conf": 0.12004822492599487,
"num_tokens": 20050973.0,
"reward": 0.9060057997703552,
"reward_std": 0.20195919275283813,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.6508910059928894,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8470580577850342,
"step": 83
},
{
"adv/mean_abs_final_conf": 0.6155233383178711,
"adv/mean_abs_reasoning": 0.42707884311676025,
"adv/mean_abs_step_conf": 0.7596107721328735,
"adv/ratio_final_to_reasoning": 1.4412405302633815,
"adv/ratio_step_to_reasoning": 1.778619532143861,
"adv/std_final_conf": 0.8287082314491272,
"adv/std_reasoning": 0.7204527854919434,
"adv/std_step_conf": 0.9342764616012573,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7035398230088495,
"calib/avg_num_step_conf": 5.83984375,
"calib/ece": 0.3858232931726907,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.8755020080321285,
"calib/gap": 0.12827368558042673,
"calib/mean_conf": 0.932008032128514,
"calib/mu_c": 0.990220588235294,
"calib/mu_w": 0.8619469026548673,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.3858232931726907,
"calib/std_conf": 0.18901983474612602,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.6036246786632391,
"calib/step_q_c_n": 778.0,
"calib/step_q_gap": 0.10646986694775795,
"calib/step_q_w": 0.49715481171548115,
"calib/step_q_w_n": 717.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2676.0,
"completions/max_terminated_length": 2676.0,
"completions/mean_length": 501.8203125,
"completions/mean_terminated_length": 505.7716369628906,
"completions/min_length": 0.0,
"completions/min_terminated_length": 156.0,
"epoch": 0.0896,
"grad_norm": 0.05019519478082657,
"kl": 0.30936431884765625,
"learning_rate": 3.2222222222222227e-06,
"loss": -0.0528,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03413093462586403,
"mask/share_reasoning": 0.8325973153114319,
"mask/share_step_conf": 0.12545925378799438,
"num_tokens": 20285359.0,
"reward": 0.862108588218689,
"reward_std": 0.21464627981185913,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.6097398996353149,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8152584433555603,
"step": 84
},
{
"adv/mean_abs_final_conf": 0.6300166249275208,
"adv/mean_abs_reasoning": 0.5347933769226074,
"adv/mean_abs_step_conf": 0.7569372057914734,
"adv/ratio_final_to_reasoning": 1.1780561467549617,
"adv/ratio_step_to_reasoning": 1.415382535489054,
"adv/std_final_conf": 0.8321962356567383,
"adv/std_reasoning": 0.7927958369255066,
"adv/std_step_conf": 0.9346948266029358,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.6801801801801802,
"calib/avg_num_step_conf": 6.0234375,
"calib/ece": 0.37691358024691374,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.8436213991769548,
"calib/gap": 0.14998771498771502,
"calib/mean_conf": 0.9201234567901234,
"calib/mu_c": 0.9886363636363636,
"calib/mu_w": 0.8386486486486486,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.9765625,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.37691358024691374,
"calib/std_conf": 0.19790566294345305,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.5567567567567567,
"calib/step_q_c_n": 740.0,
"calib/step_q_gap": 0.09856473680663203,
"calib/step_q_w": 0.4581920199501247,
"calib/step_q_w_n": 802.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2524.0,
"completions/max_terminated_length": 2524.0,
"completions/mean_length": 580.41015625,
"completions/mean_terminated_length": 582.686279296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 152.0,
"epoch": 0.09066666666666667,
"grad_norm": 0.0329107902944088,
"kl": 0.065185546875,
"learning_rate": 3.1944444444444443e-06,
"loss": 0.0589,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.03241187334060669,
"mask/share_reasoning": 0.8383945226669312,
"mask/share_step_conf": 0.12528733909130096,
"num_tokens": 20541768.0,
"reward": 0.8522720336914062,
"reward_std": 0.2573317587375641,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.6064144372940063,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.8067232370376587,
"step": 85
},
{
"adv/mean_abs_final_conf": 0.6762087941169739,
"adv/mean_abs_reasoning": 0.4965750277042389,
"adv/mean_abs_step_conf": 0.733121931552887,
"adv/ratio_final_to_reasoning": 1.3617454692460396,
"adv/ratio_step_to_reasoning": 1.476356825558163,
"adv/std_final_conf": 0.8725427985191345,
"adv/std_reasoning": 0.7575492262840271,
"adv/std_step_conf": 0.9345494508743286,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.67103125,
"calib/avg_num_step_conf": 5.9453125,
"calib/ece": 0.3817786561264823,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.7628458498023716,
"calib/gap": 0.16379812500000002,
"calib/mean_conf": 0.8758498023715414,
"calib/mu_c": 0.95872,
"calib/mu_w": 0.794921875,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.3817786561264823,
"calib/std_conf": 0.2462280049163768,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.5691800878477306,
"calib/step_q_c_n": 683.0,
"calib/step_q_gap": 0.1282504096594112,
"calib/step_q_w": 0.44092967818831935,
"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": 0.00390625,
"completions/max_length": 2411.0,
"completions/max_terminated_length": 2411.0,
"completions/mean_length": 547.69921875,
"completions/mean_terminated_length": 549.8471069335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 131.0,
"epoch": 0.09173333333333333,
"grad_norm": 0.05290186405181885,
"kl": 0.07259368896484375,
"learning_rate": 3.1666666666666667e-06,
"loss": -0.0715,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.032537199556827545,
"mask/share_reasoning": 0.8427928686141968,
"mask/share_step_conf": 0.1207636296749115,
"num_tokens": 20787491.0,
"reward": 0.8759980201721191,
"reward_std": 0.21968460083007812,
"rewards/accuracy_reward_step": 0.48828125,
"rewards/final_brier_reward_step": 0.615178108215332,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8430678248405457,
"step": 86
},
{
"adv/mean_abs_final_conf": 0.6011219620704651,
"adv/mean_abs_reasoning": 0.4481174647808075,
"adv/mean_abs_step_conf": 0.7793911695480347,
"adv/ratio_final_to_reasoning": 1.341438371219248,
"adv/ratio_step_to_reasoning": 1.7392564021785373,
"adv/std_final_conf": 0.7962128520011902,
"adv/std_reasoning": 0.7205802798271179,
"adv/std_step_conf": 0.9328499436378479,
"calib/answer_extract_rate": 0.99609375,
"calib/auroc": 0.6277848729956008,
"calib/avg_num_step_conf": 5.1875,
"calib/ece": 0.2805490196078431,
"calib/final_conf_rate": 0.99609375,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.9137254901960784,
"calib/gap": 0.07362494678586606,
"calib/mean_conf": 0.951843137254902,
"calib/mu_c": 0.9752298850574712,
"calib/mu_w": 0.9016049382716051,
"calib/nonempty_final_conf_rate": 0.99609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2750196078431372,
"calib/std_conf": 0.15608486537137664,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.5707511210762332,
"calib/step_q_c_n": 892.0,
"calib/step_q_gap": 0.04361809355329738,
"calib/step_q_w": 0.5271330275229358,
"calib/step_q_w_n": 436.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1697.0,
"completions/max_terminated_length": 1697.0,
"completions/mean_length": 485.70703125,
"completions/mean_terminated_length": 487.6117858886719,
"completions/min_length": 0.0,
"completions/min_terminated_length": 141.0,
"epoch": 0.0928,
"grad_norm": 0.03262199088931084,
"kl": 0.0754241943359375,
"learning_rate": 3.138888888888889e-06,
"loss": 0.0488,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03634430468082428,
"mask/share_reasoning": 0.8396223783493042,
"mask/share_step_conf": 0.12012706696987152,
"num_tokens": 21017328.0,
"reward": 0.9499078989028931,
"reward_std": 0.1929827332496643,
"rewards/accuracy_reward_step": 0.6796875,
"rewards/final_brier_reward_step": 0.7115480303764343,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8546739816665649,
"step": 87
},
{
"adv/mean_abs_final_conf": 0.6190221309661865,
"adv/mean_abs_reasoning": 0.5069084167480469,
"adv/mean_abs_step_conf": 0.7607830166816711,
"adv/ratio_final_to_reasoning": 1.2211715381200003,
"adv/ratio_step_to_reasoning": 1.5008293244809343,
"adv/std_final_conf": 0.8372948169708252,
"adv/std_reasoning": 0.7575567960739136,
"adv/std_step_conf": 0.933971107006073,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8295369805436918,
"calib/avg_num_step_conf": 6.265625,
"calib/ece": 0.2845564516129033,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.7661290322580645,
"calib/gap": 0.22631075859263772,
"calib/mean_conf": 0.8853629032258065,
"calib/mu_c": 0.9757046979865771,
"calib/mu_w": 0.7493939393939394,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2845564516129033,
"calib/std_conf": 0.2239128515191619,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.536989853438557,
"calib/step_q_c_n": 887.0,
"calib/step_q_gap": 0.10948636668820833,
"calib/step_q_w": 0.42750348675034866,
"calib/step_q_w_n": 717.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2901.0,
"completions/max_terminated_length": 2901.0,
"completions/mean_length": 576.59765625,
"completions/mean_terminated_length": 578.85888671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 194.0,
"epoch": 0.09386666666666667,
"grad_norm": 0.043756153434515,
"kl": 0.06731414794921875,
"learning_rate": 3.1111111111111116e-06,
"loss": -0.0369,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.030716445297002792,
"mask/share_reasoning": 0.8459240794181824,
"mask/share_step_conf": 0.11945319920778275,
"num_tokens": 21274785.0,
"reward": 0.9229838848114014,
"reward_std": 0.20685866475105286,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.7071710824966431,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8302028775215149,
"step": 88
},
{
"adv/mean_abs_final_conf": 0.6980656385421753,
"adv/mean_abs_reasoning": 0.548902690410614,
"adv/mean_abs_step_conf": 0.736992597579956,
"adv/ratio_final_to_reasoning": 1.27174752599587,
"adv/ratio_step_to_reasoning": 1.3426653038057417,
"adv/std_final_conf": 0.8767775297164917,
"adv/std_reasoning": 0.8098074793815613,
"adv/std_step_conf": 0.9346106648445129,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7770247933884298,
"calib/avg_num_step_conf": 5.9765625,
"calib/ece": 0.3142682926829268,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.6260162601626016,
"calib/gap": 0.22904661157024808,
"calib/mean_conf": 0.8211788617886179,
"calib/mu_c": 0.93384,
"calib/mu_w": 0.7047933884297519,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.3136585365853658,
"calib/std_conf": 0.266238157720389,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.5375903614457831,
"calib/step_q_c_n": 664.0,
"calib/step_q_gap": 0.13856033835109488,
"calib/step_q_w": 0.3990300230946882,
"calib/step_q_w_n": 866.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2784.0,
"completions/max_terminated_length": 2784.0,
"completions/mean_length": 598.29296875,
"completions/mean_terminated_length": 600.6392211914062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 136.0,
"epoch": 0.09493333333333333,
"grad_norm": 0.03296685218811035,
"kl": 0.08054351806640625,
"learning_rate": 3.0833333333333336e-06,
"loss": -0.084,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03184279054403305,
"mask/share_reasoning": 0.8469675183296204,
"mask/share_step_conf": 0.11728344857692719,
"num_tokens": 21536836.0,
"reward": 0.886458158493042,
"reward_std": 0.24242013692855835,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/final_brier_reward_step": 0.6579160690307617,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8267190456390381,
"step": 89
},
{
"adv/mean_abs_final_conf": 0.6241170763969421,
"adv/mean_abs_reasoning": 0.5056729912757874,
"adv/mean_abs_step_conf": 0.7552558183670044,
"adv/ratio_final_to_reasoning": 1.2342305940096314,
"adv/ratio_step_to_reasoning": 1.4935656667395507,
"adv/std_final_conf": 0.8450374603271484,
"adv/std_reasoning": 0.7576045393943787,
"adv/std_step_conf": 0.9340672492980957,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.717675046614801,
"calib/avg_num_step_conf": 6.19140625,
"calib/ece": 0.2933858267716536,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.7913385826771654,
"calib/gap": 0.16305021539252884,
"calib/mean_conf": 0.8856692913385827,
"calib/mu_c": 0.9517880794701987,
"calib/mu_w": 0.7887378640776699,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.29228346456692916,
"calib/std_conf": 0.23862826673908388,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5008129175946548,
"calib/step_q_c_n": 898.0,
"calib/step_q_gap": 0.06686823054953112,
"calib/step_q_w": 0.43394468704512373,
"calib/step_q_w_n": 687.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2457.0,
"completions/max_terminated_length": 2457.0,
"completions/mean_length": 532.34375,
"completions/mean_terminated_length": 534.431396484375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 131.0,
"epoch": 0.096,
"grad_norm": 0.04197279363870621,
"kl": 0.0772247314453125,
"learning_rate": 3.055555555555556e-06,
"loss": 0.0217,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03273506462574005,
"mask/share_reasoning": 0.8372526168823242,
"mask/share_step_conf": 0.12610609829425812,
"num_tokens": 21776436.0,
"reward": 0.9231204986572266,
"reward_std": 0.2121354043483734,
"rewards/accuracy_reward_step": 0.58984375,
"rewards/final_brier_reward_step": 0.6835886240005493,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8485898375511169,
"step": 90
},
{
"adv/mean_abs_final_conf": 0.5773174166679382,
"adv/mean_abs_reasoning": 0.4142797589302063,
"adv/mean_abs_step_conf": 0.753173291683197,
"adv/ratio_final_to_reasoning": 1.393544831055091,
"adv/ratio_step_to_reasoning": 1.818030631349489,
"adv/std_final_conf": 0.7948840856552124,
"adv/std_reasoning": 0.7013592720031738,
"adv/std_step_conf": 0.9329512715339661,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6713343732274532,
"calib/avg_num_step_conf": 5.71484375,
"calib/ece": 0.23400000000000004,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.764,
"calib/gap": 0.16991917186613714,
"calib/mean_conf": 0.86856,
"calib/mu_c": 0.9270121951219512,
"calib/mu_w": 0.7570930232558141,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.22328000000000003,
"calib/std_conf": 0.2525397521183546,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4832835820895522,
"calib/step_q_c_n": 871.0,
"calib/step_q_gap": 0.08330047398144408,
"calib/step_q_w": 0.3999831081081081,
"calib/step_q_w_n": 592.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1869.0,
"completions/max_terminated_length": 1869.0,
"completions/mean_length": 528.02734375,
"completions/mean_terminated_length": 530.0980834960938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 241.0,
"epoch": 0.09706666666666666,
"grad_norm": 0.030806271359324455,
"kl": 0.08370208740234375,
"learning_rate": 3.0277777777777776e-06,
"loss": -0.0532,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.030275100842118263,
"mask/share_reasoning": 0.8504114747047424,
"mask/share_step_conf": 0.11540718376636505,
"num_tokens": 22019323.0,
"reward": 0.9466753005981445,
"reward_std": 0.17928370833396912,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.7201827764511108,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8512927889823914,
"step": 91
},
{
"adv/mean_abs_final_conf": 0.6468561887741089,
"adv/mean_abs_reasoning": 0.44251346588134766,
"adv/mean_abs_step_conf": 0.7543888092041016,
"adv/ratio_final_to_reasoning": 1.4617774116450328,
"adv/ratio_step_to_reasoning": 1.7047815882881583,
"adv/std_final_conf": 0.8452576994895935,
"adv/std_reasoning": 0.7013332843780518,
"adv/std_step_conf": 0.9340010285377502,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7756791307126878,
"calib/avg_num_step_conf": 6.17578125,
"calib/ece": 0.2511023622047244,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.7007874015748031,
"calib/gap": 0.27194375199744325,
"calib/mean_conf": 0.8377165354330709,
"calib/mu_c": 0.9501342281879194,
"calib/mu_w": 0.6781904761904761,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2511023622047244,
"calib/std_conf": 0.2746169529161331,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.48480239520958085,
"calib/step_q_c_n": 835.0,
"calib/step_q_gap": 0.11804636303799904,
"calib/step_q_w": 0.3667560321715818,
"calib/step_q_w_n": 746.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2289.0,
"completions/max_terminated_length": 2289.0,
"completions/mean_length": 547.29296875,
"completions/mean_terminated_length": 547.29296875,
"completions/min_length": 170.0,
"completions/min_terminated_length": 170.0,
"epoch": 0.09813333333333334,
"grad_norm": 0.04027779400348663,
"kl": 0.0789642333984375,
"learning_rate": 3e-06,
"loss": -0.0147,
"mask/has_final_conf_rate": 0.9921875,
"mask/share_final_conf": 0.03349082171916962,
"mask/share_reasoning": 0.8417634963989258,
"mask/share_step_conf": 0.12474566698074341,
"num_tokens": 22266150.0,
"reward": 0.9606250524520874,
"reward_std": 0.17936253547668457,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.7411539554595947,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.866033673286438,
"step": 92
},
{
"adv/mean_abs_final_conf": 0.6630334854125977,
"adv/mean_abs_reasoning": 0.5095129609107971,
"adv/mean_abs_step_conf": 0.7587285041809082,
"adv/ratio_final_to_reasoning": 1.3013083793341973,
"adv/ratio_step_to_reasoning": 1.4891250319218914,
"adv/std_final_conf": 0.8640803694725037,
"adv/std_reasoning": 0.7575684189796448,
"adv/std_step_conf": 0.9342039823532104,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.7449599298772853,
"calib/avg_num_step_conf": 7.3046875,
"calib/ece": 0.2682213438735179,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.6126482213438735,
"calib/gap": 0.2599242424242425,
"calib/mean_conf": 0.7837944664031621,
"calib/mu_c": 0.9081060606060607,
"calib/mu_w": 0.6481818181818182,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.26513833992094876,
"calib/std_conf": 0.3035427007764735,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.4588609112709832,
"calib/step_q_c_n": 834.0,
"calib/step_q_gap": 0.08960415451422649,
"calib/step_q_w": 0.36925675675675673,
"calib/step_q_w_n": 1036.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2687.0,
"completions/max_terminated_length": 2687.0,
"completions/mean_length": 593.234375,
"completions/mean_terminated_length": 593.234375,
"completions/min_length": 130.0,
"completions/min_terminated_length": 130.0,
"epoch": 0.0992,
"grad_norm": 0.05038225278258324,
"kl": 0.07889556884765625,
"learning_rate": 2.9722222222222225e-06,
"loss": -0.0754,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.031364671885967255,
"mask/share_reasoning": 0.833544909954071,
"mask/share_step_conf": 0.1350904256105423,
"num_tokens": 22523794.0,
"reward": 0.9259089231491089,
"reward_std": 0.20582300424575806,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.7031351327896118,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8494638204574585,
"step": 93
},
{
"adv/mean_abs_final_conf": 0.6444368958473206,
"adv/mean_abs_reasoning": 0.5294855237007141,
"adv/mean_abs_step_conf": 0.732408881187439,
"adv/ratio_final_to_reasoning": 1.2171001226684743,
"adv/ratio_step_to_reasoning": 1.3832462804052506,
"adv/std_final_conf": 0.8604599833488464,
"adv/std_reasoning": 0.7928929924964905,
"adv/std_step_conf": 0.9340792298316956,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7979837653836083,
"calib/avg_num_step_conf": 6.0625,
"calib/ece": 0.2141935483870968,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.5282258064516129,
"calib/gap": 0.3021550144016759,
"calib/mean_conf": 0.7528225806451613,
"calib/mu_c": 0.8917164179104479,
"calib/mu_w": 0.589561403508772,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.21334677419354842,
"calib/std_conf": 0.3001412583667654,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4648173207036536,
"calib/step_q_c_n": 739.0,
"calib/step_q_gap": 0.12778165034695005,
"calib/step_q_w": 0.33703567035670356,
"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": 0.015625,
"completions/max_length": 3038.0,
"completions/max_terminated_length": 3038.0,
"completions/mean_length": 537.80859375,
"completions/mean_terminated_length": 546.3452758789062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 170.0,
"epoch": 0.10026666666666667,
"grad_norm": 0.0642198994755745,
"kl": 0.089935302734375,
"learning_rate": 2.944444444444445e-06,
"loss": -0.076,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.033078089356422424,
"mask/share_reasoning": 0.831333339214325,
"mask/share_step_conf": 0.11996357142925262,
"num_tokens": 22770153.0,
"reward": 0.9430240392684937,
"reward_std": 0.2067991942167282,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/final_brier_reward_step": 0.7425273656845093,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8450832366943359,
"step": 94
},
{
"adv/mean_abs_final_conf": 0.591467559337616,
"adv/mean_abs_reasoning": 0.4067404270172119,
"adv/mean_abs_step_conf": 0.753157913684845,
"adv/ratio_final_to_reasoning": 1.4541646712501164,
"adv/ratio_step_to_reasoning": 1.8516918006111396,
"adv/std_final_conf": 0.792325496673584,
"adv/std_reasoning": 0.68165522813797,
"adv/std_step_conf": 0.9338064789772034,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7732813607370659,
"calib/avg_num_step_conf": 6.16796875,
"calib/ece": 0.1265612648221344,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.5889328063241107,
"calib/gap": 0.31951736357193483,
"calib/mean_conf": 0.7577075098814229,
"calib/mu_c": 0.8625294117647059,
"calib/mu_w": 0.543012048192771,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.10616600790513835,
"calib/std_conf": 0.31944907092670294,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.43998063891577927,
"calib/step_q_c_n": 1033.0,
"calib/step_q_gap": 0.0882407121758525,
"calib/step_q_w": 0.35173992673992677,
"calib/step_q_w_n": 546.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1483.0,
"completions/max_terminated_length": 1483.0,
"completions/mean_length": 519.26953125,
"completions/mean_terminated_length": 523.3582763671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 91.0,
"epoch": 0.10133333333333333,
"grad_norm": 0.04013686254620552,
"kl": 0.08258819580078125,
"learning_rate": 2.916666666666667e-06,
"loss": -0.0758,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.033396221697330475,
"mask/share_reasoning": 0.8303040266036987,
"mask/share_step_conf": 0.12848728895187378,
"num_tokens": 23009214.0,
"reward": 0.9940088987350464,
"reward_std": 0.14248350262641907,
"rewards/accuracy_reward_step": 0.6640625,
"rewards/final_brier_reward_step": 0.7942163944244385,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8648951649665833,
"step": 95
},
{
"adv/mean_abs_final_conf": 0.5519489645957947,
"adv/mean_abs_reasoning": 0.42615142464637756,
"adv/mean_abs_step_conf": 0.7463536262512207,
"adv/ratio_final_to_reasoning": 1.295194460639911,
"adv/ratio_step_to_reasoning": 1.7513812769030361,
"adv/std_final_conf": 0.792453408241272,
"adv/std_reasoning": 0.6817641258239746,
"adv/std_step_conf": 0.9340023994445801,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.8270086083213772,
"calib/avg_num_step_conf": 6.0234375,
"calib/ece": 0.1136904761904762,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.6428571428571429,
"calib/gap": 0.4094562410329986,
"calib/mean_conf": 0.7882936507936507,
"calib/mu_c": 0.921529411764706,
"calib/mu_w": 0.5120731707317074,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.1136904761904762,
"calib/std_conf": 0.3097833745797808,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4784560570071259,
"calib/step_q_c_n": 842.0,
"calib/step_q_gap": 0.1548417712928402,
"calib/step_q_w": 0.3236142857142857,
"calib/step_q_w_n": 700.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2983.0,
"completions/max_terminated_length": 2983.0,
"completions/mean_length": 484.4296875,
"completions/mean_terminated_length": 486.3294372558594,
"completions/min_length": 0.0,
"completions/min_terminated_length": 136.0,
"epoch": 0.1024,
"grad_norm": 0.04179300367832184,
"kl": 0.1007080078125,
"learning_rate": 2.888888888888889e-06,
"loss": 0.0869,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03455401211977005,
"mask/share_reasoning": 0.8357703685760498,
"mask/share_step_conf": 0.12576934695243835,
"num_tokens": 23239044.0,
"reward": 0.9944518208503723,
"reward_std": 0.16103437542915344,
"rewards/accuracy_reward_step": 0.6640625,
"rewards/final_brier_reward_step": 0.8348710536956787,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8259074687957764,
"step": 96
},
{
"adv/mean_abs_final_conf": 0.6904168128967285,
"adv/mean_abs_reasoning": 0.5510886907577515,
"adv/mean_abs_step_conf": 0.7708159685134888,
"adv/ratio_final_to_reasoning": 1.252823410234385,
"adv/ratio_step_to_reasoning": 1.3987149100331029,
"adv/std_final_conf": 0.8760626912117004,
"adv/std_reasoning": 0.8098317384719849,
"adv/std_step_conf": 0.9342261552810669,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7776057791537668,
"calib/avg_num_step_conf": 6.1171875,
"calib/ece": 0.15060000000000007,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.46,
"calib/gap": 0.3279024767801859,
"calib/mean_conf": 0.6617999999999999,
"calib/mu_c": 0.8113235294117648,
"calib/mu_w": 0.48342105263157886,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.1342000000000001,
"calib/std_conf": 0.3494446451156463,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.44049180327868853,
"calib/step_q_c_n": 732.0,
"calib/step_q_gap": 0.12093185124031919,
"calib/step_q_w": 0.31955995203836934,
"calib/step_q_w_n": 834.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2985.0,
"completions/max_terminated_length": 2985.0,
"completions/mean_length": 535.20703125,
"completions/mean_terminated_length": 535.20703125,
"completions/min_length": 131.0,
"completions/min_terminated_length": 131.0,
"epoch": 0.10346666666666667,
"grad_norm": 0.043487440794706345,
"kl": 0.0912933349609375,
"learning_rate": 2.861111111111111e-06,
"loss": -0.0083,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03317775949835777,
"mask/share_reasoning": 0.84007728099823,
"mask/share_step_conf": 0.12674494087696075,
"num_tokens": 23481129.0,
"reward": 0.9605032205581665,
"reward_std": 0.19906312227249146,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.7593629360198975,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8608622550964355,
"step": 97
},
{
"adv/mean_abs_final_conf": 0.7032493352890015,
"adv/mean_abs_reasoning": 0.48649489879608154,
"adv/mean_abs_step_conf": 0.7983685731887817,
"adv/ratio_final_to_reasoning": 1.4455430818068544,
"adv/ratio_step_to_reasoning": 1.6410625787947362,
"adv/std_final_conf": 0.8910401463508606,
"adv/std_reasoning": 0.739440381526947,
"adv/std_step_conf": 0.9343165159225464,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.7257996632996633,
"calib/avg_num_step_conf": 5.8203125,
"calib/ece": 0.15135802469135806,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.48148148148148145,
"calib/gap": 0.30894570707070707,
"calib/mean_conf": 0.6682304526748971,
"calib/mu_c": 0.7940972222222222,
"calib/mu_w": 0.48515151515151517,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.11349794238683131,
"calib/std_conf": 0.3540601602797552,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4261679790026247,
"calib/step_q_c_n": 762.0,
"calib/step_q_gap": 0.11119545153009724,
"calib/step_q_w": 0.31497252747252746,
"calib/step_q_w_n": 728.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 3012.0,
"completions/max_terminated_length": 3012.0,
"completions/mean_length": 602.69140625,
"completions/mean_terminated_length": 607.43701171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 79.0,
"epoch": 0.10453333333333334,
"grad_norm": 0.04565083980560303,
"kl": 0.07500457763671875,
"learning_rate": 2.8333333333333335e-06,
"loss": -0.0307,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.03302436321973801,
