{ "best_metric": null, "best_model_checkpoint": null, "epoch": 0.9984, "eval_steps": 50, "global_step": 312, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "calibration/aurc": 0.6372745020551545, "calibration/batch_distribution_entropy": 0.6459174196273255, "calibration/batch_entropy_100bins": 0.47724905350322705, "calibration/batch_entropy_10bins": 0.6459174196273255, "calibration/batch_entropy_50bins": 0.5572498883154378, "calibration/batch_uniqueness": 0.7166131605566551, "calibration/confidence_entropy": 0.34035792583116287, "calibration/coverage@0%": 0.0, "calibration/coverage@1%": 0.0, "calibration/coverage@10%": 0.0, "calibration/coverage@15%": 0.0, "calibration/coverage@20%": 0.0, "calibration/coverage@25%": 0.0, "calibration/coverage@30%": 0.0, "calibration/coverage@5%": 0.0, "calibration/ece": 0.4998850169792884, "calibration/mean_confidence": 0.7930507959746237, "calibration/prompt_uniqueness": 0.5883075255362273, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0345703125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1507.2, "completions/mean_length": 267.9455078125, "completions/mean_terminated_length": 222.52838745117188, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.016, "grad_norm": 0.031000174582004547, "learning_rate": 3.1249999999999997e-07, "loss": 0.0705, "num_tokens": 17587794.0, "reward": 0.5391062259674072, "reward_std": 0.4095682919025421, "rewards/accuracy_reward": 0.2259765625, "rewards/brier_reward": 0.3801437199115753, "rewards/confidence_uniqueness_reward": 0.4887708604335785, "rewards/format_reward": 0.687109375, "rewards/frontier_aurc_reward": 0.306541383266449, "rewards/frontier_coverage_0": 0.306541383266449, "rewards/frontier_coverage_1": 0.306541383266449, "rewards/frontier_coverage_10": 0.306541383266449, "rewards/frontier_coverage_15": 0.306541383266449, "rewards/frontier_coverage_20": 0.306541383266449, "rewards/frontier_coverage_25": 0.306541383266449, "rewards/frontier_coverage_5": 0.306541383266449, "rewards/frontier_ece_reward": 0.306541383266449, "rewards/frontier_entropy_batch_reward": -0.6563648700714111, "signal/accuracy_reward/centered_abs_mean": 0.24376220703125, "signal/accuracy_reward/group_bin_occupancy": 0.210546875, "signal/accuracy_reward/group_std_mean": 0.2845919132232666, "signal/accuracy_reward/group_zero_std_frac": 0.315625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.121881103515625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.121881103515625, "signal/advantage_abs_mean": 0.34902645349502565, "signal/advantage_pre_scale_abs_mean": 0.34902645349502565, "signal/advantage_pre_scale_std": 0.41952605843544005, "signal/advantage_std": 0.41952605843544005, "signal/brier_reward/centered_abs_mean": 0.3207852005958557, "signal/brier_reward/group_bin_occupancy": 0.7453125, "signal/brier_reward/group_std_mean": 0.36561399102211, "signal/brier_reward/group_zero_std_frac": 0.003125, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.032078520581126216, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.032078520581126216, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.2937858641147614, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.59921875, "signal/confidence_uniqueness_reward/group_std_mean": 0.34655541777610777, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.029378585889935492, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.029378585889935492, "signal/format_reward/centered_abs_mean": 0.39891357421875, "signal/format_reward/group_bin_occupancy": 0.25, "signal/format_reward/group_std_mean": 0.45102897882461546, "signal/format_reward/group_zero_std_frac": 0.0, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.199456787109375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.199456787109375, "signal/frontier_aurc_reward/centered_abs_mean": 0.2946657776832581, "signal/frontier_aurc_reward/group_bin_occupancy": 0.6609375, "signal/frontier_aurc_reward/group_std_mean": 0.34457772970199585, "signal/frontier_aurc_reward/group_zero_std_frac": 0.003125, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_0/centered_abs_mean": 0.2946657776832581, "signal/frontier_coverage_0/group_bin_occupancy": 0.6609375, "signal/frontier_coverage_0/group_std_mean": 0.34457772970199585, "signal/frontier_coverage_0/group_zero_std_frac": 0.003125, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_1/centered_abs_mean": 0.2946657776832581, "signal/frontier_coverage_1/group_bin_occupancy": 0.6609375, "signal/frontier_coverage_1/group_std_mean": 0.34457772970199585, "signal/frontier_coverage_1/group_zero_std_frac": 0.003125, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_10/centered_abs_mean": 0.2946657776832581, "signal/frontier_coverage_10/group_bin_occupancy": 0.6609375, "signal/frontier_coverage_10/group_std_mean": 0.34457772970199585, "signal/frontier_coverage_10/group_zero_std_frac": 0.003125, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_15/centered_abs_mean": 0.2946657776832581, "signal/frontier_coverage_15/group_bin_occupancy": 0.6609375, "signal/frontier_coverage_15/group_std_mean": 0.34457772970199585, "signal/frontier_coverage_15/group_zero_std_frac": 0.003125, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_20/centered_abs_mean": 0.2946657776832581, "signal/frontier_coverage_20/group_bin_occupancy": 0.6609375, "signal/frontier_coverage_20/group_std_mean": 0.34457772970199585, "signal/frontier_coverage_20/group_zero_std_frac": 0.003125, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_25/centered_abs_mean": 0.2946657776832581, "signal/frontier_coverage_25/group_bin_occupancy": 0.6609375, "signal/frontier_coverage_25/group_std_mean": 0.34457772970199585, "signal/frontier_coverage_25/group_zero_std_frac": 0.003125, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_5/centered_abs_mean": 0.2946657776832581, "signal/frontier_coverage_5/group_bin_occupancy": 0.6609375, "signal/frontier_coverage_5/group_std_mean": 0.34457772970199585, "signal/frontier_coverage_5/group_zero_std_frac": 0.003125, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0036833221558481453, "signal/frontier_ece_reward/centered_abs_mean": 0.2946657776832581, "signal/frontier_ece_reward/group_bin_occupancy": 0.6609375, "signal/frontier_ece_reward/group_std_mean": 0.34457772970199585, "signal/frontier_ece_reward/group_zero_std_frac": 0.003125, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.029466577246785162, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.029466577246785162, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.42181131839752195, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.312890625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.4700261831283569, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.04218113347887993, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.04218113347887993, "step": 5 }, { "calibration/aurc": 0.6832300553855127, "calibration/batch_distribution_entropy": 0.6450415065753978, "calibration/batch_entropy_100bins": 0.4788736890772074, "calibration/batch_entropy_10bins": 0.6450415065753978, "calibration/batch_entropy_50bins": 0.5614978981364785, "calibration/batch_uniqueness": 0.7197617937415485, "calibration/confidence_entropy": 0.34426719389983695, "calibration/coverage@0%": 0.0, "calibration/coverage@1%": 0.0, "calibration/coverage@10%": 0.0, "calibration/coverage@15%": 0.0, "calibration/coverage@20%": 0.0, "calibration/coverage@25%": 0.0, "calibration/coverage@30%": 0.0, "calibration/coverage@5%": 0.0, "calibration/ece": 0.5344832495234971, "calibration/mean_confidence": 0.7921679872149265, "calibration/prompt_uniqueness": 0.6020497320516532, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.03876953125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1512.4, "completions/mean_length": 265.57275390625, "completions/mean_terminated_length": 214.37782592773436, "completions/min_length": 2.0, "completions/min_terminated_length": 2.0, "epoch": 0.032, "grad_norm": 0.022178692743182182, "learning_rate": 6.249999999999999e-07, "loss": 0.0767, "num_tokens": 35407611.0, "reward": 0.543212604522705, "reward_std": 0.3879946231842041, "rewards/accuracy_reward": 0.2068359375, "rewards/brier_reward": 0.3759405195713043, "rewards/confidence_uniqueness_reward": 0.5140512704849243, "rewards/format_reward": 0.72099609375, "rewards/frontier_aurc_reward": 0.29520038366317747, "rewards/frontier_coverage_0": 0.29520038366317747, "rewards/frontier_coverage_1": 0.29520038366317747, "rewards/frontier_coverage_10": 0.29520038366317747, "rewards/frontier_coverage_15": 0.29520038366317747, "rewards/frontier_coverage_20": 0.29520038366317747, "rewards/frontier_coverage_25": 0.29520038366317747, "rewards/frontier_coverage_5": 0.29520038366317747, "rewards/frontier_ece_reward": 0.29520038366317747, "rewards/frontier_entropy_batch_reward": -0.6874267339706421, "signal/accuracy_reward/centered_abs_mean": 0.21993408203125, "signal/accuracy_reward/group_bin_occupancy": 0.21015625, "signal/accuracy_reward/group_std_mean": 0.2673592925071716, "signal/accuracy_reward/group_zero_std_frac": 0.31875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.109967041015625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.109967041015625, "signal/advantage_abs_mean": 0.31930952072143554, "signal/advantage_pre_scale_abs_mean": 0.31930952072143554, "signal/advantage_pre_scale_std": 0.3992431938648224, "signal/advantage_std": 0.3992431938648224, "signal/brier_reward/centered_abs_mean": 0.3064981997013092, "signal/brier_reward/group_bin_occupancy": 0.76640625, "signal/brier_reward/group_std_mean": 0.3559072256088257, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.03064982034265995, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.03064982034265995, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.2823248624801636, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.59296875, "signal/confidence_uniqueness_reward/group_std_mean": 0.34024747610092165, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.02823248691856861, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.02823248691856861, "signal/format_reward/centered_abs_mean": 0.376055908203125, "signal/format_reward/group_bin_occupancy": 0.25, "signal/format_reward/group_std_mean": 0.43738778233528136, "signal/format_reward/group_zero_std_frac": 0.0, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.1880279541015625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.1880279541015625, "signal/frontier_aurc_reward/centered_abs_mean": 0.27942715883255004, "signal/frontier_aurc_reward/group_bin_occupancy": 0.666015625, "signal/frontier_aurc_reward/group_std_mean": 0.33482242822647096, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_0/centered_abs_mean": 0.27942715883255004, "signal/frontier_coverage_0/group_bin_occupancy": 0.666015625, "signal/frontier_coverage_0/group_std_mean": 0.33482242822647096, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_1/centered_abs_mean": 0.27942715883255004, "signal/frontier_coverage_1/group_bin_occupancy": 0.666015625, "signal/frontier_coverage_1/group_std_mean": 0.33482242822647096, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_10/centered_abs_mean": 0.27942715883255004, "signal/frontier_coverage_10/group_bin_occupancy": 0.666015625, "signal/frontier_coverage_10/group_std_mean": 0.33482242822647096, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_15/centered_abs_mean": 0.27942715883255004, "signal/frontier_coverage_15/group_bin_occupancy": 0.666015625, "signal/frontier_coverage_15/group_std_mean": 0.33482242822647096, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_20/centered_abs_mean": 0.27942715883255004, "signal/frontier_coverage_20/group_bin_occupancy": 0.666015625, "signal/frontier_coverage_20/group_std_mean": 0.33482242822647096, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_25/centered_abs_mean": 0.27942715883255004, "signal/frontier_coverage_25/group_bin_occupancy": 0.666015625, "signal/frontier_coverage_25/group_std_mean": 0.33482242822647096, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_5/centered_abs_mean": 0.27942715883255004, "signal/frontier_coverage_5/group_bin_occupancy": 0.666015625, "signal/frontier_coverage_5/group_std_mean": 0.33482242822647096, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.00349283954128623, "signal/frontier_ece_reward/centered_abs_mean": 0.27942715883255004, "signal/frontier_ece_reward/group_bin_occupancy": 0.666015625, "signal/frontier_ece_reward/group_std_mean": 0.33482242822647096, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.02794271633028984, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.02794271633028984, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.40591453313827514, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.315234375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.461690616607666, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.04059145227074623, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.04059145227074623, "step": 10 }, { "calibration/aurc": 0.5954983324695844, "calibration/batch_distribution_entropy": 0.6609864397410836, "calibration/batch_entropy_100bins": 0.4879162374507433, "calibration/batch_entropy_10bins": 0.6609864397410836, "calibration/batch_entropy_50bins": 0.5712314387631505, "calibration/batch_uniqueness": 0.7153247013761879, "calibration/buffer_distribution_entropy": 0.6641450194727097, "calibration/buffer_entropy_100bins": 0.49312347248479527, "calibration/buffer_entropy_10bins": 0.6641450194727097, "calibration/buffer_entropy_50bins": 0.5757311325171839, "calibration/confidence_entropy": 0.347321713656939, "calibration/coverage@0%": 0.0, "calibration/coverage@1%": 0.0, "calibration/coverage@10%": 0.0, "calibration/coverage@15%": 0.0, "calibration/coverage@20%": 0.0, "calibration/coverage@25%": 0.0, "calibration/coverage@30%": 0.0, "calibration/coverage@5%": 0.0, "calibration/ece": 0.46304059263839337, "calibration/mean_confidence": 0.7932378053531337, "calibration/prompt_uniqueness": 0.6238357100996199, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.01376953125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1443.2, "completions/mean_length": 196.64814453125, "completions/mean_terminated_length": 178.03191833496095, "completions/min_length": 18.0, "completions/min_terminated_length": 18.0, "epoch": 0.048, "grad_norm": 0.03867306560277939, "learning_rate": 9.374999999999999e-07, "loss": 0.0429, "num_tokens": 52470024.0, "reward": 0.6718583703041077, "reward_std": 0.30501891374588014, "rewards/accuracy_reward": 0.276953125, "rewards/brier_reward": 0.48754699230194093, "rewards/confidence_uniqueness_reward": 0.6436279296875, "rewards/format_reward": 0.88701171875, "rewards/frontier_aurc_reward": 0.30400659823790194, "rewards/frontier_coverage_0": 0.31805233359336854, "rewards/frontier_coverage_1": 0.31805233359336854, "rewards/frontier_coverage_10": 0.31805233359336854, "rewards/frontier_coverage_15": 0.31805233359336854, "rewards/frontier_coverage_20": 0.31805233359336854, "rewards/frontier_coverage_25": 0.31805233359336854, "rewards/frontier_coverage_5": 0.31805233359336854, "rewards/frontier_ece_reward": 0.2908082202076912, "rewards/frontier_entropy_batch_reward": -0.8395203113555908, "signal/accuracy_reward/centered_abs_mean": 0.20394287109375, "signal/accuracy_reward/group_bin_occupancy": 0.206640625, "signal/accuracy_reward/group_std_mean": 0.2528751790523529, "signal/accuracy_reward/group_zero_std_frac": 0.346875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.101971435546875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.101971435546875, "signal/advantage_abs_mean": 0.2302687406539917, "signal/advantage_pre_scale_abs_mean": 0.2302687406539917, "signal/advantage_pre_scale_std": 0.3183952569961548, "signal/advantage_std": 0.3183952569961548, "signal/brier_reward/centered_abs_mean": 0.27347792983055114, "signal/brier_reward/group_bin_occupancy": 0.79609375, "signal/brier_reward/group_std_mean": 0.32902817726135253, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.02734779305756092, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.02734779305756092, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.1997553139925003, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.588671875, "signal/confidence_uniqueness_reward/group_std_mean": 0.2634308010339737, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.019975531846284866, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.019975531846284866, "signal/format_reward/centered_abs_mean": 0.189532470703125, "signal/format_reward/group_bin_occupancy": 0.24296875, "signal/format_reward/group_std_mean": 0.29407756924629214, "signal/format_reward/group_zero_std_frac": 0.05625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0947662353515625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0947662353515625, "signal/frontier_aurc_reward/centered_abs_mean": 0.22067007487639784, "signal/frontier_aurc_reward/group_bin_occupancy": 0.728125, "signal/frontier_aurc_reward/group_std_mean": 0.26466578487306835, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 0.0027583759831031783, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 0.0027583759831031783, "signal/frontier_coverage_0/centered_abs_mean": 0.2383599117398262, "signal/frontier_coverage_0/group_bin_occupancy": 0.7, "signal/frontier_coverage_0/group_std_mean": 0.2930622071027756, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_1/centered_abs_mean": 0.2383599117398262, "signal/frontier_coverage_1/group_bin_occupancy": 0.7, "signal/frontier_coverage_1/group_std_mean": 0.2930622071027756, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_10/centered_abs_mean": 0.2383599117398262, "signal/frontier_coverage_10/group_bin_occupancy": 0.7, "signal/frontier_coverage_10/group_std_mean": 0.2930622071027756, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_15/centered_abs_mean": 0.2383599117398262, "signal/frontier_coverage_15/group_bin_occupancy": 0.7, "signal/frontier_coverage_15/group_std_mean": 0.2930622071027756, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_20/centered_abs_mean": 0.2383599117398262, "signal/frontier_coverage_20/group_bin_occupancy": 0.7, "signal/frontier_coverage_20/group_std_mean": 0.2930622071027756, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_25/centered_abs_mean": 0.2383599117398262, "signal/frontier_coverage_25/group_bin_occupancy": 0.7, "signal/frontier_coverage_25/group_std_mean": 0.2930622071027756, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_5/centered_abs_mean": 0.2383599117398262, "signal/frontier_coverage_5/group_bin_occupancy": 0.7, "signal/frontier_coverage_5/group_std_mean": 0.2930622071027756, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0029794989386573434, "signal/frontier_ece_reward/centered_abs_mean": 0.24426527619361876, "signal/frontier_ece_reward/group_bin_occupancy": 0.6890625, "signal/frontier_ece_reward/group_std_mean": 0.2948257505893707, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.024426528066396714, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.024426528066396714, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.25622371435165403, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.337890625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.36273856163024903, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.025, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.025622372701764106, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.025622372701764106, "step": 15 }, { "calibration/aurc": 0.5358041098498495, "calibration/batch_distribution_entropy": 0.6901650644698092, "calibration/batch_entropy_100bins": 0.520654048094481, "calibration/batch_entropy_10bins": 0.6901650644698092, "calibration/batch_entropy_50bins": 0.6051452224680665, "calibration/batch_uniqueness": 0.763976916589782, "calibration/buffer_distribution_entropy": 0.6606516804343314, "calibration/buffer_entropy_100bins": 0.494632779590478, "calibration/buffer_entropy_10bins": 0.6606516804343314, "calibration/buffer_entropy_50bins": 0.5773754016487056, "calibration/confidence_entropy": 0.36734673435422305, "calibration/coverage@0%": 0.0, "calibration/coverage@1%": 0.0, "calibration/coverage@10%": 0.0, "calibration/coverage@15%": 0.0, "calibration/coverage@20%": 0.0, "calibration/coverage@25%": 0.0, "calibration/coverage@30%": 0.0, "calibration/coverage@5%": 0.0, "calibration/ece": 0.3834483744279488, "calibration/mean_confidence": 0.7857268418923627, "calibration/prompt_uniqueness": 0.6830832680904964, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0041015625, "completions/max_length": 1536.0, "completions/max_terminated_length": 1014.0, "completions/mean_length": 142.58173828125, "completions/mean_terminated_length": 136.86209259033203, "completions/min_length": 31.2, "completions/min_terminated_length": 31.2, "epoch": 0.064, "grad_norm": 0.0037215217016637325, "learning_rate": 1e-06, "loss": 0.0121, "num_tokens": 68848461.0, "reward": 0.7004066944122315, "reward_std": 0.19615896046161652, "rewards/accuracy_reward": 0.33955078125, "rewards/brier_reward": 0.5693915843963623, "rewards/confidence_uniqueness_reward": 0.7526990175247192, "rewards/format_reward": 0.97900390625, "rewards/frontier_aurc_reward": -0.006726992409676313, "rewards/frontier_coverage_0": 0.0625338762998581, "rewards/frontier_coverage_1": 0.0625338762998581, "rewards/frontier_coverage_10": 0.0625338762998581, "rewards/frontier_coverage_15": 0.0625338762998581, "rewards/frontier_coverage_20": 0.0625338762998581, "rewards/frontier_coverage_25": 0.0625338762998581, "rewards/frontier_coverage_5": 0.0625338762998581, "rewards/frontier_ece_reward": -0.05991722922772169, "rewards/frontier_entropy_batch_reward": -0.9047559976577759, "signal/accuracy_reward/centered_abs_mean": 0.191717529296875, "signal/accuracy_reward/group_bin_occupancy": 0.20546875, "signal/accuracy_reward/group_std_mean": 0.24287109673023224, "signal/accuracy_reward/group_zero_std_frac": 0.35625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0958587646484375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0958587646484375, "signal/advantage_abs_mean": 0.14673084616661072, "signal/advantage_pre_scale_abs_mean": 0.14673084616661072, "signal/advantage_pre_scale_std": 0.21250708401203156, "signal/advantage_std": 0.21250708401203156, "signal/brier_reward/centered_abs_mean": 0.2369421899318695, "signal/brier_reward/group_bin_occupancy": 0.83125, "signal/brier_reward/group_std_mean": 0.2928761005401611, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.023694219067692757, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.023694219067692757, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.11683968752622605, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.664453125, "signal/confidence_uniqueness_reward/group_std_mean": 0.15450345873832702, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.01168396882712841, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.01168396882712841, "signal/format_reward/centered_abs_mean": 0.039410400390625, "signal/format_reward/group_bin_occupancy": 0.1796875, "signal/format_reward/group_std_mean": 0.09199286699295044, "signal/format_reward/group_zero_std_frac": 0.5625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0197052001953125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0197052001953125, "signal/frontier_aurc_reward/centered_abs_mean": 0.004591509327292442, "signal/frontier_aurc_reward/group_bin_occupancy": 0.752734375, "signal/frontier_aurc_reward/group_std_mean": 0.0063032075762748715, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 5.7393866154598075e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 5.7393866154598075e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.10542762577533722, "signal/frontier_coverage_0/group_bin_occupancy": 0.6671875, "signal/frontier_coverage_0/group_std_mean": 0.166673481464386, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_1/centered_abs_mean": 0.10542762577533722, "signal/frontier_coverage_1/group_bin_occupancy": 0.6671875, "signal/frontier_coverage_1/group_std_mean": 0.166673481464386, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_10/centered_abs_mean": 0.10542762577533722, "signal/frontier_coverage_10/group_bin_occupancy": 0.6671875, "signal/frontier_coverage_10/group_std_mean": 0.166673481464386, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_15/centered_abs_mean": 0.10542762577533722, "signal/frontier_coverage_15/group_bin_occupancy": 0.6671875, "signal/frontier_coverage_15/group_std_mean": 0.166673481464386, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_20/centered_abs_mean": 0.10542762577533722, "signal/frontier_coverage_20/group_bin_occupancy": 0.6671875, "signal/frontier_coverage_20/group_std_mean": 0.166673481464386, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_25/centered_abs_mean": 0.10542762577533722, "signal/frontier_coverage_25/group_bin_occupancy": 0.6671875, "signal/frontier_coverage_25/group_std_mean": 0.166673481464386, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_5/centered_abs_mean": 0.10542762577533722, "signal/frontier_coverage_5/group_bin_occupancy": 0.6671875, "signal/frontier_coverage_5/group_std_mean": 0.166673481464386, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.001317845331504941, "signal/frontier_ece_reward/centered_abs_mean": 0.12625308483839034, "signal/frontier_ece_reward/group_bin_occupancy": 0.6515625, "signal/frontier_ece_reward/group_std_mean": 0.15720948278903962, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.012625308893620967, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.012625308893620967, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.16561020612716676, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.354296875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.29078150987625123, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.1, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.016561020724475385, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.016561020724475385, "step": 20 }, { "calibration/aurc": 0.6306217252752108, "calibration/batch_distribution_entropy": 0.8167210710691316, "calibration/batch_entropy_100bins": 0.6166961519676035, "calibration/batch_entropy_10bins": 0.8167210710691316, "calibration/batch_entropy_50bins": 0.7054777530492016, "calibration/batch_uniqueness": 0.8460996938764275, "calibration/buffer_distribution_entropy": 0.6889259355418624, "calibration/buffer_entropy_100bins": 0.5185451761064622, "calibration/buffer_entropy_10bins": 0.6889259355418624, "calibration/buffer_entropy_50bins": 0.6027129524123941, "calibration/confidence_entropy": 0.43321236481882996, "calibration/coverage@0%": 0.0, "calibration/coverage@1%": 0.0, "calibration/coverage@10%": 0.0, "calibration/coverage@15%": 0.0, "calibration/coverage@20%": 0.0, "calibration/coverage@25%": 0.0, "calibration/coverage@30%": 0.0, "calibration/coverage@5%": 0.0, "calibration/ece": 0.4093629408561291, "calibration/mean_confidence": 0.71680038028082, "calibration/prompt_uniqueness": 0.7771185837506503, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.001171875, "completions/max_length": 1536.0, "completions/max_terminated_length": 644.0, "completions/mean_length": 116.4271484375, "completions/mean_terminated_length": 114.76178131103515, "completions/min_length": 35.8, "completions/min_terminated_length": 35.8, "epoch": 0.08, "grad_norm": 0.008781388401985168, "learning_rate": 1e-06, "loss": 0.0027, "num_tokens": 84973827.0, "reward": 0.7413225412368775, "reward_std": 0.17491281628608704, "rewards/accuracy_reward": 0.3615234375, "rewards/brier_reward": 0.6184327125549316, "rewards/confidence_uniqueness_reward": 0.8402995109558106, "rewards/format_reward": 0.99404296875, "rewards/frontier_aurc_reward": -0.006034567672759295, "rewards/frontier_coverage_0": 0.06498393714427948, "rewards/frontier_coverage_1": 0.06498393714427948, "rewards/frontier_coverage_10": 0.06498393714427948, "rewards/frontier_coverage_15": 0.06498393714427948, "rewards/frontier_coverage_20": 0.06498393714427948, "rewards/frontier_coverage_25": 0.06498393714427948, "rewards/frontier_coverage_5": 0.06498393714427948, "rewards/frontier_ece_reward": -0.042827145755290986, "rewards/frontier_entropy_batch_reward": -0.8366182565689086, "signal/accuracy_reward/centered_abs_mean": 0.18753662109375, "signal/accuracy_reward/group_bin_occupancy": 0.2, "signal/accuracy_reward/group_std_mean": 0.23230806291103362, "signal/accuracy_reward/group_zero_std_frac": 0.4, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.093768310546875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.093768310546875, "signal/advantage_abs_mean": 0.1363419622182846, "signal/advantage_pre_scale_abs_mean": 0.1363419622182846, "signal/advantage_pre_scale_std": 0.19149267971515654, "signal/advantage_std": 0.19149267971515654, "signal/brier_reward/centered_abs_mean": 0.22923070788383484, "signal/brier_reward/group_bin_occupancy": 0.885546875, "signal/brier_reward/group_std_mean": 0.2801697671413422, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.022923070564866067, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.022923070564866067, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.0668274387717247, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.7296875, "signal/confidence_uniqueness_reward/group_std_mean": 0.09313965737819671, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.006682743970304728, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.006682743970304728, "signal/format_reward/centered_abs_mean": 0.011492919921875, "signal/format_reward/group_bin_occupancy": 0.147265625, "signal/format_reward/group_std_mean": 0.03235282339155674, "signal/format_reward/group_zero_std_frac": 0.821875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0057464599609375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0057464599609375, "signal/frontier_aurc_reward/centered_abs_mean": 0.003977674199268222, "signal/frontier_aurc_reward/group_bin_occupancy": 0.759375, "signal/frontier_aurc_reward/group_std_mean": 0.005785228312015533, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.972092792741023e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.972092792741023e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.13958872258663177, "signal/frontier_coverage_0/group_bin_occupancy": 0.77421875, "signal/frontier_coverage_0/group_std_mean": 0.2060023784637451, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_1/centered_abs_mean": 0.13958872258663177, "signal/frontier_coverage_1/group_bin_occupancy": 0.77421875, "signal/frontier_coverage_1/group_std_mean": 0.2060023784637451, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_10/centered_abs_mean": 0.13958872258663177, "signal/frontier_coverage_10/group_bin_occupancy": 0.77421875, "signal/frontier_coverage_10/group_std_mean": 0.2060023784637451, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_15/centered_abs_mean": 0.13958872258663177, "signal/frontier_coverage_15/group_bin_occupancy": 0.77421875, "signal/frontier_coverage_15/group_std_mean": 0.2060023784637451, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_20/centered_abs_mean": 0.13958872258663177, "signal/frontier_coverage_20/group_bin_occupancy": 0.77421875, "signal/frontier_coverage_20/group_std_mean": 0.2060023784637451, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_25/centered_abs_mean": 0.13958872258663177, "signal/frontier_coverage_25/group_bin_occupancy": 0.77421875, "signal/frontier_coverage_25/group_std_mean": 0.2060023784637451, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_5/centered_abs_mean": 0.13958872258663177, "signal/frontier_coverage_5/group_bin_occupancy": 0.77421875, "signal/frontier_coverage_5/group_std_mean": 0.2060023784637451, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0017448590137064458, "signal/frontier_ece_reward/centered_abs_mean": 0.11729500740766526, "signal/frontier_ece_reward/group_bin_occupancy": 0.756640625, "signal/frontier_ece_reward/group_std_mean": 0.14286334812641144, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.011729500815272331, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.011729500815272331, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2641903102397919, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.4640625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3961125731468201, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.01875, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026419031620025634, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026419031620025634, "step": 25 }, { "calibration/aurc": 0.6427295593516567, "calibration/batch_distribution_entropy": 0.9502781652430391, "calibration/batch_entropy_100bins": 0.8156107349617958, "calibration/batch_entropy_10bins": 0.9502781652430391, "calibration/batch_entropy_50bins": 0.877820955427303, "calibration/batch_uniqueness": 0.9219015127685445, "calibration/buffer_distribution_entropy": 0.7482106162289129, "calibration/buffer_entropy_100bins": 0.5745675783096178, "calibration/buffer_entropy_10bins": 0.7482106162289129, "calibration/buffer_entropy_50bins": 0.6587946726096542, "calibration/confidence_entropy": 0.5133051273632727, "calibration/coverage@0%": 0.0, "calibration/coverage@1%": 0.0, "calibration/coverage@10%": 0.0, "calibration/coverage@15%": 0.0, "calibration/coverage@20%": 0.0, "calibration/coverage@25%": 0.0, "calibration/coverage@30%": 0.0, "calibration/coverage@5%": 0.0, "calibration/ece": 0.28071672970691786, "calibration/mean_confidence": 0.5546752935239564, "calibration/prompt_uniqueness": 0.8627877188656203, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0017578125, "completions/max_length": 1536.0, "completions/max_terminated_length": 608.2, "completions/mean_length": 108.23037109375, "completions/mean_terminated_length": 105.7136215209961, "completions/min_length": 34.8, "completions/min_terminated_length": 34.8, "epoch": 0.096, "grad_norm": 0.007898028008639812, "learning_rate": 1e-06, "loss": 0.0058, "num_tokens": 101126714.0, "reward": 0.7864526510238647, "reward_std": 0.1655167043209076, "rewards/accuracy_reward": 0.35693359375, "rewards/brier_reward": 0.6801853537559509, "rewards/confidence_uniqueness_reward": 0.9192812442779541, "rewards/format_reward": 0.99453125, "rewards/frontier_aurc_reward": -0.005380383133888245, "rewards/frontier_coverage_0": 0.10250453501939774, "rewards/frontier_coverage_1": 0.10250453501939774, "rewards/frontier_coverage_10": 0.10250453501939774, "rewards/frontier_coverage_15": 0.10250453501939774, "rewards/frontier_coverage_20": 0.10250453501939774, "rewards/frontier_coverage_25": 0.10250453501939774, "rewards/frontier_coverage_5": 0.10250453501939774, "rewards/frontier_ece_reward": -0.02438057716935873, "rewards/frontier_entropy_batch_reward": -0.5569027364253998, "signal/accuracy_reward/centered_abs_mean": 0.184271240234375, "signal/accuracy_reward/group_bin_occupancy": 0.205859375, "signal/accuracy_reward/group_std_mean": 0.2364806830883026, "signal/accuracy_reward/group_zero_std_frac": 0.353125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0921356201171875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0921356201171875, "signal/advantage_abs_mean": 0.12783702611923217, "signal/advantage_pre_scale_abs_mean": 0.12783702611923217, "signal/advantage_pre_scale_std": 0.17707869410514832, "signal/advantage_std": 0.17707869410514832, "signal/brier_reward/centered_abs_mean": 0.22551969289779664, "signal/brier_reward/group_bin_occupancy": 0.9265625, "signal/brier_reward/group_std_mean": 0.2754694700241089, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.022551970183849336, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.022551970183849336, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.04714692085981369, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.752734375, "signal/confidence_uniqueness_reward/group_std_mean": 0.07181560322642326, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.004714692197740078, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.004714692197740078, "signal/format_reward/centered_abs_mean": 0.010546875, "signal/format_reward/group_bin_occupancy": 0.1453125, "signal/format_reward/group_std_mean": 0.029590688273310663, "signal/format_reward/group_zero_std_frac": 0.8375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0052734375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0052734375, "signal/frontier_aurc_reward/centered_abs_mean": 0.002689578104764223, "signal/frontier_aurc_reward/group_bin_occupancy": 0.72734375, "signal/frontier_aurc_reward/group_std_mean": 0.004204361233860254, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.361972703714855e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.361972703714855e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.22384141981601716, "signal/frontier_coverage_0/group_bin_occupancy": 0.89453125, "signal/frontier_coverage_0/group_std_mean": 0.29406129121780394, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_1/centered_abs_mean": 0.22384141981601716, "signal/frontier_coverage_1/group_bin_occupancy": 0.89453125, "signal/frontier_coverage_1/group_std_mean": 0.29406129121780394, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_10/centered_abs_mean": 0.22384141981601716, "signal/frontier_coverage_10/group_bin_occupancy": 0.89453125, "signal/frontier_coverage_10/group_std_mean": 0.29406129121780394, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_15/centered_abs_mean": 0.22384141981601716, "signal/frontier_coverage_15/group_bin_occupancy": 0.89453125, "signal/frontier_coverage_15/group_std_mean": 0.29406129121780394, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_20/centered_abs_mean": 0.22384141981601716, "signal/frontier_coverage_20/group_bin_occupancy": 0.89453125, "signal/frontier_coverage_20/group_std_mean": 0.29406129121780394, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_25/centered_abs_mean": 0.22384141981601716, "signal/frontier_coverage_25/group_bin_occupancy": 0.89453125, "signal/frontier_coverage_25/group_std_mean": 0.29406129121780394, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_5/centered_abs_mean": 0.22384141981601716, "signal/frontier_coverage_5/group_bin_occupancy": 0.89453125, "signal/frontier_coverage_5/group_std_mean": 0.29406129121780394, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002798017906025052, "signal/frontier_ece_reward/centered_abs_mean": 0.1024959847331047, "signal/frontier_ece_reward/group_bin_occupancy": 0.789453125, "signal/frontier_ece_reward/group_std_mean": 0.12489820569753647, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.01024959869682789, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.01024959869682789, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.4530464828014374, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.666015625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.5292662143707275, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.04530465006828308, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.04530465006828308, "step": 30 }, { "calibration/aurc": 0.5411829203976338, "calibration/batch_distribution_entropy": 0.9389249423001889, "calibration/batch_entropy_100bins": 0.9339604474255918, "calibration/batch_entropy_10bins": 0.9389249423001889, "calibration/batch_entropy_50bins": 0.9446006244676504, "calibration/batch_uniqueness": 0.9466819379605237, "calibration/buffer_distribution_entropy": 0.8312428685923228, "calibration/buffer_entropy_100bins": 0.6758567957119815, "calibration/buffer_entropy_10bins": 0.8312428685923228, "calibration/buffer_entropy_50bins": 0.7508919505523195, "calibration/confidence_entropy": 0.5050965040305361, "calibration/coverage@0%": 0.0011772410339381332, "calibration/coverage@1%": 0.0011772410339381332, "calibration/coverage@10%": 0.0011772410339381332, "calibration/coverage@15%": 0.0011772410339381332, "calibration/coverage@20%": 0.0011772410339381332, "calibration/coverage@25%": 0.0011772410339381332, "calibration/coverage@30%": 0.0011772410339381332, "calibration/coverage@5%": 0.0011772410339381332, "calibration/ece": 0.19170544807693887, "calibration/mean_confidence": 0.37790683017921484, "calibration/prompt_uniqueness": 0.8894818832921436, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.002734375, "completions/max_length": 1536.0, "completions/max_terminated_length": 734.6, "completions/mean_length": 104.334375, "completions/mean_terminated_length": 100.40899200439453, "completions/min_length": 39.0, "completions/min_terminated_length": 39.0, "epoch": 0.112, "grad_norm": 0.0077709960751235485, "learning_rate": 1e-06, "loss": 0.0096, "num_tokens": 117304570.0, "reward": 0.8275706052780152, "reward_std": 0.14384739100933075, "rewards/accuracy_reward": 0.38564453125, "rewards/brier_reward": 0.7133829474449158, "rewards/confidence_uniqueness_reward": 0.9383355259895325, "rewards/format_reward": 0.99326171875, "rewards/frontier_aurc_reward": -0.0046864720061421394, "rewards/frontier_coverage_0": 0.128535196185112, "rewards/frontier_coverage_1": 0.128535196185112, "rewards/frontier_coverage_10": 0.128535196185112, "rewards/frontier_coverage_15": 0.128535196185112, "rewards/frontier_coverage_20": 0.128535196185112, "rewards/frontier_coverage_25": 0.128535196185112, "rewards/frontier_coverage_5": 0.128535196185112, "rewards/frontier_ece_reward": -0.002958059194497764, "rewards/frontier_entropy_batch_reward": -0.3794680893421173, "signal/accuracy_reward/centered_abs_mean": 0.193231201171875, "signal/accuracy_reward/group_bin_occupancy": 0.20234375, "signal/accuracy_reward/group_std_mean": 0.23948768079280852, "signal/accuracy_reward/group_zero_std_frac": 0.38125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0966156005859375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0966156005859375, "signal/advantage_abs_mean": 0.10972018837928772, "signal/advantage_pre_scale_abs_mean": 0.10972018837928772, "signal/advantage_pre_scale_std": 0.15879117250442504, "signal/advantage_std": 0.15879117250442504, "signal/brier_reward/centered_abs_mean": 0.2097387820482254, "signal/brier_reward/group_bin_occupancy": 0.8890625, "signal/brier_reward/group_std_mean": 0.2628061890602112, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.02097387835383415, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.02097387835383415, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.03145063556730747, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.808203125, "signal/confidence_uniqueness_reward/group_std_mean": 0.05689419433474541, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.00314506352879107, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.00314506352879107, "signal/format_reward/centered_abs_mean": 0.012957763671875, "signal/format_reward/group_bin_occupancy": 0.149609375, "signal/format_reward/group_std_mean": 0.0359811969101429, "signal/format_reward/group_zero_std_frac": 0.803125, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0064788818359375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0064788818359375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0015049998182803392, "signal/frontier_aurc_reward/group_bin_occupancy": 0.70234375, "signal/frontier_aurc_reward/group_std_mean": 0.0025454596150666477, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 1.881249772850424e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 1.881249772850424e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.3109524190425873, "signal/frontier_coverage_0/group_bin_occupancy": 0.940234375, "signal/frontier_coverage_0/group_std_mean": 0.3855010986328125, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_1/centered_abs_mean": 0.3109524190425873, "signal/frontier_coverage_1/group_bin_occupancy": 0.940234375, "signal/frontier_coverage_1/group_std_mean": 0.3855010986328125, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_10/centered_abs_mean": 0.3109524190425873, "signal/frontier_coverage_10/group_bin_occupancy": 0.940234375, "signal/frontier_coverage_10/group_std_mean": 0.3855010986328125, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_15/centered_abs_mean": 0.3109524190425873, "signal/frontier_coverage_15/group_bin_occupancy": 0.940234375, "signal/frontier_coverage_15/group_std_mean": 0.3855010986328125, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_20/centered_abs_mean": 0.3109524190425873, "signal/frontier_coverage_20/group_bin_occupancy": 