"mask/share_reasoning": 0.8465808629989624,
"mask/share_step_conf": 0.112582266330719,
"num_tokens": 23741602.0,
"reward": 0.9159847497940063,
"reward_std": 0.20621484518051147,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.7263554334640503,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.8063952922821045,
"step": 98
},
{
"adv/mean_abs_final_conf": 0.7148996591567993,
"adv/mean_abs_reasoning": 0.6136833429336548,
"adv/mean_abs_step_conf": 0.7592363357543945,
"adv/ratio_final_to_reasoning": 1.1649324808773358,
"adv/ratio_step_to_reasoning": 1.2371793116054568,
"adv/std_final_conf": 0.8905500769615173,
"adv/std_reasoning": 0.8266958594322205,
"adv/std_step_conf": 0.9347360134124756,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7288732394366197,
"calib/avg_num_step_conf": 6.87109375,
"calib/ece": 0.1465702479338843,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.256198347107438,
"calib/gap": 0.2946746478873239,
"calib/mean_conf": 0.46599173553719014,
"calib/mu_c": 0.6388999999999999,
"calib/mu_w": 0.344225352112676,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.09966942148760331,
"calib/std_conf": 0.35996792776941633,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.3858032786885246,
"calib/step_q_c_n": 610.0,
"calib/step_q_gap": 0.0916692490975759,
"calib/step_q_w": 0.2941340295909487,
"calib/step_q_w_n": 1149.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2702.0,
"completions/max_terminated_length": 2702.0,
"completions/mean_length": 614.54296875,
"completions/mean_terminated_length": 634.366943359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 82.0,
"epoch": 0.1056,
"grad_norm": 0.04171948879957199,
"kl": 0.079315185546875,
"learning_rate": 2.805555555555556e-06,
"loss": -0.1994,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.02855312079191208,
"mask/share_reasoning": 0.8328218460083008,
"mask/share_step_conf": 0.10737504065036774,
"num_tokens": 24004725.0,
"reward": 0.9038716554641724,
"reward_std": 0.21844570338726044,
"rewards/accuracy_reward_step": 0.390625,
"rewards/final_brier_reward_step": 0.7249144315719604,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.8172037601470947,
"step": 99
},
{
"adv/mean_abs_final_conf": 0.718896210193634,
"adv/mean_abs_reasoning": 0.38725799322128296,
"adv/mean_abs_step_conf": 0.7504380345344543,
"adv/ratio_final_to_reasoning": 1.856375395156401,
"adv/ratio_step_to_reasoning": 1.937824519236319,
"adv/std_final_conf": 0.9049928188323975,
"adv/std_reasoning": 0.6815465688705444,
"adv/std_step_conf": 0.9327692985534668,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.8164866255144033,
"calib/avg_num_step_conf": 6.13671875,
"calib/ece": 0.12813492063492066,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.3412698412698413,
"calib/gap": 0.393148148148148,
"calib/mean_conf": 0.5690079365079365,
"calib/mu_c": 0.7374999999999999,
"calib/mu_w": 0.3443518518518519,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 1.0,
"calib/pce": 0.06285714285714289,
"calib/std_conf": 0.3656406205949707,
"calib/step_conf_rate": 1.0,
"calib/step_q_c": 0.41403669724770636,
"calib/step_q_c_n": 872.0,
"calib/step_q_gap": 0.11046015933640452,
"calib/step_q_w": 0.30357653791130185,
"calib/step_q_w_n": 699.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1933.0,
"completions/max_terminated_length": 1933.0,
"completions/mean_length": 566.171875,
"completions/mean_terminated_length": 570.6299438476562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 161.0,
"epoch": 0.10666666666666667,
"grad_norm": 0.05875632166862488,
"kl": 0.085693359375,
"learning_rate": 2.7777777777777783e-06,
"loss": -0.0158,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.030663391575217247,
"mask/share_reasoning": 0.8413369655609131,
"mask/share_step_conf": 0.1201871708035469,
"num_tokens": 24257073.0,
"reward": 0.9939379692077637,
"reward_std": 0.14616990089416504,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.801247239112854,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.8764723539352417,
"step": 100
},
{
"adv/mean_abs_final_conf": 0.7209446430206299,
"adv/mean_abs_reasoning": 0.42586550116539,
"adv/mean_abs_step_conf": 0.7728959321975708,
"adv/ratio_final_to_reasoning": 1.6928928054696835,
"adv/ratio_step_to_reasoning": 1.8148827037703796,
"adv/std_final_conf": 0.8900034427642822,
"adv/std_reasoning": 0.6817054152488708,
"adv/std_step_conf": 0.9328939914703369,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7362021169354839,
"calib/avg_num_step_conf": 6.328125,
"calib/ece": 0.15242063492063493,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.21825396825396826,
"calib/gap": 0.3034173387096776,
"calib/mean_conf": 0.4427380952380952,
"calib/mu_c": 0.5968548387096776,
"calib/mu_w": 0.2934375,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.05154761904761904,
"calib/std_conf": 0.34981860783843766,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.3698863636363637,
"calib/step_q_c_n": 792.0,
"calib/step_q_gap": 0.046706411945542425,
"calib/step_q_w": 0.32317995169082125,
"calib/step_q_w_n": 828.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2451.0,
"completions/max_terminated_length": 2451.0,
"completions/mean_length": 629.3203125,
"completions/mean_terminated_length": 629.3203125,
"completions/min_length": 187.0,
"completions/min_terminated_length": 187.0,
"epoch": 0.10773333333333333,
"grad_norm": 0.06255262345075607,
"kl": 0.08121490478515625,
"learning_rate": 2.7500000000000004e-06,
"loss": -0.0331,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.028444774448871613,
"mask/share_reasoning": 0.8561521172523499,
"mask/share_step_conf": 0.11540311574935913,
"num_tokens": 24525171.0,
"reward": 0.9472886919975281,
"reward_std": 0.15424081683158875,
"rewards/accuracy_reward_step": 0.48828125,
"rewards/final_brier_reward_step": 0.754944920539856,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8474448919296265,
"step": 101
},
{
"adv/mean_abs_final_conf": 0.6129434108734131,
"adv/mean_abs_reasoning": 0.38685131072998047,
"adv/mean_abs_step_conf": 0.7592980265617371,
"adv/ratio_final_to_reasoning": 1.5844418614397389,
"adv/ratio_step_to_reasoning": 1.9627645183079703,
"adv/std_final_conf": 0.8420135378837585,
"adv/std_reasoning": 0.6815320253372192,
"adv/std_step_conf": 0.9337236285209656,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.8452188552188553,
"calib/avg_num_step_conf": 5.7109375,
"calib/ece": 0.12887550200803208,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.357429718875502,
"calib/gap": 0.4105111111111111,
"calib/mean_conf": 0.5828514056224899,
"calib/mu_c": 0.7460666666666667,
"calib/mu_w": 0.33555555555555555,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.05465863453815256,
"calib/std_conf": 0.3598854988704338,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.4501040312093628,
"calib/step_q_c_n": 769.0,
"calib/step_q_gap": 0.12143159253692415,
"calib/step_q_w": 0.32867243867243867,
"calib/step_q_w_n": 693.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2673.0,
"completions/max_terminated_length": 2673.0,
"completions/mean_length": 499.125,
"completions/mean_terminated_length": 505.0434875488281,
"completions/min_length": 0.0,
"completions/min_terminated_length": 120.0,
"epoch": 0.1088,
"grad_norm": 0.09262681007385254,
"kl": 0.09638214111328125,
"learning_rate": 2.7222222222222224e-06,
"loss": -0.0662,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03711201623082161,
"mask/share_reasoning": 0.8191013336181641,
"mask/share_step_conf": 0.13206785917282104,
"num_tokens": 24759643.0,
"reward": 0.9702571034431458,
"reward_std": 0.13695275783538818,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.8009117245674133,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8294462561607361,
"step": 102
},
{
"adv/mean_abs_final_conf": 0.5940297842025757,
"adv/mean_abs_reasoning": 0.3715140223503113,
"adv/mean_abs_step_conf": 0.7678797245025635,
"adv/ratio_final_to_reasoning": 1.5989431043398084,
"adv/ratio_step_to_reasoning": 2.066892979287085,
"adv/std_final_conf": 0.8287274241447449,
"adv/std_reasoning": 0.661227285861969,
"adv/std_step_conf": 0.9333446621894836,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7930283224400871,
"calib/avg_num_step_conf": 5.66796875,
"calib/ece": 0.16393574297188757,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.4979919678714859,
"calib/gap": 0.37458946078431377,
"calib/mean_conf": 0.6471485943775099,
"calib/mu_c": 0.7915686274509804,
"calib/mu_w": 0.4169791666666667,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.09831325301204821,
"calib/std_conf": 0.3757759068069405,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.4832968369829683,
"calib/step_q_c_n": 822.0,
"calib/step_q_gap": 0.1433922264901225,
"calib/step_q_w": 0.3399046104928458,
"calib/step_q_w_n": 629.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2695.0,
"completions/max_terminated_length": 2695.0,
"completions/mean_length": 630.10546875,
"completions/mean_terminated_length": 630.10546875,
"completions/min_length": 148.0,
"completions/min_terminated_length": 148.0,
"epoch": 0.10986666666666667,
"grad_norm": 0.04487231746315956,
"kl": 0.0799102783203125,
"learning_rate": 2.6944444444444444e-06,
"loss": -0.0508,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03193045035004616,
"mask/share_reasoning": 0.8594948649406433,
"mask/share_step_conf": 0.10857468843460083,
"num_tokens": 25025502.0,
"reward": 0.9651437997817993,
"reward_std": 0.16764892637729645,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.7688324451446533,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8489552140235901,
"step": 103
},
{
"adv/mean_abs_final_conf": 0.6918395757675171,
"adv/mean_abs_reasoning": 0.4783649444580078,
"adv/mean_abs_step_conf": 0.7479958534240723,
"adv/ratio_final_to_reasoning": 1.4462589363678773,
"adv/ratio_step_to_reasoning": 1.5636510619973605,
"adv/std_final_conf": 0.9062270522117615,
"adv/std_reasoning": 0.7393895387649536,
"adv/std_step_conf": 0.9339094161987305,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.8109590809345553,
"calib/avg_num_step_conf": 6.24609375,
"calib/ece": 0.11160682730923693,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.2931726907630522,
"calib/gap": 0.39794164192590675,
"calib/mean_conf": 0.5027714859437751,
"calib/mu_c": 0.7057377049180328,
"calib/mu_w": 0.30779606299212603,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.06220923694779116,
"calib/std_conf": 0.36954813071351916,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4920183098591549,
"calib/step_q_c_n": 710.0,
"calib/step_q_gap": 0.14827252808187702,
"calib/step_q_w": 0.3437457817772779,
"calib/step_q_w_n": 889.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2712.0,
"completions/max_terminated_length": 2712.0,
"completions/mean_length": 604.265625,
"completions/mean_terminated_length": 606.6353149414062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 140.0,
"epoch": 0.11093333333333333,
"grad_norm": 0.06312678009271622,
"kl": 0.087371826171875,
"learning_rate": 2.666666666666667e-06,
"loss": -0.0637,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.030438464134931564,
"mask/share_reasoning": 0.8470672369003296,
"mask/share_step_conf": 0.11858808249235153,
"num_tokens": 25286874.0,
"reward": 0.9490529894828796,
"reward_std": 0.187879741191864,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/final_brier_reward_step": 0.7746831774711609,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8367041349411011,
"step": 104
},
{
"adv/mean_abs_final_conf": 0.757011890411377,
"adv/mean_abs_reasoning": 0.6426866054534912,
"adv/mean_abs_step_conf": 0.7516645789146423,
"adv/ratio_final_to_reasoning": 1.1778865219654233,
"adv/ratio_step_to_reasoning": 1.1695662746608113,
"adv/std_final_conf": 0.9064919948577881,
"adv/std_reasoning": 0.8431361317634583,
"adv/std_step_conf": 0.9346654415130615,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7000324464633354,
"calib/avg_num_step_conf": 5.546875,
"calib/ece": 0.23377510040160643,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.40963855421686746,
"calib/gap": 0.25921674237508113,
"calib/mean_conf": 0.5697590361445783,
"calib/mu_c": 0.6894776119402986,
"calib/mu_w": 0.4302608695652174,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.13269076305220884,
"calib/std_conf": 0.389880970107651,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.48018099547511317,
"calib/step_q_c_n": 663.0,
"calib/step_q_gap": 0.12751256746982914,
"calib/step_q_w": 0.35266842800528403,
"calib/step_q_w_n": 757.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2914.0,
"completions/max_terminated_length": 2914.0,
"completions/mean_length": 569.80859375,
"completions/mean_terminated_length": 572.0431518554688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.112,
"grad_norm": 0.040814757347106934,
"kl": 0.07910919189453125,
"learning_rate": 2.6388888888888893e-06,
"loss": 0.0893,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.033313244581222534,
"mask/share_reasoning": 0.8503282070159912,
"mask/share_step_conf": 0.11245226860046387,
"num_tokens": 25538505.0,
"reward": 0.9136497974395752,
"reward_std": 0.25790584087371826,
"rewards/accuracy_reward_step": 0.5234375,
"rewards/final_brier_reward_step": 0.6958324313163757,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8345921635627747,
"step": 105
},
{
"adv/mean_abs_final_conf": 0.6224170327186584,
"adv/mean_abs_reasoning": 0.430540531873703,
"adv/mean_abs_step_conf": 0.7094933986663818,
"adv/ratio_final_to_reasoning": 1.4456641980022487,
"adv/ratio_step_to_reasoning": 1.6479131374197966,
"adv/std_final_conf": 0.8420764803886414,
"adv/std_reasoning": 0.7014119029045105,
"adv/std_step_conf": 0.9333073496818542,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.8074145712443583,
"calib/avg_num_step_conf": 5.30859375,
"calib/ece": 0.17131474103585662,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.5219123505976095,
"calib/gap": 0.3739458413926499,
"calib/mean_conf": 0.6982470119521913,
"calib/mu_c": 0.8621276595744681,
"calib/mu_w": 0.4881818181818182,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.15390438247011956,
"calib/std_conf": 0.35938764145933155,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5282806052269601,
"calib/step_q_c_n": 727.0,
"calib/step_q_gap": 0.12972047864468156,
"calib/step_q_w": 0.3985601265822785,
"calib/step_q_w_n": 632.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2456.0,
"completions/max_terminated_length": 2456.0,
"completions/mean_length": 528.19921875,
"completions/mean_terminated_length": 528.19921875,
"completions/min_length": 132.0,
"completions/min_terminated_length": 132.0,
"epoch": 0.11306666666666666,
"grad_norm": 0.04006750136613846,
"kl": 0.08618927001953125,
"learning_rate": 2.6111111111111113e-06,
"loss": -0.0261,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03361281752586365,
"mask/share_reasoning": 0.8508884310722351,
"mask/share_step_conf": 0.11549879610538483,
"num_tokens": 25778308.0,
"reward": 0.9733978509902954,
"reward_std": 0.1720713973045349,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/final_brier_reward_step": 0.7717820405960083,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8695447444915771,
"step": 106
},
{
"adv/mean_abs_final_conf": 0.6800111532211304,
"adv/mean_abs_reasoning": 0.5244839191436768,
"adv/mean_abs_step_conf": 0.7717577815055847,
"adv/ratio_final_to_reasoning": 1.2965338467028358,
"adv/ratio_step_to_reasoning": 1.4714612847723363,
"adv/std_final_conf": 0.8938814997673035,
"adv/std_reasoning": 0.7752963900566101,
"adv/std_step_conf": 0.9338552355766296,
"calib/answer_extract_rate": 1.0,
"calib/auroc": 0.6842861705525392,
"calib/avg_num_step_conf": 5.85546875,
"calib/ece": 0.21613281250000008,
"calib/final_conf_rate": 1.0,
"calib/format_rate": 0.98828125,
"calib/frac_conf_gt_0.9": 0.61328125,
"calib/gap": 0.23877419354838714,
"calib/mean_conf": 0.7645703125000001,
"calib/mu_c": 0.8587741935483871,
"calib/mu_w": 0.62,
"calib/nonempty_final_conf_rate": 1.0,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.18761718750000006,
"calib/std_conf": 0.32905494847920513,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5311538461538462,
"calib/step_q_c_n": 858.0,
"calib/step_q_gap": 0.09338473538941561,
"calib/step_q_w": 0.4377691107644306,
"calib/step_q_w_n": 641.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1170.0,
"completions/max_terminated_length": 1170.0,
"completions/mean_length": 485.51171875,
"completions/mean_terminated_length": 487.41571044921875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 123.0,
"epoch": 0.11413333333333334,
"grad_norm": 32082.970703125,
"kl": 260096.09117126465,
"learning_rate": 2.5833333333333337e-06,
"loss": 1538.7279,
"mask/has_final_conf_rate": 0.99609375,
"mask/share_final_conf": 0.03437138348817825,
"mask/share_reasoning": 0.8336739540100098,
"mask/share_step_conf": 0.12804844975471497,
"num_tokens": 26007215.0,
"reward": 0.9511657953262329,
"reward_std": 0.18870285153388977,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.730099618434906,
"rewards/format_reward_step": 0.98828125,
"rewards/step_l2_reward": 0.8534818887710571,
"step": 107
},
{
"adv/mean_abs_final_conf": 0.5230342745780945,
"adv/mean_abs_reasoning": 0.45457059144973755,
"adv/mean_abs_step_conf": 0.735778272151947,
"adv/ratio_final_to_reasoning": 1.150611773872149,
"adv/ratio_step_to_reasoning": 1.6186226869744673,
"adv/std_final_conf": 0.7766129374504089,
"adv/std_reasoning": 0.7206103801727295,
"adv/std_step_conf": 0.9345055818557739,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.7777818772136955,
"calib/avg_num_step_conf": 5.91015625,
"calib/ece": 0.15731225296442697,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.6837944664031621,
"calib/gap": 0.34880681818181825,
"calib/mean_conf": 0.7835573122529644,
"calib/mu_c": 0.8897159090909091,
"calib/mu_w": 0.5409090909090909,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.12260869565217403,
"calib/std_conf": 0.3358041402787481,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.5856619144602851,
"calib/step_q_c_n": 982.0,
"calib/step_q_gap": 0.12730033253938117,
"calib/step_q_w": 0.458361581920904,
"calib/step_q_w_n": 531.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2072.0,
"completions/max_terminated_length": 2072.0,
"completions/mean_length": 541.06640625,
"completions/mean_terminated_length": 543.1882934570312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 97.0,
"epoch": 0.1152,
"grad_norm": 0.02788160741329193,
"kl": 0.0900421142578125,
"learning_rate": 2.5555555555555557e-06,
"loss": -0.031,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03286145627498627,
"mask/share_reasoning": 0.8382688164710999,
"mask/share_step_conf": 0.12496345490217209,
"num_tokens": 26248960.0,
"reward": 0.9895857572555542,
"reward_std": 0.1817062944173813,
"rewards/accuracy_reward_step": 0.6875,
"rewards/final_brier_reward_step": 0.7988402843475342,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8467375040054321,
"step": 108
},
{
"adv/mean_abs_final_conf": 0.6572158336639404,
"adv/mean_abs_reasoning": 0.49283796548843384,
"adv/mean_abs_step_conf": 0.7203115224838257,
"adv/ratio_final_to_reasoning": 1.3335332902216608,
"adv/ratio_step_to_reasoning": 1.4615585099454564,
"adv/std_final_conf": 0.863275408744812,
"adv/std_reasoning": 0.7575352191925049,
"adv/std_step_conf": 0.9334295392036438,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.8282544774124566,
"calib/avg_num_step_conf": 6.390625,
"calib/ece": 0.15795918367346948,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.4775510204081633,
"calib/gap": 0.47229751403368075,
"calib/mean_conf": 0.608734693877551,
"calib/mu_c": 0.8574137931034482,
"calib/mu_w": 0.3851162790697675,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.14661224489795927,
"calib/std_conf": 0.4024958245700629,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.5868759571209802,
"calib/step_q_c_n": 653.0,
"calib/step_q_gap": 0.20669284420134637,
"calib/step_q_w": 0.3801831129196338,
"calib/step_q_w_n": 983.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2668.0,
"completions/max_terminated_length": 2668.0,
"completions/mean_length": 580.15234375,
"completions/mean_terminated_length": 587.0316162109375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 39.0,
"epoch": 0.11626666666666667,
"grad_norm": 0.02776014618575573,
"kl": 0.0984039306640625,
"learning_rate": 2.5277777777777778e-06,
"loss": -0.0688,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.029165200889110565,
"mask/share_reasoning": 0.8402441740036011,
"mask/share_step_conf": 0.11887191236019135,
"num_tokens": 26502079.0,
"reward": 0.9375416040420532,
"reward_std": 0.19762474298477173,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.7645038366317749,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8301106095314026,
"step": 109
},
{
"adv/mean_abs_final_conf": 0.7246900200843811,
"adv/mean_abs_reasoning": 0.526329517364502,
"adv/mean_abs_step_conf": 0.7481073141098022,
"adv/ratio_final_to_reasoning": 1.3768751251366877,
"adv/ratio_step_to_reasoning": 1.4213668233083558,
"adv/std_final_conf": 0.9108691215515137,
"adv/std_reasoning": 0.7754833102226257,
"adv/std_step_conf": 0.9347033500671387,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.6953282828282827,
"calib/avg_num_step_conf": 5.1953125,
"calib/ece": 0.23301587301587312,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.47619047619047616,
"calib/gap": 0.2693863636363637,
"calib/mean_conf": 0.6821428571428572,
"calib/mu_c": 0.8232500000000001,
"calib/mu_w": 0.5538636363636364,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.2194841269841271,
"calib/std_conf": 0.3538812175006936,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.6097535211267605,
"calib/step_q_c_n": 568.0,
"calib/step_q_gap": 0.11592150012938518,
"calib/step_q_w": 0.49383202099737533,
"calib/step_q_w_n": 762.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1743.0,
"completions/max_terminated_length": 1743.0,
"completions/mean_length": 502.828125,
"completions/mean_terminated_length": 504.8000183105469,
"completions/min_length": 0.0,
"completions/min_terminated_length": 89.0,
"epoch": 0.11733333333333333,
"grad_norm": 0.055533409118652344,
"kl": 0.113800048828125,
"learning_rate": 2.5e-06,
"loss": -0.0275,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03461432084441185,
"mask/share_reasoning": 0.8436667919158936,
"mask/share_step_conf": 0.11781267821788788,
"num_tokens": 26735723.0,
"reward": 0.8851801156997681,
"reward_std": 0.21155953407287598,
"rewards/accuracy_reward_step": 0.46875,
"rewards/final_brier_reward_step": 0.694117546081543,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.7887426614761353,
"step": 110
},
{
"adv/mean_abs_final_conf": 0.7247201204299927,
"adv/mean_abs_reasoning": 0.5945273637771606,
"adv/mean_abs_step_conf": 0.7475019693374634,
"adv/ratio_final_to_reasoning": 1.218985305950746,
"adv/ratio_step_to_reasoning": 1.257304566417973,
"adv/std_final_conf": 0.8908548355102539,
"adv/std_reasoning": 0.8100446462631226,
"adv/std_step_conf": 0.9354099035263062,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.759775641025641,
"calib/avg_num_step_conf": 5.8203125,
"calib/ece": 0.22508000000000006,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.548,
"calib/gap": 0.30952564102564106,
"calib/mean_conf": 0.6821200000000001,
"calib/mu_c": 0.8306923076923077,
"calib/mu_w": 0.5211666666666667,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.96484375,
"calib/pce": 0.19360000000000005,
"calib/std_conf": 0.37661532841879924,
"calib/step_conf_rate": 0.96484375,
"calib/step_q_c": 0.5931643625192012,
"calib/step_q_c_n": 651.0,
"calib/step_q_gap": 0.16203206216163268,
"calib/step_q_w": 0.43113230035756855,
"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": 0.0,
"completions/max_length": 2714.0,
"completions/max_terminated_length": 2714.0,
"completions/mean_length": 575.546875,
"completions/mean_terminated_length": 575.546875,
"completions/min_length": 110.0,
"completions/min_terminated_length": 110.0,
"epoch": 0.1184,
"grad_norm": 0.03187034651637077,
"kl": 0.11324310302734375,
"learning_rate": 2.4722222222222226e-06,
"loss": 0.0254,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03376447781920433,
"mask/share_reasoning": 0.8519448041915894,
"mask/share_step_conf": 0.1142907366156578,
"num_tokens": 26990471.0,
"reward": 0.8868539929389954,
"reward_std": 0.24743622541427612,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.6920046806335449,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.7910782098770142,
"step": 111
},
{
"adv/mean_abs_final_conf": 0.6615912318229675,
"adv/mean_abs_reasoning": 0.582384467124939,
"adv/mean_abs_step_conf": 0.7320601940155029,
"adv/ratio_final_to_reasoning": 1.1360042534943438,
"adv/ratio_step_to_reasoning": 1.2570050118772382,
"adv/std_final_conf": 0.8920271396636963,
"adv/std_reasoning": 0.8428117632865906,
"adv/std_step_conf": 0.9354498982429504,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.8028235294117647,
"calib/avg_num_step_conf": 5.109375,
"calib/ece": 0.1567213114754099,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 0.42213114754098363,
"calib/gap": 0.42403092436974793,
"calib/mean_conf": 0.5799180327868853,
"calib/mu_c": 0.78672,
"calib/mu_w": 0.36268907563025204,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.98046875,
"calib/nonempty_step_conf_rate": 0.9453125,
"calib/pce": 0.11217213114754104,
"calib/std_conf": 0.3957654875683685,
"calib/step_conf_rate": 0.9453125,
"calib/step_q_c": 0.563958,
"calib/step_q_c_n": 700.0,
"calib/step_q_gap": 0.16862905263157896,
"calib/step_q_w": 0.395328947368421,
"calib/step_q_w_n": 608.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2994.0,
"completions/max_terminated_length": 2994.0,
"completions/mean_length": 599.2265625,
"completions/mean_terminated_length": 603.9448852539062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 122.0,
"epoch": 0.11946666666666667,
"grad_norm": 0.04947257414460182,
"kl": 0.1170196533203125,
"learning_rate": 2.4444444444444447e-06,
"loss": -0.1128,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.02901388704776764,
"mask/share_reasoning": 0.8649810552597046,
"mask/share_step_conf": 0.09819258004426956,
"num_tokens": 27251793.0,
"reward": 0.8923689126968384,
"reward_std": 0.2590276598930359,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.7300676107406616,
"rewards/format_reward_step": 0.9140625,
"rewards/step_l2_reward": 0.7710765600204468,
"step": 112
},
{
"adv/mean_abs_final_conf": 0.7774215340614319,
"adv/mean_abs_reasoning": 0.6023589372634888,
"adv/mean_abs_step_conf": 0.7576698064804077,
"adv/ratio_final_to_reasoning": 1.290628371172263,
"adv/ratio_step_to_reasoning": 1.257837743592707,
"adv/std_final_conf": 0.921485960483551,
"adv/std_reasoning": 0.8266767859458923,
"adv/std_step_conf": 0.9352647662162781,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7012999487704917,
"calib/avg_num_step_conf": 5.98046875,
"calib/ece": 0.22724,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.464,
"calib/gap": 0.2638716700819671,
"calib/mean_conf": 0.6508399999999999,
"calib/mu_c": 0.7796093749999999,
"calib/mu_w": 0.5157377049180328,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.18303999999999998,
"calib/std_conf": 0.37026651806502836,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.5322963951935915,
"calib/step_q_c_n": 749.0,
"calib/step_q_gap": 0.07087695785343806,
"calib/step_q_w": 0.46141943734015345,
"calib/step_q_w_n": 782.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2579.0,
"completions/max_terminated_length": 2579.0,
"completions/mean_length": 524.98046875,
"completions/mean_terminated_length": 524.98046875,
"completions/min_length": 151.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.12053333333333334,