0.940234375, "signal/frontier_coverage_20/group_std_mean": 0.3855010986328125, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_25/centered_abs_mean": 0.3109524190425873, "signal/frontier_coverage_25/group_bin_occupancy": 0.940234375, "signal/frontier_coverage_25/group_std_mean": 0.3855010986328125, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_5/centered_abs_mean": 0.3109524190425873, "signal/frontier_coverage_5/group_bin_occupancy": 0.940234375, "signal/frontier_coverage_5/group_std_mean": 0.3855010986328125, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.003886905312538147, "signal/frontier_ece_reward/centered_abs_mean": 0.06080967858433724, "signal/frontier_ece_reward/group_bin_occupancy": 0.696484375, "signal/frontier_ece_reward/group_std_mean": 0.0861856073141098, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.006080967932939529, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.006080967932939529, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.42676566243171693, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.77265625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.4873059630393982, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.042676566541194914, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.042676566541194914, "step": 35 }, { "calibration/aurc": 0.5621712244329501, "calibration/batch_distribution_entropy": 0.9156689725109626, "calibration/batch_entropy_100bins": 0.9339182959471882, "calibration/batch_entropy_10bins": 0.9156689725109626, "calibration/batch_entropy_50bins": 0.939358189213392, "calibration/batch_uniqueness": 0.9436545189473259, "calibration/buffer_distribution_entropy": 0.8943865499464592, "calibration/buffer_entropy_100bins": 0.7588073879203389, "calibration/buffer_entropy_10bins": 0.8943865499464592, "calibration/buffer_entropy_50bins": 0.8228671782101946, "calibration/confidence_entropy": 0.5041235058768451, "calibration/coverage@0%": 0.0015625, "calibration/coverage@1%": 0.0015625, "calibration/coverage@10%": 0.0015625, "calibration/coverage@15%": 0.003125, "calibration/coverage@20%": 0.006254586594911937, "calibration/coverage@25%": 0.006254586594911937, "calibration/coverage@30%": 0.007817086594911937, "calibration/coverage@5%": 0.0015625, "calibration/ece": 0.17390477690514514, "calibration/mean_confidence": 0.34076602887171203, "calibration/prompt_uniqueness": 0.8836721244860133, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00166015625, "completions/max_length": 1536.0, "completions/max_terminated_length": 655.8, "completions/mean_length": 102.474609375, "completions/mean_terminated_length": 100.09086303710937, "completions/min_length": 39.0, "completions/min_terminated_length": 39.0, "epoch": 0.128, "grad_norm": 0.0018654018640518188, "learning_rate": 1e-06, "loss": 0.0065, "num_tokens": 133270582.0, "reward": 0.830666720867157, "reward_std": 0.124204021692276, "rewards/accuracy_reward": 0.38642578125, "rewards/brier_reward": 0.7227137207984924, "rewards/confidence_uniqueness_reward": 0.9403947472572327, "rewards/format_reward": 0.99697265625, "rewards/frontier_aurc_reward": -0.004450871236622333, "rewards/frontier_coverage_0": 0.13474646359682083, "rewards/frontier_coverage_1": 0.13474646359682083, "rewards/frontier_coverage_10": 0.13474646359682083, "rewards/frontier_coverage_15": 0.13474646359682083, "rewards/frontier_coverage_20": 0.13474646359682083, "rewards/frontier_coverage_25": 0.13474646359682083, "rewards/frontier_coverage_5": 0.13474646359682083, "rewards/frontier_ece_reward": 0.0017460158444009722, "rewards/frontier_entropy_batch_reward": -0.39252622723579406, "signal/accuracy_reward/centered_abs_mean": 0.164849853515625, "signal/accuracy_reward/group_bin_occupancy": 0.194140625, "signal/accuracy_reward/group_std_mean": 0.20658794641494752, "signal/accuracy_reward/group_zero_std_frac": 0.446875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0824249267578125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0824249267578125, "signal/advantage_abs_mean": 0.09753365814685822, "signal/advantage_pre_scale_abs_mean": 0.09753365814685822, "signal/advantage_pre_scale_std": 0.13996124565601348, "signal/advantage_std": 0.13996124565601348, "signal/brier_reward/centered_abs_mean": 0.19820670485496522, "signal/brier_reward/group_bin_occupancy": 0.892578125, "signal/brier_reward/group_std_mean": 0.24779627323150635, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.019820670410990716, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.019820670410990716, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.02475261837244034, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.91328125, "signal/confidence_uniqueness_reward/group_std_mean": 0.03843157291412354, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.002475261874496937, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.002475261874496937, "signal/format_reward/centered_abs_mean": 0.005792236328125, "signal/format_reward/group_bin_occupancy": 0.1359375, "signal/format_reward/group_std_mean": 0.015965567296370865, "signal/format_reward/group_zero_std_frac": 0.9125, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0028961181640625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0028961181640625, "signal/frontier_aurc_reward/centered_abs_mean": 0.0013649149099364878, "signal/frontier_aurc_reward/group_bin_occupancy": 0.763671875, "signal/frontier_aurc_reward/group_std_mean": 0.002124586421996355, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 1.7061437392840163e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 1.7061437392840163e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.3053977251052856, "signal/frontier_coverage_0/group_bin_occupancy": 0.951171875, "signal/frontier_coverage_0/group_std_mean": 0.3747739315032959, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_1/centered_abs_mean": 0.3053977251052856, "signal/frontier_coverage_1/group_bin_occupancy": 0.951171875, "signal/frontier_coverage_1/group_std_mean": 0.3747739315032959, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_10/centered_abs_mean": 0.3053977251052856, "signal/frontier_coverage_10/group_bin_occupancy": 0.951171875, "signal/frontier_coverage_10/group_std_mean": 0.3747739315032959, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_15/centered_abs_mean": 0.3053977251052856, "signal/frontier_coverage_15/group_bin_occupancy": 0.951171875, "signal/frontier_coverage_15/group_std_mean": 0.3747739315032959, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_20/centered_abs_mean": 0.3053977251052856, "signal/frontier_coverage_20/group_bin_occupancy": 0.951171875, "signal/frontier_coverage_20/group_std_mean": 0.3747739315032959, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_25/centered_abs_mean": 0.3053977251052856, "signal/frontier_coverage_25/group_bin_occupancy": 0.951171875, "signal/frontier_coverage_25/group_std_mean": 0.3747739315032959, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_5/centered_abs_mean": 0.3053977251052856, "signal/frontier_coverage_5/group_bin_occupancy": 0.951171875, "signal/frontier_coverage_5/group_std_mean": 0.3747739315032959, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0038174715358763934, "signal/frontier_ece_reward/centered_abs_mean": 0.054702017456293106, "signal/frontier_ece_reward/group_bin_occupancy": 0.72421875, "signal/frontier_ece_reward/group_std_mean": 0.07811730355024338, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00547020211815834, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00547020211815834, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.4245151877403259, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.77734375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.485267174243927, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.04245151951909065, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.04245151951909065, "step": 40 }, { "calibration/aurc": 0.4208307215680261, "calibration/batch_distribution_entropy": 0.9829187590716526, "calibration/batch_entropy_100bins": 0.9637267024150405, "calibration/batch_entropy_10bins": 0.9829187590716526, "calibration/batch_entropy_50bins": 0.9739330766176637, "calibration/batch_uniqueness": 0.9555517787087421, "calibration/buffer_distribution_entropy": 0.9264816083582318, "calibration/buffer_entropy_100bins": 0.8119042075651748, "calibration/buffer_entropy_10bins": 0.9264816083582318, "calibration/buffer_entropy_50bins": 0.8658613481846114, "calibration/confidence_entropy": 0.5302018213292158, "calibration/coverage@0%": 0.006256115459882583, "calibration/coverage@1%": 0.006256115459882583, "calibration/coverage@10%": 0.006256115459882583, "calibration/coverage@15%": 0.010943615459882584, "calibration/coverage@20%": 0.10391848091976516, "calibration/coverage@25%": 0.19376835837074555, "calibration/coverage@30%": 0.21174935727715746, "calibration/coverage@5%": 0.006256115459882583, "calibration/ece": 0.24131579933067834, "calibration/mean_confidence": 0.482310625258885, "calibration/prompt_uniqueness": 0.8989570800273153, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0009765625, "completions/max_length": 1516.6, "completions/max_terminated_length": 786.8, "completions/mean_length": 99.28193359375, "completions/mean_terminated_length": 97.87714538574218, "completions/min_length": 40.2, "completions/min_terminated_length": 40.2, "epoch": 0.144, "grad_norm": 0.0031734241638332605, "learning_rate": 1e-06, "loss": 0.0046, "num_tokens": 149237661.0, "reward": 0.8918002605438232, "reward_std": 0.13444311916828156, "rewards/accuracy_reward": 0.49228515625, "rewards/brier_reward": 0.7026604413986206, "rewards/confidence_uniqueness_reward": 0.953247320652008, "rewards/format_reward": 0.99833984375, "rewards/frontier_aurc_reward": -0.004078696388751269, "rewards/frontier_coverage_0": 0.027176900766789913, "rewards/frontier_coverage_1": 0.027176900766789913, "rewards/frontier_coverage_10": 0.027176900766789913, "rewards/frontier_coverage_15": 0.027176900766789913, "rewards/frontier_coverage_20": 0.027176900766789913, "rewards/frontier_coverage_25": 0.027176900766789913, "rewards/frontier_coverage_5": 0.027176900766789913, "rewards/frontier_ece_reward": 0.007179011206608266, "rewards/frontier_entropy_batch_reward": -0.22147920429706575, "signal/accuracy_reward/centered_abs_mean": 0.162103271484375, "signal/accuracy_reward/group_bin_occupancy": 0.20234375, "signal/accuracy_reward/group_std_mean": 0.21457574367523194, "signal/accuracy_reward/group_zero_std_frac": 0.38125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0810516357421875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0810516357421875, "signal/advantage_abs_mean": 0.10489514172077179, "signal/advantage_pre_scale_abs_mean": 0.10489514172077179, "signal/advantage_pre_scale_std": 0.14698890447616578, "signal/advantage_std": 0.14698890447616578, "signal/brier_reward/centered_abs_mean": 0.21170293986797334, "signal/brier_reward/group_bin_occupancy": 0.9359375, "signal/brier_reward/group_std_mean": 0.2590953230857849, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.021170295029878616, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.021170295029878616, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.016459671035408973, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.937890625, "signal/confidence_uniqueness_reward/group_std_mean": 0.025219134986400604, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0016459671780467034, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0016459671780467034, "signal/format_reward/centered_abs_mean": 0.003204345703125, "signal/format_reward/group_bin_occupancy": 0.13125, "signal/format_reward/group_std_mean": 0.0090549532789737, "signal/format_reward/group_zero_std_frac": 0.95, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0016021728515625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0016021728515625, "signal/frontier_aurc_reward/centered_abs_mean": 0.002039462886750698, "signal/frontier_aurc_reward/group_bin_occupancy": 0.793359375, "signal/frontier_aurc_reward/group_std_mean": 0.0029616751708090305, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.5493287830613552e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.5493287830613552e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.2594360411167145, "signal/frontier_coverage_0/group_bin_occupancy": 0.935546875, "signal/frontier_coverage_0/group_std_mean": 0.3276914834976196, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_1/centered_abs_mean": 0.2594360411167145, "signal/frontier_coverage_1/group_bin_occupancy": 0.935546875, "signal/frontier_coverage_1/group_std_mean": 0.3276914834976196, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_10/centered_abs_mean": 0.2594360411167145, "signal/frontier_coverage_10/group_bin_occupancy": 0.935546875, "signal/frontier_coverage_10/group_std_mean": 0.3276914834976196, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_15/centered_abs_mean": 0.2594360411167145, "signal/frontier_coverage_15/group_bin_occupancy": 0.935546875, "signal/frontier_coverage_15/group_std_mean": 0.3276914834976196, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_20/centered_abs_mean": 0.2594360411167145, "signal/frontier_coverage_20/group_bin_occupancy": 0.935546875, "signal/frontier_coverage_20/group_std_mean": 0.3276914834976196, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_25/centered_abs_mean": 0.2594360411167145, "signal/frontier_coverage_25/group_bin_occupancy": 0.935546875, "signal/frontier_coverage_25/group_std_mean": 0.3276914834976196, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_5/centered_abs_mean": 0.2594360411167145, "signal/frontier_coverage_5/group_bin_occupancy": 0.935546875, "signal/frontier_coverage_5/group_std_mean": 0.3276914834976196, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0032429506070911886, "signal/frontier_ece_reward/centered_abs_mean": 0.07746845185756683, "signal/frontier_ece_reward/group_bin_occupancy": 0.82578125, "signal/frontier_ece_reward/group_std_mean": 0.100053970515728, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00774684539064765, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00774684539064765, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.31556748151779174, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.762890625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.39420434832572937, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.03155674859881401, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.03155674859881401, "step": 45 }, { "calibration/aurc": 0.5193843102498966, "calibration/batch_distribution_entropy": 0.9845463686201054, "calibration/batch_entropy_100bins": 0.9659996922551809, "calibration/batch_entropy_10bins": 0.9845463686201054, "calibration/batch_entropy_50bins": 0.9780120236968278, "calibration/batch_uniqueness": 0.9582082870463354, "calibration/buffer_distribution_entropy": 0.9413099677688151, "calibration/buffer_entropy_100bins": 0.8490924077319223, "calibration/buffer_entropy_10bins": 0.9413099677688151, "calibration/buffer_entropy_50bins": 0.8940998957854502, "calibration/confidence_entropy": 0.519955771932801, "calibration/coverage@0%": 0.0015625, "calibration/coverage@1%": 0.0015625, "calibration/coverage@10%": 0.0015625, "calibration/coverage@15%": 0.0015625, "calibration/coverage@20%": 0.0015625, "calibration/coverage@25%": 0.0015625, "calibration/coverage@30%": 0.004296875, "calibration/coverage@5%": 0.0015625, "calibration/ece": 0.21486626669589093, "calibration/mean_confidence": 0.5483419950280221, "calibration/prompt_uniqueness": 0.8982646453889178, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0005859375, "completions/max_length": 1105.4, "completions/max_terminated_length": 637.8, "completions/mean_length": 95.578125, "completions/mean_terminated_length": 94.7332550048828, "completions/min_length": 37.0, "completions/min_terminated_length": 37.0, "epoch": 0.16, "grad_norm": 0.005004190374165773, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 165237309.0, "reward": 0.8668585777282715, "reward_std": 0.14406978189945222, "rewards/accuracy_reward": 0.43408203125, "rewards/brier_reward": 0.6894308924674988, "rewards/confidence_uniqueness_reward": 0.9561283230781555, "rewards/format_reward": 0.9982421875, "rewards/frontier_aurc_reward": -0.0047391817905008795, "rewards/frontier_coverage_0": 0.05797448009252548, "rewards/frontier_coverage_1": 0.05797448009252548, "rewards/frontier_coverage_10": 0.05797448009252548, "rewards/frontier_coverage_15": 0.05797448009252548, "rewards/frontier_coverage_20": 0.05797448009252548, "rewards/frontier_coverage_25": 0.05797448009252548, "rewards/frontier_coverage_5": 0.05797448009252548, "rewards/frontier_ece_reward": -0.002871785045135766, "rewards/frontier_entropy_batch_reward": -0.1858581006526947, "signal/accuracy_reward/centered_abs_mean": 0.165142822265625, "signal/accuracy_reward/group_bin_occupancy": 0.19765625, "signal/accuracy_reward/group_std_mean": 0.21089341938495637, "signal/accuracy_reward/group_zero_std_frac": 0.41875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0825714111328125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0825714111328125, "signal/advantage_abs_mean": 0.1148581862449646, "signal/advantage_pre_scale_abs_mean": 0.1148581862449646, "signal/advantage_pre_scale_std": 0.1592826873064041, "signal/advantage_std": 0.1592826873064041, "signal/brier_reward/centered_abs_mean": 0.22412009835243224, "signal/brier_reward/group_bin_occupancy": 0.9421875, "signal/brier_reward/group_std_mean": 0.27176443338394163, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.02241201028227806, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.02241201028227806, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.014607389643788337, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.921484375, "signal/confidence_uniqueness_reward/group_std_mean": 0.02377874292433262, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0014607389690354466, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0014607389690354466, "signal/format_reward/centered_abs_mean": 0.00340576171875, "signal/format_reward/group_bin_occupancy": 0.13203125, "signal/format_reward/group_std_mean": 0.009943688800558449, "signal/format_reward/group_zero_std_frac": 0.94375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.001702880859375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.001702880859375, "signal/frontier_aurc_reward/centered_abs_mean": 0.00291399909183383, "signal/frontier_aurc_reward/group_bin_occupancy": 0.811328125, "signal/frontier_aurc_reward/group_std_mean": 0.004128833580762148, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.6424990685191e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.6424990685191e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.22706362307071687, "signal/frontier_coverage_0/group_bin_occupancy": 0.918359375, "signal/frontier_coverage_0/group_std_mean": 0.2930738806724548, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_1/centered_abs_mean": 0.22706362307071687, "signal/frontier_coverage_1/group_bin_occupancy": 0.918359375, "signal/frontier_coverage_1/group_std_mean": 0.2930738806724548, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_10/centered_abs_mean": 0.22706362307071687, "signal/frontier_coverage_10/group_bin_occupancy": 0.918359375, "signal/frontier_coverage_10/group_std_mean": 0.2930738806724548, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_15/centered_abs_mean": 0.22706362307071687, "signal/frontier_coverage_15/group_bin_occupancy": 0.918359375, "signal/frontier_coverage_15/group_std_mean": 0.2930738806724548, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_20/centered_abs_mean": 0.22706362307071687, "signal/frontier_coverage_20/group_bin_occupancy": 0.918359375, "signal/frontier_coverage_20/group_std_mean": 0.2930738806724548, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_25/centered_abs_mean": 0.22706362307071687, "signal/frontier_coverage_25/group_bin_occupancy": 0.918359375, "signal/frontier_coverage_25/group_std_mean": 0.2930738806724548, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_5/centered_abs_mean": 0.22706362307071687, "signal/frontier_coverage_5/group_bin_occupancy": 0.918359375, "signal/frontier_coverage_5/group_std_mean": 0.2930738806724548, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0028382953256368636, "signal/frontier_ece_reward/centered_abs_mean": 0.08994001597166061, "signal/frontier_ece_reward/group_bin_occupancy": 0.861328125, "signal/frontier_ece_reward/group_std_mean": 0.11356599926948548, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.008994001522660255, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.008994001522660255, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2729247510433197, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.762890625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.35239235162734983, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027292475476861, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027292475476861, "step": 50 }, { "epoch": 0.16, "eval_calibration/aurc": 0.621319442199115, "eval_calibration/batch_distribution_entropy": 0.9132544026317804, "eval_calibration/batch_entropy_100bins": 0.6958206487495924, "eval_calibration/batch_entropy_10bins": 0.9132544026317804, "eval_calibration/batch_entropy_50bins": 0.7861990777075228, "eval_calibration/batch_uniqueness": 0.90234375, "eval_calibration/buffer_distribution_entropy": 0.9470907389146515, "eval_calibration/buffer_entropy_100bins": 0.8662498305989694, "eval_calibration/buffer_entropy_10bins": 0.9470907389146515, "eval_calibration/buffer_entropy_50bins": 0.9065459909621784, "eval_calibration/confidence_entropy": 0.4808260142016664, "eval_calibration/coverage@0%": 0.0078125, "eval_calibration/coverage@1%": 0.0078125, "eval_calibration/coverage@10%": 0.0078125, "eval_calibration/coverage@15%": 0.0078125, "eval_calibration/coverage@20%": 0.0859375, "eval_calibration/coverage@25%": 0.0859375, "eval_calibration/coverage@30%": 0.09375, "eval_calibration/coverage@5%": 0.0078125, "eval_calibration/ece": 0.3920049062265072, "eval_calibration/mean_confidence": 0.5687587711772344, "eval_calibration/prompt_uniqueness": 0.90234375, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 262.75, "eval_completions/max_terminated_length": 262.75, "eval_completions/mean_length": 96.15510559082031, "eval_completions/mean_terminated_length": 96.15510559082031, "eval_completions/min_length": 47.0, "eval_completions/min_terminated_length": 47.0, "eval_loss": 0.0, "eval_num_tokens": 165237309.0, "eval_reward": 0.7550473213195801, "eval_reward_std": 0.23834459483623505, "eval_rewards/accuracy_reward": 0.380859375, "eval_rewards/brier_reward": 0.6821498572826385, "eval_rewards/confidence_uniqueness_reward": 0.898681640625, "eval_rewards/format_reward": 1.0, "eval_rewards/frontier_aurc_reward": -0.0051134476670995355, "eval_rewards/frontier_coverage_0": 0.08652125298976898, "eval_rewards/frontier_coverage_1": 0.08652125298976898, "eval_rewards/frontier_coverage_10": 0.08652125298976898, "eval_rewards/frontier_coverage_15": 0.08652125298976898, "eval_rewards/frontier_coverage_20": 0.08652125298976898, "eval_rewards/frontier_coverage_25": 0.08652125298976898, "eval_rewards/frontier_coverage_5": 0.08652125298976898, "eval_rewards/frontier_ece_reward": -0.009721951093524694, "eval_rewards/frontier_entropy_batch_reward": -1.0, "eval_runtime": 15.7167, "eval_samples_per_second": 31.813, "eval_signal/accuracy_reward/centered_abs_mean": 0.4539794921875, "eval_signal/accuracy_reward/group_bin_occupancy": 0.25, "eval_signal/accuracy_reward/group_std_mean": 0.48320043087005615, "eval_signal/accuracy_reward/group_zero_std_frac": 0.0, "eval_signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.22698974609375, "eval_signal/accuracy_reward/weight": 0.5, "eval_signal/accuracy_reward/weighted_centered_abs_mean": 0.22698974609375, "eval_signal/advantage_abs_mean": 0.2166653871536255, "eval_signal/advantage_pre_scale_abs_mean": 0.2166653871536255, "eval_signal/advantage_pre_scale_std": 0.2358204498887062, "eval_signal/advantage_std": 0.2358204498887062, "eval_signal/brier_reward/centered_abs_mean": 0.23992039263248444, "eval_signal/brier_reward/group_bin_occupancy": 0.9609375, "eval_signal/brier_reward/group_std_mean": 0.2882373705506325, "eval_signal/brier_reward/group_zero_std_frac": 0.0, "eval_signal/brier_reward/scaled_weighted_centered_abs_mean": 0.023992038797587156, "eval_signal/brier_reward/weight": 0.10000000149011612, "eval_signal/brier_reward/weighted_centered_abs_mean": 0.023992038797587156, "eval_signal/confidence_uniqueness_reward/centered_abs_mean": 0.04217529296875, "eval_signal/confidence_uniqueness_reward/group_bin_occupancy": 0.3984375, "eval_signal/confidence_uniqueness_reward/group_std_mean": 0.0487216105684638, "eval_signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "eval_signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.004217529320158064, "eval_signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "eval_signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.004217529320158064, "eval_signal/format_reward/centered_abs_mean": 0.0, "eval_signal/format_reward/group_bin_occupancy": 0.125, "eval_signal/format_reward/group_std_mean": 0.0, "eval_signal/format_reward/group_zero_std_frac": 1.0, "eval_signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/format_reward/weight": 0.5, "eval_signal/format_reward/weighted_centered_abs_mean": 0.0, "eval_signal/frontier_aurc_reward/centered_abs_mean": 0.003932049381546676, "eval_signal/frontier_aurc_reward/group_bin_occupancy": 0.9140625, "eval_signal/frontier_aurc_reward/group_std_mean": 0.005480810417793691, "eval_signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.915061981591862e-05, "eval_signal/frontier_aurc_reward/weight": 0.012500000186264515, "eval_signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.915061981591862e-05, "eval_signal/frontier_coverage_0/centered_abs_mean": 0.27994445711374283, "eval_signal/frontier_coverage_0/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_0/group_std_mean": 0.3757804185152054, "eval_signal/frontier_coverage_0/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_0/weight": 0.012500000186264515, "eval_signal/frontier_coverage_0/weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_1/centered_abs_mean": 0.27994445711374283, "eval_signal/frontier_coverage_1/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_1/group_std_mean": 0.3757804185152054, "eval_signal/frontier_coverage_1/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_1/weight": 0.012500000186264515, "eval_signal/frontier_coverage_1/weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_10/centered_abs_mean": 0.27994445711374283, "eval_signal/frontier_coverage_10/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_10/group_std_mean": 0.3757804185152054, "eval_signal/frontier_coverage_10/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_10/weight": 0.012500000186264515, "eval_signal/frontier_coverage_10/weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_15/centered_abs_mean": 0.27994445711374283, "eval_signal/frontier_coverage_15/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_15/group_std_mean": 0.3757804185152054, "eval_signal/frontier_coverage_15/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_15/weight": 0.012500000186264515, "eval_signal/frontier_coverage_15/weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_20/centered_abs_mean": 0.27994445711374283, "eval_signal/frontier_coverage_20/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_20/group_std_mean": 0.3757804185152054, "eval_signal/frontier_coverage_20/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_20/weight": 0.012500000186264515, "eval_signal/frontier_coverage_20/weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_25/centered_abs_mean": 0.27994445711374283, "eval_signal/frontier_coverage_25/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_25/group_std_mean": 0.3757804185152054, "eval_signal/frontier_coverage_25/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_25/weight": 0.012500000186264515, "eval_signal/frontier_coverage_25/weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_5/centered_abs_mean": 0.27994445711374283, "eval_signal/frontier_coverage_5/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_5/group_std_mean": 0.3757804185152054, "eval_signal/frontier_coverage_5/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_coverage_5/weight": 0.012500000186264515, "eval_signal/frontier_coverage_5/weighted_centered_abs_mean": 0.003499305807054043, "eval_signal/frontier_ece_reward/centered_abs_mean": 0.09657513163983822, "eval_signal/frontier_ece_reward/group_bin_occupancy": 0.9140625, "eval_signal/frontier_ece_reward/group_std_mean": 0.13151323422789574, "eval_signal/frontier_ece_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.009657513117417693, "eval_signal/frontier_ece_reward/weight": 0.10000000149011612, "eval_signal/frontier_ece_reward/weighted_centered_abs_mean": 0.009657513117417693, "eval_signal/frontier_entropy_batch_reward/centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.125, "eval_signal/frontier_entropy_batch_reward/group_std_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_zero_std_frac": 1.0, "eval_signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "eval_signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.0, "eval_steps_per_second": 0.255, "step": 50 }, { "calibration/aurc": 0.4484659090115003, "calibration/batch_distribution_entropy": 0.9940617067812163, "calibration/batch_entropy_100bins": 0.9714517989941335, "calibration/batch_entropy_10bins": 0.9940617067812163, "calibration/batch_entropy_50bins": 0.9825196031249748, "calibration/batch_uniqueness": 0.9587009277049233, "calibration/buffer_distribution_entropy": 0.9508062746412843, "calibration/buffer_entropy_100bins": 0.8756113098904116, "calibration/buffer_entropy_10bins": 0.9508062746412843, "calibration/buffer_entropy_50bins": 0.9132868440661269, "calibration/confidence_entropy": 0.5075993835619301, "calibration/coverage@0%": 0.0011726394324853228, "calibration/coverage@1%": 0.0011726394324853228, "calibration/coverage@10%": 0.0011726394324853228, "calibration/coverage@15%": 0.0011726394324853228, "calibration/coverage@20%": 0.010938264432485323, "calibration/coverage@25%": 0.015235139432485324, "calibration/coverage@30%": 0.11489267367906067, "calibration/coverage@5%": 0.0011726394324853228, "calibration/ece": 0.19384218728007083, "calibration/mean_confidence": 0.5123358770717605, "calibration/prompt_uniqueness": 0.8979440361602498, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.001171875, "completions/max_length": 1536.0, "completions/max_terminated_length": 757.6, "completions/mean_length": 102.0427734375, "completions/mean_terminated_length": 100.36075744628906, "completions/min_length": 41.2, "completions/min_terminated_length": 41.2, "epoch": 0.176, "grad_norm": 0.0035601698327809572, "learning_rate": 1e-06, "loss": 0.0047, "num_tokens": 181519347.0, "reward": 0.8793681740760804, "reward_std": 0.13194239139556885, "rewards/accuracy_reward": 0.444921875, "rewards/brier_reward": 0.704438054561615, "rewards/confidence_uniqueness_reward": 0.9573986887931824, "rewards/format_reward": 0.9986328125, "rewards/frontier_aurc_reward": -0.004294648673385382, "rewards/frontier_coverage_0": 0.07248619496822357, "rewards/frontier_coverage_1": 0.07248619496822357, "rewards/frontier_coverage_10": 0.07248619496822357, "rewards/frontier_coverage_15": 0.07248619496822357, "rewards/frontier_coverage_20": 0.07248619496822357, "rewards/frontier_coverage_25": 0.07248619496822357, "rewards/frontier_coverage_5": 0.07248619496822357, "rewards/frontier_ece_reward": 0.005045533180236816, "rewards/frontier_entropy_batch_reward": -0.15386252403259276, "signal/accuracy_reward/centered_abs_mean": 0.1537109375, "signal/accuracy_reward/group_bin_occupancy": 0.190234375, "signal/accuracy_reward/group_std_mean": 0.19389514327049256, "signal/accuracy_reward/group_zero_std_frac": 0.478125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.07685546875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.07685546875, "signal/advantage_abs_mean": 0.10513403117656708, "signal/advantage_pre_scale_abs_mean": 0.10513403117656708, "signal/advantage_pre_scale_std": 0.14737688302993773, "signal/advantage_std": 0.14737688302993773, "signal/brier_reward/centered_abs_mean": 0.2251437395811081, "signal/brier_reward/group_bin_occupancy": 0.935546875, "signal/brier_reward/group_std_mean": 0.27382948994636536, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.022514374181628227, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.022514374181628227, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.013177918456494808, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.923046875, "signal/confidence_uniqueness_reward/group_std_mean": 0.020791875198483467, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001317791873589158, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001317791873589158, "signal/format_reward/centered_abs_mean": 0.00264892578125, "signal/format_reward/group_bin_occupancy": 0.13046875, "signal/format_reward/group_std_mean": 0.0077339802403002976, "signal/format_reward/group_zero_std_frac": 0.95625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.001324462890625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.001324462890625, "signal/frontier_aurc_reward/centered_abs_mean": 0.0026623731013387443, "signal/frontier_aurc_reward/group_bin_occupancy": 0.807421875, "signal/frontier_aurc_reward/group_std_mean": 0.0038204194977879526, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.3279662602581084e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.3279662602581084e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.25665857791900637, "signal/frontier_coverage_0/group_bin_occupancy": 0.93125, "signal/frontier_coverage_0/group_std_mean": 0.32424872517585757, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_1/centered_abs_mean": 0.25665857791900637, "signal/frontier_coverage_1/group_bin_occupancy": 0.93125, "signal/frontier_coverage_1/group_std_mean": 0.32424872517585757, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_10/centered_abs_mean": 0.25665857791900637, "signal/frontier_coverage_10/group_bin_occupancy": 0.93125, "signal/frontier_coverage_10/group_std_mean": 0.32424872517585757, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_15/centered_abs_mean": 0.25665857791900637, "signal/frontier_coverage_15/group_bin_occupancy": 0.93125, "signal/frontier_coverage_15/group_std_mean": 0.32424872517585757, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_20/centered_abs_mean": 0.25665857791900637, "signal/frontier_coverage_20/group_bin_occupancy": 0.93125, "signal/frontier_coverage_20/group_std_mean": 0.32424872517585757, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_25/centered_abs_mean": 0.25665857791900637, "signal/frontier_coverage_25/group_bin_occupancy": 0.93125, "signal/frontier_coverage_25/group_std_mean": 0.32424872517585757, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_5/centered_abs_mean": 0.25665857791900637, "signal/frontier_coverage_5/group_bin_occupancy": 0.93125, "signal/frontier_coverage_5/group_std_mean": 0.32424872517585757, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0032082322984933855, "signal/frontier_ece_reward/centered_abs_mean": 0.08179719746112823, "signal/frontier_ece_reward/group_bin_occupancy": 0.839453125, "signal/frontier_ece_reward/group_std_mean": 0.1045481413602829, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.008179719932377339, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.008179719932377339, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.24001478850841523, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.750390625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.322251570224762, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.024001479148864746, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.024001479148864746, "step": 55 }, { "calibration/aurc": 0.41783120815670893, "calibration/batch_distribution_entropy": 0.9869078466858362, "calibration/batch_entropy_100bins": 0.9711698902482093, "calibration/batch_entropy_10bins": 0.9869078466858362, "calibration/batch_entropy_50bins": 0.9817291450072908, "calibration/batch_uniqueness": 0.9577475658033132, "calibration/buffer_distribution_entropy": 0.9596721764788925, "calibration/buffer_entropy_100bins": 0.8948005626820563, "calibration/buffer_entropy_10bins": 0.9596721764788925, "calibration/buffer_entropy_50bins": 0.9275950682583417, "calibration/confidence_entropy": 0.4884909952526025, "calibration/coverage@0%": 0.007421875, "calibration/coverage@1%": 0.007421875, "calibration/coverage@10%": 0.010546875, "calibration/coverage@15%": 0.010546875, "calibration/coverage@20%": 0.032438692514677105, "calibration/coverage@25%": 0.042611760029354205, "calibration/coverage@30%": 0.09693691903131116, "calibration/coverage@5%": 0.007421875, "calibration/ece": 0.18198722595987377, "calibration/mean_confidence": 0.45150705174973893, "calibration/prompt_uniqueness": 0.893008432378382, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00048828125, "completions/max_length": 1314.4, "completions/max_terminated_length": 537.4, "completions/mean_length": 108.8171875, "completions/mean_terminated_length": 108.11964111328125, "completions/min_length": 42.4, "completions/min_terminated_length": 42.4, "epoch": 0.192, "grad_norm": 0.003048258600756526, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 197448451.0, "reward": 0.895499849319458, "reward_std": 0.1226910799741745, "rewards/accuracy_reward": 0.476953125, "rewards/brier_reward": 0.7108715415000916, "rewards/confidence_uniqueness_reward": 0.9559177875518798, "rewards/format_reward": 0.99892578125, "rewards/frontier_aurc_reward": -0.0037818335462361575, "rewards/frontier_coverage_0": 0.06929659023880959, "rewards/frontier_coverage_1": 0.06929659023880959, "rewards/frontier_coverage_10": 0.06929659023880959, "rewards/frontier_coverage_15": 0.06929659023880959, "rewards/frontier_coverage_20": 0.06929659023880959, "rewards/frontier_coverage_25": 0.06929659023880959, "rewards/frontier_coverage_5": 0.06929659023880959, "rewards/frontier_ece_reward": 0.012393449433147907, "rewards/frontier_entropy_batch_reward": -0.16374078691005706, "signal/accuracy_reward/centered_abs_mean": 0.14683837890625, "signal/accuracy_reward/group_bin_occupancy": 0.18984375, "signal/accuracy_reward/group_std_mean": 0.18846068680286407, "signal/accuracy_reward/group_zero_std_frac": 0.48125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.073419189453125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.073419189453125, "signal/advantage_abs_mean": 0.09704132080078125, "signal/advantage_pre_scale_abs_mean": 0.09704132080078125, "signal/advantage_pre_scale_std": 0.1375767856836319, "signal/advantage_std": 0.1375767856836319, "signal/brier_reward/centered_abs_mean": 0.2248259723186493, "signal/brier_reward/group_bin_occupancy": 0.9109375, "signal/brier_reward/group_std_mean": 0.27439039945602417, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.02248259782791138, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.02248259782791138, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.014231042377650738, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9109375, "signal/confidence_uniqueness_reward/group_std_mean": 0.021543417125940323, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0014231042703613638, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0014231042703613638, "signal/format_reward/centered_abs_mean": 0.002081298828125, "signal/format_reward/group_bin_occupancy": 0.129296875, "signal/format_reward/group_std_mean": 0.006076698750257492, "signal/format_reward/group_zero_std_frac": 0.965625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0010406494140625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0010406494140625, "signal/frontier_aurc_reward/centered_abs_mean": 0.002347089909017086, "signal/frontier_aurc_reward/group_bin_occupancy": 0.7875, "signal/frontier_aurc_reward/group_std_mean": 0.0034443451557308437, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.933862342615612e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.933862342615612e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.2860648810863495, "signal/frontier_coverage_0/group_bin_occupancy": 0.927734375, "signal/frontier_coverage_0/group_std_mean": 0.35489784479141234, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_1/centered_abs_mean": 0.2860648810863495, "signal/frontier_coverage_1/group_bin_occupancy": 0.927734375, "signal/frontier_coverage_1/group_std_mean": 0.35489784479141234, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_10/centered_abs_mean": 0.2860648810863495, "signal/frontier_coverage_10/group_bin_occupancy": 0.927734375, "signal/frontier_coverage_10/group_std_mean": 0.35489784479141234, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_15/centered_abs_mean": 0.2860648810863495, "signal/frontier_coverage_15/group_bin_occupancy": 0.927734375, "signal/frontier_coverage_15/group_std_mean": 0.35489784479141234, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_20/centered_abs_mean": 0.2860648810863495, "signal/frontier_coverage_20/group_bin_occupancy": 0.927734375, "signal/frontier_coverage_20/group_std_mean": 0.35489784479141234, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_25/centered_abs_mean": 0.2860648810863495, "signal/frontier_coverage_25/group_bin_occupancy": 0.927734375, "signal/frontier_coverage_25/group_std_mean": 0.35489784479141234, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_5/centered_abs_mean": 0.2860648810863495, "signal/frontier_coverage_5/group_bin_occupancy": 0.927734375, "signal/frontier_coverage_5/group_std_mean": 0.35489784479141234, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.003575811069458723, "signal/frontier_ece_reward/centered_abs_mean": 0.0717797726392746, "signal/frontier_ece_reward/group_bin_occupancy": 0.799609375, "signal/frontier_ece_reward/group_std_mean": 0.09460163861513138, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.007177977636456489, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.007177977636456489, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.25209241509437563, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.750390625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3328286588191986, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02520924247801304, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02520924247801304, "step": 60 }, { "calibration/aurc": 0.3645308740204516, "calibration/batch_distribution_entropy": 0.9879292022069827, "calibration/batch_entropy_100bins": 0.965284477133628, "calibration/batch_entropy_10bins": 0.9879292022069827, "calibration/batch_entropy_50bins": 0.9796218807025528, "calibration/batch_uniqueness": 0.9583140479075298, "calibration/buffer_distribution_entropy": 0.9666525654131892, "calibration/buffer_entropy_100bins": 0.9093552217555393, "calibration/buffer_entropy_10bins": 0.9666525654131892, "calibration/buffer_entropy_50bins": 0.9384331274980686, "calibration/confidence_entropy": 0.47878302776198745, "calibration/coverage@0%": 0.00234375, "calibration/coverage@1%": 0.00234375, "calibration/coverage@10%": 0.0046875, "calibration/coverage@15%": 0.059765625, "calibration/coverage@20%": 0.134765625, "calibration/coverage@25%": 0.387109375, "calibration/coverage@30%": 0.4625, "calibration/coverage@5%": 0.00234375, "calibration/ece": 0.21075495403127378, "calibration/mean_confidence": 0.48336310832626495, "calibration/prompt_uniqueness": 0.891297009137617, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 902.6, "completions/max_terminated_length": 491.6, "completions/mean_length": 120.6796875, "completions/mean_terminated_length": 120.40347442626953, "completions/min_length": 47.2, "completions/min_terminated_length": 47.2, "epoch": 0.208, "grad_norm": 0.0027220940683037043, "learning_rate": 1e-06, "loss": -0.0001, "num_tokens": 213716435.0, "reward": 0.9156344532966614, "reward_std": 0.11813800930976867, "rewards/accuracy_reward": 0.5166015625, "rewards/brier_reward": 0.7230738759040832, "rewards/confidence_uniqueness_reward": 0.9579548478126526, "rewards/format_reward": 0.99951171875, "rewards/frontier_aurc_reward": -0.0034141582436859608, "rewards/frontier_coverage_0": 0.054309183917939666, "rewards/frontier_coverage_1": 0.054309183917939666, "rewards/frontier_coverage_10": 0.054309183917939666, "rewards/frontier_coverage_15": 0.054309183917939666, "rewards/frontier_coverage_20": 0.054309183917939666, "rewards/frontier_coverage_25": 0.054309183917939666, "rewards/frontier_coverage_5": 0.054309183917939666, "rewards/frontier_ece_reward": 0.020394877111539245, "rewards/frontier_entropy_batch_reward": -0.1727394014596939, "signal/accuracy_reward/centered_abs_mean": 0.13463134765625, "signal/accuracy_reward/group_bin_occupancy": 0.1890625, "signal/accuracy_reward/group_std_mean": 0.17806513905525206, "signal/accuracy_reward/group_zero_std_frac": 0.4875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.067315673828125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.067315673828125, "signal/advantage_abs_mean": 0.09260295033454895, "signal/advantage_pre_scale_abs_mean": 0.09260295033454895, "signal/advantage_pre_scale_std": 0.13304563760757446, "signal/advantage_std": 0.13304563760757446, "signal/brier_reward/centered_abs_mean": 0.21528592705726624, "signal/brier_reward/group_bin_occupancy": 0.9125, "signal/brier_reward/group_std_mean": 0.26363363265991213, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.021528592705726622, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.021528592705726622, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012894866801798344, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.95, "signal/confidence_uniqueness_reward/group_std_mean": 0.017705611325800417, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012894866755232214, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012894866755232214, "signal/format_reward/centered_abs_mean": 0.000946044921875, "signal/format_reward/group_bin_occupancy": 0.126953125, "signal/format_reward/group_std_mean": 0.0027621358167380095, "signal/format_reward/group_zero_std_frac": 0.984375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0004730224609375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0004730224609375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0023530222941190006, "signal/frontier_aurc_reward/group_bin_occupancy": 0.793359375, "signal/frontier_aurc_reward/group_std_mean": 0.003421846404671669, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.9412779986159877e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.9412779986159877e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.26574974954128266, "signal/frontier_coverage_0/group_bin_occupancy": 0.911328125, "signal/frontier_coverage_0/group_std_mean": 0.33279095888137816, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_1/centered_abs_mean": 0.26574974954128266, "signal/frontier_coverage_1/group_bin_occupancy": 0.911328125, "signal/frontier_coverage_1/group_std_mean": 0.33279095888137816, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_10/centered_abs_mean": 0.26574974954128266, "signal/frontier_coverage_10/group_bin_occupancy": 0.911328125, "signal/frontier_coverage_10/group_std_mean": 0.33279095888137816, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_15/centered_abs_mean": 0.26574974954128266, "signal/frontier_coverage_15/group_bin_occupancy": 0.911328125, "signal/frontier_coverage_15/group_std_mean": 0.33279095888137816, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_20/centered_abs_mean": 0.26574974954128266, "signal/frontier_coverage_20/group_bin_occupancy": 0.911328125, "signal/frontier_coverage_20/group_std_mean": 0.33279095888137816, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_25/centered_abs_mean": 0.26574974954128266, "signal/frontier_coverage_25/group_bin_occupancy": 0.911328125, "signal/frontier_coverage_25/group_std_mean": 0.33279095888137816, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_5/centered_abs_mean": 0.26574974954128266, "signal/frontier_coverage_5/group_bin_occupancy": 0.911328125, "signal/frontier_coverage_5/group_std_mean": 0.33279095888137816, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0033218719996511935, "signal/frontier_ece_reward/centered_abs_mean": 0.07035605311393738, "signal/frontier_ece_reward/group_bin_occupancy": 0.81171875, "signal/frontier_ece_reward/group_std_mean": 0.09136128723621369, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.007035605423152447, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.007035605423152447, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.261078929901123, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.761328125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.33954098224639895, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026107892394065857, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026107892394065857, "step": 65 }, { "calibration/aurc": 0.3570083259648462, "calibration/batch_distribution_entropy": 0.9905399679161887, "calibration/batch_entropy_100bins": 0.968378256663429, "calibration/batch_entropy_10bins": 0.9905399679161887, "calibration/batch_entropy_50bins": 0.9805429086034797, "calibration/batch_uniqueness": 0.9607635498046875, "calibration/buffer_distribution_entropy": 0.9716154569256871, "calibration/buffer_entropy_100bins": 0.9211265341579427, "calibration/buffer_entropy_10bins": 0.9716154569256871, "calibration/buffer_entropy_50bins": 0.9470639203068393, "calibration/confidence_entropy": 0.48205357702134116, "calibration/coverage@0%": 0.007421875, "calibration/coverage@1%": 0.007421875, "calibration/coverage@10%": 0.007421875, "calibration/coverage@15%": 0.064453125, "calibration/coverage@20%": 0.20546875, "calibration/coverage@25%": 0.33984375, "calibration/coverage@30%": 0.4140625, "calibration/coverage@5%": 0.007421875, "calibration/ece": 0.17945634368871197, "calibration/mean_confidence": 0.5049372992219131, "calibration/prompt_uniqueness": 0.887060546875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1168.6, "completions/max_terminated_length": 588.8, "completions/mean_length": 133.042578125, "completions/mean_terminated_length": 132.49482421875, "completions/min_length": 51.6, "completions/min_terminated_length": 51.6, "epoch": 0.224, "grad_norm": 0.0030637825839221478, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 230231975.0, "reward": 0.9022814154624939, "reward_std": 0.11243438869714736, "rewards/accuracy_reward": 0.4802734375, "rewards/brier_reward": 0.7444987654685974, "rewards/confidence_uniqueness_reward": 0.9600777387619018, "rewards/format_reward": 0.9994140625, "rewards/frontier_aurc_reward": -0.0034460213501006365, "rewards/frontier_coverage_0": 0.09215732216835022, "rewards/frontier_coverage_1": 0.09215732216835022, "rewards/frontier_coverage_10": 0.09215732216835022, "rewards/frontier_coverage_15": 0.09215732216835022, "rewards/frontier_coverage_20": 0.09215732216835022, "rewards/frontier_coverage_25": 0.09215732216835022, "rewards/frontier_coverage_5": 0.09215732216835022, "rewards/frontier_ece_reward": 0.02385485954582691, "rewards/frontier_entropy_batch_reward": -0.1842617779970169, "signal/accuracy_reward/centered_abs_mean": 0.11500244140625, "signal/accuracy_reward/group_bin_occupancy": 0.18359375, "signal/accuracy_reward/group_std_mean": 0.15716808140277863, "signal/accuracy_reward/group_zero_std_frac": 0.53125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.057501220703125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.057501220703125, "signal/advantage_abs_mean": 0.08655558079481125, "signal/advantage_pre_scale_abs_mean": 0.08655558079481125, "signal/advantage_pre_scale_std": 0.1288917526602745, "signal/advantage_std": 0.1288917526602745, "signal/brier_reward/centered_abs_mean": 0.19534781575202942, "signal/brier_reward/group_bin_occupancy": 0.8890625, "signal/brier_reward/group_std_mean": 0.2440877377986908, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.019534782320261002, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.019534782320261002, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012417678348720073, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.925, "signal/confidence_uniqueness_reward/group_std_mean": 0.01759936846792698, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012417678721249104, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012417678721249104, "signal/format_reward/centered_abs_mean": 0.00113525390625, "signal/format_reward/group_bin_occupancy": 0.12734375, "signal/format_reward/group_std_mean": 0.0033145629800856113, "signal/format_reward/group_zero_std_frac": 0.98125, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000567626953125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000567626953125, "signal/frontier_aurc_reward/centered_abs_mean": 0.0025966161396354438, "signal/frontier_aurc_reward/group_bin_occupancy": 0.78203125, "signal/frontier_aurc_reward/group_std_mean": 0.003768688952550292, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.245770421926864e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.245770421926864e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.22448850572109222, "signal/frontier_coverage_0/group_bin_occupancy": 0.89375, "signal/frontier_coverage_0/group_std_mean": 0.28729032874107363, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_1/centered_abs_mean": 0.22448850572109222, "signal/frontier_coverage_1/group_bin_occupancy": 0.89375, "signal/frontier_coverage_1/group_std_mean": 0.28729032874107363, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_10/centered_abs_mean": 0.22448850572109222, "signal/frontier_coverage_10/group_bin_occupancy": 0.89375, "signal/frontier_coverage_10/group_std_mean": 0.28729032874107363, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_15/centered_abs_mean": 0.22448850572109222, "signal/frontier_coverage_15/group_bin_occupancy": 0.89375, "signal/frontier_coverage_15/group_std_mean": 0.28729032874107363, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_20/centered_abs_mean": 0.22448850572109222, "signal/frontier_coverage_20/group_bin_occupancy": 0.89375, "signal/frontier_coverage_20/group_std_mean": 0.28729032874107363, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_25/centered_abs_mean": 0.22448850572109222, "signal/frontier_coverage_25/group_bin_occupancy": 0.89375, "signal/frontier_coverage_25/group_std_mean": 0.28729032874107363, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_5/centered_abs_mean": 0.22448850572109222, "signal/frontier_coverage_5/group_bin_occupancy": 0.89375, "signal/frontier_coverage_5/group_std_mean": 0.28729032874107363, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0028061064425855876, "signal/frontier_ece_reward/centered_abs_mean": 0.06906488835811615, "signal/frontier_ece_reward/group_bin_occupancy": 0.755859375, "signal/frontier_ece_reward/group_std_mean": 0.08880563378334046, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.006906488910317421, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.006906488910317421, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.271158504486084, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.749609375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3494957447052002, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027115850895643234, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027115850895643234, "step": 70 }, { "calibration/aurc": 0.38536240811982125, "calibration/batch_distribution_entropy": 0.9758133901457793, "calibration/batch_entropy_100bins": 0.958513133717658, "calibration/batch_entropy_10bins": 0.9758133901457793, "calibration/batch_entropy_50bins": 0.9753266753138741, "calibration/batch_uniqueness": 0.960302734375, "calibration/buffer_distribution_entropy": 0.9749793514744705, "calibration/buffer_entropy_100bins": 0.9304407069331182, "calibration/buffer_entropy_10bins": 0.9749793514744705, "calibration/buffer_entropy_50bins": 0.9537239020071407, "calibration/confidence_entropy": 0.5108846399720195, "calibration/coverage@0%": 0.00078125, "calibration/coverage@1%": 0.00078125, "calibration/coverage@10%": 0.00078125, "calibration/coverage@15%": 0.062890625, "calibration/coverage@20%": 0.191015625, "calibration/coverage@25%": 0.20390625, "calibration/coverage@30%": 0.21953125, "calibration/coverage@5%": 0.00078125, "calibration/ece": 0.17052226836955228, "calibration/mean_confidence": 0.5127479551726574, "calibration/prompt_uniqueness": 0.8896484375, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 830.0, "completions/max_terminated_length": 623.6, "completions/mean_length": 144.61328125, "completions/mean_terminated_length": 144.34223327636718, "completions/min_length": 58.0, "completions/min_terminated_length": 58.0, "epoch": 0.24, "grad_norm": 0.0018532777903601527, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 246964495.0, "reward": 0.9268603563308716, "reward_std": 0.11645922064781189, "rewards/accuracy_reward": 0.54091796875, "rewards/brier_reward": 0.7435102462768555, "rewards/confidence_uniqueness_reward": 0.9611909866333008, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.0031730765011161568, "rewards/frontier_coverage_0": 0.04975779242813587, "rewards/frontier_coverage_1": 0.04975779242813587, "rewards/frontier_coverage_10": 0.04975779242813587, "rewards/frontier_coverage_15": 0.04975779242813587, "rewards/frontier_coverage_20": 0.04975779242813587, "rewards/frontier_coverage_25": 0.04975779242813587, "rewards/frontier_coverage_5": 0.04975779242813587, "rewards/frontier_ece_reward": 0.023415421321988105, "rewards/frontier_entropy_batch_reward": -0.20626775026321412, "signal/accuracy_reward/centered_abs_mean": 0.133074951171875, "signal/accuracy_reward/group_bin_occupancy": 0.1890625, "signal/accuracy_reward/group_std_mean": 0.1774646073579788, "signal/accuracy_reward/group_zero_std_frac": 0.4875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0665374755859375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0665374755859375, "signal/advantage_abs_mean": 0.0911499559879303, "signal/advantage_pre_scale_abs_mean": 0.0911499559879303, "signal/advantage_pre_scale_std": 0.13383136242628096, "signal/advantage_std": 0.13383136242628096, "signal/brier_reward/centered_abs_mean": 0.186209100484848, "signal/brier_reward/group_bin_occupancy": 0.9015625, "signal/brier_reward/group_std_mean": 0.23213159739971162, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01862091086804867, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01862091086804867, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011991414427757262, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.941796875, "signal/confidence_uniqueness_reward/group_std_mean": 0.015595928765833378, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011991414707154036, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011991414707154036, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.002488786540925503, "signal/frontier_aurc_reward/group_bin_occupancy": 0.784765625, "signal/frontier_aurc_reward/group_std_mean": 0.0036117580719292166, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.110983234364539e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.110983234364539e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.21226902008056642, "signal/frontier_coverage_0/group_bin_occupancy": 0.894921875, "signal/frontier_coverage_0/group_std_mean": 0.27439537048339846, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_1/centered_abs_mean": 0.21226902008056642, "signal/frontier_coverage_1/group_bin_occupancy": 0.894921875, "signal/frontier_coverage_1/group_std_mean": 0.27439537048339846, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_10/centered_abs_mean": 0.21226902008056642, "signal/frontier_coverage_10/group_bin_occupancy": 0.894921875, "signal/frontier_coverage_10/group_std_mean": 0.27439537048339846, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_15/centered_abs_mean": 0.21226902008056642, "signal/frontier_coverage_15/group_bin_occupancy": 0.894921875, "signal/frontier_coverage_15/group_std_mean": 0.27439537048339846, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_20/centered_abs_mean": 0.21226902008056642, "signal/frontier_coverage_20/group_bin_occupancy": 0.894921875, "signal/frontier_coverage_20/group_std_mean": 0.27439537048339846, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_25/centered_abs_mean": 0.21226902008056642, "signal/frontier_coverage_25/group_bin_occupancy": 0.894921875, "signal/frontier_coverage_25/group_std_mean": 0.27439537048339846, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_5/centered_abs_mean": 0.21226902008056642, "signal/frontier_coverage_5/group_bin_occupancy": 0.894921875, "signal/frontier_coverage_5/group_std_mean": 0.27439537048339846, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0026533628813922407, "signal/frontier_ece_reward/centered_abs_mean": 0.06265156567096711, "signal/frontier_ece_reward/group_bin_occupancy": 0.74765625, "signal/frontier_ece_reward/group_std_mean": 0.0817591667175293, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.006265156529843807, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.006265156529843807, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2907168686389923, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.762109375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3683573842048645, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.029071687534451485, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.029071687534451485, "step": 75 }, { "calibration/aurc": 0.3158580042037348, "calibration/batch_distribution_entropy": 0.9823403673256449, "calibration/batch_entropy_100bins": 0.9640062520840773, "calibration/batch_entropy_10bins": 0.9823403673256449, "calibration/batch_entropy_50bins": 0.9766481913725391, "calibration/batch_uniqueness": 0.959661865234375, "calibration/buffer_distribution_entropy": 0.9781583045272951, "calibration/buffer_entropy_100bins": 0.9380783537839212, "calibration/buffer_entropy_10bins": 0.9781583045272951, "calibration/buffer_entropy_50bins": 0.9593439714401006, "calibration/confidence_entropy": 0.4970008225128133, "calibration/coverage@0%": 0.003125, "calibration/coverage@1%": 0.003125, "calibration/coverage@10%": 0.047265625, "calibration/coverage@15%": 0.22421875, "calibration/coverage@20%": 0.346484375, "calibration/coverage@25%": 0.437109375, "calibration/coverage@30%": 0.51015625, "calibration/coverage@5%": 0.003125, "calibration/ece": 0.15148089804395615, "calibration/mean_confidence": 0.47737086449137217, "calibration/prompt_uniqueness": 0.885302734375, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 941.4, "completions/max_terminated_length": 574.6, "completions/mean_length": 149.9041015625, "completions/mean_terminated_length": 149.63403930664063, "completions/min_length": 58.4, "completions/min_terminated_length": 58.4, "epoch": 0.256, "grad_norm": 0.0013435143046081066, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 263554329.0, "reward": 0.9198351383209229, "reward_std": 0.1078387126326561, "rewards/accuracy_reward": 0.51611328125, "rewards/brier_reward": 0.7574279069900512, "rewards/confidence_uniqueness_reward": 0.9602339506149292, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.0029626599047333003, "rewards/frontier_coverage_0": 0.08376037068665028, "rewards/frontier_coverage_1": 0.08376037068665028, "rewards/frontier_coverage_10": 0.08376037068665028, "rewards/frontier_coverage_15": 0.08376037068665028, "rewards/frontier_coverage_20": 0.08376037068665028, "rewards/frontier_coverage_25": 0.08376037068665028, "rewards/frontier_coverage_5": 0.08376037068665028, "rewards/frontier_ece_reward": 0.023681937158107756, "rewards/frontier_entropy_batch_reward": -0.19550244212150575, "signal/accuracy_reward/centered_abs_mean": 0.127130126953125, "signal/accuracy_reward/group_bin_occupancy": 0.18359375, "signal/accuracy_reward/group_std_mean": 0.16637569069862365, "signal/accuracy_reward/group_zero_std_frac": 0.53125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0635650634765625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0635650634765625, "signal/advantage_abs_mean": 0.08491947501897812, "signal/advantage_pre_scale_abs_mean": 0.08491947501897812, "signal/advantage_pre_scale_std": 0.12484004199504853, "signal/advantage_std": 0.12484004199504853, "signal/brier_reward/centered_abs_mean": 0.17298209071159362, "signal/brier_reward/group_bin_occupancy": 0.87578125, "signal/brier_reward/group_std_mean": 0.218139910697937, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01729821041226387, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01729821041226387, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.013145623542368413, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.937890625, "signal/confidence_uniqueness_reward/group_std_mean": 0.01699326727539301, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0013145623495802283, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0013145623495802283, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.002108518313616514, "signal/frontier_aurc_reward/group_bin_occupancy": 0.76484375, "signal/frontier_aurc_reward/group_std_mean": 0.003169979201629758, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.635647906572558e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.635647906572558e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.22474998235702515, "signal/frontier_coverage_0/group_bin_occupancy": 0.898828125, "signal/frontier_coverage_0/group_std_mean": 0.28549715876579285, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_1/centered_abs_mean": 0.22474998235702515, "signal/frontier_coverage_1/group_bin_occupancy": 0.898828125, "signal/frontier_coverage_1/group_std_mean": 0.28549715876579285, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_10/centered_abs_mean": 0.22474998235702515, "signal/frontier_coverage_10/group_bin_occupancy": 0.898828125, "signal/frontier_coverage_10/group_std_mean": 0.28549715876579285, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_15/centered_abs_mean": 0.22474998235702515, "signal/frontier_coverage_15/group_bin_occupancy": 0.898828125, "signal/frontier_coverage_15/group_std_mean": 0.28549715876579285, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_20/centered_abs_mean": 0.22474998235702515, "signal/frontier_coverage_20/group_bin_occupancy": 0.898828125, "signal/frontier_coverage_20/group_std_mean": 0.28549715876579285, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_25/centered_abs_mean": 0.22474998235702515, "signal/frontier_coverage_25/group_bin_occupancy": 0.898828125, "signal/frontier_coverage_25/group_std_mean": 0.28549715876579285, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_5/centered_abs_mean": 0.22474998235702515, "signal/frontier_coverage_5/group_bin_occupancy": 0.898828125, "signal/frontier_coverage_5/group_std_mean": 0.28549715876579285, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002809374686330557, "signal/frontier_ece_reward/centered_abs_mean": 0.05240138992667198, "signal/frontier_ece_reward/group_bin_occupancy": 0.712890625, "signal/frontier_ece_reward/group_std_mean": 0.07018679082393646, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.005240139458328486, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.005240139458328486, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.28227958679199217, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.76328125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.35941845178604126, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02822795957326889, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02822795957326889, "step": 80 }, { "calibration/aurc": 0.394429960364209, "calibration/batch_distribution_entropy": 0.9882168109829511, "calibration/batch_entropy_100bins": 0.9647778146117572, "calibration/batch_entropy_10bins": 0.9882168109829511, "calibration/batch_entropy_50bins": 0.9796198563395535, "calibration/batch_uniqueness": 0.9600006103515625, "calibration/buffer_distribution_entropy": 0.9812990512986719, "calibration/buffer_entropy_100bins": 0.9444644026260851, "calibration/buffer_entropy_10bins": 0.9812990512986719, "calibration/buffer_entropy_50bins": 0.9640980870518174, "calibration/confidence_entropy": 0.5071811942098001, "calibration/coverage@0%": 0.005859375, "calibration/coverage@1%": 0.005859375, "calibration/coverage@10%": 0.013671875, "calibration/coverage@15%": 0.046875, "calibration/coverage@20%": 0.106640625, "calibration/coverage@25%": 0.2375, "calibration/coverage@30%": 0.322265625, "calibration/coverage@5%": 0.005859375, "calibration/ece": 0.13428001011321392, "calibration/mean_confidence": 0.46757799401224603, "calibration/prompt_uniqueness": 0.88369140625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 696.0, "completions/max_terminated_length": 469.0, "completions/mean_length": 161.1001953125, "completions/mean_terminated_length": 160.96529846191407, "completions/min_length": 52.2, "completions/min_terminated_length": 52.2, "epoch": 0.272, "grad_norm": 0.0017471429891884327, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 280169691.0, "reward": 0.9123312473297119, "reward_std": 0.10262252241373063, "rewards/accuracy_reward": 0.4990234375, "rewards/brier_reward": 0.7567744970321655, "rewards/confidence_uniqueness_reward": 0.958899450302124, "rewards/format_reward": 0.99921875, "rewards/frontier_aurc_reward": -0.0029471824411302803, "rewards/frontier_coverage_0": 0.09450841639190913, "rewards/frontier_coverage_1": 0.09450841639190913, "rewards/frontier_coverage_10": 0.09450841639190913, "rewards/frontier_coverage_15": 0.09450841639190913, "rewards/frontier_coverage_20": 0.09450841639190913, "rewards/frontier_coverage_25": 0.09450841639190913, "rewards/frontier_coverage_5": 0.09450841639190913, "rewards/frontier_ece_reward": 0.021173277497291566, "rewards/frontier_entropy_batch_reward": -0.18707219362258912, "signal/accuracy_reward/centered_abs_mean": 0.118701171875, "signal/accuracy_reward/group_bin_occupancy": 0.177734375, "signal/accuracy_reward/group_std_mean": 0.15330464243888856, "signal/accuracy_reward/group_zero_std_frac": 0.578125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0593505859375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0593505859375, "signal/advantage_abs_mean": 0.08100139200687409, "signal/advantage_pre_scale_abs_mean": 0.08100139200687409, "signal/advantage_pre_scale_std": 0.12124869823455811, "signal/advantage_std": 0.12124869823455811, "signal/brier_reward/centered_abs_mean": 0.1675354540348053, "signal/brier_reward/group_bin_occupancy": 0.884765625, "signal/brier_reward/group_std_mean": 0.21173399388790132, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01675354540348053, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01675354540348053, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.013952768594026565, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.940625, "signal/confidence_uniqueness_reward/group_std_mean": 0.018382455967366696, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0013952769106253982, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0013952769106253982, "signal/format_reward/centered_abs_mean": 0.0013916015625, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.002810100419446826, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00069580078125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00069580078125, "signal/frontier_aurc_reward/centered_abs_mean": 0.001968752057291567, "signal/frontier_aurc_reward/group_bin_occupancy": 0.773828125, "signal/frontier_aurc_reward/group_std_mean": 0.0029303951654583217, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.4609401953057387e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.4609401953057387e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.22191445231437684, "signal/frontier_coverage_0/group_bin_occupancy": 0.91171875, "signal/frontier_coverage_0/group_std_mean": 0.2817406952381134, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_1/centered_abs_mean": 0.22191445231437684, "signal/frontier_coverage_1/group_bin_occupancy": 0.91171875, "signal/frontier_coverage_1/group_std_mean": 0.2817406952381134, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_10/centered_abs_mean": 0.22191445231437684, "signal/frontier_coverage_10/group_bin_occupancy": 0.91171875, "signal/frontier_coverage_10/group_std_mean": 0.2817406952381134, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_15/centered_abs_mean": 0.22191445231437684, "signal/frontier_coverage_15/group_bin_occupancy": 0.91171875, "signal/frontier_coverage_15/group_std_mean": 0.2817406952381134, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_20/centered_abs_mean": 0.22191445231437684, "signal/frontier_coverage_20/group_bin_occupancy": 0.91171875, "signal/frontier_coverage_20/group_std_mean": 0.2817406952381134, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_25/centered_abs_mean": 0.22191445231437684, "signal/frontier_coverage_25/group_bin_occupancy": 0.91171875, "signal/frontier_coverage_25/group_std_mean": 0.2817406952381134, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_5/centered_abs_mean": 0.22191445231437684, "signal/frontier_coverage_5/group_bin_occupancy": 0.91171875, "signal/frontier_coverage_5/group_std_mean": 0.2817406952381134, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0027739306446164847, "signal/frontier_ece_reward/centered_abs_mean": 0.046379628032445906, "signal/frontier_ece_reward/group_bin_occupancy": 0.692578125, "signal/frontier_ece_reward/group_std_mean": 0.06191401258111, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.004637962859123945, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.004637962859123945, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2656592845916748, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.75546875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34435033798217773, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026565929502248765, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026565929502248765, "step": 85 }, { "calibration/aurc": 0.34339292028696106, "calibration/batch_distribution_entropy": 0.9861516860406185, "calibration/batch_entropy_100bins": 0.9631372903955786, "calibration/batch_entropy_10bins": 0.9861516860406185, "calibration/batch_entropy_50bins": 0.9765421354775385, "calibration/batch_uniqueness": 0.9614471435546875, "calibration/buffer_distribution_entropy": 0.9837682259335165, "calibration/buffer_entropy_100bins": 0.9497297029403435, "calibration/buffer_entropy_10bins": 0.9837682259335165, "calibration/buffer_entropy_50bins": 0.9679628603835398, "calibration/confidence_entropy": 0.5009262760712196, "calibration/coverage@0%": 0.002734375, "calibration/coverage@1%": 0.002734375, "calibration/coverage@10%": 0.05859375, "calibration/coverage@15%": 0.0921875, "calibration/coverage@20%": 0.12578125, "calibration/coverage@25%": 0.208203125, "calibration/coverage@30%": 0.330078125, "calibration/coverage@5%": 0.021484375, "calibration/ece": 0.13718419607767185, "calibration/mean_confidence": 0.5285074179511803, "calibration/prompt_uniqueness": 0.887109375, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 875.2, "completions/max_terminated_length": 472.4, "completions/mean_length": 163.47216796875, "completions/mean_terminated_length": 163.20425415039062, "completions/min_length": 66.0, "completions/min_terminated_length": 66.0, "epoch": 0.288, "grad_norm": 0.0012677995255216956, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 296801822.0, "reward": 0.9243088364601135, "reward_std": 0.10623209476470948, "rewards/accuracy_reward": 0.51796875, "rewards/brier_reward": 0.7644232630729675, "rewards/confidence_uniqueness_reward": 0.9606874465942383, "rewards/format_reward": 0.99970703125, "rewards/frontier_aurc_reward": -0.002895602444186807, "rewards/frontier_coverage_0": 0.09357523247599601, "rewards/frontier_coverage_1": 0.09357523247599601, "rewards/frontier_coverage_10": 0.09357523247599601, "rewards/frontier_coverage_15": 0.09357523247599601, "rewards/frontier_coverage_20": 0.09357523247599601, "rewards/frontier_coverage_25": 0.09357523247599601, "rewards/frontier_coverage_5": 0.09357523247599601, "rewards/frontier_ece_reward": 0.025733930245041847, "rewards/frontier_entropy_batch_reward": -0.1776515692472458, "signal/accuracy_reward/centered_abs_mean": 0.12696533203125, "signal/accuracy_reward/group_bin_occupancy": 0.183984375, "signal/accuracy_reward/group_std_mean": 0.16651269495487214, "signal/accuracy_reward/group_zero_std_frac": 0.528125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.063482666015625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.063482666015625, "signal/advantage_abs_mean": 0.08275850415229798, "signal/advantage_pre_scale_abs_mean": 0.08275850415229798, "signal/advantage_pre_scale_std": 0.12433976083993911, "signal/advantage_std": 0.12433976083993911, "signal/brier_reward/centered_abs_mean": 0.16671029329299927, "signal/brier_reward/group_bin_occupancy": 0.86953125, "signal/brier_reward/group_std_mean": 0.21041498482227325, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01667102985084057, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01667102985084057, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01224772147834301, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.93671875, "signal/confidence_uniqueness_reward/group_std_mean": 0.01621842011809349, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001224772143177688, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001224772143177688, "signal/format_reward/centered_abs_mean": 0.000567626953125, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0016572814900428056, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0002838134765625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0002838134765625, "signal/frontier_aurc_reward/centered_abs_mean": 0.002230943040922284, "signal/frontier_aurc_reward/group_bin_occupancy": 0.77578125, "signal/frontier_aurc_reward/group_std_mean": 0.003288121521472931, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.788679048535414e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.788679048535414e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.2148287147283554, "signal/frontier_coverage_0/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_0/group_std_mean": 0.27277125120162965, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_1/centered_abs_mean": 0.2148287147283554, "signal/frontier_coverage_1/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_1/group_std_mean": 0.27277125120162965, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_10/centered_abs_mean": 0.2148287147283554, "signal/frontier_coverage_10/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_10/group_std_mean": 0.27277125120162965, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_15/centered_abs_mean": 0.2148287147283554, "signal/frontier_coverage_15/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_15/group_std_mean": 0.27277125120162965, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_20/centered_abs_mean": 0.2148287147283554, "signal/frontier_coverage_20/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_20/group_std_mean": 0.27277125120162965, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_25/centered_abs_mean": 0.2148287147283554, "signal/frontier_coverage_25/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_25/group_std_mean": 0.27277125120162965, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_5/centered_abs_mean": 0.2148287147283554, "signal/frontier_coverage_5/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_5/group_std_mean": 0.27277125120162965, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0026853590272367, "signal/frontier_ece_reward/centered_abs_mean": 0.049359950423240664, "signal/frontier_ece_reward/group_bin_occupancy": 0.684375, "signal/frontier_ece_reward/group_std_mean": 0.06494068205356598, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00493599520996213, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00493599520996213, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2646158039569855, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.742578125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34475996494293215, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026461581513285638, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026461581513285638, "step": 90 }, { "calibration/aurc": 0.3242596520493021, "calibration/batch_distribution_entropy": 0.9689421000707247, "calibration/batch_entropy_100bins": 0.9573357889459958, "calibration/batch_entropy_10bins": 0.9689421000707247, "calibration/batch_entropy_50bins": 0.9692127380588558, "calibration/batch_uniqueness": 0.9587066650390625, "calibration/buffer_distribution_entropy": 0.9846685958521523, "calibration/buffer_entropy_100bins": 0.9539386537974168, "calibration/buffer_entropy_10bins": 0.9846685958521523, "calibration/buffer_entropy_50bins": 0.970798853899867, "calibration/confidence_entropy": 0.47673372489425797, "calibration/coverage@0%": 0.001953125, "calibration/coverage@1%": 0.001953125, "calibration/coverage@10%": 0.024609375, "calibration/coverage@15%": 0.069921875, "calibration/coverage@20%": 0.178125, "calibration/coverage@25%": 0.249609375, "calibration/coverage@30%": 0.519140625, "calibration/coverage@5%": 0.009375, "calibration/ece": 0.12447883619694644, "calibration/mean_confidence": 0.56840598031393, "calibration/prompt_uniqueness": 0.877783203125, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 915.8, "completions/max_terminated_length": 482.0, "completions/mean_length": 171.12587890625, "completions/mean_terminated_length": 170.8596405029297, "completions/min_length": 74.4, "completions/min_terminated_length": 74.4, "epoch": 0.304, "grad_norm": 0.0014622469898313284, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 313484103.0, "reward": 0.917207932472229, "reward_std": 0.10467681884765626, "rewards/accuracy_reward": 0.50869140625, "rewards/brier_reward": 0.7559807538986206, "rewards/confidence_uniqueness_reward": 0.9588512420654297, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.003155782353132963, "rewards/frontier_coverage_0": 0.09547070190310478, "rewards/frontier_coverage_1": 0.09547070190310478, "rewards/frontier_coverage_10": 0.09547070190310478, "rewards/frontier_coverage_15": 0.09547070190310478, "rewards/frontier_coverage_20": 0.09547070190310478, "rewards/frontier_coverage_25": 0.09547070190310478, "rewards/frontier_coverage_5": 0.09547070190310478, "rewards/frontier_ece_reward": 0.023610943742096425, "rewards/frontier_entropy_batch_reward": -0.19100989103317262, "signal/accuracy_reward/centered_abs_mean": 0.120941162109375, "signal/accuracy_reward/group_bin_occupancy": 0.183984375, "signal/accuracy_reward/group_std_mean": 0.16227840185165404, "signal/accuracy_reward/group_zero_std_frac": 0.528125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0604705810546875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0604705810546875, "signal/advantage_abs_mean": 0.0804534524679184, "signal/advantage_pre_scale_abs_mean": 0.0804534524679184, "signal/advantage_pre_scale_std": 0.12147486358880996, "signal/advantage_std": 0.12147486358880996, "signal/brier_reward/centered_abs_mean": 0.16698463261127472, "signal/brier_reward/group_bin_occupancy": 0.879296875, "signal/brier_reward/group_std_mean": 0.21043302416801452, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.016698462888598442, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.016698462888598442, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012571757286787033, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9234375, "signal/confidence_uniqueness_reward/group_std_mean": 0.016940852627158165, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001257175812497735, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001257175812497735, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086068242787, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.00244970191270113, "signal/frontier_aurc_reward/group_bin_occupancy": 0.7890625, "signal/frontier_aurc_reward/group_std_mean": 0.0035024710465222596, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.062127361772582e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.062127361772582e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.20646621882915497, "signal/frontier_coverage_0/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_0/group_std_mean": 0.2654423028230667, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_1/centered_abs_mean": 0.20646621882915497, "signal/frontier_coverage_1/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_1/group_std_mean": 0.2654423028230667, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_10/centered_abs_mean": 0.20646621882915497, "signal/frontier_coverage_10/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_10/group_std_mean": 0.2654423028230667, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_15/centered_abs_mean": 0.20646621882915497, "signal/frontier_coverage_15/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_15/group_std_mean": 0.2654423028230667, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_20/centered_abs_mean": 0.20646621882915497, "signal/frontier_coverage_20/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_20/group_std_mean": 0.2654423028230667, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_25/centered_abs_mean": 0.20646621882915497, "signal/frontier_coverage_25/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_25/group_std_mean": 0.2654423028230667, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_5/centered_abs_mean": 0.20646621882915497, "signal/frontier_coverage_5/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_5/group_std_mean": 0.2654423028230667, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002580827847123146, "signal/frontier_ece_reward/centered_abs_mean": 0.05028548315167427, "signal/frontier_ece_reward/group_bin_occupancy": 0.664453125, "signal/frontier_ece_reward/group_std_mean": 0.06461961716413497, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0050285480916500095, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0050285480916500095, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27090482115745546, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.75390625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34876925945281984, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027090481668710708, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027090481668710708, "step": 95 }, { "calibration/aurc": 0.2548086250052631, "calibration/batch_distribution_entropy": 0.9832089330249761, "calibration/batch_entropy_100bins": 0.9654193413067593, "calibration/batch_entropy_10bins": 0.9832089330249761, "calibration/batch_entropy_50bins": 0.977077582632314, "calibration/batch_uniqueness": 0.9564571354801263, "calibration/buffer_distribution_entropy": 0.9850274222605417, "calibration/buffer_entropy_100bins": 0.9573997216357878, "calibration/buffer_entropy_10bins": 0.9850274222605417, "calibration/buffer_entropy_50bins": 0.9729485825087311, "calibration/confidence_entropy": 0.4840517561929529, "calibration/coverage@0%": 0.019532014432485322, "calibration/coverage@1%": 0.019532014432485322, "calibration/coverage@10%": 0.20705647627201565, "calibration/coverage@15%": 0.36352128180039134, "calibration/coverage@20%": 0.4444425758317026, "calibration/coverage@25%": 0.5226119129158513, "calibration/coverage@30%": 0.5921722113502936, "calibration/coverage@5%": 0.05937576443248532, "calibration/ece": 0.14736111301391036, "calibration/mean_confidence": 0.5551109373295257, "calibration/prompt_uniqueness": 0.8770231915810353, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 717.0, "completions/max_terminated_length": 540.8, "completions/mean_length": 176.12119140625, "completions/mean_terminated_length": 175.98831481933593, "completions/min_length": 62.6, "completions/min_terminated_length": 62.6, "epoch": 0.32, "grad_norm": 0.0011271568946540356, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 330376288.0, "reward": 0.9313054084777832, "reward_std": 0.08764051198959351, "rewards/accuracy_reward": 0.52919921875, "rewards/brier_reward": 0.7781755566596985, "rewards/confidence_uniqueness_reward": 