"grad_norm": 0.033785078674554825,
"kl": 0.1403656005859375,
"learning_rate": 2.4166666666666667e-06,
"loss": -0.0499,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03336464241147041,
"mask/share_reasoning": 0.8438401222229004,
"mask/share_step_conf": 0.12279525399208069,
"num_tokens": 27491388.0,
"reward": 0.9093856811523438,
"reward_std": 0.25819456577301025,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.7008984088897705,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8241229057312012,
"step": 113
},
{
"adv/mean_abs_final_conf": 0.6561824679374695,
"adv/mean_abs_reasoning": 0.5457957983016968,
"adv/mean_abs_step_conf": 0.7403881549835205,
"adv/ratio_final_to_reasoning": 1.2022490278951448,
"adv/ratio_step_to_reasoning": 1.3565295982257817,
"adv/std_final_conf": 0.863419234752655,
"adv/std_reasoning": 0.7754489183425903,
"adv/std_step_conf": 0.9345146417617798,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.8388490590193317,
"calib/avg_num_step_conf": 5.6328125,
"calib/ece": 0.15166007905138346,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.5335968379446641,
"calib/gap": 0.4147695557547048,
"calib/mean_conf": 0.7094466403162055,
"calib/mu_c": 0.8848630136986301,
"calib/mu_w": 0.47009345794392526,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.14201581027667992,
"calib/std_conf": 0.3577273595261266,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.603311345646438,
"calib/step_q_c_n": 758.0,
"calib/step_q_gap": 0.16277040997392334,
"calib/step_q_w": 0.4405409356725146,
"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": 0.0,
"completions/max_length": 2523.0,
"completions/max_terminated_length": 2523.0,
"completions/mean_length": 523.515625,
"completions/mean_terminated_length": 523.515625,
"completions/min_length": 138.0,
"completions/min_terminated_length": 138.0,
"epoch": 0.1216,
"grad_norm": 0.032592494040727615,
"kl": 0.1292572021484375,
"learning_rate": 2.388888888888889e-06,
"loss": -0.0674,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.036023080348968506,
"mask/share_reasoning": 0.8391364216804504,
"mask/share_step_conf": 0.12484048306941986,
"num_tokens": 27730432.0,
"reward": 0.9680483341217041,
"reward_std": 0.21276646852493286,
"rewards/accuracy_reward_step": 0.5703125,
"rewards/final_brier_reward_step": 0.7889527082443237,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8393313884735107,
"step": 114
},
{
"adv/mean_abs_final_conf": 0.7770059704780579,
"adv/mean_abs_reasoning": 0.6370693445205688,
"adv/mean_abs_step_conf": 0.7334473133087158,
"adv/ratio_final_to_reasoning": 1.2196568194044863,
"adv/ratio_step_to_reasoning": 1.1512833251467733,
"adv/std_final_conf": 0.9263416528701782,
"adv/std_reasoning": 0.8590085506439209,
"adv/std_step_conf": 0.9356162548065186,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6730133502860776,
"calib/avg_num_step_conf": 5.7734375,
"calib/ece": 0.29470119521912347,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.4860557768924303,
"calib/gap": 0.21137698664971394,
"calib/mean_conf": 0.6537450199203186,
"calib/mu_c": 0.7632231404958677,
"calib/mu_w": 0.5518461538461538,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.23318725099601592,
"calib/std_conf": 0.38411941868882943,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.559277108433735,
"calib/step_q_c_n": 664.0,
"calib/step_q_gap": 0.11944909860572522,
"calib/step_q_w": 0.4398280098280098,
"calib/step_q_w_n": 814.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2949.0,
"completions/max_terminated_length": 2949.0,
"completions/mean_length": 509.8515625,
"completions/mean_terminated_length": 511.85101318359375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 79.0,
"epoch": 0.12266666666666666,
"grad_norm": 0.03090801276266575,
"kl": 0.1342315673828125,
"learning_rate": 2.361111111111111e-06,
"loss": -0.0815,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03292369842529297,
"mask/share_reasoning": 0.8425235748291016,
"mask/share_step_conf": 0.12064643204212189,
"num_tokens": 27966218.0,
"reward": 0.8426992893218994,
"reward_std": 0.26547738909721375,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/final_brier_reward_step": 0.6314203143119812,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.7711657285690308,
"step": 115
},
{
"adv/mean_abs_final_conf": 0.7391272187232971,
"adv/mean_abs_reasoning": 0.6228404641151428,
"adv/mean_abs_step_conf": 0.7303839921951294,
"adv/ratio_final_to_reasoning": 1.1867039174684333,
"adv/ratio_step_to_reasoning": 1.1726662512731436,
"adv/std_final_conf": 0.9049105644226074,
"adv/std_reasoning": 0.8430219888687134,
"adv/std_step_conf": 0.9353899955749512,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7813555953090837,
"calib/avg_num_step_conf": 6.15234375,
"calib/ece": 0.1734146341463415,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 0.45121951219512196,
"calib/gap": 0.3678552971576228,
"calib/mean_conf": 0.6584552845528456,
"calib/mu_c": 0.8334108527131783,
"calib/mu_w": 0.4655555555555555,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.9453125,
"calib/pce": 0.153739837398374,
"calib/std_conf": 0.36947673946902104,
"calib/step_conf_rate": 0.9453125,
"calib/step_q_c": 0.5523137876386688,
"calib/step_q_c_n": 631.0,
"calib/step_q_gap": 0.17788476221493998,
"calib/step_q_w": 0.3744290254237288,
"calib/step_q_w_n": 944.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3007.0,
"completions/max_terminated_length": 3007.0,
"completions/mean_length": 623.42578125,
"completions/mean_terminated_length": 625.87060546875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 122.0,
"epoch": 0.12373333333333333,
"grad_norm": 0.036150332540273666,
"kl": 0.125274658203125,
"learning_rate": 2.3333333333333336e-06,
"loss": -0.0167,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03112485446035862,
"mask/share_reasoning": 0.8542196750640869,
"mask/share_step_conf": 0.11074923723936081,
"num_tokens": 28230335.0,
"reward": 0.9023261070251465,
"reward_std": 0.274005651473999,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.7138823866844177,
"rewards/format_reward_step": 0.91796875,
"rewards/step_l2_reward": 0.8056135773658752,
"step": 116
},
{
"adv/mean_abs_final_conf": 0.7267533540725708,
"adv/mean_abs_reasoning": 0.6318784952163696,
"adv/mean_abs_step_conf": 0.7376343011856079,
"adv/ratio_final_to_reasoning": 1.1501473140397251,
"adv/ratio_step_to_reasoning": 1.1673673131303908,
"adv/std_final_conf": 0.906491756439209,
"adv/std_reasoning": 0.8590298891067505,
"adv/std_step_conf": 0.9355345368385315,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.722940797940798,
"calib/avg_num_step_conf": 6.35546875,
"calib/ece": 0.2083665338645419,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.3784860557768924,
"calib/gap": 0.2679929214929216,
"calib/mean_conf": 0.6077290836653387,
"calib/mu_c": 0.7572072072072074,
"calib/mu_w": 0.48921428571428577,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.18693227091633471,
"calib/std_conf": 0.3590126210351204,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.5319690846286702,
"calib/step_q_c_n": 579.0,
"calib/step_q_gap": 0.136272519743174,
"calib/step_q_w": 0.3956965648854962,
"calib/step_q_w_n": 1048.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2740.0,
"completions/max_terminated_length": 2740.0,
"completions/mean_length": 530.328125,
"completions/mean_terminated_length": 536.6166381835938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 143.0,
"epoch": 0.1248,
"grad_norm": 0.03738539293408394,
"kl": 0.1431121826171875,
"learning_rate": 2.305555555555556e-06,
"loss": -0.14,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03171270340681076,
"mask/share_reasoning": 0.8342646360397339,
"mask/share_step_conf": 0.12230387330055237,
"num_tokens": 28472699.0,
"reward": 0.8889857530593872,
"reward_std": 0.25273334980010986,
"rewards/accuracy_reward_step": 0.43359375,
"rewards/final_brier_reward_step": 0.6912992596626282,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8101098537445068,
"step": 117
},
{
"adv/mean_abs_final_conf": 0.7259243726730347,
"adv/mean_abs_reasoning": 0.5425605773925781,
"adv/mean_abs_step_conf": 0.7316884994506836,
"adv/ratio_final_to_reasoning": 1.3379600415527073,
"adv/ratio_step_to_reasoning": 1.3485839737325018,
"adv/std_final_conf": 0.9222829341888428,
"adv/std_reasoning": 0.7928994297981262,
"adv/std_step_conf": 0.9354918003082275,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7732473544973544,
"calib/avg_num_step_conf": 6.796875,
"calib/ece": 0.19256097560975607,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.45934959349593496,
"calib/gap": 0.3434523809523811,
"calib/mean_conf": 0.6615853658536586,
"calib/mu_c": 0.8375000000000001,
"calib/mu_w": 0.494047619047619,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.18317073170731707,
"calib/std_conf": 0.3702310951971517,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.5220641562064157,
"calib/step_q_c_n": 717.0,
"calib/step_q_gap": 0.13714724516047233,
"calib/step_q_w": 0.38491691104594333,
"calib/step_q_w_n": 1023.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2908.0,
"completions/max_terminated_length": 2908.0,
"completions/mean_length": 602.109375,
"completions/mean_terminated_length": 604.4706420898438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 112.0,
"epoch": 0.12586666666666665,
"grad_norm": 0.02952709048986435,
"kl": 0.1322174072265625,
"learning_rate": 2.277777777777778e-06,
"loss": -0.0579,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03140158951282501,
"mask/share_reasoning": 0.8391228914260864,
"mask/share_step_conf": 0.12556925415992737,
"num_tokens": 28730847.0,
"reward": 0.8890880346298218,
"reward_std": 0.27732986211776733,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/final_brier_reward_step": 0.7022339701652527,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.7954732179641724,
"step": 118
},
{
"adv/mean_abs_final_conf": 0.734850287437439,
"adv/mean_abs_reasoning": 0.6560388207435608,
"adv/mean_abs_step_conf": 0.7462862133979797,
"adv/ratio_final_to_reasoning": 1.120132321749729,
"adv/ratio_step_to_reasoning": 1.1375641041365991,
"adv/std_final_conf": 0.9075103402137756,
"adv/std_reasoning": 0.8748774528503418,
"adv/std_step_conf": 0.9354383945465088,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7938529742605703,
"calib/avg_num_step_conf": 5.625,
"calib/ece": 0.19991869918699193,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.3983739837398374,
"calib/gap": 0.38824852775755964,
"calib/mean_conf": 0.5930081300813008,
"calib/mu_c": 0.7934453781512605,
"calib/mu_w": 0.40519685039370085,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.93359375,
"calib/pce": 0.1545934959349594,
"calib/std_conf": 0.3947746809439399,
"calib/step_conf_rate": 0.93359375,
"calib/step_q_c": 0.5407395498392283,
"calib/step_q_c_n": 622.0,
"calib/step_q_gap": 0.1718886940935071,
"calib/step_q_w": 0.36885085574572124,
"calib/step_q_w_n": 818.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2510.0,
"completions/max_terminated_length": 2510.0,
"completions/mean_length": 588.47265625,
"completions/mean_terminated_length": 590.7804565429688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 92.0,
"epoch": 0.12693333333333334,
"grad_norm": 0.033969517797231674,
"kl": 0.1355133056640625,
"learning_rate": 2.25e-06,
"loss": -0.1945,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03004167228937149,
"mask/share_reasoning": 0.8571265935897827,
"mask/share_step_conf": 0.10892552137374878,
"num_tokens": 28986560.0,
"reward": 0.8886620998382568,
"reward_std": 0.2910217344760895,
"rewards/accuracy_reward_step": 0.46875,
"rewards/final_brier_reward_step": 0.7106913924217224,
"rewards/format_reward_step": 0.91015625,
"rewards/step_l2_reward": 0.7908514738082886,
"step": 119
},
{
"adv/mean_abs_final_conf": 0.7197772264480591,
"adv/mean_abs_reasoning": 0.43707719445228577,
"adv/mean_abs_step_conf": 0.7259597778320312,
"adv/ratio_final_to_reasoning": 1.6467965741155473,
"adv/ratio_step_to_reasoning": 1.6609417902522978,
"adv/std_final_conf": 0.909675121307373,
"adv/std_reasoning": 0.7208017706871033,
"adv/std_step_conf": 0.9347007870674133,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7929282596835788,
"calib/avg_num_step_conf": 5.8984375,
"calib/ece": 0.14759999999999998,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.368,
"calib/gap": 0.42851882160392807,
"calib/mean_conf": 0.5608,
"calib/mu_c": 0.721923076923077,
"calib/mu_w": 0.29340425531914893,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.95703125,
"calib/pce": 0.0422,
"calib/std_conf": 0.392979592345455,
"calib/step_conf_rate": 0.95703125,
"calib/step_q_c": 0.486977829638273,
"calib/step_q_c_n": 857.0,
"calib/step_q_gap": 0.15777415429370945,
"calib/step_q_w": 0.3292036753445636,
"calib/step_q_w_n": 653.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2930.0,
"completions/max_terminated_length": 2930.0,
"completions/mean_length": 514.51953125,
"completions/mean_terminated_length": 520.6205444335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 145.0,
"epoch": 0.128,
"grad_norm": 0.048599161207675934,
"kl": 0.1397857666015625,
"learning_rate": 2.222222222222222e-06,
"loss": -0.1298,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03219369053840637,
"mask/share_reasoning": 0.8342239260673523,
"mask/share_step_conf": 0.12186359614133835,
"num_tokens": 29224965.0,
"reward": 0.9581763744354248,
"reward_std": 0.22807276248931885,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.7615699172019958,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.8438452482223511,
"step": 120
},
{
"adv/mean_abs_final_conf": 0.781947135925293,
"adv/mean_abs_reasoning": 0.6653647422790527,
"adv/mean_abs_step_conf": 0.7622631788253784,
"adv/ratio_final_to_reasoning": 1.1752157669895678,
"adv/ratio_step_to_reasoning": 1.1456320577108168,
"adv/std_final_conf": 0.9308911561965942,
"adv/std_reasoning": 0.8591954112052917,
"adv/std_step_conf": 0.9346886873245239,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.6882398353062275,
"calib/avg_num_step_conf": 5.71875,
"calib/ece": 0.21704,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.91015625,
"calib/frac_conf_gt_0.9": 0.372,
"calib/gap": 0.257421513124035,
"calib/mean_conf": 0.5561600000000001,
"calib/mu_c": 0.6941379310344827,
"calib/mu_w": 0.43671641791044774,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.91796875,
"calib/pce": 0.15460000000000002,
"calib/std_conf": 0.3859612084134881,
"calib/step_conf_rate": 0.91796875,
"calib/step_q_c": 0.4705891980360065,
"calib/step_q_c_n": 611.0,
"calib/step_q_gap": 0.09674394598442387,
"calib/step_q_w": 0.37384525205158264,
"calib/step_q_w_n": 853.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2591.0,
"completions/max_terminated_length": 2591.0,
"completions/mean_length": 612.71484375,
"completions/mean_terminated_length": 615.11767578125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 107.0,
"epoch": 0.12906666666666666,
"grad_norm": 0.031079187989234924,
"kl": 0.1669921875,
"learning_rate": 2.1944444444444445e-06,
"loss": -0.1448,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.028758013620972633,
"mask/share_reasoning": 0.8577010631561279,
"mask/share_step_conf": 0.10963468253612518,
"num_tokens": 29486876.0,
"reward": 0.863829493522644,
"reward_std": 0.26976099610328674,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.6560484170913696,
"rewards/format_reward_step": 0.91015625,
"rewards/step_l2_reward": 0.7989541888237,
"step": 121
},
{
"adv/mean_abs_final_conf": 0.7424711585044861,
"adv/mean_abs_reasoning": 0.5156237483024597,
"adv/mean_abs_step_conf": 0.7362110018730164,
"adv/ratio_final_to_reasoning": 1.4399475605009564,
"adv/ratio_step_to_reasoning": 1.4278066211976768,
"adv/std_final_conf": 0.92244553565979,
"adv/std_reasoning": 0.7754589319229126,
"adv/std_step_conf": 0.9350674152374268,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8241998983997968,
"calib/avg_num_step_conf": 6.1484375,
"calib/ece": 0.16760956175298813,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.46613545816733065,
"calib/gap": 0.43796355092710176,
"calib/mean_conf": 0.622191235059761,
"calib/mu_c": 0.8437903225806451,
"calib/mu_w": 0.40582677165354336,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.14788844621513952,
"calib/std_conf": 0.3943085875731503,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.5155096418732782,
"calib/step_q_c_n": 726.0,
"calib/step_q_gap": 0.1537407739487499,
"calib/step_q_w": 0.3617688679245283,
"calib/step_q_w_n": 848.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2990.0,
"completions/max_terminated_length": 2990.0,
"completions/mean_length": 537.39453125,
"completions/mean_terminated_length": 541.6259765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 156.0,
"epoch": 0.13013333333333332,
"grad_norm": 0.03416607528924942,
"kl": 0.1433258056640625,
"learning_rate": 2.166666666666667e-06,
"loss": -0.058,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.031357355415821075,
"mask/share_reasoning": 0.8387341499328613,
"mask/share_step_conf": 0.122095987200737,
"num_tokens": 29731793.0,
"reward": 0.9425933361053467,
"reward_std": 0.2181842178106308,
"rewards/accuracy_reward_step": 0.484375,
"rewards/final_brier_reward_step": 0.7617121338844299,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8351932764053345,
"step": 122
},
{
"adv/mean_abs_final_conf": 0.7870303392410278,
"adv/mean_abs_reasoning": 0.6281484365463257,
"adv/mean_abs_step_conf": 0.7771914601325989,
"adv/ratio_final_to_reasoning": 1.2529368751887113,
"adv/ratio_step_to_reasoning": 1.2372735724787263,
"adv/std_final_conf": 0.9210789203643799,
"adv/std_reasoning": 0.8268639445304871,
"adv/std_step_conf": 0.934878945350647,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7303689064558629,
"calib/avg_num_step_conf": 6.3515625,
"calib/ece": 0.18068825910931174,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.29959514170040485,
"calib/gap": 0.30387088274044816,
"calib/mean_conf": 0.5317813765182184,
"calib/mu_c": 0.6941739130434784,
"calib/mu_w": 0.39030303030303026,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.94921875,
"calib/pce": 0.12344129554655871,
"calib/std_conf": 0.37659144732146543,
"calib/step_conf_rate": 0.94921875,
"calib/step_q_c": 0.45963909774436085,
"calib/step_q_c_n": 665.0,
"calib/step_q_gap": 0.13994086673499556,
"calib/step_q_w": 0.3196982310093653,
"calib/step_q_w_n": 961.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2887.0,
"completions/max_terminated_length": 2887.0,
"completions/mean_length": 668.88671875,
"completions/mean_terminated_length": 668.88671875,
"completions/min_length": 182.0,
"completions/min_terminated_length": 182.0,
"epoch": 0.1312,
"grad_norm": 0.039897166192531586,
"kl": 0.130279541015625,
"learning_rate": 2.138888888888889e-06,
"loss": -0.0651,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.027274832129478455,
"mask/share_reasoning": 0.8642194867134094,
"mask/share_step_conf": 0.10850568860769272,
"num_tokens": 30008316.0,
"reward": 0.8898562788963318,
"reward_std": 0.2588973343372345,
"rewards/accuracy_reward_step": 0.44921875,
"rewards/final_brier_reward_step": 0.7077487707138062,
"rewards/format_reward_step": 0.92578125,
"rewards/step_l2_reward": 0.7969638109207153,
"step": 123
},
{
"adv/mean_abs_final_conf": 0.7224191427230835,
"adv/mean_abs_reasoning": 0.43076780438423157,
"adv/mean_abs_step_conf": 0.7511848211288452,
"adv/ratio_final_to_reasoning": 1.6770499915975798,
"adv/ratio_step_to_reasoning": 1.7438276804429227,
"adv/std_final_conf": 0.8989644050598145,
"adv/std_reasoning": 0.7015025615692139,
"adv/std_step_conf": 0.9342661499977112,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7611060448150079,
"calib/avg_num_step_conf": 5.6640625,
"calib/ece": 0.18237154150197626,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.391304347826087,
"calib/gap": 0.3329872329338197,
"calib/mean_conf": 0.6119367588932806,
"calib/mu_c": 0.7448684210526316,
"calib/mu_w": 0.4118811881188119,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.953125,
"calib/pce": 0.09675889328063239,
"calib/std_conf": 0.3763510628160573,
"calib/step_conf_rate": 0.953125,
"calib/step_q_c": 0.46166259168704155,
"calib/step_q_c_n": 818.0,
"calib/step_q_gap": 0.08511195877564914,
"calib/step_q_w": 0.3765506329113924,
"calib/step_q_w_n": 632.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1879.0,
"completions/max_terminated_length": 1879.0,
"completions/mean_length": 535.0078125,
"completions/mean_terminated_length": 537.1058959960938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.13226666666666667,
"grad_norm": 0.0538254976272583,
"kl": 0.146026611328125,
"learning_rate": 2.1111111111111114e-06,
"loss": -0.135,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.030877020210027695,
"mask/share_reasoning": 0.8481591939926147,
"mask/share_step_conf": 0.11705756187438965,
"num_tokens": 30252094.0,
"reward": 0.9374511241912842,
"reward_std": 0.21843618154525757,
"rewards/accuracy_reward_step": 0.59765625,
"rewards/final_brier_reward_step": 0.7460086345672607,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8195186853408813,
"step": 124
},
{
"adv/mean_abs_final_conf": 0.7272511124610901,
"adv/mean_abs_reasoning": 0.5495245456695557,
"adv/mean_abs_step_conf": 0.7719345092773438,
"adv/ratio_final_to_reasoning": 1.323418795742759,
"adv/ratio_step_to_reasoning": 1.404731627295734,
"adv/std_final_conf": 0.9073876142501831,
"adv/std_reasoning": 0.792988121509552,
"adv/std_step_conf": 0.9352187514305115,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.662,
"calib/avg_num_step_conf": 6.171875,
"calib/ece": 0.2656275303643724,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.4251012145748988,
"calib/gap": 0.2216177049180329,
"calib/mean_conf": 0.638582995951417,
"calib/mu_c": 0.7507377049180328,
"calib/mu_w": 0.5291199999999999,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.2051417004048583,
"calib/std_conf": 0.37554496999452713,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4674045801526717,
"calib/step_q_c_n": 655.0,
"calib/step_q_gap": 0.10480998555807713,
"calib/step_q_w": 0.3625945945945946,
"calib/step_q_w_n": 925.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2824.0,
"completions/max_terminated_length": 2824.0,
"completions/mean_length": 586.6875,
"completions/mean_terminated_length": 591.3070678710938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.13333333333333333,
"grad_norm": 0.03176412731409073,
"kl": 0.1277313232421875,
"learning_rate": 2.0833333333333334e-06,
"loss": -0.1226,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.030877090990543365,
"mask/share_reasoning": 0.8451837301254272,
"mask/share_step_conf": 0.1161266565322876,
"num_tokens": 30507094.0,
"reward": 0.8768726587295532,
"reward_std": 0.2407217025756836,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/final_brier_reward_step": 0.6630710959434509,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8055179119110107,
"step": 125
},
{
"adv/mean_abs_final_conf": 0.7116903066635132,
"adv/mean_abs_reasoning": 0.6008867621421814,
"adv/mean_abs_step_conf": 0.7131746411323547,
"adv/ratio_final_to_reasoning": 1.1844000425742671,
"adv/ratio_step_to_reasoning": 1.1868702824969273,
"adv/std_final_conf": 0.9004759788513184,
"adv/std_reasoning": 0.8428918719291687,
"adv/std_step_conf": 0.9355723857879639,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7877714512711864,
"calib/avg_num_step_conf": 6.53515625,
"calib/ece": 0.18051490514905144,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 0.45528455284552843,
"calib/gap": 0.4178425141242937,
"calib/mean_conf": 0.5843631436314363,
"calib/mu_c": 0.7847916666666666,
"calib/mu_w": 0.36694915254237287,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.94140625,
"calib/pce": 0.1222764227642276,
"calib/std_conf": 0.4152145166846133,
"calib/step_conf_rate": 0.94140625,
"calib/step_q_c": 0.4872044506258692,
"calib/step_q_c_n": 719.0,
"calib/step_q_gap": 0.18707834999693834,
"calib/step_q_w": 0.30012610062893086,
"calib/step_q_w_n": 954.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2949.0,
"completions/max_terminated_length": 2949.0,
"completions/mean_length": 600.23046875,
"completions/mean_terminated_length": 604.9566650390625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 115.0,
"epoch": 0.1344,
"grad_norm": 0.02776472456753254,
"kl": 0.122283935546875,
"learning_rate": 2.0555555555555555e-06,
"loss": -0.139,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03054964542388916,
"mask/share_reasoning": 0.8413506746292114,
"mask/share_step_conf": 0.12028719484806061,
"num_tokens": 30766217.0,
"reward": 0.8982517123222351,
"reward_std": 0.28006142377853394,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.716654896736145,
"rewards/format_reward_step": 0.9140625,
"rewards/step_l2_reward": 0.7954734563827515,
"step": 126
},
{
"adv/mean_abs_final_conf": 0.7346285581588745,
"adv/mean_abs_reasoning": 0.6094740629196167,
"adv/mean_abs_step_conf": 0.7600812315940857,
"adv/ratio_final_to_reasoning": 1.2053483533650626,
"adv/ratio_step_to_reasoning": 1.2471100541227338,
"adv/std_final_conf": 0.9214181900024414,
"adv/std_reasoning": 0.8266828060150146,
"adv/std_step_conf": 0.9350873231887817,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.7964355332776385,
"calib/avg_num_step_conf": 6.65625,
"calib/ece": 0.1770833333333333,
"calib/final_conf_rate": 0.9375,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.42083333333333334,
"calib/gap": 0.41462823725981607,
"calib/mean_conf": 0.5909166666666666,
"calib/mu_c": 0.8085964912280701,
"calib/mu_w": 0.393968253968254,
"calib/nonempty_final_conf_rate": 0.9375,
"calib/nonempty_reasoning_rate": 0.97265625,
"calib/nonempty_step_conf_rate": 0.96484375,
"calib/pce": 0.14649999999999996,
"calib/std_conf": 0.4043946006755722,
"calib/step_conf_rate": 0.96484375,
"calib/step_q_c": 0.4948462664714495,
"calib/step_q_c_n": 683.0,
"calib/step_q_gap": 0.1757375495272771,
"calib/step_q_w": 0.3191087169441724,
"calib/step_q_w_n": 1021.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2889.0,
"completions/max_terminated_length": 2889.0,
"completions/mean_length": 619.01171875,
"completions/mean_terminated_length": 626.351806640625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.13546666666666668,
"grad_norm": 0.029080456122756004,
"kl": 0.13339996337890625,
"learning_rate": 2.027777777777778e-06,
"loss": -0.0763,
"mask/has_final_conf_rate": 0.9375,
"mask/share_final_conf": 0.030351759865880013,
"mask/share_reasoning": 0.8314367532730103,
"mask/share_step_conf": 0.1264927238225937,
"num_tokens": 31028356.0,
"reward": 0.9085278511047363,
"reward_std": 0.23661072552204132,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/final_brier_reward_step": 0.7260656356811523,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.8159900903701782,
"step": 127
},
{
"adv/mean_abs_final_conf": 0.717476487159729,
"adv/mean_abs_reasoning": 0.6013306379318237,
"adv/mean_abs_step_conf": 0.7633137106895447,
"adv/ratio_final_to_reasoning": 1.1931480651432789,
"adv/ratio_step_to_reasoning": 1.269374388297983,
"adv/std_final_conf": 0.8905273675918579,
"adv/std_reasoning": 0.8101703524589539,
"adv/std_step_conf": 0.9351193308830261,
"calib/answer_extract_rate": 0.93359375,
"calib/auroc": 0.7363270209157716,
"calib/avg_num_step_conf": 5.4609375,
"calib/ece": 0.21970588235294114,
"calib/final_conf_rate": 0.9296875,
"calib/format_rate": 0.8984375,
"calib/frac_conf_gt_0.9": 0.37815126050420167,
"calib/gap": 0.3302925381571509,
"calib/mean_conf": 0.5816386554621849,
"calib/mu_c": 0.7509482758620689,
"calib/mu_w": 0.42065573770491804,
"calib/nonempty_final_conf_rate": 0.9296875,
"calib/nonempty_reasoning_rate": 0.98046875,
"calib/nonempty_step_conf_rate": 0.94921875,
"calib/pce": 0.15697478991596636,
"calib/std_conf": 0.3872845615500299,
"calib/step_conf_rate": 0.94921875,
"calib/step_q_c": 0.47670175438596496,
"calib/step_q_c_n": 570.0,
"calib/step_q_gap": 0.17309064327485385,
"calib/step_q_w": 0.3036111111111111,
"calib/step_q_w_n": 828.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2997.0,
"completions/max_terminated_length": 2997.0,
"completions/mean_length": 606.43359375,
"completions/mean_terminated_length": 611.2086791992188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 62.0,
"epoch": 0.13653333333333334,
"grad_norm": 0.02509251795709133,
"kl": 0.1387786865234375,
"learning_rate": 2.0000000000000003e-06,
"loss": -0.0994,
"mask/has_final_conf_rate": 0.9296875,
"mask/share_final_conf": 0.03150024265050888,
"mask/share_reasoning": 0.8502755165100098,
"mask/share_step_conf": 0.11041173338890076,
"num_tokens": 31290267.0,
"reward": 0.8657711744308472,
"reward_std": 0.2659390866756439,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.6803550720214844,
"rewards/format_reward_step": 0.8984375,
"rewards/step_l2_reward": 0.7808746099472046,
"step": 128
},
{
"adv/mean_abs_final_conf": 0.717547595500946,
"adv/mean_abs_reasoning": 0.4793207049369812,
"adv/mean_abs_step_conf": 0.7387598752975464,
"adv/ratio_final_to_reasoning": 1.4970093887249993,
"adv/ratio_step_to_reasoning": 1.5412642677196158,
"adv/std_final_conf": 0.8871719837188721,
"adv/std_reasoning": 0.7394227385520935,
"adv/std_step_conf": 0.934581995010376,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7197559709241953,
"calib/avg_num_step_conf": 6.1171875,
"calib/ece": 0.20844621513944228,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.4063745019920319,
"calib/gap": 0.28078725337487015,
"calib/mean_conf": 0.6385657370517929,
"calib/mu_c": 0.7582638888888888,
"calib/mu_w": 0.4774766355140187,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.1366533864541833,
"calib/std_conf": 0.36887512687148855,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4696181384248211,
"calib/step_q_c_n": 838.0,
"calib/step_q_gap": 0.11189835820504085,
"calib/step_q_w": 0.35771978021978024,
"calib/step_q_w_n": 728.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2642.0,
"completions/max_terminated_length": 2642.0,
"completions/mean_length": 546.26953125,
"completions/mean_terminated_length": 546.26953125,
"completions/min_length": 136.0,
"completions/min_terminated_length": 136.0,
"epoch": 0.1376,
"grad_norm": 0.0597921647131443,
"kl": 0.1490020751953125,
"learning_rate": 1.9722222222222224e-06,
"loss": 0.0022,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03373418003320694,
"mask/share_reasoning": 0.837746262550354,
"mask/share_step_conf": 0.12851959466934204,
"num_tokens": 31532496.0,
"reward": 0.9151557087898254,
"reward_std": 0.21401160955429077,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.7176144123077393,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8087906837463379,
"step": 129
},
{
"adv/mean_abs_final_conf": 0.6780507564544678,
"adv/mean_abs_reasoning": 0.5096803307533264,
"adv/mean_abs_step_conf": 0.7484921216964722,
"adv/ratio_final_to_reasoning": 1.330345150758091,
"adv/ratio_step_to_reasoning": 1.468552103217664,
"adv/std_final_conf": 0.8762167096138,
"adv/std_reasoning": 0.7575826644897461,
"adv/std_step_conf": 0.9352434277534485,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7863839285714286,
"calib/avg_num_step_conf": 5.875,
"calib/ece": 0.18468253968253967,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.5198412698412699,
"calib/gap": 0.36919642857142865,
"calib/mean_conf": 0.7069841269841269,
"calib/mu_c": 0.8710714285714286,
"calib/mu_w": 0.501875,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.16805555555555554,
"calib/std_conf": 0.36071637242485816,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.5037385620915033,
"calib/step_q_c_n": 765.0,
"calib/step_q_gap": 0.13467225627282936,
"calib/step_q_w": 0.3690663058186739,
"calib/step_q_w_n": 739.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2839.0,
"completions/max_terminated_length": 2839.0,
"completions/mean_length": 521.73828125,
"completions/mean_terminated_length": 521.73828125,
"completions/min_length": 65.0,
"completions/min_terminated_length": 65.0,
"epoch": 0.13866666666666666,
"grad_norm": 0.05114923045039177,
"kl": 0.1488189697265625,
"learning_rate": 1.944444444444445e-06,
"loss": -0.021,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.034622110426425934,
"mask/share_reasoning": 0.8403658270835876,
"mask/share_step_conf": 0.12501206994056702,
"num_tokens": 31771349.0,
"reward": 0.9483653903007507,
"reward_std": 0.2190546840429306,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.7516831755638123,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8427038192749023,
"step": 130
},
{
"adv/mean_abs_final_conf": 0.720390796661377,
"adv/mean_abs_reasoning": 0.5134299397468567,
"adv/mean_abs_step_conf": 0.7589023113250732,
"adv/ratio_final_to_reasoning": 1.4030946403642939,
"adv/ratio_step_to_reasoning": 1.4781029553890939,
"adv/std_final_conf": 0.9076294898986816,
"adv/std_reasoning": 0.7752957344055176,
"adv/std_step_conf": 0.9347165822982788,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8260456918175039,
"calib/avg_num_step_conf": 5.77734375,
"calib/ece": 0.16827956989247317,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.36693548387096775,
"calib/gap": 0.43755316023772395,
"calib/mean_conf": 0.5579569892473119,
"calib/mu_c": 0.8208417508417508,
"calib/mu_w": 0.38328859060402687,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.9765625,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.1635215053763441,
"calib/std_conf": 0.3985211708279797,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.47491039426523296,
"calib/step_q_c_n": 558.0,
"calib/step_q_gap": 0.14866718036729593,
"calib/step_q_w": 0.32624321389793703,
"calib/step_q_w_n": 921.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2967.0,
"completions/max_terminated_length": 2967.0,
"completions/mean_length": 554.61328125,
"completions/mean_terminated_length": 554.61328125,
"completions/min_length": 188.0,
"completions/min_terminated_length": 188.0,
"epoch": 0.13973333333333332,
"grad_norm": 0.04064655676484108,
"kl": 0.139007568359375,
"learning_rate": 1.916666666666667e-06,
"loss": -0.0294,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.031148342415690422,
"mask/share_reasoning": 0.8485188484191895,
"mask/share_step_conf": 0.12033282965421677,
"num_tokens": 32019538.0,
"reward": 0.9247491359710693,
"reward_std": 0.21482647955417633,
"rewards/accuracy_reward_step": 0.38671875,
"rewards/final_brier_reward_step": 0.7481170296669006,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8334124088287354,
"step": 131
},
{
"adv/mean_abs_final_conf": 0.6843870878219604,
"adv/mean_abs_reasoning": 0.609015941619873,
"adv/mean_abs_step_conf": 0.7606111764907837,
"adv/ratio_final_to_reasoning": 1.123758905229334,
"adv/ratio_step_to_reasoning": 1.248918336140257,
"adv/std_final_conf": 0.8533295392990112,
"adv/std_reasoning": 0.8100547194480896,
"adv/std_step_conf": 0.9349188804626465,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7723787680209698,
"calib/avg_num_step_conf": 6.56640625,
"calib/ece": 0.2086345381526105,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.5261044176706827,
"calib/gap": 0.3636952817824377,
"calib/mean_conf": 0.655863453815261,
"calib/mu_c": 0.8150714285714284,
"calib/mu_w": 0.45137614678899074,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.15112449799196792,
"calib/std_conf": 0.40146938625769135,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.4591356542617047,
"calib/step_q_c_n": 833.0,
"calib/step_q_gap": 0.1655743335069877,
"calib/step_q_w": 0.293561320754717,
"calib/step_q_w_n": 848.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2947.0,
"completions/max_terminated_length": 2947.0,
"completions/mean_length": 584.30859375,
"completions/mean_terminated_length": 584.30859375,
"completions/min_length": 125.0,
"completions/min_terminated_length": 125.0,
"epoch": 0.1408,
"grad_norm": 0.03089536912739277,
"kl": 0.143707275390625,
"learning_rate": 1.888888888888889e-06,
"loss": 0.0929,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03282865881919861,
"mask/share_reasoning": 0.8385095596313477,
"mask/share_step_conf": 0.12866178154945374,
"num_tokens": 32274713.0,
"reward": 0.9386829733848572,
"reward_std": 0.23207098245620728,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/final_brier_reward_step": 0.7341241836547852,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8416792154312134,
"step": 132
},
{
"adv/mean_abs_final_conf": 0.7646936178207397,
"adv/mean_abs_reasoning": 0.6460624933242798,
"adv/mean_abs_step_conf": 0.721850574016571,
"adv/ratio_final_to_reasoning": 1.1836217482399418,
"adv/ratio_step_to_reasoning": 1.1173076621463163,
"adv/std_final_conf": 0.9069320559501648,
"adv/std_reasoning": 0.8592305183410645,
"adv/std_step_conf": 0.9352459907531738,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7253191489361701,
"calib/avg_num_step_conf": 6.62109375,
"calib/ece": 0.2306557377049181,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 0.3319672131147541,
"calib/gap": 0.3234411347517731,
"calib/mean_conf": 0.5007377049180328,
"calib/mu_c": 0.6995744680851065,
"calib/mu_w": 0.3761333333333334,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.95703125,
"calib/pce": 0.17307377049180334,
"calib/std_conf": 0.4016985821152735,
"calib/step_conf_rate": 0.95703125,
"calib/step_q_c": 0.3731669266770671,
"calib/step_q_c_n": 641.0,
"calib/step_q_gap": 0.0708728090300082,
"calib/step_q_w": 0.3022941176470589,
"calib/step_q_w_n": 1054.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2916.0,
"completions/max_terminated_length": 2916.0,
"completions/mean_length": 623.90625,
"completions/mean_terminated_length": 628.8189086914062,
"completions/min_length": 0.0,
"completions/min_terminated_length": 96.0,
"epoch": 0.14186666666666667,
"grad_norm": 0.0348166786134243,
"kl": 0.1337127685546875,
"learning_rate": 1.8611111111111113e-06,
"loss": -0.1335,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.02746543101966381,
"mask/share_reasoning": 0.8509870767593384,
"mask/share_step_conf": 0.11373503506183624,
"num_tokens": 32540777.0,
"reward": 0.8784596920013428,
"reward_std": 0.28305697441101074,
"rewards/accuracy_reward_step": 0.3671875,
"rewards/final_brier_reward_step": 0.6788663864135742,
"rewards/format_reward_step": 0.91796875,
"rewards/step_l2_reward": 0.8210216760635376,
"step": 133
},
{
"adv/mean_abs_final_conf": 0.7420914769172668,
"adv/mean_abs_reasoning": 0.5315876007080078,
"adv/mean_abs_step_conf": 0.7382508516311646,
"adv/ratio_final_to_reasoning": 1.3959909447265029,
"adv/ratio_step_to_reasoning": 1.3887661236791589,
"adv/std_final_conf": 0.9098237156867981,
"adv/std_reasoning": 0.7755208015441895,
"adv/std_step_conf": 0.9349122047424316,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7413217623497997,
"calib/avg_num_step_conf": 6.3984375,
"calib/ece": 0.24170040485829958,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.94140625,
"calib/frac_conf_gt_0.9": 0.4331983805668016,
"calib/gap": 0.35083711615487323,
"calib/mean_conf": 0.5846963562753036,
"calib/mu_c": 0.783551401869159,
"calib/mu_w": 0.4327142857142857,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.19659919028340078,
"calib/std_conf": 0.41255279248347543,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.40404494382022466,
"calib/step_q_c_n": 712.0,
"calib/step_q_gap": 0.11917453345305401,
"calib/step_q_w": 0.28487041036717065,
"calib/step_q_w_n": 926.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2303.0,
"completions/max_terminated_length": 2303.0,
"completions/mean_length": 635.078125,
"completions/mean_terminated_length": 637.5686645507812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 186.0,
"epoch": 0.14293333333333333,
"grad_norm": 0.027463382109999657,
"kl": 0.124725341796875,
"learning_rate": 1.8333333333333333e-06,
"loss": -0.1329,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.027732379734516144,
"mask/share_reasoning": 0.859741747379303,
"mask/share_step_conf": 0.10861961543560028,
"num_tokens": 32812309.0,
"reward": 0.8942639827728271,
"reward_std": 0.24431997537612915,
"rewards/accuracy_reward_step": 0.41796875,
"rewards/final_brier_reward_step": 0.6869269609451294,
"rewards/format_reward_step": 0.94140625,
"rewards/step_l2_reward": 0.829725980758667,
"step": 134
},
{
"adv/mean_abs_final_conf": 0.7548559308052063,
"adv/mean_abs_reasoning": 0.6046093106269836,
"adv/mean_abs_step_conf": 0.6980199813842773,
"adv/ratio_final_to_reasoning": 1.2485019954826961,
"adv/ratio_step_to_reasoning": 1.1544975724247883,
"adv/std_final_conf": 0.9214678406715393,
"adv/std_reasoning": 0.8430864214897156,
"adv/std_step_conf": 0.9351249933242798,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.6043043043043043,
"calib/avg_num_step_conf": 6.984375,
"calib/ece": 0.26634146341463416,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.3821138211382114,
"calib/gap": 0.18251051051051054,
"calib/mean_conf": 0.5685365853658536,
"calib/mu_c": 0.650888888888889,
"calib/mu_w": 0.4683783783783784,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.9765625,
"calib/nonempty_step_conf_rate": 0.96484375,
"calib/pce": 0.14304878048780492,
"calib/std_conf": 0.3985658903954455,
"calib/step_conf_rate": 0.96484375,
"calib/step_q_c": 0.3298093220338983,
"calib/step_q_c_n": 944.0,
"calib/step_q_gap": 0.039714535304040455,
"calib/step_q_w": 0.29009478672985783,
"calib/step_q_w_n": 844.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2364.0,
"completions/max_terminated_length": 2364.0,
"completions/mean_length": 596.37890625,
"completions/mean_terminated_length": 598.7177124023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 101.0,
"epoch": 0.144,
"grad_norm": 0.041083741933107376,
"kl": 0.1374969482421875,
"learning_rate": 1.8055555555555557e-06,
"loss": -0.1289,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03051985614001751,
"mask/share_reasoning": 0.8427517414093018,
"mask/share_step_conf": 0.12282220274209976,
"num_tokens": 33070862.0,
"reward": 0.8903567790985107,
"reward_std": 0.24555067718029022,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.6455437541007996,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8398572206497192,
"step": 135
},
{
"adv/mean_abs_final_conf": 0.6729371547698975,
"adv/mean_abs_reasoning": 0.6114941835403442,
"adv/mean_abs_step_conf": 0.7410287261009216,
"adv/ratio_final_to_reasoning": 1.1004800583283054,
"adv/ratio_step_to_reasoning": 1.2118328285816493,
"adv/std_final_conf": 0.8736879229545593,
"adv/std_reasoning": 0.8429493308067322,
"adv/std_step_conf": 0.935082197189331,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8081365337462898,
"calib/avg_num_step_conf": 6.83203125,
"calib/ece": 0.1584337349397591,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.91796875,
"calib/frac_conf_gt_0.9": 0.3333333333333333,
"calib/gap": 0.44314556716995746,
"calib/mean_conf": 0.4978714859437751,
"calib/mu_c": 0.7221138211382114,
"calib/mu_w": 0.278968253968254,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9453125,
"calib/pce": 0.08116465863453821,
"calib/std_conf": 0.40774712558630793,
"calib/step_conf_rate": 0.9453125,
"calib/step_q_c": 0.3916142857142857,
"calib/step_q_c_n": 700.0,
"calib/step_q_gap": 0.15982210268282718,
"calib/step_q_w": 0.23179218303145852,
"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": 0.00390625,
"completions/max_length": 2936.0,
"completions/max_terminated_length": 2936.0,
"completions/mean_length": 570.48828125,
"completions/mean_terminated_length": 572.7255249023438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 59.0,
"epoch": 0.14506666666666668,
"grad_norm": 0.038291603326797485,
"kl": 0.137298583984375,
"learning_rate": 1.777777777777778e-06,
"loss": -0.1012,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03168662637472153,
"mask/share_reasoning": 0.8357338905334473,
"mask/share_step_conf": 0.1286732256412506,
"num_tokens": 33325395.0,
"reward": 0.9253720641136169,
"reward_std": 0.2551695704460144,
"rewards/accuracy_reward_step": 0.484375,
"rewards/final_brier_reward_step": 0.7408660054206848,
"rewards/format_reward_step": 0.91796875,
"rewards/step_l2_reward": 0.8294093608856201,
"step": 136
},
{
"adv/mean_abs_final_conf": 0.7062271237373352,
"adv/mean_abs_reasoning": 0.5881997346878052,
"adv/mean_abs_step_conf": 0.7120691537857056,
"adv/ratio_final_to_reasoning": 1.200658691408922,
"adv/ratio_step_to_reasoning": 1.2105907428939011,
"adv/std_final_conf": 0.8991337418556213,
"adv/std_reasoning": 0.8267337083816528,
"adv/std_step_conf": 0.9343873262405396,
"calib/answer_extract_rate": 0.953125,
"calib/auroc": 0.7438038793103448,
"calib/avg_num_step_conf": 6.62890625,
"calib/ece": 0.1884016393442623,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.9140625,
"calib/frac_conf_gt_0.9": 0.35655737704918034,
"calib/gap": 0.3403933189655174,
"calib/mean_conf": 0.5397950819672132,
"calib/mu_c": 0.7183620689655174,
"calib/mu_w": 0.37796874999999996,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.9453125,
"calib/pce": 0.12639344262295082,
"calib/std_conf": 0.39818795377709504,
"calib/step_conf_rate": 0.9453125,
"calib/step_q_c": 0.3570138888888889,
"calib/step_q_c_n": 720.0,
"calib/step_q_gap": 0.07836189298305468,
"calib/step_q_w": 0.2786519959058342,
"calib/step_q_w_n": 977.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2628.0,
"completions/max_terminated_length": 2628.0,
"completions/mean_length": 589.61328125,
"completions/mean_terminated_length": 594.2559204101562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 133.0,
"epoch": 0.14613333333333334,
"grad_norm": 0.03523816913366318,
"kl": 0.13226318359375,
"learning_rate": 1.75e-06,
"loss": -0.0867,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.029486628249287605,
"mask/share_reasoning": 0.8430880308151245,
"mask/share_step_conf": 0.11961278319358826,
"num_tokens": 33583320.0,
"reward": 0.8976198434829712,
"reward_std": 0.2519051134586334,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.697229266166687,
"rewards/format_reward_step": 0.9140625,
"rewards/step_l2_reward": 0.8245728611946106,
"step": 137
},
{
"adv/mean_abs_final_conf": 0.6920844316482544,
"adv/mean_abs_reasoning": 0.59056556224823,
"adv/mean_abs_step_conf": 0.7281466126441956,
"adv/ratio_final_to_reasoning": 1.171901099369139,
"adv/ratio_step_to_reasoning": 1.2329649054919607,
"adv/std_final_conf": 0.8724373579025269,
"adv/std_reasoning": 0.8266708254814148,
"adv/std_step_conf": 0.9192236661911011,
"calib/answer_extract_rate": 0.9453125,
"calib/auroc": 0.816345600920069,
"calib/avg_num_step_conf": 6.30078125,
"calib/ece": 0.1528512396694215,
"calib/final_conf_rate": 0.9453125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.3760330578512397,
"calib/gap": 0.46143617021276595,
"calib/mean_conf": 0.5332644628099174,
"calib/mu_c": 0.7125,
"calib/mu_w": 0.25106382978723407,
"calib/nonempty_final_conf_rate": 0.9453125,
"calib/nonempty_reasoning_rate": 0.96484375,
"calib/nonempty_step_conf_rate": 0.94921875,
"calib/pce": 0.03727272727272726,
"calib/std_conf": 0.41157087233183076,
"calib/step_conf_rate": 0.94921875,
"calib/step_q_c": 0.37403567447045705,
"calib/step_q_c_n": 897.0,
"calib/step_q_gap": 0.1139379091073286,
"calib/step_q_w": 0.26009776536312845,
"calib/step_q_w_n": 716.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2931.0,
"completions/max_terminated_length": 2931.0,
"completions/mean_length": 573.640625,
"completions/mean_terminated_length": 575.8902587890625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 18.0,
"epoch": 0.1472,
"grad_norm": 0.047151170670986176,
"kl": 0.141998291015625,
"learning_rate": 1.7222222222222224e-06,
"loss": -0.1102,
"mask/has_final_conf_rate": 0.9453125,
"mask/share_final_conf": 0.03197343647480011,
"mask/share_reasoning": 0.8420336246490479,
"mask/share_step_conf": 0.12208672612905502,
"num_tokens": 33834508.0,
"reward": 0.9239314794540405,
"reward_std": 0.19209349155426025,
"rewards/accuracy_reward_step": 0.578125,
"rewards/final_brier_reward_step": 0.7442609667778015,
"rewards/format_reward_step": 0.92578125,
"rewards/step_l2_reward": 0.8028206825256348,
"step": 138
},
{
"adv/mean_abs_final_conf": 0.7184191942214966,
"adv/mean_abs_reasoning": 0.5137280225753784,
"adv/mean_abs_step_conf": 0.7425556778907776,
"adv/ratio_final_to_reasoning": 1.3984426829978585,
"adv/ratio_step_to_reasoning": 1.4454256829679244,
"adv/std_final_conf": 0.9192709922790527,
"adv/std_reasoning": 0.7754294872283936,
"adv/std_step_conf": 0.9344884753227234,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8256553079947575,
"calib/avg_num_step_conf": 6.0703125,
"calib/ece": 0.12702811244979922,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.3132530120481928,
"calib/gap": 0.44588401048492793,
"calib/mean_conf": 0.5270281124497993,
"calib/mu_c": 0.7222142857142857,
"calib/mu_w": 0.27633027522935777,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.04590361445783132,
"calib/std_conf": 0.389914009690165,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.3720024570024571,
"calib/step_q_c_n": 814.0,
"calib/step_q_gap": 0.09763759213759221,
"calib/step_q_w": 0.2743648648648649,
"calib/step_q_w_n": 740.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3070.0,
"completions/max_terminated_length": 3070.0,
"completions/mean_length": 518.35546875,
"completions/mean_terminated_length": 524.5020141601562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 142.0,
"epoch": 0.14826666666666666,
"grad_norm": 0.05162464454770088,
"kl": 0.1543731689453125,
"learning_rate": 1.6944444444444446e-06,
"loss": -0.0241,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03312474489212036,
"mask/share_reasoning": 0.8317595720291138,
"mask/share_step_conf": 0.12339693307876587,
"num_tokens": 34070303.0,
"reward": 0.9758179187774658,
"reward_std": 0.17906738817691803,
"rewards/accuracy_reward_step": 0.546875,
"rewards/final_brier_reward_step": 0.793757438659668,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8547533750534058,
"step": 139
},
{
"adv/mean_abs_final_conf": 0.6915757060050964,
"adv/mean_abs_reasoning": 0.5147801637649536,
"adv/mean_abs_step_conf": 0.7419753074645996,
"adv/ratio_final_to_reasoning": 1.3434389175898138,
"adv/ratio_step_to_reasoning": 1.4413440137980575,
"adv/std_final_conf": 0.8760222792625427,
"adv/std_reasoning": 0.7577025890350342,
"adv/std_step_conf": 0.9336758255958557,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7467915365938258,
"calib/avg_num_step_conf": 5.85546875,
"calib/ece": 0.19032096774193544,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.5645161290322581,
"calib/gap": 0.38188817204301057,
"calib/mean_conf": 0.6862112903225808,
"calib/mu_c": 0.8294193548387095,
"calib/mu_w": 0.447531182795699,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.125766129032258,
"calib/std_conf": 0.39289094095617916,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.42450601092896173,
"calib/step_q_c_n": 915.0,
"calib/step_q_gap": 0.09055053147690689,
"calib/step_q_w": 0.33395547945205484,
"calib/step_q_w_n": 584.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2782.0,
"completions/max_terminated_length": 2782.0,
"completions/mean_length": 543.66015625,
"completions/mean_terminated_length": 543.66015625,
"completions/min_length": 42.0,
"completions/min_terminated_length": 42.0,
"epoch": 0.14933333333333335,
"grad_norm": 0.03653091937303543,
"kl": 0.1446380615234375,
"learning_rate": 1.6666666666666667e-06,
"loss": 0.0809,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.03261617571115494,
"mask/share_reasoning": 0.8453826308250427,
"mask/share_step_conf": 0.12200117111206055,
"num_tokens": 34314496.0,
"reward": 0.9528766870498657,
"reward_std": 0.21324776113033295,
"rewards/accuracy_reward_step": 0.60546875,
"rewards/final_brier_reward_step": 0.7541335821151733,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8399009704589844,
"step": 140
},
{
"adv/mean_abs_final_conf": 0.7118234038352966,
"adv/mean_abs_reasoning": 0.5580950379371643,
"adv/mean_abs_step_conf": 0.7270782589912415,
"adv/ratio_final_to_reasoning": 1.2754519489482372,
"adv/ratio_step_to_reasoning": 1.3027857435871038,
"adv/std_final_conf": 0.9149222373962402,
"adv/std_reasoning": 0.8099385499954224,
"adv/std_step_conf": 0.9347255825996399,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8293533389687235,
"calib/avg_num_step_conf": 6.03125,
"calib/ece": 0.12299595141700409,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.5465587044534413,
"calib/gap": 0.5450915750915752,
"calib/mean_conf": 0.6455870445344128,
"calib/mu_c": 0.8464102564102566,
"calib/mu_w": 0.3013186813186813,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.06850202429149801,
"calib/std_conf": 0.42056396432797855,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.41401574803149604,
"calib/step_q_c_n": 889.0,
"calib/step_q_gap": 0.16306918314599983,
"calib/step_q_w": 0.2509465648854962,
"calib/step_q_w_n": 655.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2371.0,
"completions/max_terminated_length": 2371.0,
"completions/mean_length": 558.2734375,
"completions/mean_terminated_length": 562.6693115234375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 109.0,
"epoch": 0.1504,
"grad_norm": 0.04100858420133591,
"kl": 0.148345947265625,
"learning_rate": 1.638888888888889e-06,
"loss": -0.1071,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.031309887766838074,
"mask/share_reasoning": 0.8481752872467041,
"mask/share_step_conf": 0.11270233988761902,
"num_tokens": 34564510.0,
"reward": 0.9690626859664917,
"reward_std": 0.2599467635154724,
"rewards/accuracy_reward_step": 0.609375,
"rewards/final_brier_reward_step": 0.7910523414611816,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.8376979231834412,
"step": 141
},
{
"adv/mean_abs_final_conf": 0.73613041639328,
"adv/mean_abs_reasoning": 0.5790297985076904,
"adv/mean_abs_step_conf": 0.7079474925994873,
"adv/ratio_final_to_reasoning": 1.271316982805511,
"adv/ratio_step_to_reasoning": 1.2226443171388608,
"adv/std_final_conf": 0.910010814666748,
"adv/std_reasoning": 0.8266434073448181,
"adv/std_step_conf": 0.9345359802246094,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.8122915832999866,
"calib/avg_num_step_conf": 6.58203125,
"calib/ece": 0.16934693877551027,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.42448979591836733,
"calib/gap": 0.45676937441643306,
"calib/mean_conf": 0.579795918367347,
"calib/mu_c": 0.8147058823529411,
"calib/mu_w": 0.357936507936508,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.98046875,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.13171428571428576,
"calib/std_conf": 0.41512668239023115,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.42689896373057,
"calib/step_q_c_n": 772.0,
"calib/step_q_gap": 0.15124178957941997,
"calib/step_q_w": 0.27565717415115004,
"calib/step_q_w_n": 913.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2918.0,
"completions/max_terminated_length": 2918.0,
"completions/mean_length": 592.953125,
"completions/mean_terminated_length": 595.2784423828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 2.0,
"epoch": 0.15146666666666667,
"grad_norm": 0.037801120430231094,
"kl": 0.13177490234375,
"learning_rate": 1.6111111111111113e-06,
"loss": -0.0517,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.031895555555820465,
"mask/share_reasoning": 0.834732174873352,
"mask/share_step_conf": 0.1294659972190857,
"num_tokens": 34821466.0,
"reward": 0.9343152046203613,
"reward_std": 0.235845148563385,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/final_brier_reward_step": 0.7522230744361877,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.8343761563301086,
"step": 142
},
{
"adv/mean_abs_final_conf": 0.7163206934928894,
"adv/mean_abs_reasoning": 0.6285260915756226,