0.9566641449928284, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.002697052760049701, "rewards/frontier_coverage_0": 0.10069872457534075, "rewards/frontier_coverage_1": 0.10069872457534075, "rewards/frontier_coverage_10": 0.10069872457534075, "rewards/frontier_coverage_15": 0.10069872457534075, "rewards/frontier_coverage_20": 0.10069872457534075, "rewards/frontier_coverage_25": 0.10069872457534075, "rewards/frontier_coverage_5": 0.10069872457534075, "rewards/frontier_ece_reward": 0.027973872423171998, "rewards/frontier_entropy_batch_reward": -0.1815766215324402, "signal/accuracy_reward/centered_abs_mean": 0.084625244140625, "signal/accuracy_reward/group_bin_occupancy": 0.169921875, "signal/accuracy_reward/group_std_mean": 0.11722688525915145, "signal/accuracy_reward/group_zero_std_frac": 0.640625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0423126220703125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0423126220703125, "signal/advantage_abs_mean": 0.06639857962727547, "signal/advantage_pre_scale_abs_mean": 0.06639857962727547, "signal/advantage_pre_scale_std": 0.1043279305100441, "signal/advantage_std": 0.1043279305100441, "signal/brier_reward/centered_abs_mean": 0.1507304072380066, "signal/brier_reward/group_bin_occupancy": 0.842578125, "signal/brier_reward/group_std_mean": 0.1937762975692749, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.015073040872812271, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.015073040872812271, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012130972370505333, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.93984375, "signal/confidence_uniqueness_reward/group_std_mean": 0.016449279338121413, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012130972696468233, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012130972696468233, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086533904076, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0021575469290837646, "signal/frontier_aurc_reward/group_bin_occupancy": 0.771484375, "signal/frontier_aurc_reward/group_std_mean": 0.0031614691484719514, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.6969336613547056e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.6969336613547056e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.18550774753093718, "signal/frontier_coverage_0/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_0/group_std_mean": 0.23913069665431977, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_1/centered_abs_mean": 0.18550774753093718, "signal/frontier_coverage_1/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_1/group_std_mean": 0.23913069665431977, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_10/centered_abs_mean": 0.18550774753093718, "signal/frontier_coverage_10/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_10/group_std_mean": 0.23913069665431977, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_15/centered_abs_mean": 0.18550774753093718, "signal/frontier_coverage_15/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_15/group_std_mean": 0.23913069665431977, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_20/centered_abs_mean": 0.18550774753093718, "signal/frontier_coverage_20/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_20/group_std_mean": 0.23913069665431977, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_25/centered_abs_mean": 0.18550774753093718, "signal/frontier_coverage_25/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_25/group_std_mean": 0.23913069665431977, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_5/centered_abs_mean": 0.18550774753093718, "signal/frontier_coverage_5/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_5/group_std_mean": 0.23913069665431977, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0023188469000160693, "signal/frontier_ece_reward/centered_abs_mean": 0.04614235162734985, "signal/frontier_ece_reward/group_bin_occupancy": 0.650390625, "signal/frontier_ece_reward/group_std_mean": 0.05914493575692177, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.004614235367625952, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.004614235367625952, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.261948898434639, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.746875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34005898237228394, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026194890215992927, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026194890215992927, "step": 100 }, { "epoch": 0.32, "eval_calibration/aurc": 0.4821336016995468, "eval_calibration/batch_distribution_entropy": 0.9068297642190777, "eval_calibration/batch_entropy_100bins": 0.6990601166122135, "eval_calibration/batch_entropy_10bins": 0.9068297642190777, "eval_calibration/batch_entropy_50bins": 0.7761700409131593, "eval_calibration/batch_uniqueness": 0.892578125, "eval_calibration/buffer_distribution_entropy": 0.9854115328781001, "eval_calibration/buffer_entropy_100bins": 0.9594245454033348, "eval_calibration/buffer_entropy_10bins": 0.9854115328781001, "eval_calibration/buffer_entropy_50bins": 0.9741906618665293, "eval_calibration/confidence_entropy": 0.472960458671501, "eval_calibration/coverage@0%": 0.0234375, "eval_calibration/coverage@1%": 0.0234375, "eval_calibration/coverage@10%": 0.0234375, "eval_calibration/coverage@15%": 0.078125, "eval_calibration/coverage@20%": 0.109375, "eval_calibration/coverage@25%": 0.140625, "eval_calibration/coverage@30%": 0.2265625, "eval_calibration/coverage@5%": 0.0234375, "eval_calibration/ece": 0.22956069164282372, "eval_calibration/mean_confidence": 0.45870384307265843, "eval_calibration/prompt_uniqueness": 0.892578125, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 383.0, "eval_completions/max_terminated_length": 383.0, "eval_completions/mean_length": 179.88496780395508, "eval_completions/mean_terminated_length": 179.88496780395508, "eval_completions/min_length": 91.5, "eval_completions/min_terminated_length": 91.5, "eval_loss": 0.0, "eval_num_tokens": 330376288.0, "eval_reward": 0.7866154909133911, "eval_reward_std": 0.22616717219352722, "eval_rewards/accuracy_reward": 0.408203125, "eval_rewards/brier_reward": 0.766254648566246, "eval_rewards/confidence_uniqueness_reward": 0.89599609375, "eval_rewards/format_reward": 1.0, "eval_rewards/frontier_aurc_reward": -0.0034958955366164446, "eval_rewards/frontier_coverage_0": 0.1697857677936554, "eval_rewards/frontier_coverage_1": 0.1697857677936554, "eval_rewards/frontier_coverage_10": 0.1697857677936554, "eval_rewards/frontier_coverage_15": 0.1697857677936554, "eval_rewards/frontier_coverage_20": 0.1697857677936554, "eval_rewards/frontier_coverage_25": 0.1697857677936554, "eval_rewards/frontier_coverage_5": 0.1697857677936554, "eval_rewards/frontier_ece_reward": 0.014763034647330642, "eval_rewards/frontier_entropy_batch_reward": -1.0, "eval_runtime": 20.3578, "eval_samples_per_second": 24.561, "eval_signal/accuracy_reward/centered_abs_mean": 0.4749755859375, "eval_signal/accuracy_reward/group_bin_occupancy": 0.25, "eval_signal/accuracy_reward/group_std_mean": 0.49487074464559555, "eval_signal/accuracy_reward/group_zero_std_frac": 0.0, "eval_signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.23748779296875, "eval_signal/accuracy_reward/weight": 0.5, "eval_signal/accuracy_reward/weighted_centered_abs_mean": 0.23748779296875, "eval_signal/advantage_abs_mean": 0.210348479449749, "eval_signal/advantage_pre_scale_abs_mean": 0.210348479449749, "eval_signal/advantage_pre_scale_std": 0.22372927144169807, "eval_signal/advantage_std": 0.22372927144169807, "eval_signal/brier_reward/centered_abs_mean": 0.20850331708788872, "eval_signal/brier_reward/group_bin_occupancy": 0.8984375, "eval_signal/brier_reward/group_std_mean": 0.2590029910206795, "eval_signal/brier_reward/group_zero_std_frac": 0.0, "eval_signal/brier_reward/scaled_weighted_centered_abs_mean": 0.020850331522524357, "eval_signal/brier_reward/weight": 0.10000000149011612, "eval_signal/brier_reward/weighted_centered_abs_mean": 0.020850331522524357, "eval_signal/confidence_uniqueness_reward/centered_abs_mean": 0.042022705078125, "eval_signal/confidence_uniqueness_reward/group_bin_occupancy": 0.421875, "eval_signal/confidence_uniqueness_reward/group_std_mean": 0.050302863121032715, "eval_signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "eval_signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.004202270647510886, "eval_signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "eval_signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.004202270647510886, "eval_signal/format_reward/centered_abs_mean": 0.0, "eval_signal/format_reward/group_bin_occupancy": 0.125, "eval_signal/format_reward/group_std_mean": 0.0, "eval_signal/format_reward/group_zero_std_frac": 1.0, "eval_signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/format_reward/weight": 0.5, "eval_signal/format_reward/weighted_centered_abs_mean": 0.0, "eval_signal/frontier_aurc_reward/centered_abs_mean": 0.003264666476752609, "eval_signal/frontier_aurc_reward/group_bin_occupancy": 0.78125, "eval_signal/frontier_aurc_reward/group_std_mean": 0.005248750909231603, "eval_signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.080833059560973e-05, "eval_signal/frontier_aurc_reward/weight": 0.012500000186264515, "eval_signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.080833059560973e-05, "eval_signal/frontier_coverage_0/centered_abs_mean": 0.36298905313014984, "eval_signal/frontier_coverage_0/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_0/group_std_mean": 0.4504668563604355, "eval_signal/frontier_coverage_0/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_0/weight": 0.012500000186264515, "eval_signal/frontier_coverage_0/weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_1/centered_abs_mean": 0.36298905313014984, "eval_signal/frontier_coverage_1/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_1/group_std_mean": 0.4504668563604355, "eval_signal/frontier_coverage_1/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_1/weight": 0.012500000186264515, "eval_signal/frontier_coverage_1/weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_10/centered_abs_mean": 0.36298905313014984, "eval_signal/frontier_coverage_10/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_10/group_std_mean": 0.4504668563604355, "eval_signal/frontier_coverage_10/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_10/weight": 0.012500000186264515, "eval_signal/frontier_coverage_10/weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_15/centered_abs_mean": 0.36298905313014984, "eval_signal/frontier_coverage_15/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_15/group_std_mean": 0.4504668563604355, "eval_signal/frontier_coverage_15/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_15/weight": 0.012500000186264515, "eval_signal/frontier_coverage_15/weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_20/centered_abs_mean": 0.36298905313014984, "eval_signal/frontier_coverage_20/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_20/group_std_mean": 0.4504668563604355, "eval_signal/frontier_coverage_20/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_20/weight": 0.012500000186264515, "eval_signal/frontier_coverage_20/weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_25/centered_abs_mean": 0.36298905313014984, "eval_signal/frontier_coverage_25/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_25/group_std_mean": 0.4504668563604355, "eval_signal/frontier_coverage_25/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_25/weight": 0.012500000186264515, "eval_signal/frontier_coverage_25/weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_5/centered_abs_mean": 0.36298905313014984, "eval_signal/frontier_coverage_5/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_5/group_std_mean": 0.4504668563604355, "eval_signal/frontier_coverage_5/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_coverage_5/weight": 0.012500000186264515, "eval_signal/frontier_coverage_5/weighted_centered_abs_mean": 0.004537363303825259, "eval_signal/frontier_ece_reward/centered_abs_mean": 0.04522951692342758, "eval_signal/frontier_ece_reward/group_bin_occupancy": 0.78125, "eval_signal/frontier_ece_reward/group_std_mean": 0.06876218784600496, "eval_signal/frontier_ece_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.004522952018305659, "eval_signal/frontier_ece_reward/weight": 0.10000000149011612, "eval_signal/frontier_ece_reward/weighted_centered_abs_mean": 0.004522952018305659, "eval_signal/frontier_entropy_batch_reward/centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.125, "eval_signal/frontier_entropy_batch_reward/group_std_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_zero_std_frac": 1.0, "eval_signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "eval_signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.0, "eval_steps_per_second": 0.196, "step": 100 }, { "calibration/aurc": 0.2894661246726904, "calibration/batch_distribution_entropy": 0.9825675566635621, "calibration/batch_entropy_100bins": 0.9710254527417408, "calibration/batch_entropy_10bins": 0.9825675566635621, "calibration/batch_entropy_50bins": 0.9787532687697368, "calibration/batch_uniqueness": 0.954045575920319, "calibration/buffer_distribution_entropy": 0.9876786313621434, "calibration/buffer_entropy_100bins": 0.963686605417004, "calibration/buffer_entropy_10bins": 0.9876786313621434, "calibration/buffer_entropy_50bins": 0.9772690399288407, "calibration/confidence_entropy": 0.5025666251043774, "calibration/coverage@0%": 0.018360906862745097, "calibration/coverage@1%": 0.018360906862745097, "calibration/coverage@10%": 0.0601577818627451, "calibration/coverage@15%": 0.10198835784313727, "calibration/coverage@20%": 0.18480392156862746, "calibration/coverage@25%": 0.4481709558823529, "calibration/coverage@30%": 0.6224295343137255, "calibration/coverage@5%": 0.019142156862745098, "calibration/ece": 0.14327281797118085, "calibration/mean_confidence": 0.4667254899239661, "calibration/prompt_uniqueness": 0.8665733506944445, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 917.8, "completions/max_terminated_length": 478.8, "completions/mean_length": 181.013671875, "completions/mean_terminated_length": 180.61538696289062, "completions/min_length": 81.8, "completions/min_terminated_length": 81.8, "epoch": 0.336, "grad_norm": 0.001394961727783084, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 346952300.0, "reward": 0.9360666036605835, "reward_std": 0.09443065226078033, "rewards/accuracy_reward": 0.54248046875, "rewards/brier_reward": 0.7745235323905945, "rewards/confidence_uniqueness_reward": 0.9543019294738769, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.002458465192466974, "rewards/frontier_coverage_0": 0.08697182945907116, "rewards/frontier_coverage_1": 0.08697182945907116, "rewards/frontier_coverage_10": 0.08697182945907116, "rewards/frontier_coverage_15": 0.08697182945907116, "rewards/frontier_coverage_20": 0.08697182945907116, "rewards/frontier_coverage_25": 0.08697182945907116, "rewards/frontier_coverage_5": 0.08697182945907116, "rewards/frontier_ece_reward": 0.023020536825060844, "rewards/frontier_entropy_batch_reward": -0.17742236256599425, "signal/accuracy_reward/centered_abs_mean": 0.104608154296875, "signal/accuracy_reward/group_bin_occupancy": 0.1765625, "signal/accuracy_reward/group_std_mean": 0.14065491855144502, "signal/accuracy_reward/group_zero_std_frac": 0.5875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0523040771484375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0523040771484375, "signal/advantage_abs_mean": 0.07322461903095245, "signal/advantage_pre_scale_abs_mean": 0.07322461903095245, "signal/advantage_pre_scale_std": 0.11356867402791977, "signal/advantage_std": 0.11356867402791977, "signal/brier_reward/centered_abs_mean": 0.15120540261268617, "signal/brier_reward/group_bin_occupancy": 0.8671875, "signal/brier_reward/group_std_mean": 0.191496080160141, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.015120540931820869, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.015120540931820869, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011666122265160084, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.960546875, "signal/confidence_uniqueness_reward/group_std_mean": 0.015439752489328384, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011666122358292342, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011666122358292342, "signal/format_reward/centered_abs_mean": 0.00074462890625, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0018734002020210027, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000372314453125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000372314453125, "signal/frontier_aurc_reward/centered_abs_mean": 0.001908647594973445, "signal/frontier_aurc_reward/group_bin_occupancy": 0.762890625, "signal/frontier_aurc_reward/group_std_mean": 0.002805905370041728, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.38580942095723e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.38580942095723e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.19497144222259521, "signal/frontier_coverage_0/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_0/group_std_mean": 0.24864284694194794, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_1/centered_abs_mean": 0.19497144222259521, "signal/frontier_coverage_1/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_1/group_std_mean": 0.24864284694194794, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_10/centered_abs_mean": 0.19497144222259521, "signal/frontier_coverage_10/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_10/group_std_mean": 0.24864284694194794, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_15/centered_abs_mean": 0.19497144222259521, "signal/frontier_coverage_15/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_15/group_std_mean": 0.24864284694194794, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_20/centered_abs_mean": 0.19497144222259521, "signal/frontier_coverage_20/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_20/group_std_mean": 0.24864284694194794, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_25/centered_abs_mean": 0.19497144222259521, "signal/frontier_coverage_25/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_25/group_std_mean": 0.24864284694194794, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_5/centered_abs_mean": 0.19497144222259521, "signal/frontier_coverage_5/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_5/group_std_mean": 0.24864284694194794, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.00243714302778244, "signal/frontier_ece_reward/centered_abs_mean": 0.03795410767197609, "signal/frontier_ece_reward/group_bin_occupancy": 0.64765625, "signal/frontier_ece_reward/group_std_mean": 0.05044463649392128, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0037954108323901893, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0037954108323901893, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2537665843963623, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.759765625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3288058042526245, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.025376660004258155, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.025376660004258155, "step": 105 }, { "calibration/aurc": 0.33301172540451696, "calibration/batch_distribution_entropy": 0.9639292106349362, "calibration/batch_entropy_100bins": 0.9571268156475009, "calibration/batch_entropy_10bins": 0.9639292106349362, "calibration/batch_entropy_50bins": 0.9647205629822991, "calibration/batch_uniqueness": 0.9494294977181352, "calibration/buffer_distribution_entropy": 0.9935574181993925, "calibration/buffer_entropy_100bins": 0.9760804708835622, "calibration/buffer_entropy_10bins": 0.9935574181993925, "calibration/buffer_entropy_50bins": 0.9859973743512296, "calibration/confidence_entropy": 0.4677467497869757, "calibration/coverage@0%": 0.025025226272015654, "calibration/coverage@1%": 0.025025226272015654, "calibration/coverage@10%": 0.10756482387475538, "calibration/coverage@15%": 0.2276036570450098, "calibration/coverage@20%": 0.36519309564579255, "calibration/coverage@25%": 0.4253952115949119, "calibration/coverage@30%": 0.4894989909491193, "calibration/coverage@5%": 0.04458323140900196, "calibration/ece": 0.12474812521435319, "calibration/mean_confidence": 0.43935241871517305, "calibration/prompt_uniqueness": 0.8580839254357441, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1111.0, "completions/max_terminated_length": 593.6, "completions/mean_length": 185.26474609375, "completions/mean_terminated_length": 184.73795471191406, "completions/min_length": 85.0, "completions/min_terminated_length": 85.0, "epoch": 0.352, "grad_norm": 0.0014269910752773285, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 364109827.0, "reward": 0.9007545948028565, "reward_std": 0.09560614228248596, "rewards/accuracy_reward": 0.46845703125, "rewards/brier_reward": 0.7802028894424439, "rewards/confidence_uniqueness_reward": 0.9526374697685241, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.0028678081929683686, "rewards/frontier_coverage_0": 0.142903470993042, "rewards/frontier_coverage_1": 0.142903470993042, "rewards/frontier_coverage_10": 0.142903470993042, "rewards/frontier_coverage_15": 0.142903470993042, "rewards/frontier_coverage_20": 0.142903470993042, "rewards/frontier_coverage_25": 0.142903470993042, "rewards/frontier_coverage_5": 0.142903470993042, "rewards/frontier_ece_reward": 0.01925478279590607, "rewards/frontier_entropy_batch_reward": -0.20956314206123353, "signal/accuracy_reward/centered_abs_mean": 0.107208251953125, "signal/accuracy_reward/group_bin_occupancy": 0.176953125, "signal/accuracy_reward/group_std_mean": 0.14296328127384186, "signal/accuracy_reward/group_zero_std_frac": 0.584375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0536041259765625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0536041259765625, "signal/advantage_abs_mean": 0.07362588495016098, "signal/advantage_pre_scale_abs_mean": 0.07362588495016098, "signal/advantage_pre_scale_std": 0.11353260725736618, "signal/advantage_std": 0.11353260725736618, "signal/brier_reward/centered_abs_mean": 0.146582955121994, "signal/brier_reward/group_bin_occupancy": 0.852734375, "signal/brier_reward/group_std_mean": 0.1880713015794754, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01465829610824585, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01465829610824585, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.013015845604240895, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.94765625, "signal/confidence_uniqueness_reward/group_std_mean": 0.0174133975058794, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001301584579050541, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001301584579050541, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086533904076, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0021364759188145397, "signal/frontier_aurc_reward/group_bin_occupancy": 0.752734375, "signal/frontier_aurc_reward/group_std_mean": 0.0032660451252013445, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.670594876690302e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.670594876690302e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.20172029435634614, "signal/frontier_coverage_0/group_bin_occupancy": 0.88828125, "signal/frontier_coverage_0/group_std_mean": 0.2565846711397171, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_1/centered_abs_mean": 0.20172029435634614, "signal/frontier_coverage_1/group_bin_occupancy": 0.88828125, "signal/frontier_coverage_1/group_std_mean": 0.2565846711397171, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_10/centered_abs_mean": 0.20172029435634614, "signal/frontier_coverage_10/group_bin_occupancy": 0.88828125, "signal/frontier_coverage_10/group_std_mean": 0.2565846711397171, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_15/centered_abs_mean": 0.20172029435634614, "signal/frontier_coverage_15/group_bin_occupancy": 0.88828125, "signal/frontier_coverage_15/group_std_mean": 0.2565846711397171, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_20/centered_abs_mean": 0.20172029435634614, "signal/frontier_coverage_20/group_bin_occupancy": 0.88828125, "signal/frontier_coverage_20/group_std_mean": 0.2565846711397171, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_25/centered_abs_mean": 0.20172029435634614, "signal/frontier_coverage_25/group_bin_occupancy": 0.88828125, "signal/frontier_coverage_25/group_std_mean": 0.2565846711397171, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_5/centered_abs_mean": 0.20172029435634614, "signal/frontier_coverage_5/group_bin_occupancy": 0.88828125, "signal/frontier_coverage_5/group_std_mean": 0.2565846711397171, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0025215038564056157, "signal/frontier_ece_reward/centered_abs_mean": 0.03271259739995003, "signal/frontier_ece_reward/group_bin_occupancy": 0.66484375, "signal/frontier_ece_reward/group_std_mean": 0.04343086704611778, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0032712597865611317, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0032712597865611317, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2817768692970276, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.741796875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3562969326972961, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.028177687153220176, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.028177687153220176, "step": 110 }, { "calibration/aurc": 0.37623539308628906, "calibration/batch_distribution_entropy": 0.9745627928478031, "calibration/batch_entropy_100bins": 0.965844786615329, "calibration/batch_entropy_10bins": 0.9745627928478031, "calibration/batch_entropy_50bins": 0.9732442947550635, "calibration/batch_uniqueness": 0.9520355224609375, "calibration/buffer_distribution_entropy": 0.997464009283586, "calibration/buffer_entropy_100bins": 0.9859008644551851, "calibration/buffer_entropy_10bins": 0.997464009283586, "calibration/buffer_entropy_50bins": 0.9925307422402309, "calibration/confidence_entropy": 0.4720331040903247, "calibration/coverage@0%": 0.008984375, "calibration/coverage@1%": 0.008984375, "calibration/coverage@10%": 0.0203125, "calibration/coverage@15%": 0.07890625, "calibration/coverage@20%": 0.253125, "calibration/coverage@25%": 0.314453125, "calibration/coverage@30%": 0.368359375, "calibration/coverage@5%": 0.016796875, "calibration/ece": 0.13170447739165136, "calibration/mean_confidence": 0.5181385939072771, "calibration/prompt_uniqueness": 0.8658203125, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0005859375, "completions/max_length": 1361.8, "completions/max_terminated_length": 566.2, "completions/mean_length": 186.66845703125, "completions/mean_terminated_length": 185.87582092285157, "completions/min_length": 89.0, "completions/min_terminated_length": 89.0, "epoch": 0.368, "grad_norm": 0.0010730287758633494, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 381086784.0, "reward": 0.912794041633606, "reward_std": 0.0936992883682251, "rewards/accuracy_reward": 0.49140625, "rewards/brier_reward": 0.7846210598945618, "rewards/confidence_uniqueness_reward": 0.9529194116592408, "rewards/format_reward": 0.99931640625, "rewards/frontier_aurc_reward": -0.0031751598697155716, "rewards/frontier_coverage_0": 0.1327473521232605, "rewards/frontier_coverage_1": 0.1327473521232605, "rewards/frontier_coverage_10": 0.1327473521232605, "rewards/frontier_coverage_15": 0.1327473521232605, "rewards/frontier_coverage_20": 0.1327473521232605, "rewards/frontier_coverage_25": 0.1327473521232605, "rewards/frontier_coverage_5": 0.1327473521232605, "rewards/frontier_ece_reward": 0.018432273901998995, "rewards/frontier_entropy_batch_reward": -0.19740256667137146, "signal/accuracy_reward/centered_abs_mean": 0.09879150390625, "signal/accuracy_reward/group_bin_occupancy": 0.175, "signal/accuracy_reward/group_std_mean": 0.13410551398992537, "signal/accuracy_reward/group_zero_std_frac": 0.6, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.049395751953125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.049395751953125, "signal/advantage_abs_mean": 0.07162269204854965, "signal/advantage_pre_scale_abs_mean": 0.07162269204854965, "signal/advantage_pre_scale_std": 0.11210574209690094, "signal/advantage_std": 0.11210574209690094, "signal/brier_reward/centered_abs_mean": 0.1440090835094452, "signal/brier_reward/group_bin_occupancy": 0.851953125, "signal/brier_reward/group_std_mean": 0.1850114107131958, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.014400908350944519, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.014400908350944519, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.013013198785483838, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.952734375, "signal/confidence_uniqueness_reward/group_std_mean": 0.018024154752492905, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0013013198971748353, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0013013198971748353, "signal/format_reward/centered_abs_mean": 0.001312255859375, "signal/format_reward/group_bin_occupancy": 0.12734375, "signal/format_reward/group_std_mean": 0.0035306816920638085, "signal/format_reward/group_zero_std_frac": 0.98125, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0006561279296875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0006561279296875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0028695956338196994, "signal/frontier_aurc_reward/group_bin_occupancy": 0.769140625, "signal/frontier_aurc_reward/group_std_mean": 0.004302969761192798, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.586994644138031e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.586994644138031e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.18317914605140687, "signal/frontier_coverage_0/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_0/group_std_mean": 0.23540749847888948, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_1/centered_abs_mean": 0.18317914605140687, "signal/frontier_coverage_1/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_1/group_std_mean": 0.23540749847888948, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_10/centered_abs_mean": 0.18317914605140687, "signal/frontier_coverage_10/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_10/group_std_mean": 0.23540749847888948, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_15/centered_abs_mean": 0.18317914605140687, "signal/frontier_coverage_15/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_15/group_std_mean": 0.23540749847888948, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_20/centered_abs_mean": 0.18317914605140687, "signal/frontier_coverage_20/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_20/group_std_mean": 0.23540749847888948, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_25/centered_abs_mean": 0.18317914605140687, "signal/frontier_coverage_25/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_25/group_std_mean": 0.23540749847888948, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_5/centered_abs_mean": 0.18317914605140687, "signal/frontier_coverage_5/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_5/group_std_mean": 0.23540749847888948, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002289739390835166, "signal/frontier_ece_reward/centered_abs_mean": 0.03010953702032566, "signal/frontier_ece_reward/group_bin_occupancy": 0.675, "signal/frontier_ece_reward/group_std_mean": 0.03895880356431007, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0030109537299722435, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0030109537299722435, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27456563115119936, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.74609375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.35324918031692504, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027456562221050262, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027456562221050262, "step": 115 }, { "calibration/aurc": 0.332168550776088, "calibration/batch_distribution_entropy": 0.9732204164523963, "calibration/batch_entropy_100bins": 0.9668091939265284, "calibration/batch_entropy_10bins": 0.9732204164523963, "calibration/batch_entropy_50bins": 0.9738037395356592, "calibration/batch_uniqueness": 0.9515895713442595, "calibration/buffer_distribution_entropy": 0.9991584257697899, "calibration/buffer_entropy_100bins": 0.9923016998857822, "calibration/buffer_entropy_10bins": 0.9991584257697899, "calibration/buffer_entropy_50bins": 0.9962901607061493, "calibration/confidence_entropy": 0.46353524848162697, "calibration/coverage@0%": 0.017205081947162425, "calibration/coverage@1%": 0.017205081947162425, "calibration/coverage@10%": 0.1830364787181996, "calibration/coverage@15%": 0.25612463307240707, "calibration/coverage@20%": 0.3045873593444227, "calibration/coverage@25%": 0.34993502323874753, "calibration/coverage@30%": 0.4457551064090019, "calibration/coverage@5%": 0.04182821673189824, "calibration/ece": 0.1418321377017973, "calibration/mean_confidence": 0.48001571976273827, "calibration/prompt_uniqueness": 0.855612897128642, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1156.8, "completions/max_terminated_length": 564.4, "completions/mean_length": 187.4455078125, "completions/mean_terminated_length": 187.04983825683593, "completions/min_length": 84.4, "completions/min_terminated_length": 84.4, "epoch": 0.384, "grad_norm": 0.001253488240763545, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 397862738.0, "reward": 0.9291020154953002, "reward_std": 0.09313084334135055, "rewards/accuracy_reward": 0.5298828125, "rewards/brier_reward": 0.787251329421997, "rewards/confidence_uniqueness_reward": 0.9518122911453247, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.003218232188373804, "rewards/frontier_coverage_0": 0.10937264189124107, "rewards/frontier_coverage_1": 0.10937264189124107, "rewards/frontier_coverage_10": 0.10937264189124107, "rewards/frontier_coverage_15": 0.10937264189124107, "rewards/frontier_coverage_20": 0.10937264189124107, "rewards/frontier_coverage_25": 0.10937264189124107, "rewards/frontier_coverage_5": 0.10937264189124107, "rewards/frontier_ece_reward": 0.01700060982257128, "rewards/frontier_entropy_batch_reward": -0.20780385434627532, "signal/accuracy_reward/centered_abs_mean": 0.1021728515625, "signal/accuracy_reward/group_bin_occupancy": 0.176171875, "signal/accuracy_reward/group_std_mean": 0.13775901049375533, "signal/accuracy_reward/group_zero_std_frac": 0.590625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.05108642578125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.05108642578125, "signal/advantage_abs_mean": 0.07189656347036362, "signal/advantage_pre_scale_abs_mean": 0.07189656347036362, "signal/advantage_pre_scale_std": 0.11328525096178055, "signal/advantage_std": 0.11328525096178055, "signal/brier_reward/centered_abs_mean": 0.1383306473493576, "signal/brier_reward/group_bin_occupancy": 0.846875, "signal/brier_reward/group_std_mean": 0.177926504611969, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.013833064958453179, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.013833064958453179, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01332956999540329, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9484375, "signal/confidence_uniqueness_reward/group_std_mean": 0.017860040813684464, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0013329570181667804, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0013329570181667804, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086533904076, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0034921379294246433, "signal/frontier_aurc_reward/group_bin_occupancy": 0.75, "signal/frontier_aurc_reward/group_std_mean": 0.005252148490399122, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.365172499092296e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.365172499092296e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17253225445747375, "signal/frontier_coverage_0/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_0/group_std_mean": 0.22278705537319182, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_1/centered_abs_mean": 0.17253225445747375, "signal/frontier_coverage_1/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_1/group_std_mean": 0.22278705537319182, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_10/centered_abs_mean": 0.17253225445747375, "signal/frontier_coverage_10/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_10/group_std_mean": 0.22278705537319182, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_15/centered_abs_mean": 0.17253225445747375, "signal/frontier_coverage_15/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_15/group_std_mean": 0.22278705537319182, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_20/centered_abs_mean": 0.17253225445747375, "signal/frontier_coverage_20/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_20/group_std_mean": 0.22278705537319182, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_25/centered_abs_mean": 0.17253225445747375, "signal/frontier_coverage_25/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_25/group_std_mean": 0.22278705537319182, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_5/centered_abs_mean": 0.17253225445747375, "signal/frontier_coverage_5/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_5/group_std_mean": 0.22278705537319182, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002156653138808906, "signal/frontier_ece_reward/centered_abs_mean": 0.02535000443458557, "signal/frontier_ece_reward/group_bin_occupancy": 0.6796875, "signal/frontier_ece_reward/group_std_mean": 0.03262428790330887, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0025350003968924286, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0025350003968924286, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27065656185150144, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.741796875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34475955963134763, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027065655589103697, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027065655589103697, "step": 120 }, { "calibration/aurc": 0.43681933728340033, "calibration/batch_distribution_entropy": 0.988045670005367, "calibration/batch_entropy_100bins": 0.9741192793515381, "calibration/batch_entropy_10bins": 0.988045670005367, "calibration/batch_entropy_50bins": 0.9833814189803034, "calibration/batch_uniqueness": 0.9547985693707745, "calibration/buffer_distribution_entropy": 0.9993214752227338, "calibration/buffer_entropy_100bins": 0.9950994140443342, "calibration/buffer_entropy_10bins": 0.9993214752227338, "calibration/buffer_entropy_50bins": 0.9976288744140552, "calibration/confidence_entropy": 0.4980310577613645, "calibration/coverage@0%": 0.003125, "calibration/coverage@1%": 0.003125, "calibration/coverage@10%": 0.00625, "calibration/coverage@15%": 0.009375, "calibration/coverage@20%": 0.032421875, "calibration/coverage@25%": 0.124609375, "calibration/coverage@30%": 0.18515625, "calibration/coverage@5%": 0.003125, "calibration/ece": 0.16192984329388163, "calibration/mean_confidence": 0.5234467935428648, "calibration/prompt_uniqueness": 0.8683541924102498, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 911.8, "completions/max_terminated_length": 543.6, "completions/mean_length": 192.1666015625, "completions/mean_terminated_length": 191.77420043945312, "completions/min_length": 77.8, "completions/min_terminated_length": 77.8, "epoch": 0.4, "grad_norm": 0.0013689384795725346, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 414866972.0, "reward": 0.9105149269104004, "reward_std": 0.09931065142154694, "rewards/accuracy_reward": 0.49443359375, "rewards/brier_reward": 0.7720793724060059, "rewards/confidence_uniqueness_reward": 0.952759063243866, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.003909049881622195, "rewards/frontier_coverage_0": 0.11478218827396632, "rewards/frontier_coverage_1": 0.11478218827396632, "rewards/frontier_coverage_10": 0.11478218827396632, "rewards/frontier_coverage_15": 0.11478218827396632, "rewards/frontier_coverage_20": 0.11478218827396632, "rewards/frontier_coverage_25": 0.11478218827396632, "rewards/frontier_coverage_5": 0.11478218827396632, "rewards/frontier_ece_reward": 0.012619380280375481, "rewards/frontier_entropy_batch_reward": -0.20246887803077698, "signal/accuracy_reward/centered_abs_mean": 0.115435791015625, "signal/accuracy_reward/group_bin_occupancy": 0.179296875, "signal/accuracy_reward/group_std_mean": 0.15186140239238738, "signal/accuracy_reward/group_zero_std_frac": 0.565625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0577178955078125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0577178955078125, "signal/advantage_abs_mean": 0.07725874483585357, "signal/advantage_pre_scale_abs_mean": 0.07725874483585357, "signal/advantage_pre_scale_std": 0.11946925520896912, "signal/advantage_std": 0.11946925520896912, "signal/brier_reward/centered_abs_mean": 0.1468850463628769, "signal/brier_reward/group_bin_occupancy": 0.855859375, "signal/brier_reward/group_std_mean": 0.18767115473747253, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.014688505418598652, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.014688505418598652, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01240917332470417, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.951171875, "signal/confidence_uniqueness_reward/group_std_mean": 0.01641183942556381, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012409173417836429, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012409173417836429, "signal/format_reward/centered_abs_mean": 0.00074462890625, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0018734002020210027, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000372314453125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000372314453125, "signal/frontier_aurc_reward/centered_abs_mean": 0.004115176573395729, "signal/frontier_aurc_reward/group_bin_occupancy": 0.71484375, "signal/frontier_aurc_reward/group_std_mean": 0.0066203194670379165, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 5.143971065990627e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 5.143971065990627e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17863820791244506, "signal/frontier_coverage_0/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_0/group_std_mean": 0.2312378168106079, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_1/centered_abs_mean": 0.17863820791244506, "signal/frontier_coverage_1/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_1/group_std_mean": 0.2312378168106079, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_10/centered_abs_mean": 0.17863820791244506, "signal/frontier_coverage_10/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_10/group_std_mean": 0.2312378168106079, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_15/centered_abs_mean": 0.17863820791244506, "signal/frontier_coverage_15/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_15/group_std_mean": 0.2312378168106079, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_20/centered_abs_mean": 0.17863820791244506, "signal/frontier_coverage_20/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_20/group_std_mean": 0.2312378168106079, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_25/centered_abs_mean": 0.17863820791244506, "signal/frontier_coverage_25/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_25/group_std_mean": 0.2312378168106079, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_5/centered_abs_mean": 0.17863820791244506, "signal/frontier_coverage_5/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_5/group_std_mean": 0.2312378168106079, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0022329777013510464, "signal/frontier_ece_reward/centered_abs_mean": 0.02163628302514553, "signal/frontier_ece_reward/group_bin_occupancy": 0.691796875, "signal/frontier_ece_reward/group_std_mean": 0.027761366963386536, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0021636283956468104, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0021636283956468104, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2772179365158081, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.74296875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.35126439929008485, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027721792832016943, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027721792832016943, "step": 125 }, { "calibration/aurc": 0.3366690998831715, "calibration/batch_distribution_entropy": 0.978672238861005, "calibration/batch_entropy_100bins": 0.9684483719611915, "calibration/batch_entropy_10bins": 0.978672238861005, "calibration/batch_entropy_50bins": 0.9755411950876463, "calibration/batch_uniqueness": 0.953064274002753, "calibration/buffer_distribution_entropy": 0.9991570425982642, "calibration/buffer_entropy_100bins": 0.9958530326193508, "calibration/buffer_entropy_10bins": 0.9991570425982642, "calibration/buffer_entropy_50bins": 0.9978587931070985, "calibration/confidence_entropy": 0.5176350071083632, "calibration/coverage@0%": 0.003125764432485323, "calibration/coverage@1%": 0.003125764432485323, "calibration/coverage@10%": 0.019922639432485324, "calibration/coverage@15%": 0.05390701443248532, "calibration/coverage@20%": 0.1269538894324853, "calibration/coverage@25%": 0.19961013943248532, "calibration/coverage@30%": 0.3175788894324853, "calibration/coverage@5%": 0.003125764432485323, "calibration/ece": 0.0972213917075034, "calibration/mean_confidence": 0.49765389477515853, "calibration/prompt_uniqueness": 0.8708848451139583, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00048828125, "completions/max_length": 1388.0, "completions/max_terminated_length": 564.4, "completions/mean_length": 196.0033203125, "completions/mean_terminated_length": 195.34762878417968, "completions/min_length": 76.4, "completions/min_terminated_length": 76.4, "epoch": 0.416, "grad_norm": 0.0009428579360246658, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 431755230.0, "reward": 0.9199765920639038, "reward_std": 0.09755838513374329, "rewards/accuracy_reward": 0.508984375, "rewards/brier_reward": 0.7872162818908691, "rewards/confidence_uniqueness_reward": 0.952919864654541, "rewards/format_reward": 0.9994140625, "rewards/frontier_aurc_reward": -0.002840971061959863, "rewards/frontier_coverage_0": 0.11873868256807327, "rewards/frontier_coverage_1": 0.11873868256807327, "rewards/frontier_coverage_10": 0.11873868256807327, "rewards/frontier_coverage_15": 0.11873868256807327, "rewards/frontier_coverage_20": 0.11873868256807327, "rewards/frontier_coverage_25": 0.11869797259569168, "rewards/frontier_coverage_5": 0.11873868256807327, "rewards/frontier_ece_reward": 0.011243033222854137, "rewards/frontier_entropy_batch_reward": -0.1971413940191269, "signal/accuracy_reward/centered_abs_mean": 0.11490478515625, "signal/accuracy_reward/group_bin_occupancy": 0.176953125, "signal/accuracy_reward/group_std_mean": 0.14871509969234467, "signal/accuracy_reward/group_zero_std_frac": 0.584375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.057452392578125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.057452392578125, "signal/advantage_abs_mean": 0.07633275091648102, "signal/advantage_pre_scale_abs_mean": 0.07633275091648102, "signal/advantage_pre_scale_std": 0.11720214486122131, "signal/advantage_std": 0.11720214486122131, "signal/brier_reward/centered_abs_mean": 0.13782235085964203, "signal/brier_reward/group_bin_occupancy": 0.858203125, "signal/brier_reward/group_std_mean": 0.17689369916915892, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01378223542124033, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01378223542124033, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012818338163197041, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.947265625, "signal/confidence_uniqueness_reward/group_std_mean": 0.017543331906199455, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.00128183385822922, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.00128183385822922, "signal/format_reward/centered_abs_mean": 0.001123046875, "signal/format_reward/group_bin_occupancy": 0.126953125, "signal/format_reward/group_std_mean": 0.0029782545287162067, "signal/format_reward/group_zero_std_frac": 0.984375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0005615234375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0005615234375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0030377992894500493, "signal/frontier_aurc_reward/group_bin_occupancy": 0.691015625, "signal/frontier_aurc_reward/group_std_mean": 0.005163288209587336, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.7972491554683074e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.7972491554683074e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.19069285094738006, "signal/frontier_coverage_0/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_0/group_std_mean": 0.24125858545303344, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_1/centered_abs_mean": 0.19069285094738006, "signal/frontier_coverage_1/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_1/group_std_mean": 0.24125858545303344, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_10/centered_abs_mean": 0.19069285094738006, "signal/frontier_coverage_10/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_10/group_std_mean": 0.24125858545303344, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_15/centered_abs_mean": 0.19069285094738006, "signal/frontier_coverage_15/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_15/group_std_mean": 0.24125858545303344, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_20/centered_abs_mean": 0.19069285094738006, "signal/frontier_coverage_20/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_20/group_std_mean": 0.24125858545303344, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_25/centered_abs_mean": 0.19041914045810698, "signal/frontier_coverage_25/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_25/group_std_mean": 0.24093502759933472, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0023802392184734344, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0023802392184734344, "signal/frontier_coverage_5/centered_abs_mean": 0.19069285094738006, "signal/frontier_coverage_5/group_bin_occupancy": 0.89140625, "signal/frontier_coverage_5/group_std_mean": 0.24125858545303344, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0023836606182157994, "signal/frontier_ece_reward/centered_abs_mean": 0.016844046488404273, "signal/frontier_ece_reward/group_bin_occupancy": 0.682421875, "signal/frontier_ece_reward/group_std_mean": 0.021952753886580467, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.001684404700063169, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.001684404700063169, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2732956767082214, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.748828125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3495099425315857, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027329567074775695, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027329567074775695, "step": 130 }, { "calibration/aurc": 0.2643433620623352, "calibration/batch_distribution_entropy": 0.9868445927370566, "calibration/batch_entropy_100bins": 0.971041103150353, "calibration/batch_entropy_10bins": 0.9868445927370566, "calibration/batch_entropy_50bins": 0.9808039028810509, "calibration/batch_uniqueness": 0.9541412353515625, "calibration/buffer_distribution_entropy": 0.9992612840827857, "calibration/buffer_entropy_100bins": 0.9963089662826825, "calibration/buffer_entropy_10bins": 0.9992612840827857, "calibration/buffer_entropy_50bins": 0.9980831630936541, "calibration/confidence_entropy": 0.48090818324843704, "calibration/coverage@0%": 0.0171875, "calibration/coverage@1%": 0.0171875, "calibration/coverage@10%": 0.1796875, "calibration/coverage@15%": 0.28671875, "calibration/coverage@20%": 0.36796875, "calibration/coverage@25%": 0.437890625, "calibration/coverage@30%": 0.5328125, "calibration/coverage@5%": 0.07890625, "calibration/ece": 0.11728917089115898, "calibration/mean_confidence": 0.5129007686611773, "calibration/prompt_uniqueness": 0.84853515625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1127.2, "completions/max_terminated_length": 756.2, "completions/mean_length": 195.61923828125, "completions/mean_terminated_length": 195.22649230957032, "completions/min_length": 96.4, "completions/min_terminated_length": 96.4, "epoch": 0.432, "grad_norm": 0.00106110330671072, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 448772707.0, "reward": 0.9428177952766419, "reward_std": 0.08899335712194442, "rewards/accuracy_reward": 0.5517578125, "rewards/brier_reward": 0.7984528422355652, "rewards/confidence_uniqueness_reward": 0.9533126950263977, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.002184620266780257, "rewards/frontier_coverage_0": 0.10767800435423851, "rewards/frontier_coverage_1": 0.10767800435423851, "rewards/frontier_coverage_10": 0.10767800435423851, "rewards/frontier_coverage_15": 0.10767800435423851, "rewards/frontier_coverage_20": 0.10767800435423851, "rewards/frontier_coverage_25": 0.10708100348711014, "rewards/frontier_coverage_5": 0.10767800435423851, "rewards/frontier_ece_reward": 0.011894653365015984, "rewards/frontier_entropy_batch_reward": -0.18618870973587037, "signal/accuracy_reward/centered_abs_mean": 0.1056640625, "signal/accuracy_reward/group_bin_occupancy": 0.173828125, "signal/accuracy_reward/group_std_mean": 0.13766344636678696, "signal/accuracy_reward/group_zero_std_frac": 0.609375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.05283203125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.05283203125, "signal/advantage_abs_mean": 0.06958579570055008, "signal/advantage_pre_scale_abs_mean": 0.06958579570055008, "signal/advantage_pre_scale_std": 0.10905162990093231, "signal/advantage_std": 0.10905162990093231, "signal/brier_reward/centered_abs_mean": 0.12985509037971496, "signal/brier_reward/group_bin_occupancy": 0.840234375, "signal/brier_reward/group_std_mean": 0.1670115441083908, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01298550982028246, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01298550982028246, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.0121300233528018, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.956640625, "signal/confidence_uniqueness_reward/group_std_mean": 0.016346600838005543, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012130024144425989, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012130024144425989, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086533904076, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.00269233169965446, "signal/frontier_aurc_reward/group_bin_occupancy": 0.7046875, "signal/frontier_aurc_reward/group_std_mean": 0.004477695794776082, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.3654147409833966e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.3654147409833966e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.1841978460550308, "signal/frontier_coverage_0/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_0/group_std_mean": 0.23535178899765014, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_1/centered_abs_mean": 0.1841978460550308, "signal/frontier_coverage_1/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_1/group_std_mean": 0.23535178899765014, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_10/centered_abs_mean": 0.1841978460550308, "signal/frontier_coverage_10/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_10/group_std_mean": 0.23535178899765014, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_15/centered_abs_mean": 0.1841978460550308, "signal/frontier_coverage_15/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_15/group_std_mean": 0.23535178899765014, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_20/centered_abs_mean": 0.1841978460550308, "signal/frontier_coverage_20/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_20/group_std_mean": 0.23535178899765014, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_25/centered_abs_mean": 0.18254570066928863, "signal/frontier_coverage_25/group_bin_occupancy": 0.8796875, "signal/frontier_coverage_25/group_std_mean": 0.23327820003032684, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0022818212863057854, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0022818212863057854, "signal/frontier_coverage_5/centered_abs_mean": 0.1841978460550308, "signal/frontier_coverage_5/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_5/group_std_mean": 0.23535178899765014, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.00230247313156724, "signal/frontier_ece_reward/centered_abs_mean": 0.015106850303709507, "signal/frontier_ece_reward/group_bin_occupancy": 0.650390625, "signal/frontier_ece_reward/group_std_mean": 0.01951742023229599, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0015106850536540152, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0015106850536540152, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2573384702205658, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.740625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3308428943157196, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02573384717106819, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02573384717106819, "step": 135 }, { "calibration/aurc": 0.29552044099527325, "calibration/batch_distribution_entropy": 0.9754323678807462, "calibration/batch_entropy_100bins": 0.9681519824987224, "calibration/batch_entropy_10bins": 0.9754323678807462, "calibration/batch_entropy_50bins": 0.9759841993085555, "calibration/batch_uniqueness": 0.9530426025390625, "calibration/buffer_distribution_entropy": 0.9990916881733458, "calibration/buffer_entropy_100bins": 0.9965307544028359, "calibration/buffer_entropy_10bins": 0.9990916881733458, "calibration/buffer_entropy_50bins": 0.9980794004963804, "calibration/confidence_entropy": 0.5055189308277404, "calibration/coverage@0%": 0.008203125, "calibration/coverage@1%": 0.008203125, "calibration/coverage@10%": 0.091796875, "calibration/coverage@15%": 0.1484375, "calibration/coverage@20%": 0.193359375, "calibration/coverage@25%": 0.258203125, "calibration/coverage@30%": 0.452734375, "calibration/coverage@5%": 0.0203125, "calibration/ece": 0.1189624627585604, "calibration/mean_confidence": 0.5567726011112513, "calibration/prompt_uniqueness": 0.86884765625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 667.6, "completions/max_terminated_length": 667.6, "completions/mean_length": 201.92978515625, "completions/mean_terminated_length": 201.92978515625, "completions/min_length": 94.2, "completions/min_terminated_length": 94.2, "epoch": 0.448, "grad_norm": 0.0010867634555324912, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 465793268.0, "reward": 0.9269709944725036, "reward_std": 0.09072813540697097, "rewards/accuracy_reward": 0.52060546875, "rewards/brier_reward": 0.795162844657898, "rewards/confidence_uniqueness_reward": 0.9530165433883667, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.0030272399075329305, "rewards/frontier_coverage_0": 0.12207410931587219, "rewards/frontier_coverage_1": 0.12207410931587219, "rewards/frontier_coverage_10": 0.12207410931587219, "rewards/frontier_coverage_15": 0.12207410931587219, "rewards/frontier_coverage_20": 0.12203729003667832, "rewards/frontier_coverage_25": 0.12047196030616761, "rewards/frontier_coverage_5": 0.12207410931587219, "rewards/frontier_ece_reward": 0.01023321207612753, "rewards/frontier_entropy_batch_reward": -0.19698509871959685, "signal/accuracy_reward/centered_abs_mean": 0.102618408203125, "signal/accuracy_reward/group_bin_occupancy": 0.173046875, "signal/accuracy_reward/group_std_mean": 0.13459138572216034, "signal/accuracy_reward/group_zero_std_frac": 0.615625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0513092041015625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0513092041015625, "signal/advantage_abs_mean": 0.0711361289024353, "signal/advantage_pre_scale_abs_mean": 0.0711361289024353, "signal/advantage_pre_scale_std": 0.11110936403274536, "signal/advantage_std": 0.11110936403274536, "signal/brier_reward/centered_abs_mean": 0.1334667831659317, "signal/brier_reward/group_bin_occupancy": 0.851171875, "signal/brier_reward/group_std_mean": 0.17104584276676177, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.013346678577363491, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.013346678577363491, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012166332639753818, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9578125, "signal/confidence_uniqueness_reward/group_std_mean": 0.015520927682518958, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012166333151981234, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012166333151981234, "signal/format_reward/centered_abs_mean": 0.0003662109375, "signal/format_reward/group_bin_occupancy": 0.125390625, "signal/format_reward/group_std_mean": 0.000768545875325799, "signal/format_reward/group_zero_std_frac": 0.996875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00018310546875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00018310546875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0035023584961891173, "signal/frontier_aurc_reward/group_bin_occupancy": 0.694921875, "signal/frontier_aurc_reward/group_std_mean": 0.00593244144693017, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.377948280307464e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.377948280307464e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17804237306118012, "signal/frontier_coverage_0/group_bin_occupancy": 0.881640625, "signal/frontier_coverage_0/group_std_mean": 0.22660138607025146, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_1/centered_abs_mean": 0.17804237306118012, "signal/frontier_coverage_1/group_bin_occupancy": 0.881640625, "signal/frontier_coverage_1/group_std_mean": 0.22660138607025146, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_10/centered_abs_mean": 0.17804237306118012, "signal/frontier_coverage_10/group_bin_occupancy": 0.881640625, "signal/frontier_coverage_10/group_std_mean": 0.22660138607025146, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_15/centered_abs_mean": 0.17804237306118012, "signal/frontier_coverage_15/group_bin_occupancy": 0.881640625, "signal/frontier_coverage_15/group_std_mean": 0.22660138607025146, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_20/centered_abs_mean": 0.1778087705373764, "signal/frontier_coverage_20/group_bin_occupancy": 0.881640625, "signal/frontier_coverage_20/group_std_mean": 0.2263088881969452, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0022226096130907535, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0022226096130907535, "signal/frontier_coverage_25/centered_abs_mean": 0.17493513822555543, "signal/frontier_coverage_25/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_25/group_std_mean": 0.222748264670372, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0021866892231628297, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0021866892231628297, "signal/frontier_coverage_5/centered_abs_mean": 0.17804237306118012, "signal/frontier_coverage_5/group_bin_occupancy": 0.881640625, "signal/frontier_coverage_5/group_std_mean": 0.22660138607025146, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002225529635325074, "signal/frontier_ece_reward/centered_abs_mean": 0.01471986025571823, "signal/frontier_ece_reward/group_bin_occupancy": 0.6609375, "signal/frontier_ece_reward/group_std_mean": 0.01903526335954666, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0014719860861077904, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0014719860861077904, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27238709926605226, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.739453125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3480784296989441, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02723870985209942, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02723870985209942, "step": 140 }, { "calibration/aurc": 0.4201061174765079, "calibration/batch_distribution_entropy": 0.9868428061888945, "calibration/batch_entropy_100bins": 0.9744967626271691, "calibration/batch_entropy_10bins": 0.9868428061888945, "calibration/batch_entropy_50bins": 0.9829525268019108, "calibration/batch_uniqueness": 0.9549041748046875, "calibration/buffer_distribution_entropy": 0.9988032965826472, "calibration/buffer_entropy_100bins": 0.9967418936261814, "calibration/buffer_entropy_10bins": 0.9988032965826472, "calibration/buffer_entropy_50bins": 0.9980709014706515, "calibration/confidence_entropy": 0.5102673801505668, "calibration/coverage@0%": 0.003515625, "calibration/coverage@1%": 0.003515625, "calibration/coverage@10%": 0.00703125, "calibration/coverage@15%": 0.00703125, "calibration/coverage@20%": 0.099609375, "calibration/coverage@25%": 0.1390625, "calibration/coverage@30%": 0.2421875, "calibration/coverage@5%": 0.003515625, "calibration/ece": 0.12862265896128908, "calibration/mean_confidence": 0.5089816658825093, "calibration/prompt_uniqueness": 0.865234375, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 701.8, "completions/max_terminated_length": 507.8, "completions/mean_length": 204.779296875, "completions/mean_terminated_length": 204.64966735839843, "completions/min_length": 97.8, "completions/min_terminated_length": 97.8, "epoch": 0.464, "grad_norm": 0.0014885164564475417, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 483061024.0, "reward": 0.8959267139434814, "reward_std": 0.08457860499620437, "rewards/accuracy_reward": 0.45703125, "rewards/brier_reward": 0.7741384506225586, "rewards/confidence_uniqueness_reward": 0.9536405801773071, "rewards/format_reward": 0.99990234375, "rewards/frontier_aurc_reward": -0.003999204328283668, "rewards/frontier_coverage_0": 0.1452971801161766, "rewards/frontier_coverage_1": 0.1452971801161766, "rewards/frontier_coverage_10": 0.1452971801161766, "rewards/frontier_coverage_15": 0.1452971801161766, "rewards/frontier_coverage_20": 0.1444725900888443, "rewards/frontier_coverage_25": 0.14294103980064393, "rewards/frontier_coverage_5": 0.1452971801161766, "rewards/frontier_ece_reward": 0.007464424055069685, "rewards/frontier_entropy_batch_reward": -0.1868817090988159, "signal/accuracy_reward/centered_abs_mean": 0.0845947265625, "signal/accuracy_reward/group_bin_occupancy": 0.16875, "signal/accuracy_reward/group_std_mean": 0.11651781499385834, "signal/accuracy_reward/group_zero_std_frac": 0.65, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.04229736328125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.04229736328125, "signal/advantage_abs_mean": 0.06507015377283096, "signal/advantage_pre_scale_abs_mean": 0.06507015377283096, "signal/advantage_pre_scale_std": 0.10322362333536148, "signal/advantage_std": 0.10322362333536148, "signal/brier_reward/centered_abs_mean": 0.13624679744243623, "signal/brier_reward/group_bin_occupancy": 0.855859375, "signal/brier_reward/group_std_mean": 0.17391125857830048, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01362467985600233, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01362467985600233, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011658880673348904, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.958203125, "signal/confidence_uniqueness_reward/group_std_mean": 0.014889470301568508, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001165888085961342, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001165888085961342, "signal/format_reward/centered_abs_mean": 0.000189208984375, "signal/format_reward/group_bin_occupancy": 0.125390625, "signal/format_reward/group_std_mean": 0.0005524271633476019, "signal/format_reward/group_zero_std_frac": 0.996875, "signal/format_reward/scaled_weighted_centered_abs_mean": 9.46044921875e-05, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 9.46044921875e-05, "signal/frontier_aurc_reward/centered_abs_mean": 0.0036771357990801334, "signal/frontier_aurc_reward/group_bin_occupancy": 0.691796875, "signal/frontier_aurc_reward/group_std_mean": 0.006164277251809835, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.596419385052286e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.596419385052286e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.171611425280571, "signal/frontier_coverage_0/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_0/group_std_mean": 0.219672092795372, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_1/centered_abs_mean": 0.171611425280571, "signal/frontier_coverage_1/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_1/group_std_mean": 0.219672092795372, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_10/centered_abs_mean": 0.171611425280571, "signal/frontier_coverage_10/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_10/group_std_mean": 0.219672092795372, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_15/centered_abs_mean": 0.171611425280571, "signal/frontier_coverage_15/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_15/group_std_mean": 0.219672092795372, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_20/centered_abs_mean": 0.17026718258857726, "signal/frontier_coverage_20/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_20/group_std_mean": 0.21796101033687593, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002128339884802699, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002128339884802699, "signal/frontier_coverage_25/centered_abs_mean": 0.16843552887439728, "signal/frontier_coverage_25/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_25/group_std_mean": 0.21566588282585145, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0021054442040622234, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0021054442040622234, "signal/frontier_coverage_5/centered_abs_mean": 0.171611425280571, "signal/frontier_coverage_5/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_5/group_std_mean": 0.219672092795372, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0021451428066939114, "signal/frontier_ece_reward/centered_abs_mean": 0.01299218013882637, "signal/frontier_ece_reward/group_bin_occupancy": 0.65, "signal/frontier_ece_reward/group_std_mean": 0.01686352901160717, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0012992180651053785, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0012992180651053785, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.259442126750946, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.728515625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.33634223937988283, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.025944213569164275, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.025944213569164275, "step": 145 }, { "calibration/aurc": 0.2736750914975225, "calibration/batch_distribution_entropy": 0.979895433797069, "calibration/batch_entropy_100bins": 0.9683092226635296, "calibration/batch_entropy_10bins": 0.979895433797069, "calibration/batch_entropy_50bins": 0.976651207232614, "calibration/batch_uniqueness": 0.9530262916969872, "calibration/buffer_distribution_entropy": 0.9986458433631796, "calibration/buffer_entropy_100bins": 0.9970068023524898, "calibration/buffer_entropy_10bins": 0.9986458433631796, "calibration/buffer_entropy_50bins": 0.998109788604495, "calibration/confidence_entropy": 0.47700435559482424, "calibration/coverage@0%": 0.01446076932485323, "calibration/coverage@1%": 0.01446076932485323, "calibration/coverage@10%": 0.09735277030332681, "calibration/coverage@15%": 0.26653085249510766, "calibration/coverage@20%": 0.3611072040117417, "calibration/coverage@25%": 0.4451190985812133, "calibration/coverage@30%": 0.5553189212328766, "calibration/coverage@5%": 0.01446076932485323, "calibration/ece": 0.13376888797207984, "calibration/mean_confidence": 0.5207233986198011, "calibration/prompt_uniqueness": 0.8551470737675599, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0005859375, "completions/max_length": 974.6, "completions/max_terminated_length": 596.2, "completions/mean_length": 203.0658203125, "completions/mean_terminated_length": 202.28616333007812, "completions/min_length": 97.6, "completions/min_terminated_length": 97.6, "epoch": 0.48, "grad_norm": 0.0009758519590832293, "learning_rate": 1e-06, "loss": 0.0018, "num_tokens": 500188450.0, "reward": 0.9241597652435303, "reward_std": 0.09158220142126083, "rewards/accuracy_reward": 0.52060546875, "rewards/brier_reward": 0.7813067078590393, "rewards/confidence_uniqueness_reward": 0.9521458625793457, "rewards/format_reward": 0.99931640625, "rewards/frontier_aurc_reward": -0.0033276322763413193, "rewards/frontier_coverage_0": 0.11744775027036666, "rewards/frontier_coverage_1": 0.11744775027036666, "rewards/frontier_coverage_10": 0.11744775027036666, "rewards/frontier_coverage_15": 0.11742079854011536, "rewards/frontier_coverage_20": 0.11692409068346024, "rewards/frontier_coverage_25": 0.11456361413002014, "rewards/frontier_coverage_5": 0.11744775027036666, "rewards/frontier_ece_reward": 0.008499128930270671, "rewards/frontier_entropy_batch_reward": -0.2018849402666092, "signal/accuracy_reward/centered_abs_mean": 0.111077880859375, "signal/accuracy_reward/group_bin_occupancy": 0.1765625, "signal/accuracy_reward/group_std_mean": 0.14600239396095277, "signal/accuracy_reward/group_zero_std_frac": 0.5875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0555389404296875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0555389404296875, "signal/advantage_abs_mean": 0.07034167349338531, "signal/advantage_pre_scale_abs_mean": 0.07034167349338531, "signal/advantage_pre_scale_std": 0.11202311217784881, "signal/advantage_std": 0.11202311217784881, "signal/brier_reward/centered_abs_mean": 0.13615815937519074, "signal/brier_reward/group_bin_occupancy": 0.8421875, "signal/brier_reward/group_std_mean": 0.17466709315776824, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.013615815714001656, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.013615815714001656, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012963654100894928, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.940625, "signal/confidence_uniqueness_reward/group_std_mean": 0.01802794709801674, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012963654939085246, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012963654939085246, "signal/format_reward/centered_abs_mean": 0.001312255859375, "signal/format_reward/group_bin_occupancy": 0.12734375, "signal/format_reward/group_std_mean": 0.003530681785196066, "signal/format_reward/group_zero_std_frac": 0.98125, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0006561279296875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0006561279296875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0036876605823636054, "signal/frontier_aurc_reward/group_bin_occupancy": 0.72890625, "signal/frontier_aurc_reward/group_std_mean": 0.0060782676562666895, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.609575698850676e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.609575698850676e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.18423805236816407, "signal/frontier_coverage_0/group_bin_occupancy": 0.8625, "signal/frontier_coverage_0/group_std_mean": 0.23684498071670532, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_coverage_1/centered_abs_mean": 0.18423805236816407, "signal/frontier_coverage_1/group_bin_occupancy": 0.8625, "signal/frontier_coverage_1/group_std_mean": 0.23684498071670532, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_coverage_10/centered_abs_mean": 0.18423805236816407, "signal/frontier_coverage_10/group_bin_occupancy": 0.8625, "signal/frontier_coverage_10/group_std_mean": 0.23684498071670532, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_coverage_15/centered_abs_mean": 0.18408388793468475, "signal/frontier_coverage_15/group_bin_occupancy": 0.862109375, "signal/frontier_coverage_15/group_std_mean": 0.2366536408662796, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002301048533990979, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002301048533990979, "signal/frontier_coverage_20/centered_abs_mean": 0.1824486345052719, "signal/frontier_coverage_20/group_bin_occupancy": 0.86328125, "signal/frontier_coverage_20/group_std_mean": 0.2346276342868805, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002280608005821705, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002280608005821705, "signal/frontier_coverage_25/centered_abs_mean": 0.17726730406284333, "signal/frontier_coverage_25/group_bin_occupancy": 0.862109375, "signal/frontier_coverage_25/group_std_mean": 0.22807096540927888, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.002215841319411993, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.002215841319411993, "signal/frontier_coverage_5/centered_abs_mean": 0.18423805236816407, "signal/frontier_coverage_5/group_bin_occupancy": 0.8625, "signal/frontier_coverage_5/group_std_mean": 0.23684498071670532, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0023029757663607596, "signal/frontier_ece_reward/centered_abs_mean": 0.01261788047850132, "signal/frontier_ece_reward/group_bin_occupancy": 0.65234375, "signal/frontier_ece_reward/group_std_mean": 0.01614722013473511, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0012617880711331964, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0012617880711331964, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26360839009284975, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.73359375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3400068938732147, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02636084109544754, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02636084109544754, "step": 150 }, { "epoch": 0.48, "eval_calibration/aurc": 0.526235934319538, "eval_calibration/batch_distribution_entropy": 0.9093259472487039, "eval_calibration/batch_entropy_100bins": 0.6925811808869712, "eval_calibration/batch_entropy_10bins": 0.9093259472487039, "eval_calibration/batch_entropy_50bins": 0.779983490066886, "eval_calibration/batch_uniqueness": 0.8955078125, "eval_calibration/buffer_distribution_entropy": 0.9984767511025705, "eval_calibration/buffer_entropy_100bins": 0.9971319443431579, "eval_calibration/buffer_entropy_10bins": 0.9984767511025705, "eval_calibration/buffer_entropy_50bins": 0.9981330737123422, "eval_calibration/confidence_entropy": 0.4570937384077164, "eval_calibration/coverage@0%": 0.0390625, "eval_calibration/coverage@1%": 0.0390625, "eval_calibration/coverage@10%": 0.0390625, "eval_calibration/coverage@15%": 0.0390625, "eval_calibration/coverage@20%": 0.1640625, "eval_calibration/coverage@25%": 0.171875, "eval_calibration/coverage@30%": 0.25, "eval_calibration/coverage@5%": 0.0390625, "eval_calibration/ece": 0.23224618837958527, "eval_calibration/mean_confidence": 0.4527100395085424, "eval_calibration/prompt_uniqueness": 0.8955078125, "eval_completions/clipped_ratio": 0.001953125, "eval_completions/max_length": 655.75, "eval_completions/max_terminated_length": 374.25, "eval_completions/mean_length": 203.42618560791016, "eval_completions/mean_terminated_length": 200.80736923217773, "eval_completions/min_length": 106.75, "eval_completions/min_terminated_length": 106.75, "eval_loss": 0.0, "eval_num_tokens": 500188450.0, "eval_reward": 0.7950556129217148, "eval_reward_std": 0.22721097245812416, "eval_rewards/accuracy_reward": 0.423828125, "eval_rewards/brier_reward": 0.7800905853509903, "eval_rewards/confidence_uniqueness_reward": 0.8937279582023621, "eval_rewards/format_reward": 0.998046875, "eval_rewards/frontier_aurc_reward": -0.004944973974488676, "eval_rewards/frontier_coverage_0": 0.18303139507770538, "eval_rewards/frontier_coverage_1": 0.18303139507770538, "eval_rewards/frontier_coverage_10": 0.18297437578439713, "eval_rewards/frontier_coverage_15": 0.1826879307627678, "eval_rewards/frontier_coverage_20": 0.18085048720240593, "eval_rewards/frontier_coverage_25": 0.17026891186833382, "eval_rewards/frontier_coverage_5": 0.18303139507770538, "eval_rewards/frontier_ece_reward": 0.007793309981934726, "eval_rewards/frontier_entropy_batch_reward": -0.998046875, "eval_runtime": 28.4158, "eval_samples_per_second": 17.596, "eval_signal/accuracy_reward/centered_abs_mean": 0.4681396484375, "eval_signal/accuracy_reward/group_bin_occupancy": 0.25, "eval_signal/accuracy_reward/group_std_mean": 0.4908996820449829, "eval_signal/accuracy_reward/group_zero_std_frac": 0.0, "eval_signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.23406982421875, "eval_signal/accuracy_reward/weight": 0.5, "eval_signal/accuracy_reward/weighted_centered_abs_mean": 0.23406982421875, "eval_signal/advantage_abs_mean": 0.2096048854291439, "eval_signal/advantage_pre_scale_abs_mean": 0.2096048854291439, "eval_signal/advantage_pre_scale_std": 0.22496796399354935, "eval_signal/advantage_std": 0.22496796399354935, "eval_signal/brier_reward/centered_abs_mean": 0.21101701259613037, "eval_signal/brier_reward/group_bin_occupancy": 0.875, "eval_signal/brier_reward/group_std_mean": 0.26776499301195145, "eval_signal/brier_reward/group_zero_std_frac": 0.0, "eval_signal/brier_reward/scaled_weighted_centered_abs_mean": 0.02110170153900981, "eval_signal/brier_reward/weight": 0.10000000149011612, "eval_signal/brier_reward/weighted_centered_abs_mean": 0.02110170153900981, "eval_signal/confidence_uniqueness_reward/centered_abs_mean": 0.0438153650611639, "eval_signal/confidence_uniqueness_reward/group_bin_occupancy": 0.3984375, "eval_signal/confidence_uniqueness_reward/group_std_mean": 0.05629879981279373, "eval_signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "eval_signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.004381536622531712, "eval_signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "eval_signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.004381536622531712, "eval_signal/format_reward/centered_abs_mean": 0.0037841796875, "eval_signal/format_reward/group_bin_occupancy": 0.1328125, "eval_signal/format_reward/group_std_mean": 0.011048543266952038, "eval_signal/format_reward/group_zero_std_frac": 0.9375, "eval_signal/format_reward/scaled_weighted_centered_abs_mean": 0.00189208984375, "eval_signal/format_reward/weight": 0.5, "eval_signal/format_reward/weighted_centered_abs_mean": 0.00189208984375, "eval_signal/frontier_aurc_reward/centered_abs_mean": 0.0070755783235654235, "eval_signal/frontier_aurc_reward/group_bin_occupancy": 0.640625, "eval_signal/frontier_aurc_reward/group_std_mean": 0.01427709311246872, "eval_signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 8.84447326825466e-05, "eval_signal/frontier_aurc_reward/weight": 0.012500000186264515, "eval_signal/frontier_aurc_reward/weighted_centered_abs_mean": 8.84447326825466e-05, "eval_signal/frontier_coverage_0/centered_abs_mean": 0.35811903327703476, "eval_signal/frontier_coverage_0/group_bin_occupancy": 0.9453125, "eval_signal/frontier_coverage_0/group_std_mean": 0.43483443558216095, "eval_signal/frontier_coverage_0/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.004476487752981484, "eval_signal/frontier_coverage_0/weight": 0.012500000186264515, "eval_signal/frontier_coverage_0/weighted_centered_abs_mean": 0.004476487752981484, "eval_signal/frontier_coverage_1/centered_abs_mean": 0.35811903327703476, "eval_signal/frontier_coverage_1/group_bin_occupancy": 0.9453125, "eval_signal/frontier_coverage_1/group_std_mean": 0.43483443558216095, "eval_signal/frontier_coverage_1/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.004476487752981484, "eval_signal/frontier_coverage_1/weight": 0.012500000186264515, "eval_signal/frontier_coverage_1/weighted_centered_abs_mean": 0.004476487752981484, "eval_signal/frontier_coverage_10/centered_abs_mean": 0.3580302596092224, "eval_signal/frontier_coverage_10/group_bin_occupancy": 0.9453125, "eval_signal/frontier_coverage_10/group_std_mean": 0.4347318038344383, "eval_signal/frontier_coverage_10/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.004475378314964473, "eval_signal/frontier_coverage_10/weight": 0.012500000186264515, "eval_signal/frontier_coverage_10/weighted_centered_abs_mean": 0.004475378314964473, "eval_signal/frontier_coverage_15/centered_abs_mean": 0.3575844541192055, "eval_signal/frontier_coverage_15/group_bin_occupancy": 0.9453125, "eval_signal/frontier_coverage_15/group_std_mean": 0.4342164024710655, "eval_signal/frontier_coverage_15/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.004469805862754583, "eval_signal/frontier_coverage_15/weight": 0.012500000186264515, "eval_signal/frontier_coverage_15/weighted_centered_abs_mean": 0.004469805862754583, "eval_signal/frontier_coverage_20/centered_abs_mean": 0.35324113816022873, "eval_signal/frontier_coverage_20/group_bin_occupancy": 0.9453125, "eval_signal/frontier_coverage_20/group_std_mean": 0.42919736355543137, "eval_signal/frontier_coverage_20/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.004415514413267374, "eval_signal/frontier_coverage_20/weight": 0.012500000186264515, "eval_signal/frontier_coverage_20/weighted_centered_abs_mean": 0.004415514413267374, "eval_signal/frontier_coverage_25/centered_abs_mean": 0.32904020696878433, "eval_signal/frontier_coverage_25/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_25/group_std_mean": 0.40118255466222763, "eval_signal/frontier_coverage_25/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.00411300256382674, "eval_signal/frontier_coverage_25/weight": 0.012500000186264515, "eval_signal/frontier_coverage_25/weighted_centered_abs_mean": 0.00411300256382674, "eval_signal/frontier_coverage_5/centered_abs_mean": 0.35811903327703476, "eval_signal/frontier_coverage_5/group_bin_occupancy": 0.9453125, "eval_signal/frontier_coverage_5/group_std_mean": 0.43483443558216095, "eval_signal/frontier_coverage_5/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.004476487752981484, "eval_signal/frontier_coverage_5/weight": 0.012500000186264515, "eval_signal/frontier_coverage_5/weighted_centered_abs_mean": 0.004476487752981484, "eval_signal/frontier_ece_reward/centered_abs_mean": 0.01556184096261859, "eval_signal/frontier_ece_reward/group_bin_occupancy": 0.875, "eval_signal/frontier_ece_reward/group_std_mean": 0.020720474421977997, "eval_signal/frontier_ece_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0015561840846203268, "eval_signal/frontier_ece_reward/weight": 0.10000000149011612, "eval_signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0015561840846203268, "eval_signal/frontier_entropy_batch_reward/centered_abs_mean": 0.0037841796875, "eval_signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.1328125, "eval_signal/frontier_entropy_batch_reward/group_std_mean": 0.011048543266952038, "eval_signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.9375, "eval_signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.0003784179862122983, "eval_signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "eval_signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.0003784179862122983, "eval_steps_per_second": 0.141, "step": 150 }, { "calibration/aurc": 0.37673132248406577, "calibration/batch_distribution_entropy": 0.9807509997094209, "calibration/batch_entropy_100bins": 0.9707388839043682, "calibration/batch_entropy_10bins": 0.9807509997094209, "calibration/batch_entropy_50bins": 0.9793094257561638, "calibration/batch_uniqueness": 0.9534393310546875, "calibration/buffer_distribution_entropy": 0.998310856594891, "calibration/buffer_entropy_100bins": 0.997157838232434, "calibration/buffer_entropy_10bins": 0.998310856594891, "calibration/buffer_entropy_50bins": 0.998086773812687, "calibration/confidence_entropy": 0.4785268084200866, "calibration/coverage@0%": 0.004296875, "calibration/coverage@1%": 0.004296875, "calibration/coverage@10%": 0.0796875, "calibration/coverage@15%": 0.130859375, "calibration/coverage@20%": 0.174609375, "calibration/coverage@25%": 0.238671875, "calibration/coverage@30%": 0.27578125, "calibration/coverage@5%": 0.03515625, "calibration/ece": 0.13422602179767124, "calibration/mean_confidence": 0.5212211718200898, "calibration/prompt_uniqueness": 0.8513671875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 511.0, "completions/max_terminated_length": 511.0, "completions/mean_length": 199.3521484375, "completions/mean_terminated_length": 199.3521484375, "completions/min_length": 99.6, "completions/min_terminated_length": 99.6, "epoch": 0.496, "grad_norm": 