"adv/mean_abs_step_conf": 0.759658694267273,
"adv/ratio_final_to_reasoning": 1.1396833052661006,
"adv/ratio_step_to_reasoning": 1.208635097968519,
"adv/std_final_conf": 0.8896148204803467,
"adv/std_reasoning": 0.8430455923080444,
"adv/std_step_conf": 0.9334934949874878,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7944293478260871,
"calib/avg_num_step_conf": 6.28125,
"calib/ece": 0.17691358024691362,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.921875,
"calib/frac_conf_gt_0.9": 0.43621399176954734,
"calib/gap": 0.42025747282608705,
"calib/mean_conf": 0.5980658436213991,
"calib/mu_c": 0.796953125,
"calib/mu_w": 0.376695652173913,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.97265625,
"calib/nonempty_step_conf_rate": 0.94921875,
"calib/pce": 0.1241152263374486,
"calib/std_conf": 0.4057670222221783,
"calib/step_conf_rate": 0.94921875,
"calib/step_q_c": 0.4118918918918919,
"calib/step_q_c_n": 740.0,
"calib/step_q_gap": 0.15042875825133895,
"calib/step_q_w": 0.26146313364055296,
"calib/step_q_w_n": 868.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2854.0,
"completions/max_terminated_length": 2854.0,
"completions/mean_length": 607.9296875,
"completions/mean_terminated_length": 607.9296875,
"completions/min_length": 201.0,
"completions/min_terminated_length": 201.0,
"epoch": 0.15253333333333333,
"grad_norm": 0.030441317707300186,
"kl": 0.1333465576171875,
"learning_rate": 1.5833333333333333e-06,
"loss": -0.0023,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.029380368068814278,
"mask/share_reasoning": 0.8472878932952881,
"mask/share_step_conf": 0.12333173304796219,
"num_tokens": 35084432.0,
"reward": 0.9215575456619263,
"reward_std": 0.26581722497940063,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.734161376953125,
"rewards/format_reward_step": 0.921875,
"rewards/step_l2_reward": 0.8245788812637329,
"step": 143
},
{
"adv/mean_abs_final_conf": 0.7154229879379272,
"adv/mean_abs_reasoning": 0.5836465358734131,
"adv/mean_abs_step_conf": 0.7379428148269653,
"adv/ratio_final_to_reasoning": 1.2257812630847091,
"adv/ratio_step_to_reasoning": 1.2643659637637557,
"adv/std_final_conf": 0.9021844267845154,
"adv/std_reasoning": 0.8100001811981201,
"adv/std_step_conf": 0.9352454543113708,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7495052117693626,
"calib/avg_num_step_conf": 6.140625,
"calib/ece": 0.22224859437751004,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.4819277108433735,
"calib/gap": 0.3838138672648106,
"calib/mean_conf": 0.5932534136546185,
"calib/mu_c": 0.7566440559440559,
"calib/mu_w": 0.37283018867924533,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.12060240963855423,
"calib/std_conf": 0.4278204237038552,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.41317606444188715,
"calib/step_q_c_n": 869.0,
"calib/step_q_gap": 0.11862414410049316,
"calib/step_q_w": 0.294551920341394,
"calib/step_q_w_n": 703.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2855.0,
"completions/max_terminated_length": 2855.0,
"completions/mean_length": 544.4375,
"completions/mean_terminated_length": 546.5725708007812,
"completions/min_length": 0.0,
"completions/min_terminated_length": 79.0,
"epoch": 0.1536,
"grad_norm": 0.024527657777071,
"kl": 0.16070556640625,
"learning_rate": 1.5555555555555558e-06,
"loss": -0.1027,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03337983787059784,
"mask/share_reasoning": 0.8375337719917297,
"mask/share_step_conf": 0.12518012523651123,
"num_tokens": 35327936.0,
"reward": 0.9219741225242615,
"reward_std": 0.24387477338314056,
"rewards/accuracy_reward_step": 0.5625,
"rewards/final_brier_reward_step": 0.7212362885475159,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.821149468421936,
"step": 144
},
{
"adv/mean_abs_final_conf": 0.6954442262649536,
"adv/mean_abs_reasoning": 0.6353583931922913,
"adv/mean_abs_step_conf": 0.7420624494552612,
"adv/ratio_final_to_reasoning": 1.0945699839908738,
"adv/ratio_step_to_reasoning": 1.167943097008362,
"adv/std_final_conf": 0.8785400390625,
"adv/std_reasoning": 0.8430765867233276,
"adv/std_step_conf": 0.9350871443748474,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7888645391042323,
"calib/avg_num_step_conf": 6.890625,
"calib/ece": 0.15010796221322534,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.4534412955465587,
"calib/gap": 0.4093300004565585,
"calib/mean_conf": 0.6260053981106612,
"calib/mu_c": 0.7884116331096197,
"calib/mu_w": 0.3790816326530612,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.08643724696356272,
"calib/std_conf": 0.39533294414075476,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4404112554112555,
"calib/step_q_c_n": 924.0,
"calib/step_q_gap": 0.1496493506493507,
"calib/step_q_w": 0.2907619047619048,
"calib/step_q_w_n": 840.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 3033.0,
"completions/max_terminated_length": 3033.0,
"completions/mean_length": 548.89453125,
"completions/mean_terminated_length": 551.047119140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.15466666666666667,
"grad_norm": 0.029142454266548157,
"kl": 0.1400146484375,
"learning_rate": 1.527777777777778e-06,
"loss": 0.006,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.033541686832904816,
"mask/share_reasoning": 0.8256252408027649,
"mask/share_step_conf": 0.1369268149137497,
"num_tokens": 35571157.0,
"reward": 0.9419224262237549,
"reward_std": 0.22772300243377686,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.7556396722793579,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.8219549655914307,
"step": 145
},
{
"adv/mean_abs_final_conf": 0.7435042858123779,
"adv/mean_abs_reasoning": 0.54443359375,
"adv/mean_abs_step_conf": 0.752627968788147,
"adv/ratio_final_to_reasoning": 1.365647333940583,
"adv/ratio_step_to_reasoning": 1.3824054529848655,
"adv/std_final_conf": 0.9029568433761597,
"adv/std_reasoning": 0.8098462820053101,
"adv/std_step_conf": 0.9342830181121826,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.7773940132737369,
"calib/avg_num_step_conf": 6.75390625,
"calib/ece": 0.19571428571428573,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.3469387755102041,
"calib/gap": 0.4054591629418936,
"calib/mean_conf": 0.519469387755102,
"calib/mu_c": 0.7478504672897197,
"calib/mu_w": 0.3423913043478261,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.96484375,
"calib/pce": 0.1392244897959184,
"calib/std_conf": 0.41661919685168064,
"calib/step_conf_rate": 0.96484375,
"calib/step_q_c": 0.38796969696969696,
"calib/step_q_c_n": 660.0,
"calib/step_q_gap": 0.1105889673158148,
"calib/step_q_w": 0.27738072965388216,
"calib/step_q_w_n": 1069.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2628.0,
"completions/max_terminated_length": 2628.0,
"completions/mean_length": 609.6015625,
"completions/mean_terminated_length": 614.4015502929688,
"completions/min_length": 0.0,
"completions/min_terminated_length": 157.0,
"epoch": 0.15573333333333333,
"grad_norm": 0.057016413658857346,
"kl": 0.2330474853515625,
"learning_rate": 1.5e-06,
"loss": -0.0669,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.030465461313724518,
"mask/share_reasoning": 0.8398491144180298,
"mask/share_step_conf": 0.12187288701534271,
"num_tokens": 35834431.0,
"reward": 0.9122192859649658,
"reward_std": 0.24562488496303558,
"rewards/accuracy_reward_step": 0.41796875,
"rewards/final_brier_reward_step": 0.7225179672241211,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.8316080570220947,
"step": 146
},
{
"adv/mean_abs_final_conf": 0.7490389347076416,
"adv/mean_abs_reasoning": 0.575813353061676,
"adv/mean_abs_step_conf": 0.7271679043769836,
"adv/ratio_final_to_reasoning": 1.3008363399787486,
"adv/ratio_step_to_reasoning": 1.2628534932552977,
"adv/std_final_conf": 0.9133417010307312,
"adv/std_reasoning": 0.8266557455062866,
"adv/std_step_conf": 0.9348275065422058,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7377620013522652,
"calib/avg_num_step_conf": 6.15234375,
"calib/ece": 0.24530364372469646,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.47368421052631576,
"calib/gap": 0.3844475997295469,
"calib/mean_conf": 0.5774493927125506,
"calib/mu_c": 0.8031372549019608,
"calib/mu_w": 0.41868965517241385,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.20489878542510132,
"calib/std_conf": 0.4354433411011676,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.4652090032154341,
"calib/step_q_c_n": 622.0,
"calib/step_q_gap": 0.16378193081249603,
"calib/step_q_w": 0.3014270724029381,
"calib/step_q_w_n": 953.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2338.0,
"completions/max_terminated_length": 2338.0,
"completions/mean_length": 565.53515625,
"completions/mean_terminated_length": 567.7529907226562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 166.0,
"epoch": 0.1568,
"grad_norm": 0.043372150510549545,
"kl": 0.1708984375,
"learning_rate": 1.4722222222222225e-06,
"loss": -0.1202,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.030713319778442383,
"mask/share_reasoning": 0.8421392440795898,
"mask/share_step_conf": 0.12324120104312897,
"num_tokens": 36082888.0,
"reward": 0.9022128582000732,
"reward_std": 0.2670474946498871,
"rewards/accuracy_reward_step": 0.40234375,
"rewards/final_brier_reward_step": 0.6899276971817017,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8441853523254395,
"step": 147
},
{
"adv/mean_abs_final_conf": 0.6602885127067566,
"adv/mean_abs_reasoning": 0.5762661695480347,
"adv/mean_abs_step_conf": 0.7480043172836304,
"adv/ratio_final_to_reasoning": 1.1458047471789305,
"adv/ratio_step_to_reasoning": 1.2980187920284993,
"adv/std_final_conf": 0.8627505898475647,
"adv/std_reasoning": 0.7929645776748657,
"adv/std_step_conf": 0.9347200989723206,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8135960591133004,
"calib/avg_num_step_conf": 6.89453125,
"calib/ece": 0.16544000000000003,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.512,
"calib/gap": 0.484623973727422,
"calib/mean_conf": 0.63032,
"calib/mu_c": 0.8338620689655172,
"calib/mu_w": 0.3492380952380952,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.10788000000000003,
"calib/std_conf": 0.4194709734892273,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.44010273972602737,
"calib/step_q_c_n": 876.0,
"calib/step_q_gap": 0.17063142364053802,
"calib/step_q_w": 0.26947131608548935,
"calib/step_q_w_n": 889.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2956.0,
"completions/max_terminated_length": 2956.0,
"completions/mean_length": 543.12109375,
"completions/mean_terminated_length": 545.2510375976562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 96.0,
"epoch": 0.15786666666666666,
"grad_norm": 0.02674427255988121,
"kl": 0.15020751953125,
"learning_rate": 1.4444444444444445e-06,
"loss": 0.0487,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03537040203809738,
"mask/share_reasoning": 0.8225604891777039,
"mask/share_step_conf": 0.13816285133361816,
"num_tokens": 36327039.0,
"reward": 0.9648141860961914,
"reward_std": 0.20383086800575256,
"rewards/accuracy_reward_step": 0.56640625,
"rewards/final_brier_reward_step": 0.7801804542541504,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.843979001045227,
"step": 148
},
{
"adv/mean_abs_final_conf": 0.7208631038665771,
"adv/mean_abs_reasoning": 0.5675036907196045,
"adv/mean_abs_step_conf": 0.7488936185836792,
"adv/ratio_final_to_reasoning": 1.2702350938942975,
"adv/ratio_step_to_reasoning": 1.319627750145676,
"adv/std_final_conf": 0.9123585224151611,
"adv/std_reasoning": 0.8098829388618469,
"adv/std_step_conf": 0.933817446231842,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.8856749311294766,
"calib/avg_num_step_conf": 6.6015625,
"calib/ece": 0.0940080971659919,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.43724696356275305,
"calib/gap": 0.5695808736717829,
"calib/mean_conf": 0.5731174089068826,
"calib/mu_c": 0.8521428571428572,
"calib/mu_w": 0.28256198347107436,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.07850202429149797,
"calib/std_conf": 0.41840017066719615,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.4510818307905687,
"calib/step_q_c_n": 721.0,
"calib/step_q_gap": 0.17016335813834987,
"calib/step_q_w": 0.2809184726522188,
"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": 0.01171875,
"completions/max_length": 2757.0,
"completions/max_terminated_length": 2757.0,
"completions/mean_length": 591.9296875,
"completions/mean_terminated_length": 598.9486694335938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 78.0,
"epoch": 0.15893333333333334,
"grad_norm": 0.05520571395754814,
"kl": 0.131378173828125,
"learning_rate": 1.4166666666666667e-06,
"loss": -0.0464,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.03246816247701645,
"mask/share_reasoning": 0.8309919834136963,
"mask/share_step_conf": 0.12482112646102905,
"num_tokens": 36583029.0,
"reward": 0.9735437035560608,
"reward_std": 0.22069458663463593,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.8104152679443359,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.8483908772468567,
"step": 149
},
{
"adv/mean_abs_final_conf": 0.7165673971176147,
"adv/mean_abs_reasoning": 0.6197329759597778,
"adv/mean_abs_step_conf": 0.7352439165115356,
"adv/ratio_final_to_reasoning": 1.1562518454143413,
"adv/ratio_step_to_reasoning": 1.1863882430539807,
"adv/std_final_conf": 0.9072695374488831,
"adv/std_reasoning": 0.8589852452278137,
"adv/std_step_conf": 0.9348389506340027,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.793323628664064,
"calib/avg_num_step_conf": 5.984375,
"calib/ece": 0.20621951219512194,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.4634146341463415,
"calib/gap": 0.3964619863693509,
"calib/mean_conf": 0.6045934959349594,
"calib/mu_c": 0.7963779527559055,
"calib/mu_w": 0.3999159663865546,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.14727642276422764,
"calib/std_conf": 0.4129662152247405,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.44794330042313124,
"calib/step_q_c_n": 709.0,
"calib/step_q_gap": 0.13494208535630253,
"calib/step_q_w": 0.3130012150668287,
"calib/step_q_w_n": 823.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2953.0,
"completions/max_terminated_length": 2953.0,
"completions/mean_length": 507.625,
"completions/mean_terminated_length": 507.625,
"completions/min_length": 117.0,
"completions/min_terminated_length": 117.0,
"epoch": 0.16,
"grad_norm": 0.027483409270644188,
"kl": 0.159912109375,
"learning_rate": 1.3888888888888892e-06,
"loss": -0.0122,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.03679945319890976,
"mask/share_reasoning": 0.8243503570556641,
"mask/share_step_conf": 0.1388501524925232,
"num_tokens": 36817941.0,
"reward": 0.934812068939209,
"reward_std": 0.23240233957767487,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.7398660182952881,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.838351845741272,
"step": 150
},
{
"adv/mean_abs_final_conf": 0.7235825657844543,
"adv/mean_abs_reasoning": 0.6277079582214355,
"adv/mean_abs_step_conf": 0.7511394023895264,
"adv/ratio_final_to_reasoning": 1.1527376008337897,
"adv/ratio_step_to_reasoning": 1.1966383292603535,
"adv/std_final_conf": 0.8992826342582703,
"adv/std_reasoning": 0.8590068221092224,
"adv/std_step_conf": 0.9348013997077942,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7804633867276887,
"calib/avg_num_step_conf": 6.67578125,
"calib/ece": 0.2073360655737705,
"calib/final_conf_rate": 0.953125,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.3319672131147541,
"calib/gap": 0.38922196796338665,
"calib/mean_conf": 0.4737295081967213,
"calib/mu_c": 0.716195652173913,
"calib/mu_w": 0.32697368421052636,
"calib/nonempty_final_conf_rate": 0.953125,
"calib/nonempty_reasoning_rate": 0.98046875,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.15200819672131147,
"calib/std_conf": 0.42256683508395787,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.44220802919708024,
"calib/step_q_c_n": 548.0,
"calib/step_q_gap": 0.15332775357261852,
"calib/step_q_w": 0.2888802756244617,
"calib/step_q_w_n": 1161.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2809.0,
"completions/max_terminated_length": 2809.0,
"completions/mean_length": 589.71875,
"completions/mean_terminated_length": 594.3621826171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.16106666666666666,
"grad_norm": 0.0312860906124115,
"kl": 0.1264190673828125,
"learning_rate": 1.3611111111111112e-06,
"loss": -0.0412,
"mask/has_final_conf_rate": 0.953125,
"mask/share_final_conf": 0.03059176355600357,
"mask/share_reasoning": 0.8369712233543396,
"mask/share_step_conf": 0.12462448328733444,
"num_tokens": 37075933.0,
"reward": 0.8848069906234741,
"reward_std": 0.24623329937458038,
"rewards/accuracy_reward_step": 0.359375,
"rewards/final_brier_reward_step": 0.7045589685440063,
"rewards/format_reward_step": 0.92578125,
"rewards/step_l2_reward": 0.8080236911773682,
"step": 151
},
{
"adv/mean_abs_final_conf": 0.7726234197616577,
"adv/mean_abs_reasoning": 0.6658381223678589,
"adv/mean_abs_step_conf": 0.7652601003646851,
"adv/ratio_final_to_reasoning": 1.160377265594298,
"adv/ratio_step_to_reasoning": 1.1493185425359258,
"adv/std_final_conf": 0.9337043762207031,
"adv/std_reasoning": 0.8431472182273865,
"adv/std_step_conf": 0.9347693920135498,
"calib/answer_extract_rate": 0.94921875,
"calib/auroc": 0.7234513274336284,
"calib/avg_num_step_conf": 6.65234375,
"calib/ece": 0.19057613168724277,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9296875,
"calib/frac_conf_gt_0.9": 0.3004115226337449,
"calib/gap": 0.34416814159292025,
"calib/mean_conf": 0.4730452674897119,
"calib/mu_c": 0.6571681415929203,
"calib/mu_w": 0.313,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.96484375,
"calib/pce": 0.09930041152263372,
"calib/std_conf": 0.40861578289777173,
"calib/step_conf_rate": 0.96484375,
"calib/step_q_c": 0.4087291981845689,
"calib/step_q_c_n": 661.0,
"calib/step_q_gap": 0.10920904463370512,
"calib/step_q_w": 0.29952015355086375,
"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": 0.00390625,
"completions/max_length": 2884.0,
"completions/max_terminated_length": 2884.0,
"completions/mean_length": 585.3671875,
"completions/mean_terminated_length": 587.6627807617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 165.0,
"epoch": 0.16213333333333332,
"grad_norm": 0.026001159101724625,
"kl": 0.13336181640625,
"learning_rate": 1.3333333333333334e-06,
"loss": -0.0659,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.03251414746046066,
"mask/share_reasoning": 0.8299437761306763,
"mask/share_step_conf": 0.1336357444524765,
"num_tokens": 37331179.0,
"reward": 0.8980525732040405,
"reward_std": 0.24627810716629028,
"rewards/accuracy_reward_step": 0.44140625,
"rewards/final_brier_reward_step": 0.7018734216690063,
"rewards/format_reward_step": 0.9296875,
"rewards/step_l2_reward": 0.8200129270553589,
"step": 152
},
{
"adv/mean_abs_final_conf": 0.7555943727493286,
"adv/mean_abs_reasoning": 0.5032080411911011,
"adv/mean_abs_step_conf": 0.7314299941062927,
"adv/ratio_final_to_reasoning": 1.5015546471809658,
"adv/ratio_step_to_reasoning": 1.4535339943594439,
"adv/std_final_conf": 0.9189577698707581,
"adv/std_reasoning": 0.7753830552101135,
"adv/std_step_conf": 0.9348416924476624,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.6744253615702478,
"calib/avg_num_step_conf": 6.59765625,
"calib/ece": 0.2887550200803213,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.30522088353413657,
"calib/gap": 0.21644176136363646,
"calib/mean_conf": 0.46489959839357425,
"calib/mu_c": 0.5700781250000001,
"calib/mu_w": 0.35363636363636364,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.11979919678714858,
"calib/std_conf": 0.41180386064046676,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.3873626373626374,
"calib/step_q_c_n": 728.0,
"calib/step_q_gap": 0.1172481732627414,
"calib/step_q_w": 0.270114464099896,
"calib/step_q_w_n": 961.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 3056.0,
"completions/max_terminated_length": 3056.0,
"completions/mean_length": 584.50390625,
"completions/mean_terminated_length": 584.50390625,
"completions/min_length": 170.0,
"completions/min_terminated_length": 170.0,
"epoch": 0.1632,
"grad_norm": 0.04574429616332054,
"kl": 0.139495849609375,
"learning_rate": 1.3055555555555556e-06,
"loss": 0.0269,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03054162487387657,
"mask/share_reasoning": 0.8452918529510498,
"mask/share_step_conf": 0.1241665631532669,
"num_tokens": 37588132.0,
"reward": 0.8886984586715698,
"reward_std": 0.22010064125061035,
"rewards/accuracy_reward_step": 0.5,
"rewards/final_brier_reward_step": 0.6561176180839539,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8298731446266174,
"step": 153
},
{
"adv/mean_abs_final_conf": 0.6796963810920715,
"adv/mean_abs_reasoning": 0.5204298496246338,
"adv/mean_abs_step_conf": 0.7172330617904663,
"adv/ratio_final_to_reasoning": 1.3060288174906005,
"adv/ratio_step_to_reasoning": 1.3781551198644335,
"adv/std_final_conf": 0.9021387100219727,
"adv/std_reasoning": 0.7927193641662598,
"adv/std_step_conf": 0.9346775412559509,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7935254638768259,
"calib/avg_num_step_conf": 6.34765625,
"calib/ece": 0.1801593625498008,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.3705179282868526,
"calib/gap": 0.44687261481773927,
"calib/mean_conf": 0.5024701195219123,
"calib/mu_c": 0.7677450980392158,
"calib/mu_w": 0.3208724832214765,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.13812749003984065,
"calib/std_conf": 0.429219937891115,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.45802980132450327,
"calib/step_q_c_n": 604.0,
"calib/step_q_gap": 0.17397495313645234,
"calib/step_q_w": 0.28405484818805093,
"calib/step_q_w_n": 1021.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2211.0,
"completions/max_terminated_length": 2211.0,
"completions/mean_length": 532.17578125,
"completions/mean_terminated_length": 534.2627563476562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 178.0,
"epoch": 0.16426666666666667,
"grad_norm": 0.026894347742199898,
"kl": 0.1571197509765625,
"learning_rate": 1.2777777777777779e-06,
"loss": -0.0654,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03247256577014923,
"mask/share_reasoning": 0.8317904472351074,
"mask/share_step_conf": 0.13183078169822693,
"num_tokens": 37828809.0,
"reward": 0.948471188545227,
"reward_std": 0.20287421345710754,
"rewards/accuracy_reward_step": 0.3984375,
"rewards/final_brier_reward_step": 0.7616676092147827,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8610559105873108,
"step": 154
},
{
"adv/mean_abs_final_conf": 0.7484700679779053,
"adv/mean_abs_reasoning": 0.46591895818710327,
"adv/mean_abs_step_conf": 0.729465901851654,
"adv/ratio_final_to_reasoning": 1.6064383189948142,
"adv/ratio_step_to_reasoning": 1.565649753103019,
"adv/std_final_conf": 0.9147512912750244,
"adv/std_reasoning": 0.7575052976608276,
"adv/std_step_conf": 0.9337078928947449,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.7547084623323014,
"calib/avg_num_step_conf": 6.0546875,
"calib/ece": 0.192244094488189,
"calib/final_conf_rate": 0.9921875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.2755905511811024,
"calib/gap": 0.38964912280701747,
"calib/mean_conf": 0.4051574803149606,
"calib/mu_c": 0.6383333333333332,
"calib/mu_w": 0.24868421052631576,
"calib/nonempty_final_conf_rate": 0.9921875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.09791338582677167,
"calib/std_conf": 0.4184680709218479,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.3828700906344411,
"calib/step_q_c_n": 662.0,
"calib/step_q_gap": 0.1125885591029096,
"calib/step_q_w": 0.2702815315315315,
"calib/step_q_w_n": 888.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1205.0,
"completions/max_terminated_length": 1205.0,
"completions/mean_length": 477.25390625,
"completions/mean_terminated_length": 479.1255187988281,
"completions/min_length": 0.0,
"completions/min_terminated_length": 160.0,
"epoch": 0.16533333333333333,
"grad_norm": 0.03665679320693016,
"kl": 0.157745361328125,
"learning_rate": 1.25e-06,
"loss": -0.1132,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03502430021762848,
"mask/share_reasoning": 0.8271464109420776,
"mask/share_step_conf": 0.1339230239391327,
"num_tokens": 38058202.0,
"reward": 0.9258972406387329,
"reward_std": 0.214618980884552,
"rewards/accuracy_reward_step": 0.3984375,
"rewards/final_brier_reward_step": 0.7388640642166138,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8402740955352783,
"step": 155
},
{
"adv/mean_abs_final_conf": 0.7042100429534912,
"adv/mean_abs_reasoning": 0.5381708145141602,
"adv/mean_abs_step_conf": 0.7314833998680115,
"adv/ratio_final_to_reasoning": 1.3085251447335078,
"adv/ratio_step_to_reasoning": 1.3592030265118826,
"adv/std_final_conf": 0.9068632125854492,
"adv/std_reasoning": 0.792761504650116,
"adv/std_step_conf": 0.9347695112228394,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.7569124423963133,
"calib/avg_num_step_conf": 6.62109375,
"calib/ece": 0.21859999999999996,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.392,
"calib/gap": 0.3890847414234512,
"calib/mean_conf": 0.5012399999999999,
"calib/mu_c": 0.6973387096774194,
"calib/mu_w": 0.3082539682539682,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.11191999999999998,
"calib/std_conf": 0.4379913953492694,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.3803166869671133,
"calib/step_q_c_n": 821.0,
"calib/step_q_gap": 0.07485902106322312,
"calib/step_q_w": 0.3054576659038902,
"calib/step_q_w_n": 874.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2618.0,
"completions/max_terminated_length": 2618.0,
"completions/mean_length": 529.1640625,
"completions/mean_terminated_length": 533.3306884765625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 121.0,
"epoch": 0.1664,
"grad_norm": 0.03280864655971527,
"kl": 0.1468963623046875,
"learning_rate": 1.2222222222222223e-06,
"loss": -0.0633,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.033598169684410095,
"mask/share_reasoning": 0.8221753835678101,
"mask/share_step_conf": 0.13641397655010223,
"num_tokens": 38298428.0,
"reward": 0.9258990287780762,
"reward_std": 0.21684980392456055,
"rewards/accuracy_reward_step": 0.484375,
"rewards/final_brier_reward_step": 0.7279226779937744,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8340317010879517,
"step": 156
},
{
"adv/mean_abs_final_conf": 0.7056574821472168,
"adv/mean_abs_reasoning": 0.575699508190155,
"adv/mean_abs_step_conf": 0.7127252221107483,
"adv/ratio_final_to_reasoning": 1.2257392478336742,
"adv/ratio_step_to_reasoning": 1.2380160343568216,
"adv/std_final_conf": 0.9043259620666504,
"adv/std_reasoning": 0.8265734314918518,
"adv/std_step_conf": 0.9344764947891235,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.8329150579150579,
"calib/avg_num_step_conf": 7.14453125,
"calib/ece": 0.17561752988047816,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.40239043824701193,
"calib/gap": 0.5030920205920206,
"calib/mean_conf": 0.5243027888446214,
"calib/mu_c": 0.7467857142857143,
"calib/mu_w": 0.24369369369369365,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.07107569721115545,
"calib/std_conf": 0.43926211664983866,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4437849462365592,
"calib/step_q_c_n": 930.0,
"calib/step_q_gap": 0.17983611420096407,
"calib/step_q_w": 0.2639488320355951,
"calib/step_q_w_n": 899.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2560.0,
"completions/max_terminated_length": 2560.0,
"completions/mean_length": 547.28515625,
"completions/mean_terminated_length": 547.28515625,
"completions/min_length": 128.0,
"completions/min_terminated_length": 128.0,
"epoch": 0.16746666666666668,
"grad_norm": 0.03486839309334755,
"kl": 0.1378326416015625,
"learning_rate": 1.1944444444444446e-06,
"loss": -0.0098,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03291250020265579,