0.0009006433538161218, "learning_rate": 1e-06, "loss": -0.0003, "num_tokens": 517537656.0, "reward": 0.939317774772644, "reward_std": 0.08587512522935867, "rewards/accuracy_reward": 0.550390625, "rewards/brier_reward": 0.7851529479026794, "rewards/confidence_uniqueness_reward": 0.9537322998046875, "rewards/format_reward": 1.0, "rewards/frontier_aurc_reward": -0.003137741610407829, "rewards/frontier_coverage_0": 0.09517468363046647, "rewards/frontier_coverage_1": 0.09517468363046647, "rewards/frontier_coverage_10": 0.09508385360240937, "rewards/frontier_coverage_15": 0.0948850855231285, "rewards/frontier_coverage_20": 0.0938334509730339, "rewards/frontier_coverage_25": 0.08577444255352021, "rewards/frontier_coverage_5": 0.09517468363046647, "rewards/frontier_ece_reward": 0.007408684119582176, "rewards/frontier_entropy_batch_reward": -0.18656471371650696, "signal/accuracy_reward/centered_abs_mean": 0.091845703125, "signal/accuracy_reward/group_bin_occupancy": 0.171484375, "signal/accuracy_reward/group_std_mean": 0.1246898889541626, "signal/accuracy_reward/group_zero_std_frac": 0.628125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0459228515625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0459228515625, "signal/advantage_abs_mean": 0.06639864295721054, "signal/advantage_pre_scale_abs_mean": 0.06639864295721054, "signal/advantage_pre_scale_std": 0.10511699765920639, "signal/advantage_std": 0.10511699765920639, "signal/brier_reward/centered_abs_mean": 0.13154455572366713, "signal/brier_reward/group_bin_occupancy": 0.842578125, "signal/brier_reward/group_std_mean": 0.16837890744209288, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01315445601940155, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01315445601940155, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011860489845275879, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.95546875, "signal/confidence_uniqueness_reward/group_std_mean": 0.014825479499995709, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011860489845275878, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011860489845275878, "signal/format_reward/centered_abs_mean": 0.0, "signal/format_reward/group_bin_occupancy": 0.125, "signal/format_reward/group_std_mean": 0.0, "signal/format_reward/group_zero_std_frac": 1.0, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0, "signal/frontier_aurc_reward/centered_abs_mean": 0.0033515980932861565, "signal/frontier_aurc_reward/group_bin_occupancy": 0.70078125, "signal/frontier_aurc_reward/group_std_mean": 0.005607766285538674, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.189497740298975e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.189497740298975e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16509951949119567, "signal/frontier_coverage_0/group_bin_occupancy": 0.86015625, "signal/frontier_coverage_0/group_std_mean": 0.2131485790014267, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0020637440029531717, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0020637440029531717, "signal/frontier_coverage_1/centered_abs_mean": 0.16509951949119567, "signal/frontier_coverage_1/group_bin_occupancy": 0.86015625, "signal/frontier_coverage_1/group_std_mean": 0.2131485790014267, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0020637440029531717, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0020637440029531717, "signal/frontier_coverage_10/centered_abs_mean": 0.16501767039299012, "signal/frontier_coverage_10/group_bin_occupancy": 0.86015625, "signal/frontier_coverage_10/group_std_mean": 0.21304723024368286, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002062720968388021, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002062720968388021, "signal/frontier_coverage_15/centered_abs_mean": 0.16483201384544371, "signal/frontier_coverage_15/group_bin_occupancy": 0.860546875, "signal/frontier_coverage_15/group_std_mean": 0.21281163692474364, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002060400112532079, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002060400112532079, "signal/frontier_coverage_20/centered_abs_mean": 0.16328216195106507, "signal/frontier_coverage_20/group_bin_occupancy": 0.85859375, "signal/frontier_coverage_20/group_std_mean": 0.21082625091075896, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002041027043014765, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002041027043014765, "signal/frontier_coverage_25/centered_abs_mean": 0.14890957474708558, "signal/frontier_coverage_25/group_bin_occupancy": 0.85546875, "signal/frontier_coverage_25/group_std_mean": 0.19260528683662415, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.001861369819380343, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.001861369819380343, "signal/frontier_coverage_5/centered_abs_mean": 0.16509951949119567, "signal/frontier_coverage_5/group_bin_occupancy": 0.86015625, "signal/frontier_coverage_5/group_std_mean": 0.2131485790014267, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0020637440029531717, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0020637440029531717, "signal/frontier_ece_reward/centered_abs_mean": 0.011068322695791721, "signal/frontier_ece_reward/group_bin_occupancy": 0.63671875, "signal/frontier_ece_reward/group_std_mean": 0.0142789700999856, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0011068322928622365, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0011068322928622365, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26318793296813964, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.741015625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.33682562708854674, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02631879411637783, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02631879411637783, "step": 155 }, { "calibration/aurc": 0.3250374330543305, "calibration/batch_distribution_entropy": 0.9794119329606665, "calibration/batch_entropy_100bins": 0.9697917289089146, "calibration/batch_entropy_10bins": 0.9794119329606665, "calibration/batch_entropy_50bins": 0.9772763806222701, "calibration/batch_uniqueness": 0.9530876600886741, "calibration/buffer_distribution_entropy": 0.9982220140902178, "calibration/buffer_entropy_100bins": 0.9973190433371958, "calibration/buffer_entropy_10bins": 0.9982220140902178, "calibration/buffer_entropy_50bins": 0.9981314950384299, "calibration/confidence_entropy": 0.5201237551888573, "calibration/coverage@0%": 0.035546875, "calibration/coverage@1%": 0.035546875, "calibration/coverage@10%": 0.165625, "calibration/coverage@15%": 0.23633347602739727, "calibration/coverage@20%": 0.3312576443248532, "calibration/coverage@25%": 0.38868869251467714, "calibration/coverage@30%": 0.45431980797455973, "calibration/coverage@5%": 0.1171875, "calibration/ece": 0.13896484124688296, "calibration/mean_confidence": 0.5119458417066906, "calibration/prompt_uniqueness": 0.8669053537168313, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 916.4, "completions/max_terminated_length": 489.6, "completions/mean_length": 199.701953125, "completions/mean_terminated_length": 199.4419677734375, "completions/min_length": 97.0, "completions/min_terminated_length": 97.0, "epoch": 0.512, "grad_norm": 0.0009735728963278234, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 534728268.0, "reward": 0.9376786231994629, "reward_std": 0.0903705671429634, "rewards/accuracy_reward": 0.54384765625, "rewards/brier_reward": 0.7990026593208313, "rewards/confidence_uniqueness_reward": 0.9534576177597046, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.002713106363080442, "rewards/frontier_coverage_0": 0.10563770458102226, "rewards/frontier_coverage_1": 0.10563770458102226, "rewards/frontier_coverage_10": 0.10557092055678367, "rewards/frontier_coverage_15": 0.1054814524948597, "rewards/frontier_coverage_20": 0.10448024049401283, "rewards/frontier_coverage_25": 0.09347152709960938, "rewards/frontier_coverage_5": 0.10563770458102226, "rewards/frontier_ece_reward": 0.006927299126982689, "rewards/frontier_entropy_batch_reward": -0.19126361310482026, "signal/accuracy_reward/centered_abs_mean": 0.101068115234375, "signal/accuracy_reward/group_bin_occupancy": 0.17265625, "signal/accuracy_reward/group_std_mean": 0.13370549082756042, "signal/accuracy_reward/group_zero_std_frac": 0.61875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0505340576171875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0505340576171875, "signal/advantage_abs_mean": 0.0700280249118805, "signal/advantage_pre_scale_abs_mean": 0.0700280249118805, "signal/advantage_pre_scale_std": 0.11139876544475555, "signal/advantage_std": 0.11139876544475555, "signal/brier_reward/centered_abs_mean": 0.12366594821214676, "signal/brier_reward/group_bin_occupancy": 0.841796875, "signal/brier_reward/group_std_mean": 0.1601497322320938, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.0123665951192379, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.0123665951192379, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012105925008654594, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.962109375, "signal/confidence_uniqueness_reward/group_std_mean": 0.015737836621701716, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012105925241485238, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012105925241485238, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0030028943438082933, "signal/frontier_aurc_reward/group_bin_occupancy": 0.701953125, "signal/frontier_aurc_reward/group_std_mean": 0.005186095833778381, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.753618002519943e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.753618002519943e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15795475244522095, "signal/frontier_coverage_0/group_bin_occupancy": 0.875390625, "signal/frontier_coverage_0/group_std_mean": 0.20469961762428285, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.001974434358999133, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.001974434358999133, "signal/frontier_coverage_1/centered_abs_mean": 0.15795475244522095, "signal/frontier_coverage_1/group_bin_occupancy": 0.875390625, "signal/frontier_coverage_1/group_std_mean": 0.20469961762428285, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.001974434358999133, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.001974434358999133, "signal/frontier_coverage_10/centered_abs_mean": 0.15781155228614807, "signal/frontier_coverage_10/group_bin_occupancy": 0.875390625, "signal/frontier_coverage_10/group_std_mean": 0.204518723487854, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001972644473426044, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001972644473426044, "signal/frontier_coverage_15/centered_abs_mean": 0.15762017369270326, "signal/frontier_coverage_15/group_bin_occupancy": 0.875390625, "signal/frontier_coverage_15/group_std_mean": 0.204276967048645, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0019702521385625005, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0019702521385625005, "signal/frontier_coverage_20/centered_abs_mean": 0.15554924607276915, "signal/frontier_coverage_20/group_bin_occupancy": 0.87265625, "signal/frontier_coverage_20/group_std_mean": 0.2016424357891083, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0019443656085059047, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0019443656085059047, "signal/frontier_coverage_25/centered_abs_mean": 0.13331365585327148, "signal/frontier_coverage_25/group_bin_occupancy": 0.8640625, "signal/frontier_coverage_25/group_std_mean": 0.17340582311153413, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0016664206981658936, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0016664206981658936, "signal/frontier_coverage_5/centered_abs_mean": 0.15795475244522095, "signal/frontier_coverage_5/group_bin_occupancy": 0.875390625, "signal/frontier_coverage_5/group_std_mean": 0.20469961762428285, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.001974434358999133, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.001974434358999133, "signal/frontier_ece_reward/centered_abs_mean": 0.009351738914847373, "signal/frontier_ece_reward/group_bin_occupancy": 0.6375, "signal/frontier_ece_reward/group_std_mean": 0.0121949952095747, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0009351738961413502, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0009351738961413502, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2719482213258743, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.7453125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3479082763195038, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027194822579622267, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027194822579622267, "step": 160 }, { "calibration/aurc": 0.22114443144112092, "calibration/batch_distribution_entropy": 0.9881730583343551, "calibration/batch_entropy_100bins": 0.9738740046917483, "calibration/batch_entropy_10bins": 0.9881730583343551, "calibration/batch_entropy_50bins": 0.9820913548352396, "calibration/batch_uniqueness": 0.9550997115458519, "calibration/buffer_distribution_entropy": 0.9984131329493497, "calibration/buffer_entropy_100bins": 0.9976733881832335, "calibration/buffer_entropy_10bins": 0.9984131329493497, "calibration/buffer_entropy_50bins": 0.9983565416022511, "calibration/confidence_entropy": 0.501012683895661, "calibration/coverage@0%": 0.018767581947162427, "calibration/coverage@1%": 0.018767581947162427, "calibration/coverage@10%": 0.2706633745107632, "calibration/coverage@15%": 0.34887934197651665, "calibration/coverage@20%": 0.47473015533268104, "calibration/coverage@25%": 0.6169795743639922, "calibration/coverage@30%": 0.7318837145303327, "calibration/coverage@5%": 0.17956213307240704, "calibration/ece": 0.11014316975541685, "calibration/mean_confidence": 0.5185333569003603, "calibration/prompt_uniqueness": 0.8565758426281217, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0005859375, "completions/max_length": 1364.0, "completions/max_terminated_length": 633.2, "completions/mean_length": 201.49541015625, "completions/mean_terminated_length": 200.71401977539062, "completions/min_length": 100.8, "completions/min_terminated_length": 100.8, "epoch": 0.528, "grad_norm": 0.0013889584224671125, "learning_rate": 1e-06, "loss": 0.0021, "num_tokens": 551821117.0, "reward": 0.9383474349975586, "reward_std": 0.09054242372512818, "rewards/accuracy_reward": 0.53857421875, "rewards/brier_reward": 0.8096924185752868, "rewards/confidence_uniqueness_reward": 0.9535269737243652, "rewards/format_reward": 0.9994140625, "rewards/frontier_aurc_reward": -0.0022374443244189022, "rewards/frontier_coverage_0": 0.12531741112470626, "rewards/frontier_coverage_1": 0.12531741112470626, "rewards/frontier_coverage_10": 0.12526962161064148, "rewards/frontier_coverage_15": 0.12519316524267196, "rewards/frontier_coverage_20": 0.12377268075942993, "rewards/frontier_coverage_25": 0.10036717355251312, "rewards/frontier_coverage_5": 0.12531741112470626, "rewards/frontier_ece_reward": 0.006689018942415715, "rewards/frontier_entropy_batch_reward": -0.18241511583328246, "signal/accuracy_reward/centered_abs_mean": 0.109295654296875, "signal/accuracy_reward/group_bin_occupancy": 0.17265625, "signal/accuracy_reward/group_std_mean": 0.1399888128042221, "signal/accuracy_reward/group_zero_std_frac": 0.61875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0546478271484375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0546478271484375, "signal/advantage_abs_mean": 0.07105203717947006, "signal/advantage_pre_scale_abs_mean": 0.07105203717947006, "signal/advantage_pre_scale_std": 0.11316181421279907, "signal/advantage_std": 0.11316181421279907, "signal/brier_reward/centered_abs_mean": 0.12089329361915588, "signal/brier_reward/group_bin_occupancy": 0.85390625, "signal/brier_reward/group_std_mean": 0.15612443685531616, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012089329212903977, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012089329212903977, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012254759855568408, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.94609375, "signal/confidence_uniqueness_reward/group_std_mean": 0.01700945682823658, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012254760134965182, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012254760134965182, "signal/format_reward/centered_abs_mean": 0.00113525390625, "signal/format_reward/group_bin_occupancy": 0.12734375, "signal/format_reward/group_std_mean": 0.0033145629335194827, "signal/format_reward/group_zero_std_frac": 0.98125, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000567626953125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000567626953125, "signal/frontier_aurc_reward/centered_abs_mean": 0.002470444841310382, "signal/frontier_aurc_reward/group_bin_occupancy": 0.703515625, "signal/frontier_aurc_reward/group_std_mean": 0.004230479802936315, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.088056109845638e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.088056109845638e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17215701341629028, "signal/frontier_coverage_0/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_0/group_std_mean": 0.22058985531330108, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002151962695643306, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002151962695643306, "signal/frontier_coverage_1/centered_abs_mean": 0.17215701341629028, "signal/frontier_coverage_1/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_1/group_std_mean": 0.22058985531330108, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002151962695643306, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002151962695643306, "signal/frontier_coverage_10/centered_abs_mean": 0.17201497554779052, "signal/frontier_coverage_10/group_bin_occupancy": 0.8859375, "signal/frontier_coverage_10/group_std_mean": 0.22041167318820953, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.00215018717572093, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.00215018717572093, "signal/frontier_coverage_15/centered_abs_mean": 0.17178415656089782, "signal/frontier_coverage_15/group_bin_occupancy": 0.8859375, "signal/frontier_coverage_15/group_std_mean": 0.2201210618019104, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0021473018918186427, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0021473018918186427, "signal/frontier_coverage_20/centered_abs_mean": 0.16905337870121, "signal/frontier_coverage_20/group_bin_occupancy": 0.885546875, "signal/frontier_coverage_20/group_std_mean": 0.21666520833969116, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002113167243078351, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002113167243078351, "signal/frontier_coverage_25/centered_abs_mean": 0.12845720499753951, "signal/frontier_coverage_25/group_bin_occupancy": 0.877734375, "signal/frontier_coverage_25/group_std_mean": 0.165821772813797, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.001605715067125857, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.001605715067125857, "signal/frontier_coverage_5/centered_abs_mean": 0.17215701341629028, "signal/frontier_coverage_5/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_5/group_std_mean": 0.22058985531330108, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002151962695643306, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002151962695643306, "signal/frontier_ece_reward/centered_abs_mean": 0.008193978387862445, "signal/frontier_ece_reward/group_bin_occupancy": 0.616015625, "signal/frontier_ece_reward/group_std_mean": 0.010671781562268733, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0008193978574126959, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0008193978574126959, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.25369992852211, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.7265625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3307046353816986, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02536999322474003, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02536999322474003, "step": 165 }, { "calibration/aurc": 0.252147588510932, "calibration/batch_distribution_entropy": 0.9738425318127055, "calibration/batch_entropy_100bins": 0.9664689681370076, "calibration/batch_entropy_10bins": 0.9738425318127055, "calibration/batch_entropy_50bins": 0.9725743614789206, "calibration/batch_uniqueness": 0.952545166015625, "calibration/buffer_distribution_entropy": 0.9984631192159462, "calibration/buffer_entropy_100bins": 0.9979719389417199, "calibration/buffer_entropy_10bins": 0.9984631192159462, "calibration/buffer_entropy_50bins": 0.9985060717857308, "calibration/confidence_entropy": 0.47491404501031836, "calibration/coverage@0%": 0.034375, "calibration/coverage@1%": 0.034375, "calibration/coverage@10%": 0.08828125, "calibration/coverage@15%": 0.208984375, "calibration/coverage@20%": 0.44921875, "calibration/coverage@25%": 0.561328125, "calibration/coverage@30%": 0.701171875, "calibration/coverage@5%": 0.052734375, "calibration/ece": 0.07464836513285356, "calibration/mean_confidence": 0.5533402279576579, "calibration/prompt_uniqueness": 0.8580078125, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 775.0, "completions/max_terminated_length": 629.2, "completions/mean_length": 202.2388671875, "completions/mean_terminated_length": 201.97777709960937, "completions/min_length": 98.0, "completions/min_terminated_length": 98.0, "epoch": 0.544, "grad_norm": 0.0016668845200911164, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 569055627.0, "reward": 0.9471798181533814, "reward_std": 0.09814426302909851, "rewards/accuracy_reward": 0.5724609375, "rewards/brier_reward": 0.7898291230201722, "rewards/confidence_uniqueness_reward": 0.9525644302368164, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.002871453482657671, "rewards/frontier_coverage_0": 0.07826435565948486, "rewards/frontier_coverage_1": 0.07826435565948486, "rewards/frontier_coverage_10": 0.07819170504808426, "rewards/frontier_coverage_15": 0.07806946784257889, "rewards/frontier_coverage_20": 0.0771796777844429, "rewards/frontier_coverage_25": 0.05799282640218735, "rewards/frontier_coverage_5": 0.07826236486434937, "rewards/frontier_ece_reward": 0.0054378600791096686, "rewards/frontier_entropy_batch_reward": -0.20278048217296601, "signal/accuracy_reward/centered_abs_mean": 0.1200927734375, "signal/accuracy_reward/group_bin_occupancy": 0.186328125, "signal/accuracy_reward/group_std_mean": 0.1636652648448944, "signal/accuracy_reward/group_zero_std_frac": 0.509375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.06004638671875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.06004638671875, "signal/advantage_abs_mean": 0.07490566372871399, "signal/advantage_pre_scale_abs_mean": 0.07490566372871399, "signal/advantage_pre_scale_std": 0.11634613871574402, "signal/advantage_std": 0.11634613871574402, "signal/brier_reward/centered_abs_mean": 0.13515127897262574, "signal/brier_reward/group_bin_occupancy": 0.8546875, "signal/brier_reward/group_std_mean": 0.17328186929225922, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.013515128940343856, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.013515128940343856, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012677741795778274, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.94609375, "signal/confidence_uniqueness_reward/group_std_mean": 0.01651177238672972, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012677741935476661, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012677741935476661, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0032612524461001157, "signal/frontier_aurc_reward/group_bin_occupancy": 0.6890625, "signal/frontier_aurc_reward/group_std_mean": 0.0055005763657391075, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.07656560128089e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.07656560128089e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17763693630695343, "signal/frontier_coverage_0/group_bin_occupancy": 0.86796875, "signal/frontier_coverage_0/group_std_mean": 0.22864043712615967, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0022204618901014326, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0022204618901014326, "signal/frontier_coverage_1/centered_abs_mean": 0.17763693630695343, "signal/frontier_coverage_1/group_bin_occupancy": 0.86796875, "signal/frontier_coverage_1/group_std_mean": 0.22864043712615967, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0022204618901014326, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0022204618901014326, "signal/frontier_coverage_10/centered_abs_mean": 0.177477365732193, "signal/frontier_coverage_10/group_bin_occupancy": 0.867578125, "signal/frontier_coverage_10/group_std_mean": 0.22843942940235137, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0022184671834111214, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0022184671834111214, "signal/frontier_coverage_15/centered_abs_mean": 0.17723246812820434, "signal/frontier_coverage_15/group_bin_occupancy": 0.8671875, "signal/frontier_coverage_15/group_std_mean": 0.22813022136688232, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002215405786409974, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002215405786409974, "signal/frontier_coverage_20/centered_abs_mean": 0.17215644419193268, "signal/frontier_coverage_20/group_bin_occupancy": 0.862109375, "signal/frontier_coverage_20/group_std_mean": 0.22173726558685303, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0021519554778933526, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0021519554778933526, "signal/frontier_coverage_25/centered_abs_mean": 0.10993409007787705, "signal/frontier_coverage_25/group_bin_occupancy": 0.858203125, "signal/frontier_coverage_25/group_std_mean": 0.1427074909210205, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.001374176167882979, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.001374176167882979, "signal/frontier_coverage_5/centered_abs_mean": 0.17762283384799957, "signal/frontier_coverage_5/group_bin_occupancy": 0.86796875, "signal/frontier_coverage_5/group_std_mean": 0.2286229282617569, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0022202855441719295, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0022202855441719295, "signal/frontier_ece_reward/centered_abs_mean": 0.008673343248665333, "signal/frontier_ece_reward/group_bin_occupancy": 0.625390625, "signal/frontier_ece_reward/group_std_mean": 0.01110559906810522, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0008673343108966947, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0008673343108966947, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2686024695634842, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.73125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3455950260162354, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026860246807336806, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026860246807336806, "step": 170 }, { "calibration/aurc": 0.2515479634906527, "calibration/batch_distribution_entropy": 0.9860691051510942, "calibration/batch_entropy_100bins": 0.971378468672888, "calibration/batch_entropy_10bins": 0.9860691051510942, "calibration/batch_entropy_50bins": 0.9801984357603191, "calibration/batch_uniqueness": 0.9542152412586322, "calibration/buffer_distribution_entropy": 0.9982029436555464, "calibration/buffer_entropy_100bins": 0.998058913638045, "calibration/buffer_entropy_10bins": 0.9982029436555464, "calibration/buffer_entropy_50bins": 0.9984170704958666, "calibration/confidence_entropy": 0.4932112945407329, "calibration/coverage@0%": 0.04219055772994129, "calibration/coverage@1%": 0.09297180772994129, "calibration/coverage@10%": 0.24613961594911937, "calibration/coverage@15%": 0.33801140533268104, "calibration/coverage@20%": 0.41734038649706456, "calibration/coverage@25%": 0.5271350599315069, "calibration/coverage@30%": 0.6354138637475538, "calibration/coverage@5%": 0.16679993272994129, "calibration/ece": 0.11478470629407063, "calibration/mean_confidence": 0.5093848801888953, "calibration/prompt_uniqueness": 0.8577200771494538, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1322.6, "completions/max_terminated_length": 563.4, "completions/mean_length": 201.420703125, "completions/mean_terminated_length": 200.89898071289062, "completions/min_length": 99.2, "completions/min_terminated_length": 99.2, "epoch": 0.56, "grad_norm": 0.0008646171190775931, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 585939583.0, "reward": 0.9280932784080506, "reward_std": 0.08654378652572632, "rewards/accuracy_reward": 0.525, "rewards/brier_reward": 0.8003496289253235, "rewards/confidence_uniqueness_reward": 0.9529324412345886, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.002947695879265666, "rewards/frontier_coverage_0": 0.12158387079834938, "rewards/frontier_coverage_1": 0.12158387079834938, "rewards/frontier_coverage_10": 0.12146811336278915, "rewards/frontier_coverage_15": 0.1212589830160141, "rewards/frontier_coverage_20": 0.11846682131290436, "rewards/frontier_coverage_25": 0.07554255723953247, "rewards/frontier_coverage_5": 0.12154962569475174, "rewards/frontier_ece_reward": 0.005375421233475208, "rewards/frontier_entropy_batch_reward": -0.20058484077453614, "signal/accuracy_reward/centered_abs_mean": 0.0883544921875, "signal/accuracy_reward/group_bin_occupancy": 0.17109375, "signal/accuracy_reward/group_std_mean": 0.1225573942065239, "signal/accuracy_reward/group_zero_std_frac": 0.63125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.04417724609375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.04417724609375, "signal/advantage_abs_mean": 0.06553731113672256, "signal/advantage_pre_scale_abs_mean": 0.06553731113672256, "signal/advantage_pre_scale_std": 0.10620072782039643, "signal/advantage_std": 0.10620072782039643, "signal/brier_reward/centered_abs_mean": 0.12693677842617035, "signal/brier_reward/group_bin_occupancy": 0.84921875, "signal/brier_reward/group_std_mean": 0.16420524418354035, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01269367802888155, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01269367802888155, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012149993889033794, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9578125, "signal/confidence_uniqueness_reward/group_std_mean": 0.01637979969382286, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012149993563070893, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012149993563070893, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086533904076, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0031220025382936003, "signal/frontier_aurc_reward/group_bin_occupancy": 0.6953125, "signal/frontier_aurc_reward/group_std_mean": 0.005364646296948194, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.902503231074661e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.902503231074661e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16292471885681153, "signal/frontier_coverage_0/group_bin_occupancy": 0.8671875, "signal/frontier_coverage_0/group_std_mean": 0.21221697330474854, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002036558976396918, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002036558976396918, "signal/frontier_coverage_1/centered_abs_mean": 0.16292471885681153, "signal/frontier_coverage_1/group_bin_occupancy": 0.8671875, "signal/frontier_coverage_1/group_std_mean": 0.21221697330474854, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002036558976396918, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002036558976396918, "signal/frontier_coverage_10/centered_abs_mean": 0.16274016201496125, "signal/frontier_coverage_10/group_bin_occupancy": 0.8671875, "signal/frontier_coverage_10/group_std_mean": 0.2119736611843109, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0020342519972473385, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0020342519972473385, "signal/frontier_coverage_15/centered_abs_mean": 0.16240600347518921, "signal/frontier_coverage_15/group_bin_occupancy": 0.866796875, "signal/frontier_coverage_15/group_std_mean": 0.21153769195079802, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0020300750620663166, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0020300750620663166, "signal/frontier_coverage_20/centered_abs_mean": 0.15688030421733856, "signal/frontier_coverage_20/group_bin_occupancy": 0.86484375, "signal/frontier_coverage_20/group_std_mean": 0.2044644385576248, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001961003732867539, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001961003732867539, "signal/frontier_coverage_25/centered_abs_mean": 0.09234647154808044, "signal/frontier_coverage_25/group_bin_occupancy": 0.8671875, "signal/frontier_coverage_25/group_std_mean": 0.12112226784229278, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.001154330896679312, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.001154330896679312, "signal/frontier_coverage_5/centered_abs_mean": 0.16285774111747742, "signal/frontier_coverage_5/group_bin_occupancy": 0.8671875, "signal/frontier_coverage_5/group_std_mean": 0.21212912797927858, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002035721787251532, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002035721787251532, "signal/frontier_ece_reward/centered_abs_mean": 0.00738415727391839, "signal/frontier_ece_reward/group_bin_occupancy": 0.61015625, "signal/frontier_ece_reward/group_std_mean": 0.009608357027173043, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0007384157390333713, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0007384157390333713, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27037072777748106, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.746484375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3438795268535614, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02703707329928875, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02703707329928875, "step": 175 }, { "calibration/aurc": 0.2996449541916843, "calibration/batch_distribution_entropy": 0.9857962924982369, "calibration/batch_entropy_100bins": 0.9725112492310117, "calibration/batch_entropy_10bins": 0.9857962924982369, "calibration/batch_entropy_50bins": 0.9808299037997216, "calibration/batch_uniqueness": 0.9545145167753176, "calibration/buffer_distribution_entropy": 0.9982409057748015, "calibration/buffer_entropy_100bins": 0.9981985980748382, "calibration/buffer_entropy_10bins": 0.9982409057748015, "calibration/buffer_entropy_50bins": 0.9984704570072415, "calibration/confidence_entropy": 0.4946387752860536, "calibration/coverage@0%": 0.006643682729941291, "calibration/coverage@1%": 0.006643682729941291, "calibration/coverage@10%": 0.1425811827299413, "calibration/coverage@15%": 0.17812805772994128, "calibration/coverage@20%": 0.2937538221624266, "calibration/coverage@25%": 0.4215210677592955, "calibration/coverage@30%": 0.512568798923679, "calibration/coverage@5%": 0.06914368272994129, "calibration/ece": 0.09716097472285265, "calibration/mean_confidence": 0.48591203652211473, "calibration/prompt_uniqueness": 0.85220570003577, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 877.4, "completions/max_terminated_length": 503.6, "completions/mean_length": 197.653515625, "completions/mean_terminated_length": 197.2631042480469, "completions/min_length": 100.4, "completions/min_terminated_length": 100.4, "epoch": 0.576, "grad_norm": 0.002247588010504842, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 603150179.0, "reward": 0.9243762850761413, "reward_std": 0.08188213706016541, "rewards/accuracy_reward": 0.5177734375, "rewards/brier_reward": 0.7902379393577575, "rewards/confidence_uniqueness_reward": 0.9534862518310547, "rewards/format_reward": 0.99970703125, "rewards/frontier_aurc_reward": -0.0030144399497658014, "rewards/frontier_coverage_0": 0.12199763208627701, "rewards/frontier_coverage_1": 0.12199763208627701, "rewards/frontier_coverage_10": 0.12183189988136292, "rewards/frontier_coverage_15": 0.12161823213100434, "rewards/frontier_coverage_20": 0.11783311516046524, "rewards/frontier_coverage_25": 0.07518755495548249, "rewards/frontier_coverage_5": 0.12193977534770965, "rewards/frontier_ece_reward": 0.004343415051698685, "rewards/frontier_entropy_batch_reward": -0.1916309654712677, "signal/accuracy_reward/centered_abs_mean": 0.08441162109375, "signal/accuracy_reward/group_bin_occupancy": 0.16796875, "signal/accuracy_reward/group_std_mean": 0.11587727218866348, "signal/accuracy_reward/group_zero_std_frac": 0.65625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.042205810546875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.042205810546875, "signal/advantage_abs_mean": 0.062145890295505525, "signal/advantage_pre_scale_abs_mean": 0.062145890295505525, "signal/advantage_pre_scale_std": 0.09997193515300751, "signal/advantage_std": 0.09997193515300751, "signal/brier_reward/centered_abs_mean": 0.1247845709323883, "signal/brier_reward/group_bin_occupancy": 0.8484375, "signal/brier_reward/group_std_mean": 0.16126168072223662, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012478457018733024, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012478457018733024, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01191106867045164, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.95234375, "signal/confidence_uniqueness_reward/group_std_mean": 0.015692189522087573, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011911068810150026, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011911068810150026, "signal/format_reward/centered_abs_mean": 0.000567626953125, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0016572814900428056, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0002838134765625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0002838134765625, "signal/frontier_aurc_reward/centered_abs_mean": 0.0029186610598117115, "signal/frontier_aurc_reward/group_bin_occupancy": 0.694921875, "signal/frontier_aurc_reward/group_std_mean": 0.005101753072813154, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.6483262010733594e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.6483262010733594e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16748374998569487, "signal/frontier_coverage_0/group_bin_occupancy": 0.878125, "signal/frontier_coverage_0/group_std_mean": 0.21470698118209838, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0020935469307005405, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0020935469307005405, "signal/frontier_coverage_1/centered_abs_mean": 0.16748374998569487, "signal/frontier_coverage_1/group_bin_occupancy": 0.878125, "signal/frontier_coverage_1/group_std_mean": 0.21470698118209838, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0020935469307005405, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0020935469307005405, "signal/frontier_coverage_10/centered_abs_mean": 0.16731707453727723, "signal/frontier_coverage_10/group_bin_occupancy": 0.877734375, "signal/frontier_coverage_10/group_std_mean": 0.21449258029460908, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0020914634224027395, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0020914634224027395, "signal/frontier_coverage_15/centered_abs_mean": 0.16635558009147644, "signal/frontier_coverage_15/group_bin_occupancy": 0.8765625, "signal/frontier_coverage_15/group_std_mean": 0.21327796578407288, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0020794447511434553, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0020794447511434553, "signal/frontier_coverage_20/centered_abs_mean": 0.15995977818965912, "signal/frontier_coverage_20/group_bin_occupancy": 0.874609375, "signal/frontier_coverage_20/group_std_mean": 0.20524422228336334, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0019994973437860606, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0019994973437860606, "signal/frontier_coverage_25/centered_abs_mean": 0.09211662709712982, "signal/frontier_coverage_25/group_bin_occupancy": 0.884375, "signal/frontier_coverage_25/group_std_mean": 0.11896159499883652, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0011514578247442842, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0011514578247442842, "signal/frontier_coverage_5/centered_abs_mean": 0.1674229770898819, "signal/frontier_coverage_5/group_bin_occupancy": 0.878125, "signal/frontier_coverage_5/group_std_mean": 0.2146292507648468, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0020927871111780403, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0020927871111780403, "signal/frontier_ece_reward/centered_abs_mean": 0.006697707623243332, "signal/frontier_ece_reward/group_bin_occupancy": 0.593359375, "signal/frontier_ece_reward/group_std_mean": 0.008749750256538392, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0006697707343846559, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0006697707343846559, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26192234456539154, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.727734375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3407859206199646, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026192235574126244, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026192235574126244, "step": 180 }, { "calibration/aurc": 0.30584589303928156, "calibration/batch_distribution_entropy": 0.9866766763088485, "calibration/batch_entropy_100bins": 0.9724047659598251, "calibration/batch_entropy_10bins": 0.9866766763088485, "calibration/batch_entropy_50bins": 0.9817849770051279, "calibration/batch_uniqueness": 0.954449462890625, "calibration/buffer_distribution_entropy": 0.998152111313456, "calibration/buffer_entropy_100bins": 0.9982497260431769, "calibration/buffer_entropy_10bins": 0.998152111313456, "calibration/buffer_entropy_50bins": 0.9984519766077582, "calibration/confidence_entropy": 0.5032711790217702, "calibration/coverage@0%": 0.0390625, "calibration/coverage@1%": 0.0390625, "calibration/coverage@10%": 0.198828125, "calibration/coverage@15%": 0.328125, "calibration/coverage@20%": 0.419921875, "calibration/coverage@25%": 0.4859375, "calibration/coverage@30%": 0.565234375, "calibration/coverage@5%": 0.073828125, "calibration/ece": 0.13038858466256106, "calibration/mean_confidence": 0.47837184533592775, "calibration/prompt_uniqueness": 0.857080078125, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 451.6, "completions/max_terminated_length": 451.6, "completions/mean_length": 192.9451171875, "completions/mean_terminated_length": 192.9451171875, "completions/min_length": 93.8, "completions/min_terminated_length": 93.8, "epoch": 0.592, "grad_norm": 0.001037395792081952, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 620293649.0, "reward": 0.9257462859153748, "reward_std": 0.08210539668798447, "rewards/accuracy_reward": 0.52255859375, "rewards/brier_reward": 0.7917325615882873, "rewards/confidence_uniqueness_reward": 0.9531358003616333, "rewards/format_reward": 0.99990234375, "rewards/frontier_aurc_reward": -0.002609179844148457, "rewards/frontier_coverage_0": 0.11608581244945526, "rewards/frontier_coverage_1": 0.11608581244945526, "rewards/frontier_coverage_10": 0.11598109006881714, "rewards/frontier_coverage_15": 0.11568910479545594, "rewards/frontier_coverage_20": 0.10997468084096909, "rewards/frontier_coverage_25": 0.06685756668448448, "rewards/frontier_coverage_5": 0.11603517830371857, "rewards/frontier_ece_reward": 0.0043127467855811116, "rewards/frontier_entropy_batch_reward": -0.1982853889465332, "signal/accuracy_reward/centered_abs_mean": 0.091021728515625, "signal/accuracy_reward/group_bin_occupancy": 0.16953125, "signal/accuracy_reward/group_std_mean": 0.12103503942489624, "signal/accuracy_reward/group_zero_std_frac": 0.64375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0455108642578125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0455108642578125, "signal/advantage_abs_mean": 0.06382817998528481, "signal/advantage_pre_scale_abs_mean": 0.06382817998528481, "signal/advantage_pre_scale_std": 0.10076852142810822, "signal/advantage_std": 0.10076852142810822, "signal/brier_reward/centered_abs_mean": 0.1221377670764923, "signal/brier_reward/group_bin_occupancy": 0.850390625, "signal/brier_reward/group_std_mean": 0.1577945113182068, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012213777005672454, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012213777005672454, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012191111594438553, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.948046875, "signal/confidence_uniqueness_reward/group_std_mean": 0.015553272515535354, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001219111168757081, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001219111168757081, "signal/format_reward/centered_abs_mean": 0.000189208984375, "signal/format_reward/group_bin_occupancy": 0.125390625, "signal/format_reward/group_std_mean": 0.0005524271633476019, "signal/format_reward/group_zero_std_frac": 0.996875, "signal/format_reward/scaled_weighted_centered_abs_mean": 9.46044921875e-05, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 9.46044921875e-05, "signal/frontier_aurc_reward/centered_abs_mean": 0.002383097121492028, "signal/frontier_aurc_reward/group_bin_occupancy": 0.71328125, "signal/frontier_aurc_reward/group_std_mean": 0.004014624189585448, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.9788713436573744e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.9788713436573744e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17336324751377105, "signal/frontier_coverage_0/group_bin_occupancy": 0.875, "signal/frontier_coverage_0/group_std_mean": 0.2194622665643692, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002167040528729558, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002167040528729558, "signal/frontier_coverage_1/centered_abs_mean": 0.17336324751377105, "signal/frontier_coverage_1/group_bin_occupancy": 0.875, "signal/frontier_coverage_1/group_std_mean": 0.2194622665643692, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002167040528729558, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002167040528729558, "signal/frontier_coverage_10/centered_abs_mean": 0.1732115387916565, "signal/frontier_coverage_10/group_bin_occupancy": 0.875, "signal/frontier_coverage_10/group_std_mean": 0.21927422285079956, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002165144309401512, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002165144309401512, "signal/frontier_coverage_15/centered_abs_mean": 0.17268791794776917, "signal/frontier_coverage_15/group_bin_occupancy": 0.87578125, "signal/frontier_coverage_15/group_std_mean": 0.21862509846687317, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0021585989743471146, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0021585989743471146, "signal/frontier_coverage_20/centered_abs_mean": 0.16179357171058656, "signal/frontier_coverage_20/group_bin_occupancy": 0.8734375, "signal/frontier_coverage_20/group_std_mean": 0.2050326019525528, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002022419683635235, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002022419683635235, "signal/frontier_coverage_25/centered_abs_mean": 0.09012775868177414, "signal/frontier_coverage_25/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_25/group_std_mean": 0.11584892421960831, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0011265970300883056, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0011265970300883056, "signal/frontier_coverage_5/centered_abs_mean": 0.17328991889953613, "signal/frontier_coverage_5/group_bin_occupancy": 0.875, "signal/frontier_coverage_5/group_std_mean": 0.21937137544155122, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002166124014183879, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002166124014183879, "signal/frontier_ece_reward/centered_abs_mean": 0.006071403250098228, "signal/frontier_ece_reward/group_bin_occupancy": 0.60078125, "signal/frontier_ece_reward/group_std_mean": 0.007915554381906987, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0006071403389796614, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0006071403389796614, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2702577829360962, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.728125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3449578046798706, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027025777474045755, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027025777474045755, "step": 185 }, { "calibration/aurc": 0.22400625891389309, "calibration/batch_distribution_entropy": 0.9719454581801458, "calibration/batch_entropy_100bins": 0.9668305302966649, "calibration/batch_entropy_10bins": 0.9719454581801458, "calibration/batch_entropy_50bins": 0.9726482595542361, "calibration/batch_uniqueness": 0.95205078125, "calibration/buffer_distribution_entropy": 0.9982247184091022, "calibration/buffer_entropy_100bins": 0.9983723385641456, "calibration/buffer_entropy_10bins": 0.9982247184091022, "calibration/buffer_entropy_50bins": 0.9985019650021949, "calibration/confidence_entropy": 0.4803624366523741, "calibration/coverage@0%": 0.044921875, "calibration/coverage@1%": 0.047265625, "calibration/coverage@10%": 0.303125, "calibration/coverage@15%": 0.396484375, "calibration/coverage@20%": 0.517578125, "calibration/coverage@25%": 0.614453125, "calibration/coverage@30%": 0.701953125, "calibration/coverage@5%": 0.176953125, "calibration/ece": 0.0920661102784004, "calibration/mean_confidence": 0.4792326134411461, "calibration/prompt_uniqueness": 0.84638671875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1121.8, "completions/max_terminated_length": 489.6, "completions/mean_length": 191.39521484375, "completions/mean_terminated_length": 190.86993408203125, "completions/min_length": 98.6, "completions/min_terminated_length": 98.6, "epoch": 0.608, "grad_norm": 0.0009905729675665498, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 637253024.0, "reward": 0.9326250672340393, "reward_std": 0.07998792231082916, "rewards/accuracy_reward": 0.52998046875, "rewards/brier_reward": 0.8121574759483338, "rewards/confidence_uniqueness_reward": 0.9530542731285095, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.0022178215906023977, "rewards/frontier_coverage_0": 0.13555689454078673, "rewards/frontier_coverage_1": 0.13555689454078673, "rewards/frontier_coverage_10": 0.13542194217443465, "rewards/frontier_coverage_15": 0.13477266132831572, "rewards/frontier_coverage_20": 0.1249439612030983, "rewards/frontier_coverage_25": 0.07523647993803025, "rewards/frontier_coverage_5": 0.13551316857337953, "rewards/frontier_ece_reward": 0.0049108750186860565, "rewards/frontier_entropy_batch_reward": -0.2011692315340042, "signal/accuracy_reward/centered_abs_mean": 0.087628173828125, "signal/accuracy_reward/group_bin_occupancy": 0.167578125, "signal/accuracy_reward/group_std_mean": 0.11695856302976608, "signal/accuracy_reward/group_zero_std_frac": 0.659375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0438140869140625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0438140869140625, "signal/advantage_abs_mean": 0.06124872416257858, "signal/advantage_pre_scale_abs_mean": 0.06124872416257858, "signal/advantage_pre_scale_std": 0.0987380400300026, "signal/advantage_std": 0.0987380400300026, "signal/brier_reward/centered_abs_mean": 0.11684625893831253, "signal/brier_reward/group_bin_occupancy": 0.82890625, "signal/brier_reward/group_std_mean": 0.15167405605316162, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011684626340866089, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011684626340866089, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012386371567845344, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.953515625, "signal/confidence_uniqueness_reward/group_std_mean": 0.016383925825357436, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012386371847242117, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012386371847242117, "signal/format_reward/centered_abs_mean": 0.00074462890625, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0018734002020210027, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000372314453125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000372314453125, "signal/frontier_aurc_reward/centered_abs_mean": 0.002177320816554129, "signal/frontier_aurc_reward/group_bin_occupancy": 0.720703125, "signal/frontier_aurc_reward/group_std_mean": 0.0038774567656219006, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.721651217143517e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.721651217143517e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17268600165843964, "signal/frontier_coverage_0/group_bin_occupancy": 0.86484375, "signal/frontier_coverage_0/group_std_mean": 0.21827268302440644, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0021585751324892043, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0021585751324892043, "signal/frontier_coverage_1/centered_abs_mean": 0.17268600165843964, "signal/frontier_coverage_1/group_bin_occupancy": 0.86484375, "signal/frontier_coverage_1/group_std_mean": 0.21827268302440644, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0021585751324892043, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0021585751324892043, "signal/frontier_coverage_10/centered_abs_mean": 0.1725066900253296, "signal/frontier_coverage_10/group_bin_occupancy": 0.864453125, "signal/frontier_coverage_10/group_std_mean": 0.21804390251636505, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0021563337184488775, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0021563337184488775, "signal/frontier_coverage_15/centered_abs_mean": 0.17139166891574859, "signal/frontier_coverage_15/group_bin_occupancy": 0.8625, "signal/frontier_coverage_15/group_std_mean": 0.21663654446601868, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0021423958940431477, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0021423958940431477, "signal/frontier_coverage_20/centered_abs_mean": 0.15554386079311372, "signal/frontier_coverage_20/group_bin_occupancy": 0.858203125, "signal/frontier_coverage_20/group_std_mean": 0.19667071998119354, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0019442982971668243, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0019442982971668243, "signal/frontier_coverage_25/centered_abs_mean": 0.08549280762672425, "signal/frontier_coverage_25/group_bin_occupancy": 0.902734375, "signal/frontier_coverage_25/group_std_mean": 0.10843969881534576, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0010686601744964719, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0010686601744964719, "signal/frontier_coverage_5/centered_abs_mean": 0.17262336611747742, "signal/frontier_coverage_5/group_bin_occupancy": 0.86484375, "signal/frontier_coverage_5/group_std_mean": 0.21819275319576265, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0021577921230345964, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0021577921230345964, "signal/frontier_ece_reward/centered_abs_mean": 0.005996010545641184, "signal/frontier_ece_reward/group_bin_occupancy": 0.583203125, "signal/frontier_ece_reward/group_std_mean": 0.007664621062576771, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0005996010499075055, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0005996010499075055, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27318122386932375, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.75078125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34829328060150144, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02731812372803688, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02731812372803688, "step": 190 }, { "calibration/aurc": 0.24215358226982628, "calibration/batch_distribution_entropy": 0.9880607992688615, "calibration/batch_entropy_100bins": 0.9744133167257015, "calibration/batch_entropy_10bins": 0.9880607992688615, "calibration/batch_entropy_50bins": 0.9836194456756535, "calibration/batch_uniqueness": 0.9545318603515625, "calibration/buffer_distribution_entropy": 0.9985963168356852, "calibration/buffer_entropy_100bins": 0.9986492092964603, "calibration/buffer_entropy_10bins": 0.9985963168356852, "calibration/buffer_entropy_50bins": 0.9987528035913161, "calibration/confidence_entropy": 0.5069571505841879, "calibration/coverage@0%": 0.03046875, "calibration/coverage@1%": 0.03046875, "calibration/coverage@10%": 0.2, "calibration/coverage@15%": 0.344140625, "calibration/coverage@20%": 0.4671875, "calibration/coverage@25%": 0.551171875, "calibration/coverage@30%": 0.670703125, "calibration/coverage@5%": 0.054296875, "calibration/ece": 0.08878387812757721, "calibration/mean_confidence": 0.4981061984879608, "calibration/prompt_uniqueness": 0.852685546875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 867.4, "completions/max_terminated_length": 421.8, "completions/mean_length": 192.1154296875, "completions/mean_terminated_length": 191.85313415527344, "completions/min_length": 93.2, "completions/min_terminated_length": 93.2, "epoch": 0.624, "grad_norm": 0.0009193821460939944, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 654564190.0, "reward": 0.9324703335762023, "reward_std": 0.08379258662462234, "rewards/accuracy_reward": 0.5236328125, "rewards/brier_reward": 0.8111774682998657, "rewards/confidence_uniqueness_reward": 0.954227352142334, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.002601839741691947, "rewards/frontier_coverage_0": 0.13432081043720245, "rewards/frontier_coverage_1": 0.13432081043720245, "rewards/frontier_coverage_10": 0.1341655969619751, "rewards/frontier_coverage_15": 0.1336636573076248, "rewards/frontier_coverage_20": 0.11634680330753326, "rewards/frontier_coverage_25": 0.06917308419942855, "rewards/frontier_coverage_5": 0.13430217653512955, "rewards/frontier_ece_reward": 0.004168864293023944, "rewards/frontier_entropy_batch_reward": -0.16876917481422424, "signal/accuracy_reward/centered_abs_mean": 0.0984130859375, "signal/accuracy_reward/group_bin_occupancy": 0.16953125, "signal/accuracy_reward/group_std_mean": 0.12785917967557908, "signal/accuracy_reward/group_zero_std_frac": 0.64375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.04920654296875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.04920654296875, "signal/advantage_abs_mean": 0.06533887535333634, "signal/advantage_pre_scale_abs_mean": 0.06533887535333634, "signal/advantage_pre_scale_std": 0.10442866384983063, "signal/advantage_std": 0.10442866384983063, "signal/brier_reward/centered_abs_mean": 0.12079737186431885, "signal/brier_reward/group_bin_occupancy": 0.850390625, "signal/brier_reward/group_std_mean": 0.15655288994312286, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012079737521708012, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012079737521708012, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011342980526387691, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9546875, "signal/confidence_uniqueness_reward/group_std_mean": 0.01486249603331089, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011342980898916722, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011342980898916722, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.002568914764560759, "signal/frontier_aurc_reward/group_bin_occupancy": 0.703515625, "signal/frontier_aurc_reward/group_std_mean": 0.004492522496730089, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.2111437394632955e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.2111437394632955e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16961192786693574, "signal/frontier_coverage_0/group_bin_occupancy": 0.88125, "signal/frontier_coverage_0/group_std_mean": 0.21750209331512452, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0021201491355895994, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0021201491355895994, "signal/frontier_coverage_1/centered_abs_mean": 0.16961192786693574, "signal/frontier_coverage_1/group_bin_occupancy": 0.88125, "signal/frontier_coverage_1/group_std_mean": 0.21750209331512452, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0021201491355895994, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0021201491355895994, "signal/frontier_coverage_10/centered_abs_mean": 0.16930575370788575, "signal/frontier_coverage_10/group_bin_occupancy": 0.88125, "signal/frontier_coverage_10/group_std_mean": 0.21711575090885163, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002116321912035346, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002116321912035346, "signal/frontier_coverage_15/centered_abs_mean": 0.16842733919620514, "signal/frontier_coverage_15/group_bin_occupancy": 0.88125, "signal/frontier_coverage_15/group_std_mean": 0.21600719392299653, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002105341712012887, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002105341712012887, "signal/frontier_coverage_20/centered_abs_mean": 0.14266384541988372, "signal/frontier_coverage_20/group_bin_occupancy": 0.87734375, "signal/frontier_coverage_20/group_std_mean": 0.1834921807050705, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001783298049122095, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001783298049122095, "signal/frontier_coverage_25/centered_abs_mean": 0.07387905418872834, "signal/frontier_coverage_25/group_bin_occupancy": 0.909375, "signal/frontier_coverage_25/group_std_mean": 0.09607784301042557, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0009234882425516843, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0009234882425516843, "signal/frontier_coverage_5/centered_abs_mean": 0.16955203413963318, "signal/frontier_coverage_5/group_bin_occupancy": 0.88125, "signal/frontier_coverage_5/group_std_mean": 0.21742674708366394, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0021194005385041237, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0021194005385041237, "signal/frontier_ece_reward/centered_abs_mean": 0.00531910676509142, "signal/frontier_ece_reward/group_bin_occupancy": 0.605078125, "signal/frontier_ece_reward/group_std_mean": 0.006922620441764593, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0005319106741808355, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0005319106741808355, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2476567655801773, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.733984375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.32226383686065674, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.024765677005052566, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.024765677005052566, "step": 195 }, { "calibration/aurc": 0.2515398156768674, "calibration/batch_distribution_entropy": 0.9799172567201893, "calibration/batch_entropy_100bins": 0.9687537997758875, "calibration/batch_entropy_10bins": 0.9799172567201893, "calibration/batch_entropy_50bins": 0.9771154575805976, "calibration/batch_uniqueness": 0.9531890869140625, "calibration/buffer_distribution_entropy": 0.9987663800446954, "calibration/buffer_entropy_100bins": 0.9987640628879382, "calibration/buffer_entropy_10bins": 0.9987663800446954, "calibration/buffer_entropy_50bins": 0.9988594448171497, "calibration/confidence_entropy": 0.5068997539072493, "calibration/coverage@0%": 0.0203125, "calibration/coverage@1%": 0.092578125, "calibration/coverage@10%": 0.3078125, "calibration/coverage@15%": 0.375390625, "calibration/coverage@20%": 0.433203125, "calibration/coverage@25%": 0.488671875, "calibration/coverage@30%": 0.56328125, "calibration/coverage@5%": 0.23671875, "calibration/ece": 0.15521042633852694, "calibration/mean_confidence": 0.5374644621865269, "calibration/prompt_uniqueness": 0.856103515625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 820.4, "completions/max_terminated_length": 781.6, "completions/mean_length": 194.8408203125, "completions/mean_terminated_length": 194.31633605957032, "completions/min_length": 97.8, "completions/min_terminated_length": 97.8, "epoch": 0.64, "grad_norm": 0.0012083922047168016, "learning_rate": 1e-06, "loss": 0.0014, "num_tokens": 671902048.0, "reward": 0.9501808404922485, "reward_std": 0.07826168239116668, "rewards/accuracy_reward": 0.571484375, "rewards/brier_reward": 0.8080654382705689, "rewards/confidence_uniqueness_reward": 0.9528155326843262, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.0023103010840713977, "rewards/frontier_coverage_0": 0.0970426581799984, "rewards/frontier_coverage_1": 0.0970426581799984, "rewards/frontier_coverage_10": 0.096921756118536, "rewards/frontier_coverage_15": 0.0964154452085495, "rewards/frontier_coverage_20": 0.08248816579580306, "rewards/frontier_coverage_25": 0.054835057258605956, "rewards/frontier_coverage_5": 0.0970426581799984, "rewards/frontier_ece_reward": 0.004011458484455943, "rewards/frontier_entropy_batch_reward": -0.19598765969276427, "signal/accuracy_reward/centered_abs_mean": 0.07796630859375, "signal/accuracy_reward/group_bin_occupancy": 0.16640625, "signal/accuracy_reward/group_std_mean": 0.10859100967645645, "signal/accuracy_reward/group_zero_std_frac": 0.66875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.038983154296875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.038983154296875, "signal/advantage_abs_mean": 0.05957977846264839, "signal/advantage_pre_scale_abs_mean": 0.05957977846264839, "signal/advantage_pre_scale_std": 0.0975896254181862, "signal/advantage_std": 0.0975896254181862, "signal/brier_reward/centered_abs_mean": 0.11409203559160233, "signal/brier_reward/group_bin_occupancy": 0.855078125, "signal/brier_reward/group_std_mean": 0.14763712584972383, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01140920389443636, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01140920389443636, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012328281439840794, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.950390625, "signal/confidence_uniqueness_reward/group_std_mean": 0.015604946203529835, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012328281998634338, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012328281998634338, "signal/format_reward/centered_abs_mean": 0.00068359375, "signal/format_reward/group_bin_occupancy": 0.125390625, "signal/format_reward/group_std_mean": 0.0010500336065888405, "signal/format_reward/group_zero_std_frac": 0.996875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000341796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000341796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.002368218312039971, "signal/frontier_aurc_reward/group_bin_occupancy": 0.72265625, "signal/frontier_aurc_reward/group_std_mean": 0.003913992503657937, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.9602729773614557e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.9602729773614557e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.14712594151496888, "signal/frontier_coverage_0/group_bin_occupancy": 0.88671875, "signal/frontier_coverage_0/group_std_mean": 0.19017478823661804, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0018390743294730782, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0018390743294730782, "signal/frontier_coverage_1/centered_abs_mean": 0.14712594151496888, "signal/frontier_coverage_1/group_bin_occupancy": 0.88671875, "signal/frontier_coverage_1/group_std_mean": 0.19017478823661804, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0018390743294730782, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0018390743294730782, "signal/frontier_coverage_10/centered_abs_mean": 0.14685225784778594, "signal/frontier_coverage_10/group_bin_occupancy": 0.88671875, "signal/frontier_coverage_10/group_std_mean": 0.18982412219047545, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0018356532789766788, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0018356532789766788, "signal/frontier_coverage_15/centered_abs_mean": 0.14580887258052827, "signal/frontier_coverage_15/group_bin_occupancy": 0.8859375, "signal/frontier_coverage_15/group_std_mean": 0.18846336305141448, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001822610874660313, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001822610874660313, "signal/frontier_coverage_20/centered_abs_mean": 0.11683784425258636, "signal/frontier_coverage_20/group_bin_occupancy": 0.871484375, "signal/frontier_coverage_20/group_std_mean": 0.15163523852825164, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0014604730997234583, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0014604730997234583, "signal/frontier_coverage_25/centered_abs_mean": 0.0633377879858017, "signal/frontier_coverage_25/group_bin_occupancy": 0.92109375, "signal/frontier_coverage_25/group_std_mean": 0.0822565495967865, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007917223498225212, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007917223498225212, "signal/frontier_coverage_5/centered_abs_mean": 0.14712594151496888, "signal/frontier_coverage_5/group_bin_occupancy": 0.88671875, "signal/frontier_coverage_5/group_std_mean": 0.19017478823661804, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0018390743294730782, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0018390743294730782, "signal/frontier_ece_reward/centered_abs_mean": 0.0051858888007700445, "signal/frontier_ece_reward/group_bin_occupancy": 0.592578125, "signal/frontier_ece_reward/group_std_mean": 0.006696830876171589, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0005185888556297868, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0005185888556297868, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2651937812566757, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.7234375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3395101547241211, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026519378274679185, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026519378274679185, "step": 200 }, { "epoch": 0.64, "eval_calibration/aurc": 0.4285342195709427, "eval_calibration/batch_distribution_entropy": 0.9108287680099018, "eval_calibration/batch_entropy_100bins": 0.7052278361140916, "eval_calibration/batch_entropy_10bins": 0.9108287680099018, "eval_calibration/batch_entropy_50bins": 0.7827519872565041, "eval_calibration/batch_uniqueness": 0.900390625, "eval_calibration/buffer_distribution_entropy": 0.9987172905462296, "eval_calibration/buffer_entropy_100bins": 0.9987579879035519, "eval_calibration/buffer_entropy_10bins": 0.9987172905462296, "eval_calibration/buffer_entropy_50bins": 0.9988387937871521, "eval_calibration/confidence_entropy": 0.4819646086455892, "eval_calibration/coverage@0%": 0.0625, "eval_calibration/coverage@1%": 0.0625, "eval_calibration/coverage@10%": 0.1171875, "eval_calibration/coverage@15%": 0.171875, "eval_calibration/coverage@20%": 0.1796875, "eval_calibration/coverage@25%": 0.2734375, "eval_calibration/coverage@30%": 0.296875, "eval_calibration/coverage@5%": 0.0625, "eval_calibration/ece": 0.19295738677661167, "eval_calibration/mean_confidence": 0.5222269841977814, "eval_calibration/prompt_uniqueness": 0.900390625, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 339.5, "eval_completions/max_terminated_length": 339.5, "eval_completions/mean_length": 194.57038116455078, "eval_completions/mean_terminated_length": 194.57038116455078, "eval_completions/min_length": 111.5, "eval_completions/min_terminated_length": 111.5, "eval_loss": 0.0, "eval_num_tokens": 671902048.0, "eval_reward": 0.7919438034296036, "eval_reward_std": 0.2309967651963234, "eval_rewards/accuracy_reward": 0.416015625, "eval_rewards/brier_reward": 0.7947191894054413, "eval_rewards/confidence_uniqueness_reward": 0.900146484375, "eval_rewards/format_reward": 1.0, "eval_rewards/frontier_aurc_reward": -0.003882020653691143, "eval_rewards/frontier_coverage_0": 0.18356220796704292, "eval_rewards/frontier_coverage_1": 0.18356220796704292, "eval_rewards/frontier_coverage_10": 0.1832105666399002, "eval_rewards/frontier_coverage_15": 0.18193263188004494, "eval_rewards/frontier_coverage_20": 0.1427107285708189, "eval_rewards/frontier_coverage_25": 0.07165315933525562, "eval_rewards/frontier_coverage_5": 0.18356220796704292, "eval_rewards/frontier_ece_reward": 0.003705238050315529, "eval_rewards/frontier_entropy_batch_reward": -1.0, "eval_runtime": 18.6694, "eval_samples_per_second": 26.782, "eval_signal/accuracy_reward/centered_abs_mean": 0.4705810546875, "eval_signal/accuracy_reward/group_bin_occupancy": 0.25, "eval_signal/accuracy_reward/group_std_mean": 0.4925585761666298, "eval_signal/accuracy_reward/group_zero_std_frac": 0.0, "eval_signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.23529052734375, "eval_signal/accuracy_reward/weight": 0.5, "eval_signal/accuracy_reward/weighted_centered_abs_mean": 0.23529052734375, "eval_signal/advantage_abs_mean": 0.21665040776133537, "eval_signal/advantage_pre_scale_abs_mean": 0.21665040776133537, "eval_signal/advantage_pre_scale_std": 0.22837723419070244, "eval_signal/advantage_std": 0.22837723419070244, "eval_signal/brier_reward/centered_abs_mean": 0.1852213516831398, "eval_signal/brier_reward/group_bin_occupancy": 0.875, "eval_signal/brier_reward/group_std_mean": 0.23745058476924896, "eval_signal/brier_reward/group_zero_std_frac": 0.0, "eval_signal/brier_reward/scaled_weighted_centered_abs_mean": 0.018522135447710752, "eval_signal/brier_reward/weight": 0.10000000149011612, "eval_signal/brier_reward/weighted_centered_abs_mean": 0.018522135447710752, "eval_signal/confidence_uniqueness_reward/centered_abs_mean": 0.0397796630859375, "eval_signal/confidence_uniqueness_reward/group_bin_occupancy": 0.3828125, "eval_signal/confidence_uniqueness_reward/group_std_mean": 0.04554889351129532, "eval_signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "eval_signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.003977966320235282, "eval_signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "eval_signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.003977966320235282, "eval_signal/format_reward/centered_abs_mean": 0.0, "eval_signal/format_reward/group_bin_occupancy": 0.125, "eval_signal/format_reward/group_std_mean": 0.0, "eval_signal/format_reward/group_zero_std_frac": 1.0, "eval_signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/format_reward/weight": 0.5, "eval_signal/format_reward/weighted_centered_abs_mean": 0.0, "eval_signal/frontier_aurc_reward/centered_abs_mean": 0.0049892914248630404, "eval_signal/frontier_aurc_reward/group_bin_occupancy": 0.65625, "eval_signal/frontier_aurc_reward/group_std_mean": 0.009260304505005479, "eval_signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 6.236614262888907e-05, "eval_signal/frontier_aurc_reward/weight": 0.012500000186264515, "eval_signal/frontier_aurc_reward/weighted_centered_abs_mean": 6.236614262888907e-05, "eval_signal/frontier_coverage_0/centered_abs_mean": 0.3247213587164879, "eval_signal/frontier_coverage_0/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_0/group_std_mean": 0.3992055878043175, "eval_signal/frontier_coverage_0/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.004059016995597631, "eval_signal/frontier_coverage_0/weight": 0.012500000186264515, "eval_signal/frontier_coverage_0/weighted_centered_abs_mean": 0.004059016995597631, "eval_signal/frontier_coverage_1/centered_abs_mean": 0.3247213587164879, "eval_signal/frontier_coverage_1/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_1/group_std_mean": 0.3992055878043175, "eval_signal/frontier_coverage_1/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.004059016995597631, "eval_signal/frontier_coverage_1/weight": 0.012500000186264515, "eval_signal/frontier_coverage_1/weighted_centered_abs_mean": 0.004059016995597631, "eval_signal/frontier_coverage_10/centered_abs_mean": 0.3240225836634636, "eval_signal/frontier_coverage_10/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_10/group_std_mean": 0.39838140457868576, "eval_signal/frontier_coverage_10/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0040502825868315995, "eval_signal/frontier_coverage_10/weight": 0.012500000186264515, "eval_signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0040502825868315995, "eval_signal/frontier_coverage_15/centered_abs_mean": 0.3213137164711952, "eval_signal/frontier_coverage_15/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_15/group_std_mean": 0.3951790928840637, "eval_signal/frontier_coverage_15/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.004016421502456069, "eval_signal/frontier_coverage_15/weight": 0.012500000186264515, "eval_signal/frontier_coverage_15/weighted_centered_abs_mean": 0.004016421502456069, "eval_signal/frontier_coverage_20/centered_abs_mean": 0.24385424330830574, "eval_signal/frontier_coverage_20/group_bin_occupancy": 0.9453125, "eval_signal/frontier_coverage_20/group_std_mean": 0.3041408210992813, "eval_signal/frontier_coverage_20/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0030481782159768045, "eval_signal/frontier_coverage_20/weight": 0.012500000186264515, "eval_signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0030481782159768045, "eval_signal/frontier_coverage_25/centered_abs_mean": 0.10962977819144726, "eval_signal/frontier_coverage_25/group_bin_occupancy": 0.8984375, "eval_signal/frontier_coverage_25/group_std_mean": 0.14342036098241806, "eval_signal/frontier_coverage_25/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0013703722797799855, "eval_signal/frontier_coverage_25/weight": 0.012500000186264515, "eval_signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0013703722797799855, "eval_signal/frontier_coverage_5/centered_abs_mean": 0.3247213587164879, "eval_signal/frontier_coverage_5/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_5/group_std_mean": 0.3992055878043175, "eval_signal/frontier_coverage_5/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.004059016995597631, "eval_signal/frontier_coverage_5/weight": 0.012500000186264515, "eval_signal/frontier_coverage_5/weighted_centered_abs_mean": 0.004059016995597631, "eval_signal/frontier_ece_reward/centered_abs_mean": 0.0062738541746512055, "eval_signal/frontier_ece_reward/group_bin_occupancy": 0.8203125, "eval_signal/frontier_ece_reward/group_std_mean": 0.00850680354051292, "eval_signal/frontier_ece_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0006273854378378019, "eval_signal/frontier_ece_reward/weight": 0.10000000149011612, "eval_signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0006273854378378019, "eval_signal/frontier_entropy_batch_reward/centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.125, "eval_signal/frontier_entropy_batch_reward/group_std_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_zero_std_frac": 1.0, "eval_signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "eval_signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.0, "eval_steps_per_second": 0.214, "step": 200 }, { "calibration/aurc": 0.4069649690629715, "calibration/batch_distribution_entropy": 0.984386800490445, "calibration/batch_entropy_100bins": 0.9724784749244421, "calibration/batch_entropy_10bins": 0.984386800490445, "calibration/batch_entropy_50bins": 0.9800729545005282, "calibration/batch_uniqueness": 0.9540802001953125, "calibration/buffer_distribution_entropy": 0.998715097234584, "calibration/buffer_entropy_100bins": 0.9987760976482077, "calibration/buffer_entropy_10bins": 0.998715097234584, "calibration/buffer_entropy_50bins": 0.998847381635841, "calibration/confidence_entropy": 0.5199441633438218, "calibration/coverage@0%": 0.012109375, "calibration/coverage@1%": 0.012109375, "calibration/coverage@10%": 0.01484375, "calibration/coverage@15%": 0.029296875, "calibration/coverage@20%": 0.13046875, "calibration/coverage@25%": 0.20703125, "calibration/coverage@30%": 0.297265625, "calibration/coverage@5%": 0.012890625, "calibration/ece": 0.11827923226521812, "calibration/mean_confidence": 0.5078317764550387, "calibration/prompt_uniqueness": 0.859716796875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 886.8, "completions/max_terminated_length": 436.4, "completions/mean_length": 188.9830078125, "completions/mean_terminated_length": 188.71949157714843, "completions/min_length": 74.4, "completions/min_terminated_length": 74.4, "epoch": 0.656, "grad_norm": 0.001141530810855329, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 688693778.0, "reward": 0.914924168586731, "reward_std": 0.08795479238033295, "rewards/accuracy_reward": 0.50224609375, "rewards/brier_reward": 0.7855765461921692, "rewards/confidence_uniqueness_reward": 0.9532721877098084, "rewards/format_reward": 0.99951171875, "rewards/frontier_aurc_reward": -0.0035641049034893513, "rewards/frontier_coverage_0": 0.11712456494569778, "rewards/frontier_coverage_1": 0.11712456494569778, "rewards/frontier_coverage_10": 0.11691995114088058, "rewards/frontier_coverage_15": 0.11614772528409958, "rewards/frontier_coverage_20": 0.09290838539600373, "rewards/frontier_coverage_25": 0.05343174412846565, "rewards/frontier_coverage_5": 0.11712456494569778, "rewards/frontier_ece_reward": 0.0032050035893917083, "rewards/frontier_entropy_batch_reward": -0.19250327944755555, "signal/accuracy_reward/centered_abs_mean": 0.094586181640625, "signal/accuracy_reward/group_bin_occupancy": 0.169921875, "signal/accuracy_reward/group_std_mean": 0.125684654712677, "signal/accuracy_reward/group_zero_std_frac": 0.640625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0472930908203125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0472930908203125, "signal/advantage_abs_mean": 0.06804078668355942, "signal/advantage_pre_scale_abs_mean": 0.06804078668355942, "signal/advantage_pre_scale_std": 0.10926683694124222, "signal/advantage_std": 0.10926683694124222, "signal/brier_reward/centered_abs_mean": 0.12656457871198654, "signal/brier_reward/group_bin_occupancy": 0.8578125, "signal/brier_reward/group_std_mean": 0.16280275881290435, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012656458467245103, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012656458467245103, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012493164278566837, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.94921875, "signal/confidence_uniqueness_reward/group_std_mean": 0.016446012258529662, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012493164744228125, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012493164744228125, "signal/format_reward/centered_abs_mean": 0.000909423828125, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.002030306123197079, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0004547119140625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0004547119140625, "signal/frontier_aurc_reward/centered_abs_mean": 0.0035232894588261843, "signal/frontier_aurc_reward/group_bin_occupancy": 0.70546875, "signal/frontier_aurc_reward/group_std_mean": 0.0060936240945011376, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 4.404111969051883e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 4.404111969051883e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15928294360637665, "signal/frontier_coverage_0/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_0/group_std_mean": 0.204520520567894, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019910367904230953, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019910367904230953, "signal/frontier_coverage_1/centered_abs_mean": 0.15928294360637665, "signal/frontier_coverage_1/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_1/group_std_mean": 0.204520520567894, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019910367904230953, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019910367904230953, "signal/frontier_coverage_10/centered_abs_mean": 0.1589731901884079, "signal/frontier_coverage_10/group_bin_occupancy": 0.87578125, "signal/frontier_coverage_10/group_std_mean": 0.20413482785224915, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001987164909951389, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001987164909951389, "signal/frontier_coverage_15/centered_abs_mean": 0.15778957009315492, "signal/frontier_coverage_15/group_bin_occupancy": 0.87578125, "signal/frontier_coverage_15/group_std_mean": 0.20266322791576385, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001972369709983468, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001972369709983468, "signal/frontier_coverage_20/centered_abs_mean": 0.12203610986471176, "signal/frontier_coverage_20/group_bin_occupancy": 0.866796875, "signal/frontier_coverage_20/group_std_mean": 0.15783211588859558, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0015254513593390584, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0015254513593390584, "signal/frontier_coverage_25/centered_abs_mean": 0.06601603329181671, "signal/frontier_coverage_25/group_bin_occupancy": 0.912890625, "signal/frontier_coverage_25/group_std_mean": 0.08583395630121231, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0008252004394307732, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0008252004394307732, "signal/frontier_coverage_5/centered_abs_mean": 0.15928294360637665, "signal/frontier_coverage_5/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_5/group_std_mean": 0.204520520567894, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0019910367904230953, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0019910367904230953, "signal/frontier_ece_reward/centered_abs_mean": 0.004856765456497669, "signal/frontier_ece_reward/group_bin_occupancy": 0.60625, "signal/frontier_ece_reward/group_std_mean": 0.0063993423245847225, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004856765328440815, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004856765328440815, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27011048793792725, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.73984375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34519472122192385, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027011049166321754, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027011049166321754, "step": 205 }, { "calibration/aurc": 0.31224763370014774, "calibration/batch_distribution_entropy": 0.9812707489143555, "calibration/batch_entropy_100bins": 0.9685574613192488, "calibration/batch_entropy_10bins": 0.9812707489143555, "calibration/batch_entropy_50bins": 0.9778195044036864, "calibration/batch_uniqueness": 0.9537811279296875, "calibration/buffer_distribution_entropy": 0.9986814917327997, "calibration/buffer_entropy_100bins": 0.9987910254909181, "calibration/buffer_entropy_10bins": 0.9986814917327997, "calibration/buffer_entropy_50bins": 0.9988459369425353, "calibration/confidence_entropy": 0.4835208385082536, "calibration/coverage@0%": 0.028125, "calibration/coverage@1%": 0.028125, "calibration/coverage@10%": 0.12734375, "calibration/coverage@15%": 0.155078125, "calibration/coverage@20%": 0.258203125, "calibration/coverage@25%": 0.336328125, "calibration/coverage@30%": 0.43984375, "calibration/coverage@5%": 0.073828125, "calibration/ece": 0.1390745187229425, "calibration/mean_confidence": 0.5119650249220349, "calibration/prompt_uniqueness": 0.85556640625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 738.0, "completions/max_terminated_length": 507.8, "completions/mean_length": 189.82353515625, "completions/mean_terminated_length": 189.56029052734374, "completions/min_length": 94.0, "completions/min_terminated_length": 94.0, "epoch": 0.672, "grad_norm": 0.0010166987776756287, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 705551011.0, "reward": 0.9231850385665894, "reward_std": 0.08314592391252518, "rewards/accuracy_reward": 0.5166015625, "rewards/brier_reward": 0.7946917176246643, "rewards/confidence_uniqueness_reward": 0.9523738622665405, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.003188123693689704, "rewards/frontier_coverage_0": 0.13523748219013215, "rewards/frontier_coverage_1": 0.13523748219013215, "rewards/frontier_coverage_10": 0.13507505059242247, "rewards/frontier_coverage_15": 0.13440233916044236, "rewards/frontier_coverage_20": 0.11007508486509324, "rewards/frontier_coverage_25": 0.06688552126288413, "rewards/frontier_coverage_5": 0.13523748219013215, "rewards/frontier_ece_reward": 0.0035264060832560064, "rewards/frontier_entropy_batch_reward": -0.20689319372177123, "signal/accuracy_reward/centered_abs_mean": 0.09715576171875, "signal/accuracy_reward/group_bin_occupancy": 0.16953125, "signal/accuracy_reward/group_std_mean": 0.1267818480730057, "signal/accuracy_reward/group_zero_std_frac": 0.64375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.048577880859375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.048577880859375, "signal/advantage_abs_mean": 0.0644009344279766, "signal/advantage_pre_scale_abs_mean": 