"mask/share_reasoning": 0.8276211619377136,
"mask/share_step_conf": 0.13946637511253357,
"num_tokens": 38542261.0,
"reward": 0.9581904411315918,
"reward_std": 0.22784942388534546,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/final_brier_reward_step": 0.7759785056114197,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8380587100982666,
"step": 157
},
{
"adv/mean_abs_final_conf": 0.7180853486061096,
"adv/mean_abs_reasoning": 0.6382078528404236,
"adv/mean_abs_step_conf": 0.7345453500747681,
"adv/ratio_final_to_reasoning": 1.1251590612841589,
"adv/ratio_step_to_reasoning": 1.1509500342334906,
"adv/std_final_conf": 0.8862758874893188,
"adv/std_reasoning": 0.8430286645889282,
"adv/std_step_conf": 0.9345474243164062,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7362024500322373,
"calib/avg_num_step_conf": 6.296875,
"calib/ece": 0.21848605577689245,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9609375,
"calib/frac_conf_gt_0.9": 0.4302788844621514,
"calib/gap": 0.35798001289490644,
"calib/mean_conf": 0.563187250996016,
"calib/mu_c": 0.7200709219858156,
"calib/mu_w": 0.36209090909090913,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.10996015936254981,
"calib/std_conf": 0.422513113963258,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.4147051597051597,
"calib/step_q_c_n": 814.0,
"calib/step_q_gap": 0.09493448301343033,
"calib/step_q_w": 0.31977067669172937,
"calib/step_q_w_n": 798.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2601.0,
"completions/max_terminated_length": 2601.0,
"completions/mean_length": 527.61328125,
"completions/mean_terminated_length": 527.61328125,
"completions/min_length": 128.0,
"completions/min_terminated_length": 128.0,
"epoch": 0.16853333333333334,
"grad_norm": 0.03904188424348831,
"kl": 0.1563262939453125,
"learning_rate": 1.1666666666666668e-06,
"loss": -0.0257,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03514157235622406,
"mask/share_reasoning": 0.8268975019454956,
"mask/share_step_conf": 0.13796088099479675,
"num_tokens": 38782570.0,
"reward": 0.9303572773933411,
"reward_std": 0.20081466436386108,
"rewards/accuracy_reward_step": 0.55078125,
"rewards/final_brier_reward_step": 0.7250785231590271,
"rewards/format_reward_step": 0.9609375,
"rewards/step_l2_reward": 0.8332923054695129,
"step": 158
},
{
"adv/mean_abs_final_conf": 0.7105034589767456,
"adv/mean_abs_reasoning": 0.549357533454895,
"adv/mean_abs_step_conf": 0.7166721820831299,
"adv/ratio_final_to_reasoning": 1.2933352429125857,
"adv/ratio_step_to_reasoning": 1.3045642199097507,
"adv/std_final_conf": 0.897257924079895,
"adv/std_reasoning": 0.8098735809326172,
"adv/std_step_conf": 0.9343751072883606,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7527901785714285,
"calib/avg_num_step_conf": 6.2109375,
"calib/ece": 0.22258064516129036,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.43548387096774194,
"calib/gap": 0.39795168067226894,
"calib/mean_conf": 0.5548387096774194,
"calib/mu_c": 0.7345588235294118,
"calib/mu_w": 0.3366071428571429,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.11451612903225813,
"calib/std_conf": 0.43352732523725346,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.4350317662007624,
"calib/step_q_c_n": 787.0,
"calib/step_q_gap": 0.16991096918955445,
"calib/step_q_w": 0.26512079701120794,
"calib/step_q_w_n": 803.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2926.0,
"completions/max_terminated_length": 2926.0,
"completions/mean_length": 500.1171875,
"completions/mean_terminated_length": 500.1171875,
"completions/min_length": 144.0,
"completions/min_terminated_length": 144.0,
"epoch": 0.1696,
"grad_norm": 0.034035515040159225,
"kl": 0.15728759765625,
"learning_rate": 1.138888888888889e-06,
"loss": 0.0961,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.03542216867208481,
"mask/share_reasoning": 0.8289942145347595,
"mask/share_step_conf": 0.13558357954025269,
"num_tokens": 39015384.0,
"reward": 0.9383366107940674,
"reward_std": 0.21544626355171204,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.7279398441314697,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8510770797729492,
"step": 159
},
{
"adv/mean_abs_final_conf": 0.6851258873939514,
"adv/mean_abs_reasoning": 0.505584716796875,
"adv/mean_abs_step_conf": 0.7561661005020142,
"adv/ratio_final_to_reasoning": 1.3551158977560813,
"adv/ratio_step_to_reasoning": 1.4956268957113539,
"adv/std_final_conf": 0.8633057475090027,
"adv/std_reasoning": 0.7754652500152588,
"adv/std_step_conf": 0.9348845481872559,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.7667584940312214,
"calib/avg_num_step_conf": 6.75,
"calib/ece": 0.21218623481781382,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.4048582995951417,
"calib/gap": 0.40031680440771367,
"calib/mean_conf": 0.5177732793522267,
"calib/mu_c": 0.7219834710743803,
"calib/mu_w": 0.32166666666666666,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.12004048582995956,
"calib/std_conf": 0.4386465903080075,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.4451233671988389,
"calib/step_q_c_n": 689.0,
"calib/step_q_gap": 0.15851124015360313,
"calib/step_q_w": 0.28661212704523575,
"calib/step_q_w_n": 1039.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2861.0,
"completions/max_terminated_length": 2861.0,
"completions/mean_length": 559.60546875,
"completions/mean_terminated_length": 564.0117797851562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 153.0,
"epoch": 0.17066666666666666,
"grad_norm": 0.025723211467266083,
"kl": 0.1482696533203125,
"learning_rate": 1.111111111111111e-06,
"loss": 0.0134,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.03189799189567566,
"mask/share_reasoning": 0.8309605121612549,
"mask/share_step_conf": 0.12932898104190826,
"num_tokens": 39263483.0,
"reward": 0.9138648509979248,
"reward_std": 0.22909203171730042,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/final_brier_reward_step": 0.7227054834365845,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8190867304801941,
"step": 160
},
{
"adv/mean_abs_final_conf": 0.6754765510559082,
"adv/mean_abs_reasoning": 0.5887191891670227,
"adv/mean_abs_step_conf": 0.7318241596221924,
"adv/ratio_final_to_reasoning": 1.1473662885214226,
"adv/ratio_step_to_reasoning": 1.2430784881628345,
"adv/std_final_conf": 0.8675520420074463,
"adv/std_reasoning": 0.8266057372093201,
"adv/std_step_conf": 0.9342828989028931,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.742180774748924,
"calib/avg_num_step_conf": 6.15625,
"calib/ece": 0.257710843373494,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.4578313253012048,
"calib/gap": 0.37000215208034437,
"calib/mean_conf": 0.5787550200803213,
"calib/mu_c": 0.7050609756097561,
"calib/mu_w": 0.33505882352941174,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.0889156626506024,
"calib/std_conf": 0.4327511414503791,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.44351409978308026,
"calib/step_q_c_n": 922.0,
"calib/step_q_gap": 0.1667862710369029,
"calib/step_q_w": 0.27672782874617735,
"calib/step_q_w_n": 654.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2987.0,
"completions/max_terminated_length": 2987.0,
"completions/mean_length": 514.06640625,
"completions/mean_terminated_length": 514.06640625,
"completions/min_length": 111.0,
"completions/min_terminated_length": 111.0,
"epoch": 0.17173333333333332,
"grad_norm": 0.024139799177646637,
"kl": 0.1499176025390625,
"learning_rate": 1.0833333333333335e-06,
"loss": 0.1239,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03729427233338356,
"mask/share_reasoning": 0.8286226987838745,
"mask/share_step_conf": 0.13408304750919342,
"num_tokens": 39499004.0,
"reward": 0.9434548616409302,
"reward_std": 0.22313153743743896,
"rewards/accuracy_reward_step": 0.640625,
"rewards/final_brier_reward_step": 0.7232269048690796,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8425890207290649,
"step": 161
},
{
"adv/mean_abs_final_conf": 0.693047285079956,
"adv/mean_abs_reasoning": 0.589208722114563,
"adv/mean_abs_step_conf": 0.7283810377120972,
"adv/ratio_final_to_reasoning": 1.1762339202867456,
"adv/ratio_step_to_reasoning": 1.236202062824308,
"adv/std_final_conf": 0.8884176015853882,
"adv/std_reasoning": 0.8100014328956604,
"adv/std_step_conf": 0.9344998598098755,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7707106812447435,
"calib/avg_num_step_conf": 5.82421875,
"calib/ece": 0.21223107569721117,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.5059760956175299,
"calib/gap": 0.397907905803196,
"calib/mean_conf": 0.6206772908366535,
"calib/mu_c": 0.7585975609756098,
"calib/mu_w": 0.3606896551724138,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.0897609561752988,
"calib/std_conf": 0.4271847285111425,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.43395248380129586,
"calib/step_q_c_n": 926.0,
"calib/step_q_gap": 0.0908374395535082,
"calib/step_q_w": 0.34311504424778766,
"calib/step_q_w_n": 565.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2955.0,
"completions/max_terminated_length": 2955.0,
"completions/mean_length": 468.21875,
"completions/mean_terminated_length": 468.21875,
"completions/min_length": 146.0,
"completions/min_terminated_length": 146.0,
"epoch": 0.1728,
"grad_norm": 0.037547219544649124,
"kl": 0.1618194580078125,
"learning_rate": 1.0555555555555557e-06,
"loss": 0.0287,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03762802481651306,
"mask/share_reasoning": 0.8248691558837891,
"mask/share_step_conf": 0.1375027745962143,
"num_tokens": 39723012.0,
"reward": 0.9379310607910156,
"reward_std": 0.22288298606872559,
"rewards/accuracy_reward_step": 0.64453125,
"rewards/final_brier_reward_step": 0.7288386821746826,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8274922370910645,
"step": 162
},
{
"adv/mean_abs_final_conf": 0.6022693514823914,
"adv/mean_abs_reasoning": 0.5638035535812378,
"adv/mean_abs_step_conf": 0.7278131246566772,
"adv/ratio_final_to_reasoning": 1.0682255329126993,
"adv/ratio_step_to_reasoning": 1.2908984344523247,
"adv/std_final_conf": 0.82899010181427,
"adv/std_reasoning": 0.8099051713943481,
"adv/std_step_conf": 0.9342523813247681,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8520267611176703,
"calib/avg_num_step_conf": 7.03125,
"calib/ece": 0.1543319838056681,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.4008097165991903,
"calib/gap": 0.5143998425816607,
"calib/mean_conf": 0.5124696356275305,
"calib/mu_c": 0.7748760330578512,
"calib/mu_w": 0.2604761904761905,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.08846153846153852,
"calib/std_conf": 0.43631041174122165,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.43778807947019865,
"calib/step_q_c_n": 755.0,
"calib/step_q_gap": 0.1376608067429259,
"calib/step_q_w": 0.30012727272727274,
"calib/step_q_w_n": 1045.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 3010.0,
"completions/max_terminated_length": 3010.0,
"completions/mean_length": 556.234375,
"completions/mean_terminated_length": 560.6141967773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 106.0,
"epoch": 0.17386666666666667,
"grad_norm": 0.03478431701660156,
"kl": 0.1400909423828125,
"learning_rate": 1.0277777777777777e-06,
"loss": -0.0236,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.033809371292591095,
"mask/share_reasoning": 0.8174687623977661,
"mask/share_step_conf": 0.14090941846370697,
"num_tokens": 39970240.0,
"reward": 0.9545302391052246,
"reward_std": 0.2042197585105896,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/final_brier_reward_step": 0.7759039402008057,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8480002880096436,
"step": 163
},
{
"adv/mean_abs_final_conf": 0.7186143398284912,
"adv/mean_abs_reasoning": 0.5782505869865417,
"adv/mean_abs_step_conf": 0.7410357594490051,
"adv/ratio_final_to_reasoning": 1.242738625780619,
"adv/ratio_step_to_reasoning": 1.2815131988205868,
"adv/std_final_conf": 0.9063172340393066,
"adv/std_reasoning": 0.8099116683006287,
"adv/std_step_conf": 0.9345961213111877,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.728862590401052,
"calib/avg_num_step_conf": 6.88671875,
"calib/ece": 0.25417004048583,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.38461538461538464,
"calib/gap": 0.3163589743589743,
"calib/mean_conf": 0.5063157894736843,
"calib/mu_c": 0.6728205128205128,
"calib/mu_w": 0.3564615384615385,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.1434008097165992,
"calib/std_conf": 0.4343376332177941,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.418381414701803,
"calib/step_q_c_n": 721.0,
"calib/step_q_gap": 0.13701865078625597,
"calib/step_q_w": 0.28136276391554704,
"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": 0.0,
"completions/max_length": 2974.0,
"completions/max_terminated_length": 2974.0,
"completions/mean_length": 590.64453125,
"completions/mean_terminated_length": 590.64453125,
"completions/min_length": 156.0,
"completions/min_terminated_length": 156.0,
"epoch": 0.17493333333333333,
"grad_norm": 0.0259377583861351,
"kl": 0.1478271484375,
"learning_rate": 1.0000000000000002e-06,
"loss": -0.0438,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.029251031577587128,
"mask/share_reasoning": 0.8463806509971619,
"mask/share_step_conf": 0.1243683397769928,
"num_tokens": 40227581.0,
"reward": 0.9030520915985107,
"reward_std": 0.21659719944000244,
"rewards/accuracy_reward_step": 0.46484375,
"rewards/final_brier_reward_step": 0.6856386661529541,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8368717432022095,
"step": 164
},
{
"adv/mean_abs_final_conf": 0.6869238018989563,
"adv/mean_abs_reasoning": 0.45050010085105896,
"adv/mean_abs_step_conf": 0.7137496471405029,
"adv/ratio_final_to_reasoning": 1.5248027705238227,
"adv/ratio_step_to_reasoning": 1.584349583478734,
"adv/std_final_conf": 0.8749675154685974,
"adv/std_reasoning": 0.7206880450248718,
"adv/std_step_conf": 0.9337314963340759,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7735745614035088,
"calib/avg_num_step_conf": 7.0546875,
"calib/ece": 0.20679999999999998,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.408,
"calib/gap": 0.43031733746130024,
"calib/mean_conf": 0.53696,
"calib/mu_c": 0.7710526315789473,
"calib/mu_w": 0.3407352941176471,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.14387999999999998,
"calib/std_conf": 0.43471135066846367,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.4286647314949202,
"calib/step_q_c_n": 689.0,
"calib/step_q_gap": 0.13150716658892203,
"calib/step_q_w": 0.29715756490599815,
"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": 0.00390625,
"completions/max_length": 2957.0,
"completions/max_terminated_length": 2957.0,
"completions/mean_length": 561.78515625,
"completions/mean_terminated_length": 563.98828125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 137.0,
"epoch": 0.176,
"grad_norm": 0.02698436565697193,
"kl": 0.147064208984375,
"learning_rate": 9.722222222222224e-07,
"loss": 0.0705,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.032273873686790466,
"mask/share_reasoning": 0.8308225870132446,
"mask/share_step_conf": 0.13299725949764252,
"num_tokens": 40476974.0,
"reward": 0.9402658939361572,
"reward_std": 0.1786276400089264,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/final_brier_reward_step": 0.744081974029541,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8536372184753418,
"step": 165
},
{
"adv/mean_abs_final_conf": 0.6894981861114502,
"adv/mean_abs_reasoning": 0.5755706429481506,
"adv/mean_abs_step_conf": 0.7388315200805664,
"adv/ratio_final_to_reasoning": 1.1979384191308773,
"adv/ratio_step_to_reasoning": 1.283650459127261,
"adv/std_final_conf": 0.8869600892066956,
"adv/std_reasoning": 0.8099501132965088,
"adv/std_step_conf": 0.9338527917861938,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.830402344790834,
"calib/avg_num_step_conf": 7.05859375,
"calib/ece": 0.14919028340080978,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.4493927125506073,
"calib/gap": 0.5527331468158806,
"calib/mean_conf": 0.5517004048582995,
"calib/mu_c": 0.7933812949640288,
"calib/mu_w": 0.24064814814814814,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.06906882591093122,
"calib/std_conf": 0.44479120257315763,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.45416470588235297,
"calib/step_q_c_n": 850.0,
"calib/step_q_gap": 0.16483032761694028,
"calib/step_q_w": 0.2893343782654127,
"calib/step_q_w_n": 957.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2567.0,
"completions/max_terminated_length": 2567.0,
"completions/mean_length": 585.5625,
"completions/mean_terminated_length": 587.85888671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 123.0,
"epoch": 0.17706666666666668,
"grad_norm": 0.02961987629532814,
"kl": 0.1471710205078125,
"learning_rate": 9.444444444444445e-07,
"loss": -0.1233,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.030628211796283722,
"mask/share_reasoning": 0.8332484364509583,
"mask/share_step_conf": 0.13221710920333862,
"num_tokens": 40733062.0,
"reward": 0.9741063117980957,
"reward_std": 0.23434357345104218,
"rewards/accuracy_reward_step": 0.54296875,
"rewards/final_brier_reward_step": 0.7911171913146973,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8570953607559204,
"step": 166
},
{
"adv/mean_abs_final_conf": 0.6714800000190735,
"adv/mean_abs_reasoning": 0.5729876756668091,
"adv/mean_abs_step_conf": 0.7475663423538208,
"adv/ratio_final_to_reasoning": 1.1718925703552783,
"adv/ratio_step_to_reasoning": 1.3046813641913806,
"adv/std_final_conf": 0.8594887852668762,
"adv/std_reasoning": 0.8099749088287354,
"adv/std_step_conf": 0.9354498386383057,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7133557046979866,
"calib/avg_num_step_conf": 6.296875,
"calib/ece": 0.22803212851405624,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.5943775100401606,
"calib/gap": 0.33976510067114096,
"calib/mean_conf": 0.6983132530120482,
"calib/mu_c": 0.834765100671141,
"calib/mu_w": 0.495,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.96484375,
"calib/pce": 0.16397590361445782,
"calib/std_conf": 0.40734171817593934,
"calib/step_conf_rate": 0.96484375,
"calib/step_q_c": 0.44661474558670816,
"calib/step_q_c_n": 963.0,
"calib/step_q_gap": 0.05849456068686226,
"calib/step_q_w": 0.3881201848998459,
"calib/step_q_w_n": 649.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2567.0,
"completions/max_terminated_length": 2567.0,
"completions/mean_length": 529.953125,
"completions/mean_terminated_length": 529.953125,
"completions/min_length": 47.0,
"completions/min_terminated_length": 47.0,
"epoch": 0.17813333333333334,
"grad_norm": 0.046685412526130676,
"kl": 0.1434783935546875,
"learning_rate": 9.166666666666666e-07,
"loss": -0.1107,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.033018916845321655,
"mask/share_reasoning": 0.8381372690200806,
"mask/share_step_conf": 0.12884379923343658,
"num_tokens": 40974338.0,
"reward": 0.9086949825286865,
"reward_std": 0.263832151889801,
"rewards/accuracy_reward_step": 0.58203125,
"rewards/final_brier_reward_step": 0.7043409943580627,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.8075802326202393,
"step": 167
},
{
"adv/mean_abs_final_conf": 0.6658957600593567,
"adv/mean_abs_reasoning": 0.5945309400558472,
"adv/mean_abs_step_conf": 0.7367751598358154,
"adv/ratio_final_to_reasoning": 1.1200355022680668,
"adv/ratio_step_to_reasoning": 1.2392545285643277,
"adv/std_final_conf": 0.8759635090827942,
"adv/std_reasoning": 0.8100911974906921,
"adv/std_step_conf": 0.9350863695144653,
"calib/answer_extract_rate": 0.96484375,
"calib/auroc": 0.8082469431153643,
"calib/avg_num_step_conf": 7.3203125,
"calib/ece": 0.17772357723577242,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.4715447154471545,
"calib/gap": 0.4370175438596491,
"calib/mean_conf": 0.6041463414634146,
"calib/mu_c": 0.8066666666666666,
"calib/mu_w": 0.36964912280701756,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.12264227642276426,
"calib/std_conf": 0.4214173275389015,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.46009720534629406,
"calib/step_q_c_n": 823.0,
"calib/step_q_gap": 0.17525419868597053,
"calib/step_q_w": 0.28484300666032353,
"calib/step_q_w_n": 1051.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01953125,
"completions/max_length": 2884.0,
"completions/max_terminated_length": 2884.0,
"completions/mean_length": 543.21484375,
"completions/mean_terminated_length": 554.035888671875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 106.0,
"epoch": 0.1792,
"grad_norm": 0.04968612268567085,
"kl": 0.1490478515625,
"learning_rate": 8.88888888888889e-07,
"loss": -0.1385,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.031077388674020767,
"mask/share_reasoning": 0.820555567741394,
"mask/share_step_conf": 0.12883584201335907,
"num_tokens": 41218073.0,
"reward": 0.9409385919570923,
"reward_std": 0.24200639128684998,
"rewards/accuracy_reward_step": 0.515625,
"rewards/final_brier_reward_step": 0.7511484622955322,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8377599716186523,
"step": 168
},
{
"adv/mean_abs_final_conf": 0.7448919415473938,
"adv/mean_abs_reasoning": 0.6670287251472473,
"adv/mean_abs_step_conf": 0.7577471733093262,
"adv/ratio_final_to_reasoning": 1.1167314291943846,
"adv/ratio_step_to_reasoning": 1.1360038102437833,
"adv/std_final_conf": 0.9090984463691711,
"adv/std_reasoning": 0.8747960329055786,
"adv/std_step_conf": 0.934616208076477,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7648950131233596,
"calib/avg_num_step_conf": 6.8359375,
"calib/ece": 0.22121457489878543,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.4777327935222672,
"calib/gap": 0.40010236220472434,
"calib/mean_conf": 0.5932793522267207,
"calib/mu_c": 0.7989999999999999,
"calib/mu_w": 0.3988976377952756,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.164331983805668,
"calib/std_conf": 0.4297803022442411,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.45931645569620255,
"calib/step_q_c_n": 790.0,
"calib/step_q_gap": 0.1249518723628692,
"calib/step_q_w": 0.33436458333333335,
"calib/step_q_w_n": 960.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2946.0,
"completions/max_terminated_length": 2946.0,
"completions/mean_length": 569.7265625,
"completions/mean_terminated_length": 571.9608154296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 77.0,
"epoch": 0.18026666666666666,
"grad_norm": 0.03378553315997124,
"kl": 0.14874267578125,
"learning_rate": 8.611111111111112e-07,
"loss": -0.0706,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.03206229209899902,
"mask/share_reasoning": 0.8360965251922607,
"mask/share_step_conf": 0.12793496251106262,
"num_tokens": 41468107.0,
"reward": 0.9184834957122803,
"reward_std": 0.2511550784111023,
"rewards/accuracy_reward_step": 0.46875,
"rewards/final_brier_reward_step": 0.7157703042030334,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8368216156959534,
"step": 169
},
{
"adv/mean_abs_final_conf": 0.6869648098945618,
"adv/mean_abs_reasoning": 0.6289730072021484,
"adv/mean_abs_step_conf": 0.7552691698074341,
"adv/ratio_final_to_reasoning": 1.0922007813187047,
"adv/ratio_step_to_reasoning": 1.2007974287594423,
"adv/std_final_conf": 0.8834342956542969,
"adv/std_reasoning": 0.8429259657859802,
"adv/std_step_conf": 0.9347723722457886,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.812171743697479,
"calib/avg_num_step_conf": 6.87890625,
"calib/ece": 0.21886639676113367,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.5465587044534413,
"calib/gap": 0.45621192226890755,
"calib/mean_conf": 0.6547368421052631,
"calib/mu_c": 0.87453125,
"calib/mu_w": 0.4183193277310924,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.17769230769230776,
"calib/std_conf": 0.4249189391366208,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.47638487208008895,
"calib/step_q_c_n": 899.0,
"calib/step_q_gap": 0.15251712266013534,
"calib/step_q_w": 0.3238677494199536,
"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": 0.0078125,
"completions/max_length": 2524.0,
"completions/max_terminated_length": 2524.0,
"completions/mean_length": 554.39453125,
"completions/mean_terminated_length": 558.7598266601562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 50.0,
"epoch": 0.18133333333333335,
"grad_norm": 0.03549743816256523,
"kl": 0.141571044921875,
"learning_rate": 8.333333333333333e-07,
"loss": -0.1443,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.030027873814105988,
"mask/share_reasoning": 0.8298295140266418,
"mask/share_step_conf": 0.13233011960983276,
"num_tokens": 41714184.0,
"reward": 0.924481213092804,
"reward_std": 0.2552022635936737,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.7320507764816284,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.8270678520202637,
"step": 170
},
{
"adv/mean_abs_final_conf": 0.7088747024536133,
"adv/mean_abs_reasoning": 0.6031726002693176,
"adv/mean_abs_step_conf": 0.7413415312767029,
"adv/ratio_final_to_reasoning": 1.1752435407992663,
"adv/ratio_step_to_reasoning": 1.2290703041645006,
"adv/std_final_conf": 0.9034948348999023,
"adv/std_reasoning": 0.8428179621696472,
"adv/std_step_conf": 0.9349961280822754,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.7144636015325669,
"calib/avg_num_step_conf": 6.56640625,
"calib/ece": 0.3018181818181818,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.4743083003952569,
"calib/gap": 0.2947420178799488,
"calib/mean_conf": 0.5753359683794467,
"calib/mu_c": 0.7442592592592592,
"calib/mu_w": 0.4495172413793104,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.22513833992094867,
"calib/std_conf": 0.43983870760345284,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.4494075144508671,
"calib/step_q_c_n": 692.0,
"calib/step_q_gap": 0.13023663477442626,
"calib/step_q_w": 0.3191708796764408,
"calib/step_q_w_n": 989.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2561.0,
"completions/max_terminated_length": 2561.0,
"completions/mean_length": 520.73828125,
"completions/mean_terminated_length": 520.73828125,
"completions/min_length": 165.0,
"completions/min_terminated_length": 165.0,
"epoch": 0.1824,
"grad_norm": 0.030517173931002617,
"kl": 0.145599365234375,
"learning_rate": 8.055555555555557e-07,
"loss": -0.0771,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03335035592317581,
"mask/share_reasoning": 0.830026388168335,
"mask/share_step_conf": 0.1366232931613922,
"num_tokens": 41954389.0,
"reward": 0.8957168459892273,
"reward_std": 0.23421956598758698,
"rewards/accuracy_reward_step": 0.42578125,
"rewards/final_brier_reward_step": 0.6607366800308228,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8510094881057739,
"step": 171
},
{
"adv/mean_abs_final_conf": 0.7520140409469604,
"adv/mean_abs_reasoning": 0.5973731279373169,
"adv/mean_abs_step_conf": 0.7457846403121948,
"adv/ratio_final_to_reasoning": 1.2588682111356542,
"adv/ratio_step_to_reasoning": 1.248440221754419,
"adv/std_final_conf": 0.9215742349624634,
"adv/std_reasoning": 0.8100011348724365,
"adv/std_step_conf": 0.9346137642860413,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7111765887631419,
"calib/avg_num_step_conf": 6.78515625,
"calib/ece": 0.2620149051490514,
"calib/final_conf_rate": 0.9609375,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.5447154471544715,
"calib/gap": 0.28152911895354804,
"calib/mean_conf": 0.6972452574525745,
"calib/mu_c": 0.8219878345498783,
"calib/mu_w": 0.5404587155963303,
"calib/nonempty_final_conf_rate": 0.9609375,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.2011747967479674,
"calib/std_conf": 0.3830020412565528,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.440875,
"calib/step_q_c_n": 880.0,
"calib/step_q_gap": 0.09097348308051345,
"calib/step_q_w": 0.34990151691948657,
"calib/step_q_w_n": 857.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.01171875,
"completions/max_length": 2696.0,
"completions/max_terminated_length": 2696.0,
"completions/mean_length": 525.3359375,
"completions/mean_terminated_length": 531.5652465820312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.18346666666666667,
"grad_norm": 0.033332545310258865,
"kl": 0.1470794677734375,
"learning_rate": 7.777777777777779e-07,
"loss": -0.078,
"mask/has_final_conf_rate": 0.9609375,
"mask/share_final_conf": 0.033402156084775925,
"mask/share_reasoning": 0.8161604404449463,
"mask/share_step_conf": 0.13871869444847107,
"num_tokens": 42192227.0,
"reward": 0.9112281799316406,
"reward_std": 0.25344720482826233,
"rewards/accuracy_reward_step": 0.5390625,
"rewards/final_brier_reward_step": 0.693537712097168,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8296998739242554,
"step": 172
},
{
"adv/mean_abs_final_conf": 0.7470325231552124,
"adv/mean_abs_reasoning": 0.6900979280471802,
"adv/mean_abs_step_conf": 0.7407292127609253,
"adv/ratio_final_to_reasoning": 1.0825021968536324,
"adv/ratio_step_to_reasoning": 1.0733682607293433,
"adv/std_final_conf": 0.9220872521400452,
"adv/std_reasoning": 0.8749023079872131,
"adv/std_step_conf": 0.9353823661804199,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.7292431561996779,
"calib/avg_num_step_conf": 6.66015625,
"calib/ece": 0.26159999999999994,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.604,
"calib/gap": 0.31702415458937194,
"calib/mean_conf": 0.70928,
"calib/mu_c": 0.855111111111111,
"calib/mu_w": 0.5380869565217391,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.2154399999999999,
"calib/std_conf": 0.3978244356496971,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.42527352297593,
"calib/step_q_c_n": 914.0,
"calib/step_q_gap": 0.03423686052333835,
"calib/step_q_w": 0.39103666245259167,
"calib/step_q_w_n": 791.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2816.0,
"completions/max_terminated_length": 2816.0,
"completions/mean_length": 568.78515625,
"completions/mean_terminated_length": 568.78515625,
"completions/min_length": 94.0,
"completions/min_terminated_length": 94.0,
"epoch": 0.18453333333333333,
"grad_norm": 0.03139273077249527,
"kl": 0.147369384765625,
"learning_rate": 7.5e-07,
"loss": -0.095,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03458606079220772,
"mask/share_reasoning": 0.834449052810669,
"mask/share_step_conf": 0.13096490502357483,
"num_tokens": 42440996.0,
"reward": 0.8875725269317627,
"reward_std": 0.2774200439453125,
"rewards/accuracy_reward_step": 0.52734375,
"rewards/final_brier_reward_step": 0.681814432144165,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.7972366809844971,
"step": 173
},
{
"adv/mean_abs_final_conf": 0.7353447675704956,
"adv/mean_abs_reasoning": 0.6441205739974976,
"adv/mean_abs_step_conf": 0.7374603748321533,
"adv/ratio_final_to_reasoning": 1.141625958330827,
"adv/ratio_step_to_reasoning": 1.1449104478302512,
"adv/std_final_conf": 0.8920470476150513,
"adv/std_reasoning": 0.8101984262466431,
"adv/std_step_conf": 0.9349736571311951,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7004191863661353,
"calib/avg_num_step_conf": 6.8046875,
"calib/ece": 0.25469135802469134,
"calib/final_conf_rate": 0.94921875,
"calib/format_rate": 0.9375,
"calib/frac_conf_gt_0.9": 0.411522633744856,
"calib/gap": 0.30320368334249603,
"calib/mean_conf": 0.5339506172839507,
"calib/mu_c": 0.7036448598130842,
"calib/mu_w": 0.4004411764705882,
"calib/nonempty_final_conf_rate": 0.94921875,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.17415637860082303,
"calib/std_conf": 0.4292907702974546,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.43064371257485035,
"calib/step_q_c_n": 668.0,
"calib/step_q_gap": 0.10949846117820233,
"calib/step_q_w": 0.321145251396648,
"calib/step_q_w_n": 1074.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2698.0,
"completions/max_terminated_length": 2698.0,
"completions/mean_length": 589.47265625,
"completions/mean_terminated_length": 594.1141967773438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 134.0,
"epoch": 0.1856,
"grad_norm": 0.06488266587257385,
"kl": 0.141143798828125,
"learning_rate": 7.222222222222222e-07,
"loss": -0.1133,
"mask/has_final_conf_rate": 0.94921875,
"mask/share_final_conf": 0.02945755049586296,
"mask/share_reasoning": 0.839420735836029,
"mask/share_step_conf": 0.12330922484397888,
"num_tokens": 42696133.0,
"reward": 0.8655682802200317,
"reward_std": 0.25477084517478943,
"rewards/accuracy_reward_step": 0.41796875,
"rewards/final_brier_reward_step": 0.6679195165634155,
"rewards/format_reward_step": 0.9375,
"rewards/step_l2_reward": 0.7921232581138611,
"step": 174
},
{
"adv/mean_abs_final_conf": 0.6823322176933289,
"adv/mean_abs_reasoning": 0.5172361135482788,
"adv/mean_abs_step_conf": 0.7226990461349487,
"adv/ratio_final_to_reasoning": 1.3191890508427153,
"adv/ratio_step_to_reasoning": 1.3972323803478814,
"adv/std_final_conf": 0.8660195469856262,
"adv/std_reasoning": 0.7754583358764648,
"adv/std_step_conf": 0.9350190758705139,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.8058183538315988,
"calib/avg_num_step_conf": 7.16015625,
"calib/ece": 0.19823293172690765,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.3534136546184739,
"calib/gap": 0.4353682930125693,
"calib/mean_conf": 0.4786345381526105,
"calib/mu_c": 0.7426530612244898,
"calib/mu_w": 0.3072847682119205,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.14164658634538152,
"calib/std_conf": 0.43176733909865955,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.4574626865671642,
"calib/step_q_c_n": 603.0,
"calib/step_q_gap": 0.18419520689236746,
"calib/step_q_w": 0.27326747967479675,
"calib/step_q_w_n": 1230.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2934.0,
"completions/max_terminated_length": 2934.0,
"completions/mean_length": 573.8203125,
"completions/mean_terminated_length": 578.3385620117188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 153.0,
"epoch": 0.18666666666666668,
"grad_norm": 0.03821876645088196,
"kl": 0.139007568359375,
"learning_rate": 6.944444444444446e-07,
"loss": -0.0861,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.030831394717097282,
"mask/share_reasoning": 0.8289897441864014,
"mask/share_step_conf": 0.1323663592338562,
"num_tokens": 42948855.0,
"reward": 0.9255420565605164,
"reward_std": 0.23187220096588135,
"rewards/accuracy_reward_step": 0.3828125,
"rewards/final_brier_reward_step": 0.7417949438095093,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8421017527580261,
"step": 175
},
{
"adv/mean_abs_final_conf": 0.7288157939910889,
"adv/mean_abs_reasoning": 0.6108537912368774,
"adv/mean_abs_step_conf": 0.7525293827056885,
"adv/ratio_final_to_reasoning": 1.193110044410722,
"adv/ratio_step_to_reasoning": 1.2319304447336596,
"adv/std_final_conf": 0.9071044325828552,
"adv/std_reasoning": 0.8099657893180847,
"adv/std_step_conf": 0.9341826438903809,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7841346153846154,
"calib/avg_num_step_conf": 6.4765625,
"calib/ece": 0.19042666666666674,
"calib/final_conf_rate": 0.9765625,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.368,
"calib/gap": 0.41957905982905985,
"calib/mean_conf": 0.5190133333333332,
"calib/mu_c": 0.7371944444444445,
"calib/mu_w": 0.31761538461538463,
"calib/nonempty_final_conf_rate": 0.9765625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.11472000000000007,
"calib/std_conf": 0.4250891721352905,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.43961379310344834,
"calib/step_q_c_n": 725.0,
"calib/step_q_gap": 0.14185387884835732,
"calib/step_q_w": 0.297759914255091,
"calib/step_q_w_n": 933.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2674.0,
"completions/max_terminated_length": 2674.0,
"completions/mean_length": 551.00390625,
"completions/mean_terminated_length": 553.1647338867188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 129.0,
"epoch": 0.18773333333333334,
"grad_norm": 0.037594687193632126,
"kl": 0.1331939697265625,
"learning_rate": 6.666666666666667e-07,
"loss": -0.0645,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03376203775405884,
"mask/share_reasoning": 0.8257086277008057,
"mask/share_step_conf": 0.1366230547428131,
"num_tokens": 43193976.0,
"reward": 0.9356297254562378,
"reward_std": 0.23185500502586365,
"rewards/accuracy_reward_step": 0.46875,
"rewards/final_brier_reward_step": 0.743251383304596,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.8436331152915955,
"step": 176
},
{
"adv/mean_abs_final_conf": 0.7497272491455078,
"adv/mean_abs_reasoning": 0.6292704343795776,
"adv/mean_abs_step_conf": 0.7332544922828674,
"adv/ratio_final_to_reasoning": 1.1914229688618587,
"adv/ratio_step_to_reasoning": 1.1652454210816559,
"adv/std_final_conf": 0.9185965061187744,
"adv/std_reasoning": 0.8429964184761047,
"adv/std_step_conf": 0.9343379139900208,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7373385012919897,
"calib/avg_num_step_conf": 6.734375,
"calib/ece": 0.21686746987951808,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.3855421686746988,
"calib/gap": 0.3612693798449613,
"calib/mean_conf": 0.5465863453815262,
"calib/mu_c": 0.7337500000000001,
"calib/mu_w": 0.3724806201550388,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.14076305220883534,
"calib/std_conf": 0.4245110506249608,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.4502612481857765,
"calib/step_q_c_n": 689.0,
"calib/step_q_gap": 0.1357974800698345,
"calib/step_q_w": 0.314463768115942,
"calib/step_q_w_n": 1035.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2611.0,
"completions/max_terminated_length": 2611.0,
"completions/mean_length": 542.15625,
"completions/mean_terminated_length": 546.4251708984375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 128.0,
"epoch": 0.1888,
"grad_norm": 0.03646094724535942,
"kl": 0.137420654296875,
"learning_rate": 6.388888888888889e-07,
"loss": -0.0679,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.033001139760017395,
"mask/share_reasoning": 0.8247432112693787,
"mask/share_step_conf": 0.13444316387176514,
"num_tokens": 43436600.0,
"reward": 0.9250853061676025,
"reward_std": 0.22668462991714478,
"rewards/accuracy_reward_step": 0.46875,
"rewards/final_brier_reward_step": 0.7222949266433716,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8411569595336914,
"step": 177
},
{
"adv/mean_abs_final_conf": 0.6748043894767761,
"adv/mean_abs_reasoning": 0.5698914527893066,
"adv/mean_abs_step_conf": 0.7398583889007568,
"adv/ratio_final_to_reasoning": 1.1840928411436566,
"adv/ratio_step_to_reasoning": 1.298244402999124,
"adv/std_final_conf": 0.8798059225082397,
"adv/std_reasoning": 0.7929153442382812,
"adv/std_step_conf": 0.9339677095413208,
"calib/answer_extract_rate": 0.9921875,
"calib/auroc": 0.8274903137107861,
"calib/avg_num_step_conf": 6.19140625,
"calib/ece": 0.13019762845849808,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.97265625,
"calib/frac_conf_gt_0.9": 0.391304347826087,
"calib/gap": 0.5169072615923009,
"calib/mean_conf": 0.5277470355731225,
"calib/mu_c": 0.7872222222222222,
"calib/mu_w": 0.2703149606299213,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.07996047430830047,
"calib/std_conf": 0.4283981368738465,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.44836879432624116,
"calib/step_q_c_n": 705.0,
"calib/step_q_gap": 0.16823243068987753,
"calib/step_q_w": 0.28013636363636363,
"calib/step_q_w_n": 880.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2337.0,
"completions/max_terminated_length": 2337.0,
"completions/mean_length": 505.48046875,
"completions/mean_terminated_length": 507.4627685546875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 94.0,
"epoch": 0.18986666666666666,
"grad_norm": 0.031120702624320984,
"kl": 0.1497802734375,
"learning_rate": 6.111111111111112e-07,
"loss": -0.0717,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03448161482810974,
"mask/share_reasoning": 0.8300520181655884,
"mask/share_step_conf": 0.13156011700630188,
"num_tokens": 43672075.0,
"reward": 0.9877314567565918,
"reward_std": 0.1797032356262207,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.8070328235626221,
"rewards/format_reward_step": 0.97265625,
"rewards/step_l2_reward": 0.8746801614761353,
"step": 178
},
{
"adv/mean_abs_final_conf": 0.7142655849456787,
"adv/mean_abs_reasoning": 0.6147536039352417,
"adv/mean_abs_step_conf": 0.7718060612678528,
"adv/ratio_final_to_reasoning": 1.1618729526324496,
"adv/ratio_step_to_reasoning": 1.2554722027284855,
"adv/std_final_conf": 0.8914516568183899,
"adv/std_reasoning": 0.8100534677505493,
"adv/std_step_conf": 0.9345685839653015,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.770054908263024,
"calib/avg_num_step_conf": 6.53125,
"calib/ece": 0.2055510204081633,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.39183673469387753,
"calib/gap": 0.40611557519753577,
"calib/mean_conf": 0.5335510204081632,
"calib/mu_c": 0.7225190839694656,
"calib/mu_w": 0.31640350877192985,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.10220408163265306,
"calib/std_conf": 0.431957153072843,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.4421166032953105,
"calib/step_q_c_n": 789.0,
"calib/step_q_gap": 0.14329440623981782,
"calib/step_q_w": 0.2988221970554927,
"calib/step_q_w_n": 883.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2931.0,
"completions/max_terminated_length": 2931.0,
"completions/mean_length": 550.17578125,
"completions/mean_terminated_length": 556.6996459960938,
"completions/min_length": 0.0,
"completions/min_terminated_length": 97.0,
"epoch": 0.19093333333333334,
"grad_norm": 0.03112025186419487,
"kl": 0.1467742919921875,
"learning_rate": 5.833333333333334e-07,
"loss": -0.0728,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.0336512066423893,
"mask/share_reasoning": 0.8248656988143921,
"mask/share_step_conf": 0.12976431846618652,
"num_tokens": 43919184.0,
"reward": 0.9264711141586304,
"reward_std": 0.23171232640743256,
"rewards/accuracy_reward_step": 0.51171875,
"rewards/final_brier_reward_step": 0.7263144254684448,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8344402313232422,
"step": 179
},
{
"adv/mean_abs_final_conf": 0.7504141926765442,
"adv/mean_abs_reasoning": 0.6587625741958618,
"adv/mean_abs_step_conf": 0.7519873380661011,
"adv/ratio_final_to_reasoning": 1.1391269359716734,
"adv/ratio_step_to_reasoning": 1.1415149668817128,
"adv/std_final_conf": 0.8830128908157349,
"adv/std_reasoning": 0.8430957198143005,
"adv/std_step_conf": 0.9344708919525146,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.6929724234337273,
"calib/avg_num_step_conf": 6.7265625,
"calib/ece": 0.2840637450199204,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.4063745019920319,
"calib/gap": 0.2824666412504766,
"calib/mean_conf": 0.5317131474103586,
"calib/mu_c": 0.6768852459016393,
"calib/mu_w": 0.39441860465116274,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9609375,
"calib/pce": 0.16486055776892433,
"calib/std_conf": 0.4366340073345297,
"calib/step_conf_rate": 0.9609375,
"calib/step_q_c": 0.41103666245259163,
"calib/step_q_c_n": 791.0,
"calib/step_q_gap": 0.10133204376301058,
"calib/step_q_w": 0.30970461868958105,
"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": 0.0,
"completions/max_length": 2782.0,
"completions/max_terminated_length": 2782.0,
"completions/mean_length": 622.55078125,
"completions/mean_terminated_length": 622.55078125,
"completions/min_length": 147.0,
"completions/min_terminated_length": 147.0,
"epoch": 0.192,
"grad_norm": 0.0328676737844944,
"kl": 0.135284423828125,
"learning_rate": 5.555555555555555e-07,
"loss": -0.063,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.02904374897480011,
"mask/share_reasoning": 0.8486019372940063,
"mask/share_step_conf": 0.12235426902770996,
"num_tokens": 44182413.0,
"reward": 0.8859990835189819,
"reward_std": 0.24790097773075104,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/final_brier_reward_step": 0.6629855632781982,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8238563537597656,
"step": 180
},
{
"adv/mean_abs_final_conf": 0.6953284740447998,
"adv/mean_abs_reasoning": 0.5843336582183838,
"adv/mean_abs_step_conf": 0.7292240858078003,
"adv/ratio_final_to_reasoning": 1.1899510908969988,
"adv/ratio_step_to_reasoning": 1.247958380544402,
"adv/std_final_conf": 0.9049976468086243,
"adv/std_reasoning": 0.8265220522880554,
"adv/std_step_conf": 0.934059739112854,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.8302599564604944,
"calib/avg_num_step_conf": 6.78125,
"calib/ece": 0.13549800796812747,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.3147410358565737,
"calib/gap": 0.49808810347035476,
"calib/mean_conf": 0.46848605577689245,
"calib/mu_c": 0.7403508771929825,
"calib/mu_w": 0.24226277372262778,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.07490039840637447,
"calib/std_conf": 0.4211595863991668,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.43901217861975644,
"calib/step_q_c_n": 739.0,
"calib/step_q_gap": 0.1599359499336983,
"calib/step_q_w": 0.27907622868605814,
"calib/step_q_w_n": 997.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1555.0,
"completions/max_terminated_length": 1555.0,
"completions/mean_length": 507.296875,
"completions/mean_terminated_length": 511.2913513183594,
"completions/min_length": 0.0,
"completions/min_terminated_length": 170.0,
"epoch": 0.19306666666666666,
"grad_norm": 0.03237759321928024,
"kl": 0.1660614013671875,
"learning_rate": 5.277777777777779e-07,
"loss": -0.1071,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.032008811831474304,
"mask/share_reasoning": 0.8223521709442139,
"mask/share_step_conf": 0.13782651722431183,
"num_tokens": 44418545.0,
"reward": 0.9796043634414673,
"reward_std": 0.191335991024971,
"rewards/accuracy_reward_step": 0.4453125,
"rewards/final_brier_reward_step": 0.8015214800834656,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8733123540878296,
"step": 181
},
{
"adv/mean_abs_final_conf": 0.7607433795928955,
"adv/mean_abs_reasoning": 0.6217296123504639,
"adv/mean_abs_step_conf": 0.7305428981781006,
"adv/ratio_final_to_reasoning": 1.2235919996103881,
"adv/ratio_step_to_reasoning": 1.1750170551089973,
"adv/std_final_conf": 0.9192807078361511,
"adv/std_reasoning": 0.8429564833641052,
"adv/std_step_conf": 0.9337859749794006,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.713858868404323,
"calib/avg_num_step_conf": 6.73046875,
"calib/ece": 0.23800796812749003,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.44223107569721115,
"calib/gap": 0.3338232676414495,
"calib/mean_conf": 0.5732270916334661,
"calib/mu_c": 0.7341538461538462,
"calib/mu_w": 0.4003305785123967,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.14665338645418324,
"calib/std_conf": 0.4292033111404569,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.40531690140845067,
"calib/step_q_c_n": 852.0,
"calib/step_q_gap": 0.08159703918112571,
"calib/step_q_w": 0.32371986222732496,
"calib/step_q_w_n": 871.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0,
"completions/max_length": 2616.0,
"completions/max_terminated_length": 2616.0,
"completions/mean_length": 572.48828125,
"completions/mean_terminated_length": 572.48828125,
"completions/min_length": 175.0,
"completions/min_terminated_length": 175.0,
"epoch": 0.19413333333333332,
"grad_norm": 0.057896681129932404,
"kl": 0.139312744140625,
"learning_rate": 5.000000000000001e-07,
"loss": -0.0456,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03143026679754257,
"mask/share_reasoning": 0.8360381126403809,
"mask/share_step_conf": 0.13253158330917358,
"num_tokens": 44671262.0,
"reward": 0.9323064088821411,
"reward_std": 0.24068082869052887,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.7115858793258667,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8561519384384155,
"step": 182
},
{
"adv/mean_abs_final_conf": 0.6727540493011475,
"adv/mean_abs_reasoning": 0.5549894571304321,
"adv/mean_abs_step_conf": 0.7398919463157654,
"adv/ratio_final_to_reasoning": 1.2121924852043426,
"adv/ratio_step_to_reasoning": 1.333163967008256,
"adv/std_final_conf": 0.8799149990081787,
"adv/std_reasoning": 0.8099837303161621,
"adv/std_step_conf": 0.9335006475448608,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8204834605597965,
"calib/avg_num_step_conf": 6.47265625,
"calib/ece": 0.1960159362549801,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.32669322709163345,
"calib/gap": 0.4512843511450381,
"calib/mean_conf": 0.4352191235059761,
"calib/mu_c": 0.67075,
"calib/mu_w": 0.21946564885496184,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.07657370517928286,
"calib/std_conf": 0.43562595412257965,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.4167121418826739,
"calib/step_q_c_n": 733.0,
"calib/step_q_gap": 0.12872512889566096,
"calib/step_q_w": 0.28798701298701296,
"calib/step_q_w_n": 924.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 1725.0,
"completions/max_terminated_length": 1725.0,
"completions/mean_length": 523.46875,
"completions/mean_terminated_length": 527.590576171875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 98.0,
"epoch": 0.1952,
"grad_norm": 0.02976737916469574,
"kl": 0.143646240234375,
"learning_rate": 4.7222222222222226e-07,
"loss": -0.1384,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03223654627799988,
"mask/share_reasoning": 0.8284767270088196,
"mask/share_step_conf": 0.13147422671318054,
"num_tokens": 44911950.0,
"reward": 0.9484266638755798,
"reward_std": 0.2068886160850525,
"rewards/accuracy_reward_step": 0.46875,
"rewards/final_brier_reward_step": 0.7585452795028687,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8515892028808594,
"step": 183
},
{
"adv/mean_abs_final_conf": 0.6752747893333435,
"adv/mean_abs_reasoning": 0.5506070852279663,
"adv/mean_abs_step_conf": 0.7362768650054932,
"adv/ratio_final_to_reasoning": 1.226418633995168,
"adv/ratio_step_to_reasoning": 1.3372092091780012,
"adv/std_final_conf": 0.8800711035728455,
"adv/std_reasoning": 0.7928036451339722,
"adv/std_step_conf": 0.9339790344238281,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.8077777777777778,
"calib/avg_num_step_conf": 6.71484375,
"calib/ece": 0.1896031746031746,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.44047619047619047,
"calib/gap": 0.48925490196078436,
"calib/mean_conf": 0.5456349206349206,
"calib/mu_c": 0.7436666666666667,
"calib/mu_w": 0.25441176470588234,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.07000000000000002,
"calib/std_conf": 0.437287809825313,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.4014785214785215,
"calib/step_q_c_n": 1001.0,
"calib/step_q_gap": 0.11227239334481676,
"calib/step_q_w": 0.2892061281337047,
"calib/step_q_w_n": 718.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2377.0,
"completions/max_terminated_length": 2377.0,
"completions/mean_length": 521.51171875,
"completions/mean_terminated_length": 525.6181030273438,
"completions/min_length": 0.0,
"completions/min_terminated_length": 151.0,
"epoch": 0.19626666666666667,
"grad_norm": 0.03793696314096451,
"kl": 0.159942626953125,
"learning_rate": 4.444444444444445e-07,
"loss": 0.0157,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.031797416508197784,
"mask/share_reasoning": 0.8241387605667114,
"mask/share_step_conf": 0.1362513303756714,
"num_tokens": 45150737.0,
"reward": 0.9748945832252502,
"reward_std": 0.18860071897506714,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.7809671759605408,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8563219308853149,
"step": 184
},
{
"adv/mean_abs_final_conf": 0.7041515111923218,
"adv/mean_abs_reasoning": 0.5635975003242493,
"adv/mean_abs_step_conf": 0.737541139125824,
"adv/ratio_final_to_reasoning": 1.2493872147892935,
"adv/ratio_step_to_reasoning": 1.3086309621698133,
"adv/std_final_conf": 0.896740734577179,
"adv/std_reasoning": 0.8265684843063354,
"adv/std_step_conf": 0.9342312216758728,
"calib/answer_extract_rate": 0.9609375,
"calib/auroc": 0.8083511063716342,
"calib/avg_num_step_conf": 6.796875,
"calib/ece": 0.18306081632653065,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.93359375,
"calib/frac_conf_gt_0.9": 0.4530612244897959,
"calib/gap": 0.4999691815515862,
"calib/mean_conf": 0.5642861224489795,
"calib/mu_c": 0.8112096774193548,
"calib/mu_w": 0.3112404958677686,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.96875,
"calib/pce": 0.1206122448979592,
"calib/std_conf": 0.4405038356191336,
"calib/step_conf_rate": 0.96875,
"calib/step_q_c": 0.44092207792207794,
"calib/step_q_c_n": 770.0,
"calib/step_q_gap": 0.18606640781898515,
"calib/step_q_w": 0.2548556701030928,
"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": 0.00390625,
"completions/max_length": 2907.0,
"completions/max_terminated_length": 2907.0,
"completions/mean_length": 556.76953125,
"completions/mean_terminated_length": 558.9530029296875,
"completions/min_length": 0.0,
"completions/min_terminated_length": 198.0,
"epoch": 0.19733333333333333,
"grad_norm": 0.028952667489647865,
"kl": 0.13585662841796875,
"learning_rate": 4.1666666666666667e-07,
"loss": 0.0111,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.03193461149930954,
"mask/share_reasoning": 0.8293842077255249,
"mask/share_step_conf": 0.13477492332458496,
"num_tokens": 45400190.0,
"reward": 0.9280073642730713,
"reward_std": 0.24359695613384247,
"rewards/accuracy_reward_step": 0.484375,
"rewards/final_brier_reward_step": 0.7500066757202148,
"rewards/format_reward_step": 0.93359375,
"rewards/step_l2_reward": 0.8224143981933594,
"step": 185
},
{
"adv/mean_abs_final_conf": 0.7179276347160339,
"adv/mean_abs_reasoning": 0.5549343824386597,
"adv/mean_abs_step_conf": 0.7462427616119385,
"adv/ratio_final_to_reasoning": 1.2937162616616045,
"adv/ratio_step_to_reasoning": 1.3447405409132769,
"adv/std_final_conf": 0.9043472409248352,
"adv/std_reasoning": 0.8097760081291199,
"adv/std_step_conf": 0.9339725375175476,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.77057535959975,
"calib/avg_num_step_conf": 6.828125,
"calib/ece": 0.20245059288537548,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.98046875,
"calib/frac_conf_gt_0.9": 0.41106719367588934,
"calib/gap": 0.42446341463414633,
"calib/mean_conf": 0.5266403162055336,
"calib/mu_c": 0.733,
"calib/mu_w": 0.30853658536585366,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.10762845849802372,
"calib/std_conf": 0.4359000695429722,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.4225626716604245,
"calib/step_q_c_n": 801.0,
"calib/step_q_gap": 0.1352553854935818,
"calib/step_q_w": 0.2873072861668427,
"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": 0.0,
"completions/max_length": 2131.0,
"completions/max_terminated_length": 2131.0,
"completions/mean_length": 553.19921875,
"completions/mean_terminated_length": 553.19921875,
"completions/min_length": 92.0,
"completions/min_terminated_length": 92.0,
"epoch": 0.1984,
"grad_norm": 0.03475416079163551,
"kl": 0.146728515625,
"learning_rate": 3.8888888888888895e-07,
"loss": -0.1265,
"mask/has_final_conf_rate": 0.98828125,
"mask/share_final_conf": 0.03105587139725685,
"mask/share_reasoning": 0.8362254500389099,
"mask/share_step_conf": 0.13271866738796234,
"num_tokens": 45646849.0,
"reward": 0.960451602935791,
"reward_std": 0.20122194290161133,
"rewards/accuracy_reward_step": 0.5078125,
"rewards/final_brier_reward_step": 0.7569808959960938,
"rewards/format_reward_step": 0.98046875,
"rewards/step_l2_reward": 0.8662659525871277,
"step": 186
},
{
"adv/mean_abs_final_conf": 0.7896785736083984,
"adv/mean_abs_reasoning": 0.679695188999176,
"adv/mean_abs_step_conf": 0.7941338419914246,
"adv/ratio_final_to_reasoning": 1.1618128043118394,
"adv/ratio_step_to_reasoning": 1.1683676077812981,
"adv/std_final_conf": 0.9174456000328064,
"adv/std_reasoning": 0.8591005206108093,
"adv/std_step_conf": 0.9346184134483337,
"calib/answer_extract_rate": 0.9375,
"calib/auroc": 0.6905064183946961,
"calib/avg_num_step_conf": 7.2734375,
"calib/ece": 0.28705394190871375,
"calib/final_conf_rate": 0.94140625,
"calib/format_rate": 0.92578125,
"calib/frac_conf_gt_0.9": 0.3941908713692946,
"calib/gap": 0.2636711807024969,
"calib/mean_conf": 0.5146887966804979,
"calib/mu_c": 0.666764705882353,
"calib/mu_w": 0.40309352517985614,
"calib/nonempty_final_conf_rate": 0.94140625,
"calib/nonempty_reasoning_rate": 0.96875,
"calib/nonempty_step_conf_rate": 0.95703125,
"calib/pce": 0.18925311203319506,
"calib/std_conf": 0.4275016762955222,
"calib/step_conf_rate": 0.95703125,
"calib/step_q_c": 0.42846666666666666,
"calib/step_q_c_n": 600.0,
"calib/step_q_gap": 0.14417982039091393,
"calib/step_q_w": 0.28428684627575274,
"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": 0.0,
"completions/max_length": 2975.0,
"completions/max_terminated_length": 2975.0,
"completions/mean_length": 667.828125,
"completions/mean_terminated_length": 667.828125,
"completions/min_length": 179.0,
"completions/min_terminated_length": 179.0,
"epoch": 0.19946666666666665,
"grad_norm": 0.029744060710072517,
"kl": 0.123138427734375,
"learning_rate": 3.611111111111111e-07,
"loss": 0.1127,
"mask/has_final_conf_rate": 0.94140625,
"mask/share_final_conf": 0.028970632702112198,
"mask/share_reasoning": 0.8479760885238647,
"mask/share_step_conf": 0.12305329740047455,
"num_tokens": 45919357.0,
"reward": 0.8574899435043335,
"reward_std": 0.24872718751430511,
"rewards/accuracy_reward_step": 0.3984375,
"rewards/final_brier_reward_step": 0.6425703167915344,
"rewards/format_reward_step": 0.92578125,
"rewards/step_l2_reward": 0.8075656890869141,
"step": 187
},
{
"adv/mean_abs_final_conf": 0.7544801235198975,
"adv/mean_abs_reasoning": 0.6453492045402527,
"adv/mean_abs_step_conf": 0.7617857456207275,
"adv/ratio_final_to_reasoning": 1.1691036700934492,
"adv/ratio_step_to_reasoning": 1.1804240870854166,
"adv/std_final_conf": 0.9098480939865112,
"adv/std_reasoning": 0.8589980602264404,
"adv/std_step_conf": 0.9342439770698547,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7976353332030487,
"calib/avg_num_step_conf": 6.9375,
"calib/ece": 0.20193548387096763,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.3588709677419355,
"calib/gap": 0.4378411829848219,
"calib/mean_conf": 0.4895967741935484,
"calib/mu_c": 0.6996899224806202,
"calib/mu_w": 0.26184873949579834,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.08568548387096765,
"calib/std_conf": 0.4361837789531521,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.41524390243902437,
"calib/step_q_c_n": 820.0,
"calib/step_q_gap": 0.11278574344320852,
"calib/step_q_w": 0.30245815899581585,
"calib/step_q_w_n": 956.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 3056.0,
"completions/max_terminated_length": 3056.0,
"completions/mean_length": 599.65625,
"completions/mean_terminated_length": 602.0078735351562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 139.0,
"epoch": 0.20053333333333334,
"grad_norm": 0.03370795398950577,
"kl": 0.1395416259765625,
"learning_rate": 3.3333333333333335e-07,
"loss": 0.0514,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.03147049620747566,
"mask/share_reasoning": 0.8312385678291321,
"mask/share_step_conf": 0.13338470458984375,
"num_tokens": 46176941.0,
"reward": 0.9409340620040894,
"reward_std": 0.22601254284381866,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.7418223023414612,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8478583097457886,
"step": 188
},
{
"adv/mean_abs_final_conf": 0.6637656092643738,
"adv/mean_abs_reasoning": 0.5426450371742249,
"adv/mean_abs_step_conf": 0.7172122001647949,
"adv/ratio_final_to_reasoning": 1.2232040538339268,
"adv/ratio_step_to_reasoning": 1.3216967834067235,
"adv/std_final_conf": 0.8593613505363464,
"adv/std_reasoning": 0.7928246259689331,
"adv/std_step_conf": 0.9344122409820557,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8002862347124642,
"calib/avg_num_step_conf": 6.57421875,
"calib/ece": 0.18821236559139787,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.94921875,
"calib/frac_conf_gt_0.9": 0.31451612903225806,
"calib/gap": 0.453576632838928,
"calib/mean_conf": 0.41638440860215054,
"calib/mu_c": 0.6468306010928963,
"calib/mu_w": 0.19325396825396826,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.05633064516129035,
"calib/std_conf": 0.43000735707708637,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.42563421828908554,
"calib/step_q_c_n": 678.0,
"calib/step_q_gap": 0.19948496455774226,
"calib/step_q_w": 0.22614925373134329,
"calib/step_q_w_n": 1005.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2374.0,
"completions/max_terminated_length": 2374.0,
"completions/mean_length": 522.41796875,
"completions/mean_terminated_length": 526.531494140625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 78.0,
"epoch": 0.2016,
"grad_norm": 0.025214437395334244,
"kl": 0.1561431884765625,
"learning_rate": 3.055555555555556e-07,
"loss": -0.1702,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.032530926167964935,
"mask/share_reasoning": 0.833452582359314,
"mask/share_step_conf": 0.1262039989233017,
"num_tokens": 46418448.0,
"reward": 0.9340362548828125,
"reward_std": 0.21618613600730896,
"rewards/accuracy_reward_step": 0.4765625,
"rewards/final_brier_reward_step": 0.7459970712661743,
"rewards/format_reward_step": 0.94921875,
"rewards/step_l2_reward": 0.8369191288948059,
"step": 189
},
{
"adv/mean_abs_final_conf": 0.736014723777771,
"adv/mean_abs_reasoning": 0.519279956817627,
"adv/mean_abs_step_conf": 0.7525043487548828,
"adv/ratio_final_to_reasoning": 1.4173755680623392,
"adv/ratio_step_to_reasoning": 1.4491303561311246,
"adv/std_final_conf": 0.9154024720191956,
"adv/std_reasoning": 0.775436520576477,
"adv/std_step_conf": 0.9342000484466553,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7823073919833421,
"calib/avg_num_step_conf": 6.703125,
"calib/ece": 0.20297188755020082,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96484375,
"calib/frac_conf_gt_0.9": 0.3092369477911647,
"calib/gap": 0.40319104633003644,
"calib/mean_conf": 0.4552610441767068,
"calib/mu_c": 0.638235294117647,
"calib/mu_w": 0.2350442477876106,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.9765625,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.05602409638554215,
"calib/std_conf": 0.4188893712544085,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.3838032786885246,
"calib/step_q_c_n": 915.0,
"calib/step_q_gap": 0.10417781052373054,
"calib/step_q_w": 0.27962546816479406,
"calib/step_q_w_n": 801.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1753.0,
"completions/max_terminated_length": 1753.0,
"completions/mean_length": 563.4609375,
"completions/mean_terminated_length": 565.6705932617188,
"completions/min_length": 0.0,
"completions/min_terminated_length": 2.0,
"epoch": 0.20266666666666666,
"grad_norm": 0.06035936623811722,
"kl": 0.1408233642578125,
"learning_rate": 2.7777777777777776e-07,
"loss": -0.1432,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.030015822499990463,
"mask/share_reasoning": 0.8396799564361572,
"mask/share_step_conf": 0.1263979971408844,
"num_tokens": 46668302.0,
"reward": 0.9438740015029907,
"reward_std": 0.20643171668052673,
"rewards/accuracy_reward_step": 0.53125,
"rewards/final_brier_reward_step": 0.739453136920929,
"rewards/format_reward_step": 0.96484375,
"rewards/step_l2_reward": 0.8490761518478394,
"step": 190
},
{
"adv/mean_abs_final_conf": 0.7668330669403076,
"adv/mean_abs_reasoning": 0.6082367300987244,
"adv/mean_abs_step_conf": 0.7396363019943237,
"adv/ratio_final_to_reasoning": 1.2607477138314898,
"adv/ratio_step_to_reasoning": 1.2160336023677354,
"adv/std_final_conf": 0.9332967400550842,
"adv/std_reasoning": 0.8266222476959229,
"adv/std_step_conf": 0.9346864819526672,
"calib/answer_extract_rate": 0.97265625,
"calib/auroc": 0.7542235437221247,
"calib/avg_num_step_conf": 7.11328125,
"calib/ece": 0.21726907630522088,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.37349397590361444,
"calib/gap": 0.40706649547236107,
"calib/mean_conf": 0.5136546184738956,
"calib/mu_c": 0.7605102040816326,
"calib/mu_w": 0.35344370860927155,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.984375,
"calib/pce": 0.16867469879518074,
"calib/std_conf": 0.438051427805417,
"calib/step_conf_rate": 0.984375,
"calib/step_q_c": 0.45821848739495796,
"calib/step_q_c_n": 595.0,
"calib/step_q_gap": 0.15850396863476218,
"calib/step_q_w": 0.2997145187601958,
"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": 0.0,
"completions/max_length": 2903.0,
"completions/max_terminated_length": 2903.0,
"completions/mean_length": 576.875,
"completions/mean_terminated_length": 576.875,
"completions/min_length": 112.0,
"completions/min_terminated_length": 112.0,
"epoch": 0.20373333333333332,
"grad_norm": 0.04359288513660431,
"kl": 0.1428985595703125,
"learning_rate": 2.5000000000000004e-07,
"loss": -0.0011,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.03338399529457092,
"mask/share_reasoning": 0.8276382684707642,
"mask/share_step_conf": 0.13897772133350372,
"num_tokens": 46920150.0,
"reward": 0.9028844833374023,
"reward_std": 0.24160155653953552,
"rewards/accuracy_reward_step": 0.38671875,
"rewards/final_brier_reward_step": 0.7154718637466431,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8215470314025879,
"step": 191
},
{
"adv/mean_abs_final_conf": 0.7420070171356201,
"adv/mean_abs_reasoning": 0.6264379024505615,
"adv/mean_abs_step_conf": 0.7037802934646606,
"adv/ratio_final_to_reasoning": 1.1844861465645102,
"adv/ratio_step_to_reasoning": 1.1234637794289641,
"adv/std_final_conf": 0.916337788105011,
"adv/std_reasoning": 0.8430317044258118,
"adv/std_step_conf": 0.9342029094696045,
"calib/answer_extract_rate": 0.98046875,
"calib/auroc": 0.8022857142857143,
"calib/avg_num_step_conf": 6.87890625,
"calib/ece": 0.19637450199203188,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.95703125,
"calib/frac_conf_gt_0.9": 0.36254980079681276,
"calib/gap": 0.4784342857142856,
"calib/mean_conf": 0.4474501992031873,
"calib/mu_c": 0.6857142857142856,
"calib/mu_w": 0.20728,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9765625,
"calib/pce": 0.07091633466135458,
"calib/std_conf": 0.4440952489605836,
"calib/step_conf_rate": 0.9765625,
"calib/step_q_c": 0.38982278481012655,
"calib/step_q_c_n": 790.0,
"calib/step_q_gap": 0.1389473986103325,
"calib/step_q_w": 0.25087538619979405,
"calib/step_q_w_n": 971.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.0078125,
"completions/max_length": 2570.0,
"completions/max_terminated_length": 2570.0,
"completions/mean_length": 567.75390625,
"completions/mean_terminated_length": 572.2244262695312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 86.0,
"epoch": 0.2048,
"grad_norm": 0.029248613864183426,
"kl": 0.1450042724609375,
"learning_rate": 2.2222222222222224e-07,
"loss": 0.0297,
"mask/has_final_conf_rate": 0.98046875,
"mask/share_final_conf": 0.03411424905061722,
"mask/share_reasoning": 0.8246064186096191,
"mask/share_step_conf": 0.13346685469150543,
"num_tokens": 47170471.0,
"reward": 0.953478217124939,
"reward_std": 0.22533637285232544,
"rewards/accuracy_reward_step": 0.4921875,
"rewards/final_brier_reward_step": 0.7501965165138245,
"rewards/format_reward_step": 0.95703125,
"rewards/step_l2_reward": 0.8669161200523376,
"step": 192
},
{
"adv/mean_abs_final_conf": 0.766300618648529,
"adv/mean_abs_reasoning": 0.6802315711975098,
"adv/mean_abs_step_conf": 0.7448439002037048,
"adv/ratio_final_to_reasoning": 1.1265290396614487,
"adv/ratio_step_to_reasoning": 1.0949857838742307,
"adv/std_final_conf": 0.9214043021202087,
"adv/std_reasoning": 0.8592211008071899,
"adv/std_step_conf": 0.9346270561218262,
"calib/answer_extract_rate": 0.984375,
"calib/auroc": 0.7196492805755397,
"calib/avg_num_step_conf": 6.125,
"calib/ece": 0.2221513944223108,
"calib/final_conf_rate": 0.98046875,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.21115537848605578,
"calib/gap": 0.3011208890030832,
"calib/mean_conf": 0.4049402390438247,
"calib/mu_c": 0.5716964285714285,
"calib/mu_w": 0.27057553956834535,
"calib/nonempty_final_conf_rate": 0.98046875,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.98828125,
"calib/pce": 0.09043824701195222,
"calib/std_conf": 0.39759573305878976,
"calib/step_conf_rate": 0.98828125,
"calib/step_q_c": 0.3862333825701625,
"calib/step_q_c_n": 677.0,
"calib/step_q_gap": 0.06997187864199189,
"calib/step_q_w": 0.3162615039281706,
"calib/step_q_w_n": 891.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 1709.0,
"completions/max_terminated_length": 1709.0,
"completions/mean_length": 509.3515625,
"completions/mean_terminated_length": 511.34906005859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 168.0,
"epoch": 0.20586666666666667,
"grad_norm": 0.03636451065540314,
"kl": 0.1459808349609375,
"learning_rate": 1.9444444444444447e-07,
"loss": -0.0495,
"mask/has_final_conf_rate": 0.9765625,
"mask/share_final_conf": 0.03263282775878906,
"mask/share_reasoning": 0.8374545574188232,
"mask/share_step_conf": 0.12600642442703247,
"num_tokens": 47406577.0,
"reward": 0.927412211894989,
"reward_std": 0.2111629843711853,
"rewards/accuracy_reward_step": 0.44140625,
"rewards/final_brier_reward_step": 0.7223039269447327,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8504891991615295,
"step": 193
},
{
"adv/mean_abs_final_conf": 0.6695102453231812,
"adv/mean_abs_reasoning": 0.5588866472244263,
"adv/mean_abs_step_conf": 0.7345038652420044,
"adv/ratio_final_to_reasoning": 1.1979356612796888,
"adv/ratio_step_to_reasoning": 1.3142268989422776,
"adv/std_final_conf": 0.8768190741539001,
"adv/std_reasoning": 0.7928152680397034,
"adv/std_step_conf": 0.9336667060852051,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8134920634920635,
"calib/avg_num_step_conf": 6.26171875,
"calib/ece": 0.16755020080321287,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.37751004016064255,
"calib/gap": 0.4849632210607821,
"calib/mean_conf": 0.5169477911646586,
"calib/mu_c": 0.7565079365079366,
"calib/mu_w": 0.27154471544715447,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 1.0,
"calib/nonempty_step_conf_rate": 0.99609375,
"calib/pce": 0.08923694779116467,
"calib/std_conf": 0.43803174373464926,
"calib/step_conf_rate": 0.99609375,
"calib/step_q_c": 0.4300995024875622,
"calib/step_q_c_n": 804.0,
"calib/step_q_gap": 0.16561890173662352,
"calib/step_q_w": 0.26448060075093865,
"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": 0.00390625,
"completions/max_length": 2510.0,
"completions/max_terminated_length": 2510.0,
"completions/mean_length": 511.00390625,
"completions/mean_terminated_length": 513.0078735351562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 177.0,
"epoch": 0.20693333333333333,
"grad_norm": 0.03281570225954056,
"kl": 0.1405181884765625,
"learning_rate": 1.6666666666666668e-07,
"loss": -0.015,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.0331403985619545,
"mask/share_reasoning": 0.8325739502906799,
"mask/share_step_conf": 0.13037940859794617,
"num_tokens": 47643338.0,
"reward": 0.9630546569824219,
"reward_std": 0.19925682246685028,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.7746968865394592,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.8584436178207397,
"step": 194
},
{
"adv/mean_abs_final_conf": 0.7317264080047607,
"adv/mean_abs_reasoning": 0.6601746082305908,
"adv/mean_abs_step_conf": 0.7371601462364197,
"adv/ratio_final_to_reasoning": 1.1083831442198966,
"adv/ratio_step_to_reasoning": 1.1166139034219547,
"adv/std_final_conf": 0.919392466545105,
"adv/std_reasoning": 0.859114408493042,
"adv/std_step_conf": 0.9345622062683105,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.7941827893948277,
"calib/avg_num_step_conf": 6.72265625,
"calib/ece": 0.21443548387096772,
"calib/final_conf_rate": 0.96875,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.3588709677419355,
"calib/gap": 0.44475278483486413,
"calib/mean_conf": 0.4640322580645161,
"calib/mu_c": 0.6774418604651162,
"calib/mu_w": 0.23268907563025207,
"calib/nonempty_final_conf_rate": 0.96875,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.07915322580645162,
"calib/std_conf": 0.4345695299704894,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.4276574307304787,
"calib/step_q_c_n": 794.0,
"calib/step_q_gap": 0.17257436708430823,
"calib/step_q_w": 0.25508306364617045,
"calib/step_q_w_n": 927.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 2930.0,
"completions/max_terminated_length": 2930.0,
"completions/mean_length": 574.30859375,
"completions/mean_terminated_length": 576.560791015625,
"completions/min_length": 0.0,
"completions/min_terminated_length": 129.0,
"epoch": 0.208,
"grad_norm": 0.04601413384079933,
"kl": 0.14105224609375,
"learning_rate": 1.3888888888888888e-07,
"loss": 0.01,
"mask/has_final_conf_rate": 0.96875,
"mask/share_final_conf": 0.03239461034536362,
"mask/share_reasoning": 0.8322038650512695,
"mask/share_step_conf": 0.13149525225162506,
"num_tokens": 47896345.0,
"reward": 0.9279034733772278,
"reward_std": 0.23658813536167145,
"rewards/accuracy_reward_step": 0.50390625,
"rewards/final_brier_reward_step": 0.7333793044090271,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.8325839042663574,
"step": 195
},
{
"adv/mean_abs_final_conf": 0.7037650346755981,
"adv/mean_abs_reasoning": 0.5706368684768677,
"adv/mean_abs_step_conf": 0.761438250541687,
"adv/ratio_final_to_reasoning": 1.2332975199343033,
"adv/ratio_step_to_reasoning": 1.3343656756250302,
"adv/std_final_conf": 0.9014235138893127,
"adv/std_reasoning": 0.7928955554962158,
"adv/std_step_conf": 0.9342949390411377,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.702474690663667,
"calib/avg_num_step_conf": 5.85546875,
"calib/ece": 0.26434782608695656,
"calib/final_conf_rate": 0.98828125,
"calib/format_rate": 0.984375,
"calib/frac_conf_gt_0.9": 0.4505928853754941,
"calib/gap": 0.29691163604549436,
"calib/mean_conf": 0.5812648221343873,
"calib/mu_c": 0.7291338582677166,
"calib/mu_w": 0.43222222222222223,
"calib/nonempty_final_conf_rate": 0.98828125,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.17181818181818181,
"calib/std_conf": 0.4268674168044643,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.44906030855539975,
"calib/step_q_c_n": 713.0,
"calib/step_q_gap": 0.09664300575641754,
"calib/step_q_w": 0.3524173027989822,
"calib/step_q_w_n": 786.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 1124.0,
"completions/max_terminated_length": 1124.0,
"completions/mean_length": 435.875,
"completions/mean_terminated_length": 437.5843505859375,
"completions/min_length": 0.0,
"completions/min_terminated_length": 140.0,
"epoch": 0.20906666666666668,
"grad_norm": 0.02775087207555771,
"kl": 0.162445068359375,
"learning_rate": 1.1111111111111112e-07,
"loss": -0.0394,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.03831184655427933,
"mask/share_reasoning": 0.8148221969604492,
"mask/share_step_conf": 0.14295971393585205,
"num_tokens": 48110473.0,
"reward": 0.9260044693946838,
"reward_std": 0.21588167548179626,
"rewards/accuracy_reward_step": 0.49609375,
"rewards/final_brier_reward_step": 0.6996980309486389,
"rewards/format_reward_step": 0.984375,
"rewards/step_l2_reward": 0.856217086315155,
"step": 196
},
{
"adv/mean_abs_final_conf": 0.7061256170272827,
"adv/mean_abs_reasoning": 0.5036745071411133,
"adv/mean_abs_step_conf": 0.7423129677772522,
"adv/ratio_final_to_reasoning": 1.4019482960042868,
"adv/ratio_step_to_reasoning": 1.4737949950865394,
"adv/std_final_conf": 0.8724972009658813,
"adv/std_reasoning": 0.7576428651809692,
"adv/std_step_conf": 0.933771014213562,
"calib/answer_extract_rate": 0.98828125,
"calib/auroc": 0.8175076064908722,
"calib/avg_num_step_conf": 7.34375,
"calib/ece": 0.1787698412698412,
"calib/final_conf_rate": 0.984375,
"calib/format_rate": 0.9765625,
"calib/frac_conf_gt_0.9": 0.3412698412698413,
"calib/gap": 0.49297920892494945,
"calib/mean_conf": 0.4517063492063492,
"calib/mu_c": 0.7177586206896553,
"calib/mu_w": 0.2247794117647059,
"calib/nonempty_final_conf_rate": 0.984375,
"calib/nonempty_reasoning_rate": 0.9921875,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.085079365079365,
"calib/std_conf": 0.4412687617749729,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.43830136986301377,
"calib/step_q_c_n": 730.0,
"calib/step_q_gap": 0.1627970220369268,
"calib/step_q_w": 0.27550434782608696,
"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": 0.00390625,
"completions/max_length": 2272.0,
"completions/max_terminated_length": 2272.0,
"completions/mean_length": 564.4765625,
"completions/mean_terminated_length": 566.6902465820312,
"completions/min_length": 0.0,
"completions/min_terminated_length": 46.0,
"epoch": 0.21013333333333334,
"grad_norm": 0.027106324210762978,
"kl": 0.147796630859375,
"learning_rate": 8.333333333333334e-08,
"loss": -0.1334,
"mask/has_final_conf_rate": 0.984375,
"mask/share_final_conf": 0.031168133020401,
"mask/share_reasoning": 0.8300639986991882,
"mask/share_step_conf": 0.13486161828041077,
"num_tokens": 48360035.0,
"reward": 0.9748759269714355,
"reward_std": 0.19350528717041016,
"rewards/accuracy_reward_step": 0.453125,
"rewards/final_brier_reward_step": 0.7844051122665405,
"rewards/format_reward_step": 0.9765625,
"rewards/step_l2_reward": 0.8794092535972595,
"step": 197
},
{
"adv/mean_abs_final_conf": 0.6879289150238037,
"adv/mean_abs_reasoning": 0.5308552980422974,
"adv/mean_abs_step_conf": 0.7343833446502686,
"adv/ratio_final_to_reasoning": 1.2958878202040494,
"adv/ratio_step_to_reasoning": 1.383396468601797,
"adv/std_final_conf": 0.8872906565666199,
"adv/std_reasoning": 0.7755385637283325,
"adv/std_step_conf": 0.9340898394584656,
"calib/answer_extract_rate": 0.9765625,
"calib/auroc": 0.8280604338842976,
"calib/avg_num_step_conf": 7.02734375,
"calib/ece": 0.1369076305220884,
"calib/final_conf_rate": 0.97265625,
"calib/format_rate": 0.96875,
"calib/frac_conf_gt_0.9": 0.2931726907630522,
"calib/gap": 0.49805397727272727,
"calib/mean_conf": 0.44124497991967876,
"calib/mu_c": 0.6972727272727273,
"calib/mu_w": 0.19921875,
"calib/nonempty_final_conf_rate": 0.97265625,
"calib/nonempty_reasoning_rate": 0.99609375,
"calib/nonempty_step_conf_rate": 0.9921875,
"calib/pce": 0.04610441767068277,
"calib/std_conf": 0.421304695929381,
"calib/step_conf_rate": 0.9921875,
"calib/step_q_c": 0.4291873589164785,
"calib/step_q_c_n": 886.0,
"calib/step_q_gap": 0.13000882660541613,
"calib/step_q_w": 0.2991785323110624,
"calib/step_q_w_n": 913.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2773.0,
"completions/max_terminated_length": 2773.0,
"completions/mean_length": 528.96484375,
"completions/mean_terminated_length": 533.1299438476562,
"completions/min_length": 0.0,
"completions/min_terminated_length": 112.0,
"epoch": 0.2112,
"grad_norm": 0.028157765045762062,
"kl": 0.1432647705078125,
"learning_rate": 5.555555555555556e-08,
"loss": -0.0526,
"mask/has_final_conf_rate": 0.97265625,
"mask/share_final_conf": 0.034415990114212036,
"mask/share_reasoning": 0.8118489384651184,
"mask/share_step_conf": 0.14592257142066956,
"num_tokens": 48600834.0,
"reward": 0.9752619862556458,
"reward_std": 0.17956218123435974,
"rewards/accuracy_reward_step": 0.47265625,
"rewards/final_brier_reward_step": 0.7932171821594238,
"rewards/format_reward_step": 0.96875,
"rewards/step_l2_reward": 0.869025468826294,
"step": 198
},
{
"adv/mean_abs_final_conf": 0.7226166725158691,
"adv/mean_abs_reasoning": 0.691262423992157,
"adv/mean_abs_step_conf": 0.7821977138519287,
"adv/ratio_final_to_reasoning": 1.0453579529791828,
"adv/ratio_step_to_reasoning": 1.1315495920270122,
"adv/std_final_conf": 0.8770599365234375,
"adv/std_reasoning": 0.8748891353607178,
"adv/std_step_conf": 0.9344019293785095,
"calib/answer_extract_rate": 0.95703125,
"calib/auroc": 0.7468637394901908,
"calib/avg_num_step_conf": 6.82421875,
"calib/ece": 0.21922448979591835,
"calib/final_conf_rate": 0.95703125,
"calib/format_rate": 0.9453125,
"calib/frac_conf_gt_0.9": 0.3224489795918367,
"calib/gap": 0.34982517015881476,
"calib/mean_conf": 0.4873061224489796,
"calib/mu_c": 0.66864406779661,
"calib/mu_w": 0.31881889763779525,
"calib/nonempty_final_conf_rate": 0.95703125,
"calib/nonempty_reasoning_rate": 0.98828125,
"calib/nonempty_step_conf_rate": 0.98046875,
"calib/pce": 0.11244897959183672,
"calib/std_conf": 0.41766221114146396,
"calib/step_conf_rate": 0.98046875,
"calib/step_q_c": 0.39829530201342284,
"calib/step_q_c_n": 745.0,
"calib/step_q_gap": 0.11073043175394182,
"calib/step_q_w": 0.287564870259481,
"calib/step_q_w_n": 1002.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 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": 2804.0,
"completions/max_terminated_length": 2804.0,
"completions/mean_length": 585.66015625,
"completions/mean_terminated_length": 592.604736328125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 115.0,
"epoch": 0.21226666666666666,
"grad_norm": 0.030716493725776672,
"kl": 0.1441192626953125,
"learning_rate": 2.777777777777778e-08,
"loss": -0.1249,
"mask/has_final_conf_rate": 0.95703125,
"mask/share_final_conf": 0.033089008182287216,
"mask/share_reasoning": 0.8238986134529114,
"mask/share_step_conf": 0.1312936544418335,
"num_tokens": 48854963.0,
"reward": 0.9157102704048157,
"reward_std": 0.23339220881462097,
"rewards/accuracy_reward_step": 0.4609375,
"rewards/final_brier_reward_step": 0.7084863185882568,
"rewards/format_reward_step": 0.9453125,
"rewards/step_l2_reward": 0.841684103012085,
"step": 199
},
{
"adv/mean_abs_final_conf": 0.6958111524581909,
"adv/mean_abs_reasoning": 0.5497347116470337,
"adv/mean_abs_step_conf": 0.7385485172271729,
"adv/ratio_final_to_reasoning": 1.2657216976048404,
"adv/ratio_step_to_reasoning": 1.3434634953547744,
"adv/std_final_conf": 0.8847732543945312,
"adv/std_reasoning": 0.7929638028144836,
"adv/std_step_conf": 0.9344330430030823,
"calib/answer_extract_rate": 0.96875,
"calib/auroc": 0.8473539518900344,
"calib/avg_num_step_conf": 6.02734375,
"calib/ece": 0.14963562753036444,
"calib/final_conf_rate": 0.96484375,
"calib/format_rate": 0.953125,
"calib/frac_conf_gt_0.9": 0.48582995951417,
"calib/gap": 0.5701237113402062,
"calib/mean_conf": 0.5821052631578948,
"calib/mu_c": 0.806,
"calib/mu_w": 0.2358762886597938,
"calib/nonempty_final_conf_rate": 0.96484375,
"calib/nonempty_reasoning_rate": 0.984375,
"calib/nonempty_step_conf_rate": 0.97265625,
"calib/pce": 0.06222672064777332,
"calib/std_conf": 0.44081955906755055,
"calib/step_conf_rate": 0.97265625,
"calib/step_q_c": 0.44430604982206406,
"calib/step_q_c_n": 843.0,
"calib/step_q_gap": 0.18290604982206404,
"calib/step_q_w": 0.2614,
"calib/step_q_w_n": 700.0,
"clip_ratio/high_max": 0.0,
"clip_ratio/high_mean": 0.0,
"clip_ratio/low_mean": 0.0,
"clip_ratio/low_min": 0.0,
"clip_ratio/region_mean": 0.0,
"completions/clipped_ratio": 0.00390625,
"completions/max_length": 3050.0,
"completions/max_terminated_length": 3050.0,
"completions/mean_length": 546.375,
"completions/mean_terminated_length": 548.5177001953125,
"completions/min_length": 0.0,
"completions/min_terminated_length": 99.0,
"epoch": 0.21333333333333335,
"grad_norm": 0.05999818071722984,
"kl": 0.1330413818359375,
"learning_rate": 0.0,
"loss": -0.0525,
"mask/has_final_conf_rate": 0.96484375,
"mask/share_final_conf": 0.03310883790254593,
"mask/share_reasoning": 0.8392375707626343,
"mask/share_step_conf": 0.12374730408191681,
"num_tokens": 49102883.0,
"reward": 0.9791679382324219,
"reward_std": 0.22837120294570923,
"rewards/accuracy_reward_step": 0.5859375,
"rewards/final_brier_reward_step": 0.7976699471473694,
"rewards/format_reward_step": 0.953125,
"rewards/step_l2_reward": 0.852853536605835,
"step": 200
},
{
"epoch": 0.21333333333333335,
"step": 200,
"total_flos": 0.0,
"train_loss": 7.663144015993457,
"train_runtime": 15537.4107,
"train_samples_per_second": 3.295,
"train_steps_per_second": 0.013
}
],
"logging_steps": 1,
"max_steps": 200,
"num_input_tokens_seen": 49102883,
"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