0.0644009344279766, "signal/advantage_pre_scale_std": 0.10230901390314102, "signal/advantage_std": 0.10230901390314102, "signal/brier_reward/centered_abs_mean": 0.1246399313211441, "signal/brier_reward/group_bin_occupancy": 0.8296875, "signal/brier_reward/group_std_mean": 0.16017679274082183, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012463993206620217, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012463993206620217, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012779261544346809, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.95625, "signal/confidence_uniqueness_reward/group_std_mean": 0.016479437798261644, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012779261451214552, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012779261451214552, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.002893506386317313, "signal/frontier_aurc_reward/group_bin_occupancy": 0.705859375, "signal/frontier_aurc_reward/group_std_mean": 0.004778713081032037, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.616883113863878e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.616883113863878e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17400326430797577, "signal/frontier_coverage_0/group_bin_occupancy": 0.85859375, "signal/frontier_coverage_0/group_std_mean": 0.2215314984321594, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.00217504077591002, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.00217504077591002, "signal/frontier_coverage_1/centered_abs_mean": 0.17400326430797577, "signal/frontier_coverage_1/group_bin_occupancy": 0.85859375, "signal/frontier_coverage_1/group_std_mean": 0.2215314984321594, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.00217504077591002, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.00217504077591002, "signal/frontier_coverage_10/centered_abs_mean": 0.17367968559265137, "signal/frontier_coverage_10/group_bin_occupancy": 0.858984375, "signal/frontier_coverage_10/group_std_mean": 0.22113430202007295, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.00217099622823298, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.00217099622823298, "signal/frontier_coverage_15/centered_abs_mean": 0.17229842245578766, "signal/frontier_coverage_15/group_bin_occupancy": 0.857421875, "signal/frontier_coverage_15/group_std_mean": 0.21944558024406433, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0021537303924560546, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0021537303924560546, "signal/frontier_coverage_20/centered_abs_mean": 0.1332439363002777, "signal/frontier_coverage_20/group_bin_occupancy": 0.84375, "signal/frontier_coverage_20/group_std_mean": 0.17046449780464173, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001665549143217504, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001665549143217504, "signal/frontier_coverage_25/centered_abs_mean": 0.07343141734600067, "signal/frontier_coverage_25/group_bin_occupancy": 0.90390625, "signal/frontier_coverage_25/group_std_mean": 0.09391821920871735, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0009178927168250084, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0009178927168250084, "signal/frontier_coverage_5/centered_abs_mean": 0.17400326430797577, "signal/frontier_coverage_5/group_bin_occupancy": 0.85859375, "signal/frontier_coverage_5/group_std_mean": 0.2215314984321594, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.00217504077591002, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.00217504077591002, "signal/frontier_ece_reward/centered_abs_mean": 0.005090892966836691, "signal/frontier_ece_reward/group_bin_occupancy": 0.58046875, "signal/frontier_ece_reward/group_std_mean": 0.00655532805249095, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0005090892896987498, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0005090892896987498, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26849772334098815, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.73828125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.343877500295639, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02684977315366268, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02684977315366268, "step": 210 }, { "calibration/aurc": 0.33956609467022403, "calibration/batch_distribution_entropy": 0.9777819756848999, "calibration/batch_entropy_100bins": 0.9697816063890488, "calibration/batch_entropy_10bins": 0.9777819756848999, "calibration/batch_entropy_50bins": 0.9775854515385525, "calibration/batch_uniqueness": 0.95316162109375, "calibration/buffer_distribution_entropy": 0.9986899187522837, "calibration/buffer_entropy_100bins": 0.998781258499992, "calibration/buffer_entropy_10bins": 0.9986899187522837, "calibration/buffer_entropy_50bins": 0.99886050495407, "calibration/confidence_entropy": 0.5071052934560327, "calibration/coverage@0%": 0.008984375, "calibration/coverage@1%": 0.008984375, "calibration/coverage@10%": 0.025, "calibration/coverage@15%": 0.167578125, "calibration/coverage@20%": 0.291015625, "calibration/coverage@25%": 0.4171875, "calibration/coverage@30%": 0.578125, "calibration/coverage@5%": 0.01015625, "calibration/ece": 0.13743849110349354, "calibration/mean_confidence": 0.4900060622097592, "calibration/prompt_uniqueness": 0.8544921875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 517.0, "completions/max_terminated_length": 517.0, "completions/mean_length": 193.9955078125, "completions/mean_terminated_length": 193.9955078125, "completions/min_length": 100.2, "completions/min_terminated_length": 100.2, "epoch": 0.688, "grad_norm": 0.0012745624408125877, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 722491445.0, "reward": 0.936010479927063, "reward_std": 0.08188026994466782, "rewards/accuracy_reward": 0.54326171875, "rewards/brier_reward": 0.7951080083847046, "rewards/confidence_uniqueness_reward": 0.9535774230957031, "rewards/format_reward": 1.0, "rewards/frontier_aurc_reward": -0.0027361839776858686, "rewards/frontier_coverage_0": 0.10980521962046623, "rewards/frontier_coverage_1": 0.10980521962046623, "rewards/frontier_coverage_10": 0.10960426107048989, "rewards/frontier_coverage_15": 0.10778558254241943, "rewards/frontier_coverage_20": 0.08492772355675697, "rewards/frontier_coverage_25": 0.0536965161561966, "rewards/frontier_coverage_5": 0.10980521962046623, "rewards/frontier_ece_reward": 0.0029097449500113727, "rewards/frontier_entropy_batch_reward": -0.19313570261001586, "signal/accuracy_reward/centered_abs_mean": 0.093548583984375, "signal/accuracy_reward/group_bin_occupancy": 0.172265625, "signal/accuracy_reward/group_std_mean": 0.12751732766628265, "signal/accuracy_reward/group_zero_std_frac": 0.621875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0467742919921875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0467742919921875, "signal/advantage_abs_mean": 0.06265989765524864, "signal/advantage_pre_scale_abs_mean": 0.06265989765524864, "signal/advantage_pre_scale_std": 0.10179222971200944, "signal/advantage_std": 0.10179222971200944, "signal/brier_reward/centered_abs_mean": 0.11799525022506714, "signal/brier_reward/group_bin_occupancy": 0.859375, "signal/brier_reward/group_std_mean": 0.1507485955953598, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011799525283277035, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011799525283277035, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011584711074829102, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9640625, "signal/confidence_uniqueness_reward/group_std_mean": 0.014366124570369721, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011584711726754903, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011584711726754903, "signal/format_reward/centered_abs_mean": 0.0, "signal/format_reward/group_bin_occupancy": 0.125, "signal/format_reward/group_std_mean": 0.0, "signal/format_reward/group_zero_std_frac": 1.0, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0, "signal/frontier_aurc_reward/centered_abs_mean": 0.002534355386160314, "signal/frontier_aurc_reward/group_bin_occupancy": 0.70859375, "signal/frontier_aurc_reward/group_std_mean": 0.004396425699815154, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.167944196320604e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.167944196320604e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16423133611679078, "signal/frontier_coverage_0/group_bin_occupancy": 0.88046875, "signal/frontier_coverage_0/group_std_mean": 0.20978615581989288, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002052891789935529, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002052891789935529, "signal/frontier_coverage_1/centered_abs_mean": 0.16423133611679078, "signal/frontier_coverage_1/group_bin_occupancy": 0.88046875, "signal/frontier_coverage_1/group_std_mean": 0.20978615581989288, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002052891789935529, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002052891789935529, "signal/frontier_coverage_10/centered_abs_mean": 0.16396727859973909, "signal/frontier_coverage_10/group_bin_occupancy": 0.880859375, "signal/frontier_coverage_10/group_std_mean": 0.2094414174556732, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002049591043032706, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002049591043032706, "signal/frontier_coverage_15/centered_abs_mean": 0.16176777482032775, "signal/frontier_coverage_15/group_bin_occupancy": 0.88046875, "signal/frontier_coverage_15/group_std_mean": 0.20658864378929137, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.002022097306326032, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.002022097306326032, "signal/frontier_coverage_20/centered_abs_mean": 0.12009549140930176, "signal/frontier_coverage_20/group_bin_occupancy": 0.8765625, "signal/frontier_coverage_20/group_std_mean": 0.15400693118572234, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0015011936426162719, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0015011936426162719, "signal/frontier_coverage_25/centered_abs_mean": 0.06709338203072548, "signal/frontier_coverage_25/group_bin_occupancy": 0.90546875, "signal/frontier_coverage_25/group_std_mean": 0.08632016777992249, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0008386673056520521, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0008386673056520521, "signal/frontier_coverage_5/centered_abs_mean": 0.16423133611679078, "signal/frontier_coverage_5/group_bin_occupancy": 0.88046875, "signal/frontier_coverage_5/group_std_mean": 0.20978615581989288, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002052891789935529, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002052891789935529, "signal/frontier_ece_reward/centered_abs_mean": 0.004593132715672255, "signal/frontier_ece_reward/group_bin_occupancy": 0.59140625, "signal/frontier_ece_reward/group_std_mean": 0.005961979646235705, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004593132820446044, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004593132820446044, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.263191482424736, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.7359375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.33676049709320066, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026319148391485213, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026319148391485213, "step": 215 }, { "calibration/aurc": 0.25694371238113783, "calibration/batch_distribution_entropy": 0.9755123216572388, "calibration/batch_entropy_100bins": 0.9691040477993896, "calibration/batch_entropy_10bins": 0.9755123216572388, "calibration/batch_entropy_50bins": 0.9767746849705027, "calibration/batch_uniqueness": 0.9531938857708202, "calibration/buffer_distribution_entropy": 0.9986875087129239, "calibration/buffer_entropy_100bins": 0.9987532406753752, "calibration/buffer_entropy_10bins": 0.9986875087129239, "calibration/buffer_entropy_50bins": 0.9988574370768362, "calibration/confidence_entropy": 0.49334564066115816, "calibration/coverage@0%": 0.007813264432485322, "calibration/coverage@1%": 0.007813264432485322, "calibration/coverage@10%": 0.11135182240704503, "calibration/coverage@15%": 0.21955571183953032, "calibration/coverage@20%": 0.40287655944227, "calibration/coverage@25%": 0.5440007338551859, "calibration/coverage@30%": 0.6354398544520548, "calibration/coverage@5%": 0.017578889432485322, "calibration/ece": 0.11242217169316457, "calibration/mean_confidence": 0.5120740217635475, "calibration/prompt_uniqueness": 0.8453241354220864, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 736.0, "completions/max_terminated_length": 528.6, "completions/mean_length": 194.155859375, "completions/mean_terminated_length": 193.89388122558594, "completions/min_length": 99.8, "completions/min_terminated_length": 99.8, "epoch": 0.704, "grad_norm": 0.0011202789610251784, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 739345745.0, "reward": 0.9372921228408814, "reward_std": 0.08125913888216019, "rewards/accuracy_reward": 0.54267578125, "rewards/brier_reward": 0.8051641941070556, "rewards/confidence_uniqueness_reward": 0.9533455967903137, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.0025887496070936323, "rewards/frontier_coverage_0": 0.11237857863306999, "rewards/frontier_coverage_1": 0.11237857863306999, "rewards/frontier_coverage_10": 0.11224546730518341, "rewards/frontier_coverage_15": 0.11079215556383133, "rewards/frontier_coverage_20": 0.08878249004483223, "rewards/frontier_coverage_25": 0.05674021542072296, "rewards/frontier_coverage_5": 0.11235189475119114, "rewards/frontier_ece_reward": 0.00307169328443706, "rewards/frontier_entropy_batch_reward": -0.18894782066345214, "signal/accuracy_reward/centered_abs_mean": 0.083758544921875, "signal/accuracy_reward/group_bin_occupancy": 0.166796875, "signal/accuracy_reward/group_std_mean": 0.11209209114313126, "signal/accuracy_reward/group_zero_std_frac": 0.665625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0418792724609375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0418792724609375, "signal/advantage_abs_mean": 0.06274550780653954, "signal/advantage_pre_scale_abs_mean": 0.06274550780653954, "signal/advantage_pre_scale_std": 0.10022006779909134, "signal/advantage_std": 0.10022006779909134, "signal/brier_reward/centered_abs_mean": 0.11893865019083023, "signal/brier_reward/group_bin_occupancy": 0.86640625, "signal/brier_reward/group_std_mean": 0.15209771692752838, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011893865279853344, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011893865279853344, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012048188224434852, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.95390625, "signal/confidence_uniqueness_reward/group_std_mean": 0.015659395046532153, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012048188596963882, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012048188596963882, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.002481231279671192, "signal/frontier_aurc_reward/group_bin_occupancy": 0.72109375, "signal/frontier_aurc_reward/group_std_mean": 0.004121466958895326, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.1015391505206936e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.1015391505206936e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15765490531921386, "signal/frontier_coverage_0/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_0/group_std_mean": 0.19848573505878447, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019706863909959792, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019706863909959792, "signal/frontier_coverage_1/centered_abs_mean": 0.15765490531921386, "signal/frontier_coverage_1/group_bin_occupancy": 0.886328125, "signal/frontier_coverage_1/group_std_mean": 0.19848573505878447, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019706863909959792, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019706863909959792, "signal/frontier_coverage_10/centered_abs_mean": 0.15719155073165894, "signal/frontier_coverage_10/group_bin_occupancy": 0.8859375, "signal/frontier_coverage_10/group_std_mean": 0.19792267084121704, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0019648944726213813, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0019648944726213813, "signal/frontier_coverage_15/centered_abs_mean": 0.15138939321041106, "signal/frontier_coverage_15/group_bin_occupancy": 0.880859375, "signal/frontier_coverage_15/group_std_mean": 0.190908744931221, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001892367540858686, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001892367540858686, "signal/frontier_coverage_20/centered_abs_mean": 0.11000291109085084, "signal/frontier_coverage_20/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_20/group_std_mean": 0.1399712085723877, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001375036407262087, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001375036407262087, "signal/frontier_coverage_25/centered_abs_mean": 0.06324589103460312, "signal/frontier_coverage_25/group_bin_occupancy": 0.915625, "signal/frontier_coverage_25/group_std_mean": 0.08118669390678405, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.000790573691483587, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.000790573691483587, "signal/frontier_coverage_5/centered_abs_mean": 0.15748987197875977, "signal/frontier_coverage_5/group_bin_occupancy": 0.8859375, "signal/frontier_coverage_5/group_std_mean": 0.19828707873821258, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0019686234649270774, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0019686234649270774, "signal/frontier_ece_reward/centered_abs_mean": 0.0045226853340864185, "signal/frontier_ece_reward/group_bin_occupancy": 0.596484375, "signal/frontier_ece_reward/group_std_mean": 0.005821056663990021, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.000452268507797271, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.000452268507797271, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26287021636962893, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.73828125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3382034659385681, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02628702260553837, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02628702260553837, "step": 220 }, { "calibration/aurc": 0.24410758068936075, "calibration/batch_distribution_entropy": 0.9902223360233717, "calibration/batch_entropy_100bins": 0.9749508985569101, "calibration/batch_entropy_10bins": 0.9902223360233717, "calibration/batch_entropy_50bins": 0.982083200775563, "calibration/batch_uniqueness": 0.9553034992913328, "calibration/buffer_distribution_entropy": 0.9986947194306925, "calibration/buffer_entropy_100bins": 0.9987722009070856, "calibration/buffer_entropy_10bins": 0.9986947194306925, "calibration/buffer_entropy_50bins": 0.9988598558545136, "calibration/confidence_entropy": 0.49363974004994526, "calibration/coverage@0%": 0.04375076443248532, "calibration/coverage@1%": 0.053125764432485324, "calibration/coverage@10%": 0.2335945144324853, "calibration/coverage@15%": 0.3425788894324853, "calibration/coverage@20%": 0.4457038894324853, "calibration/coverage@25%": 0.5449226394324853, "calibration/coverage@30%": 0.6683601394324853, "calibration/coverage@5%": 0.12929763943248532, "calibration/ece": 0.1235796989243186, "calibration/mean_confidence": 0.5142428710823418, "calibration/prompt_uniqueness": 0.8502101489334027, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1087.8, "completions/max_terminated_length": 472.2, "completions/mean_length": 194.4005859375, "completions/mean_terminated_length": 193.87747192382812, "completions/min_length": 95.6, "completions/min_terminated_length": 95.6, "epoch": 0.72, "grad_norm": 0.0009790080366656184, "learning_rate": 1e-06, "loss": 0.0018, "num_tokens": 756346263.0, "reward": 0.9469334363937378, "reward_std": 0.08140352815389633, "rewards/accuracy_reward": 0.56572265625, "rewards/brier_reward": 0.806804609298706, "rewards/confidence_uniqueness_reward": 0.9528944611549377, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.002266845665872097, "rewards/frontier_coverage_0": 0.09872241467237472, "rewards/frontier_coverage_1": 0.09872241467237472, "rewards/frontier_coverage_10": 0.09856997728347779, "rewards/frontier_coverage_15": 0.09556975364685058, "rewards/frontier_coverage_20": 0.07581645771861076, "rewards/frontier_coverage_25": 0.05162230357527733, "rewards/frontier_coverage_5": 0.09866813272237777, "rewards/frontier_ece_reward": 0.0029384993948042394, "rewards/frontier_entropy_batch_reward": -0.19689127802848816, "signal/accuracy_reward/centered_abs_mean": 0.086151123046875, "signal/accuracy_reward/group_bin_occupancy": 0.16796875, "signal/accuracy_reward/group_std_mean": 0.11651939004659653, "signal/accuracy_reward/group_zero_std_frac": 0.65625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0430755615234375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0430755615234375, "signal/advantage_abs_mean": 0.06210223212838173, "signal/advantage_pre_scale_abs_mean": 0.06210223212838173, "signal/advantage_pre_scale_std": 0.10064510852098466, "signal/advantage_std": 0.10064510852098466, "signal/brier_reward/centered_abs_mean": 0.11116381585597992, "signal/brier_reward/group_bin_occupancy": 0.8546875, "signal/brier_reward/group_std_mean": 0.14320469200611113, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011116381734609604, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011116381734609604, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012627355940639972, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.944140625, "signal/confidence_uniqueness_reward/group_std_mean": 0.017049738205969334, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012627356685698033, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012627356685698033, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086533904076, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0021892196498811247, "signal/frontier_aurc_reward/group_bin_occupancy": 0.71796875, "signal/frontier_aurc_reward/group_std_mean": 0.0037096437066793443, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.7365245114197023e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.7365245114197023e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15539000034332276, "signal/frontier_coverage_0/group_bin_occupancy": 0.8828125, "signal/frontier_coverage_0/group_std_mean": 0.19757642447948456, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019423750694841147, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019423750694841147, "signal/frontier_coverage_1/centered_abs_mean": 0.15539000034332276, "signal/frontier_coverage_1/group_bin_occupancy": 0.8828125, "signal/frontier_coverage_1/group_std_mean": 0.19757642447948456, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019423750694841147, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019423750694841147, "signal/frontier_coverage_10/centered_abs_mean": 0.1548478752374649, "signal/frontier_coverage_10/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_10/group_std_mean": 0.19690344035625457, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0019355985103175044, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0019355985103175044, "signal/frontier_coverage_15/centered_abs_mean": 0.1453756719827652, "signal/frontier_coverage_15/group_bin_occupancy": 0.87734375, "signal/frontier_coverage_15/group_std_mean": 0.18513388633728028, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0018171959090977907, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0018171959090977907, "signal/frontier_coverage_20/centered_abs_mean": 0.10124388188123704, "signal/frontier_coverage_20/group_bin_occupancy": 0.878515625, "signal/frontier_coverage_20/group_std_mean": 0.12981236577033997, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.00126554851885885, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.00126554851885885, "signal/frontier_coverage_25/centered_abs_mean": 0.058337075263261796, "signal/frontier_coverage_25/group_bin_occupancy": 0.921875, "signal/frontier_coverage_25/group_std_mean": 0.07468763291835785, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007292134570889175, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007292134570889175, "signal/frontier_coverage_5/centered_abs_mean": 0.1553172081708908, "signal/frontier_coverage_5/group_bin_occupancy": 0.8828125, "signal/frontier_coverage_5/group_std_mean": 0.19748497605323792, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0019414650974795221, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0019414650974795221, "signal/frontier_ece_reward/centered_abs_mean": 0.004282759223133326, "signal/frontier_ece_reward/group_bin_occupancy": 0.607421875, "signal/frontier_ece_reward/group_std_mean": 0.005443438142538071, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004282759327907115, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004282759327907115, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27192082405090334, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.733984375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34859787225723265, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02719208337366581, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02719208337366581, "step": 225 }, { "calibration/aurc": 0.2659348583300164, "calibration/batch_distribution_entropy": 0.9791736660714812, "calibration/batch_entropy_100bins": 0.9688359377133906, "calibration/batch_entropy_10bins": 0.9791736660714812, "calibration/batch_entropy_50bins": 0.9770796411069307, "calibration/batch_uniqueness": 0.9527679443359375, "calibration/buffer_distribution_entropy": 0.9987280854955143, "calibration/buffer_entropy_100bins": 0.9987886911789593, "calibration/buffer_entropy_10bins": 0.9987280854955143, "calibration/buffer_entropy_50bins": 0.9988642803021669, "calibration/confidence_entropy": 0.47431208190112073, "calibration/coverage@0%": 0.020703125, "calibration/coverage@1%": 0.020703125, "calibration/coverage@10%": 0.13515625, "calibration/coverage@15%": 0.1875, "calibration/coverage@20%": 0.400390625, "calibration/coverage@25%": 0.49375, "calibration/coverage@30%": 0.593359375, "calibration/coverage@5%": 0.05078125, "calibration/ece": 0.1259610923546552, "calibration/mean_confidence": 0.5462464850096745, "calibration/prompt_uniqueness": 0.846875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 511.4, "completions/max_terminated_length": 511.4, "completions/mean_length": 194.6498046875, "completions/mean_terminated_length": 194.6498046875, "completions/min_length": 100.6, "completions/min_terminated_length": 100.6, "epoch": 0.736, "grad_norm": 0.001196362660266459, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 773279061.0, "reward": 0.9465953826904296, "reward_std": 0.07829982340335846, "rewards/accuracy_reward": 0.56357421875, "rewards/brier_reward": 0.798375380039215, "rewards/confidence_uniqueness_reward": 0.9540367126464844, "rewards/format_reward": 1.0, "rewards/frontier_aurc_reward": -0.0031000898219645024, "rewards/frontier_coverage_0": 0.10112178847193717, "rewards/frontier_coverage_1": 0.10112178847193717, "rewards/frontier_coverage_10": 0.10094411000609398, "rewards/frontier_coverage_15": 0.09661712646484374, "rewards/frontier_coverage_20": 0.0787421464920044, "rewards/frontier_coverage_25": 0.056681372225284576, "rewards/frontier_coverage_5": 0.10109797269105911, "rewards/frontier_ece_reward": 0.0027812039479613304, "rewards/frontier_entropy_batch_reward": -0.1862635552883148, "signal/accuracy_reward/centered_abs_mean": 0.079498291015625, "signal/accuracy_reward/group_bin_occupancy": 0.165625, "signal/accuracy_reward/group_std_mean": 0.10849804282188416, "signal/accuracy_reward/group_zero_std_frac": 0.675, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0397491455078125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0397491455078125, "signal/advantage_abs_mean": 0.06039750427007675, "signal/advantage_pre_scale_abs_mean": 0.06039750427007675, "signal/advantage_pre_scale_std": 0.09750747084617614, "signal/advantage_std": 0.09750747084617614, "signal/brier_reward/centered_abs_mean": 0.11960845291614533, "signal/brier_reward/group_bin_occupancy": 0.846875, "signal/brier_reward/group_std_mean": 0.15357136130332946, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011960845440626144, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011960845440626144, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011609864234924317, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.964453125, "signal/confidence_uniqueness_reward/group_std_mean": 0.014429912157356738, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001160986442118883, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001160986442118883, "signal/format_reward/centered_abs_mean": 0.0, "signal/format_reward/group_bin_occupancy": 0.125, "signal/format_reward/group_std_mean": 0.0, "signal/format_reward/group_zero_std_frac": 1.0, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0, "signal/frontier_aurc_reward/centered_abs_mean": 0.003057365911081433, "signal/frontier_aurc_reward/group_bin_occupancy": 0.7234375, "signal/frontier_aurc_reward/group_std_mean": 0.005309991631656885, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.821707505267113e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.821707505267113e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.1546383649110794, "signal/frontier_coverage_0/group_bin_occupancy": 0.86875, "signal/frontier_coverage_0/group_std_mean": 0.19802598655223846, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019329796545207501, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019329796545207501, "signal/frontier_coverage_1/centered_abs_mean": 0.1546383649110794, "signal/frontier_coverage_1/group_bin_occupancy": 0.86875, "signal/frontier_coverage_1/group_std_mean": 0.19802598655223846, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019329796545207501, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019329796545207501, "signal/frontier_coverage_10/centered_abs_mean": 0.15410158336162566, "signal/frontier_coverage_10/group_bin_occupancy": 0.86875, "signal/frontier_coverage_10/group_std_mean": 0.1973479300737381, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0019262698944658041, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0019262698944658041, "signal/frontier_coverage_15/centered_abs_mean": 0.14173146486282348, "signal/frontier_coverage_15/group_bin_occupancy": 0.865234375, "signal/frontier_coverage_15/group_std_mean": 0.18179621696472167, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017716433154419064, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017716433154419064, "signal/frontier_coverage_20/centered_abs_mean": 0.09922464936971664, "signal/frontier_coverage_20/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_20/group_std_mean": 0.1279410809278488, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0012403082335367798, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0012403082335367798, "signal/frontier_coverage_25/centered_abs_mean": 0.06166926845908165, "signal/frontier_coverage_25/group_bin_occupancy": 0.93359375, "signal/frontier_coverage_25/group_std_mean": 0.07894663363695145, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007708658929914236, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007708658929914236, "signal/frontier_coverage_5/centered_abs_mean": 0.15458077192306519, "signal/frontier_coverage_5/group_bin_occupancy": 0.86875, "signal/frontier_coverage_5/group_std_mean": 0.19795386791229247, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0019322596490383148, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0019322596490383148, "signal/frontier_ece_reward/centered_abs_mean": 0.004501782171428203, "signal/frontier_ece_reward/group_bin_occupancy": 0.5984375, "signal/frontier_ece_reward/group_std_mean": 0.0057001769542694095, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00045017823576927186, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00045017823576927186, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26178981065750123, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.7328125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.338212788105011, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026178980991244317, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026178980991244317, "step": 230 }, { "calibration/aurc": 0.2772283302476903, "calibration/batch_distribution_entropy": 0.9713057418477284, "calibration/batch_entropy_100bins": 0.9642264486835341, "calibration/batch_entropy_10bins": 0.9713057418477284, "calibration/batch_entropy_50bins": 0.9722775741926162, "calibration/batch_uniqueness": 0.9512542112203214, "calibration/buffer_distribution_entropy": 0.9987307345646208, "calibration/buffer_entropy_100bins": 0.9987792165907727, "calibration/buffer_entropy_10bins": 0.9987307345646208, "calibration/buffer_entropy_50bins": 0.9988563466393927, "calibration/confidence_entropy": 0.4661328198851214, "calibration/coverage@0%": 0.008988197162426614, "calibration/coverage@1%": 0.008988197162426614, "calibration/coverage@10%": 0.060160072162426616, "calibration/coverage@15%": 0.2336006298923679, "calibration/coverage@20%": 0.37241468933463795, "calibration/coverage@25%": 0.5111683586105675, "calibration/coverage@30%": 0.6131711717221136, "calibration/coverage@5%": 0.008988197162426614, "calibration/ece": 0.12689066643542019, "calibration/mean_confidence": 0.46690607059943084, "calibration/prompt_uniqueness": 0.8444940572970865, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 902.4, "completions/max_terminated_length": 482.6, "completions/mean_length": 194.7859375, "completions/mean_terminated_length": 194.26220703125, "completions/min_length": 97.2, "completions/min_terminated_length": 97.2, "epoch": 0.752, "grad_norm": 0.0008059156243689358, "learning_rate": 1e-06, "loss": 0.0009, "num_tokens": 790500869.0, "reward": 0.9452716708183289, "reward_std": 0.07896565645933151, "rewards/accuracy_reward": 0.5607421875, "rewards/brier_reward": 0.8016233563423156, "rewards/confidence_uniqueness_reward": 0.9531951069831848, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.0028346756007522345, "rewards/frontier_coverage_0": 0.11009920351207256, "rewards/frontier_coverage_1": 0.11009920351207256, "rewards/frontier_coverage_10": 0.10985869988799095, "rewards/frontier_coverage_15": 0.10413307882845402, "rewards/frontier_coverage_20": 0.07679208926856518, "rewards/frontier_coverage_25": 0.05207032486796379, "rewards/frontier_coverage_5": 0.11007008776068687, "rewards/frontier_ece_reward": 0.0029186978936195374, "rewards/frontier_entropy_batch_reward": -0.19056436717510222, "signal/accuracy_reward/centered_abs_mean": 0.0796630859375, "signal/accuracy_reward/group_bin_occupancy": 0.167578125, "signal/accuracy_reward/group_std_mean": 0.11097259819507599, "signal/accuracy_reward/group_zero_std_frac": 0.659375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.03983154296875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.03983154296875, "signal/advantage_abs_mean": 0.05965143889188766, "signal/advantage_pre_scale_abs_mean": 0.05965143889188766, "signal/advantage_pre_scale_std": 0.09786994606256486, "signal/advantage_std": 0.09786994606256486, "signal/brier_reward/centered_abs_mean": 0.11516528874635697, "signal/brier_reward/group_bin_occupancy": 0.8328125, "signal/brier_reward/group_std_mean": 0.14950103163719178, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011516529507935047, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011516529507935047, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012155953608453274, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.953515625, "signal/confidence_uniqueness_reward/group_std_mean": 0.01616560444235802, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001215595379471779, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001215595379471779, "signal/format_reward/centered_abs_mean": 0.00074462890625, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0018734002020210027, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000372314453125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000372314453125, "signal/frontier_aurc_reward/centered_abs_mean": 0.002739709848538041, "signal/frontier_aurc_reward/group_bin_occupancy": 0.712109375, "signal/frontier_aurc_reward/group_std_mean": 0.004523551557213068, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.424637325224467e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.424637325224467e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.154658043384552, "signal/frontier_coverage_0/group_bin_occupancy": 0.85546875, "signal/frontier_coverage_0/group_std_mean": 0.20031827986240386, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019332255935296417, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019332255935296417, "signal/frontier_coverage_1/centered_abs_mean": 0.154658043384552, "signal/frontier_coverage_1/group_bin_occupancy": 0.85546875, "signal/frontier_coverage_1/group_std_mean": 0.20031827986240386, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019332255935296417, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019332255935296417, "signal/frontier_coverage_10/centered_abs_mean": 0.1541069507598877, "signal/frontier_coverage_10/group_bin_occupancy": 0.8546875, "signal/frontier_coverage_10/group_std_mean": 0.19962221682071685, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001926336926408112, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001926336926408112, "signal/frontier_coverage_15/centered_abs_mean": 0.14271292537450792, "signal/frontier_coverage_15/group_bin_occupancy": 0.849609375, "signal/frontier_coverage_15/group_std_mean": 0.1852043628692627, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001783911627717316, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001783911627717316, "signal/frontier_coverage_20/centered_abs_mean": 0.09863460063934326, "signal/frontier_coverage_20/group_bin_occupancy": 0.8640625, "signal/frontier_coverage_20/group_std_mean": 0.12887403964996338, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0012329325312748552, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0012329325312748552, "signal/frontier_coverage_25/centered_abs_mean": 0.05909756124019623, "signal/frontier_coverage_25/group_bin_occupancy": 0.91875, "signal/frontier_coverage_25/group_std_mean": 0.07642732262611389, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007387195248156786, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007387195248156786, "signal/frontier_coverage_5/centered_abs_mean": 0.15456025600433348, "signal/frontier_coverage_5/group_bin_occupancy": 0.85546875, "signal/frontier_coverage_5/group_std_mean": 0.2001951366662979, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0019320032326504589, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0019320032326504589, "signal/frontier_ece_reward/centered_abs_mean": 0.004383979924023151, "signal/frontier_ece_reward/group_bin_occupancy": 0.582421875, "signal/frontier_ece_reward/group_std_mean": 0.005572784133255482, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00043839800055138766, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00043839800055138766, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.25986859798431394, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.72734375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.33691503405570983, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.025986860692501067, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.025986860692501067, "step": 235 }, { "calibration/aurc": 0.27125758629249114, "calibration/batch_distribution_entropy": 0.9868716997354584, "calibration/batch_entropy_100bins": 0.9729941083707784, "calibration/batch_entropy_10bins": 0.9868716997354584, "calibration/batch_entropy_50bins": 0.9803523750251664, "calibration/batch_uniqueness": 0.9542999267578125, "calibration/buffer_distribution_entropy": 0.9989236501322223, "calibration/buffer_entropy_100bins": 0.99886782733087, "calibration/buffer_entropy_10bins": 0.9989236501322223, "calibration/buffer_entropy_50bins": 0.9989652423459281, "calibration/confidence_entropy": 0.4985706565252402, "calibration/coverage@0%": 0.0484375, "calibration/coverage@1%": 0.0484375, "calibration/coverage@10%": 0.251953125, "calibration/coverage@15%": 0.3109375, "calibration/coverage@20%": 0.395703125, "calibration/coverage@25%": 0.48203125, "calibration/coverage@30%": 0.537109375, "calibration/coverage@5%": 0.198046875, "calibration/ece": 0.16223442060832519, "calibration/mean_confidence": 0.4842945512823123, "calibration/prompt_uniqueness": 0.85478515625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 706.4, "completions/max_terminated_length": 488.0, "completions/mean_length": 201.5484375, "completions/mean_terminated_length": 201.4177276611328, "completions/min_length": 101.2, "completions/min_terminated_length": 101.2, "epoch": 0.768, "grad_norm": 0.0011189074721187353, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 807497429.0, "reward": 0.9230096817016602, "reward_std": 0.08094646781682968, "rewards/accuracy_reward": 0.509375, "rewards/brier_reward": 0.8112827181816101, "rewards/confidence_uniqueness_reward": 0.9531510829925537, "rewards/format_reward": 0.99990234375, "rewards/frontier_aurc_reward": -0.0026364182587713, "rewards/frontier_coverage_0": 0.1476791650056839, "rewards/frontier_coverage_1": 0.1476791650056839, "rewards/frontier_coverage_10": 0.14736358374357222, "rewards/frontier_coverage_15": 0.14083798974752426, "rewards/frontier_coverage_20": 0.10304524302482605, "rewards/frontier_coverage_25": 0.06215212792158127, "rewards/frontier_coverage_5": 0.14767136573791503, "rewards/frontier_ece_reward": 0.0030877976678311825, "rewards/frontier_entropy_batch_reward": -0.19553562998771667, "signal/accuracy_reward/centered_abs_mean": 0.08564453125, "signal/accuracy_reward/group_bin_occupancy": 0.16875, "signal/accuracy_reward/group_std_mean": 0.11769998520612716, "signal/accuracy_reward/group_zero_std_frac": 0.65, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.042822265625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.042822265625, "signal/advantage_abs_mean": 0.06164178028702736, "signal/advantage_pre_scale_abs_mean": 0.06164178028702736, "signal/advantage_pre_scale_std": 0.10060565769672394, "signal/advantage_std": 0.10060565769672394, "signal/brier_reward/centered_abs_mean": 0.11535971462726594, "signal/brier_reward/group_bin_occupancy": 0.84921875, "signal/brier_reward/group_std_mean": 0.14856164455413817, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01153597179800272, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01153597179800272, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01174762025475502, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.962890625, "signal/confidence_uniqueness_reward/group_std_mean": 0.014978520385921002, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011747620534151792, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011747620534151792, "signal/format_reward/centered_abs_mean": 0.000189208984375, "signal/format_reward/group_bin_occupancy": 0.125390625, "signal/format_reward/group_std_mean": 0.0005524271633476019, "signal/format_reward/group_zero_std_frac": 0.996875, "signal/format_reward/scaled_weighted_centered_abs_mean": 9.46044921875e-05, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 9.46044921875e-05, "signal/frontier_aurc_reward/centered_abs_mean": 0.002402938576415181, "signal/frontier_aurc_reward/group_bin_occupancy": 0.711328125, "signal/frontier_aurc_reward/group_std_mean": 0.00410534543916583, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.0036732641747222e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.0036732641747222e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16049768924713134, "signal/frontier_coverage_0/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_0/group_std_mean": 0.20674357414245606, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0020062211668118836, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0020062211668118836, "signal/frontier_coverage_1/centered_abs_mean": 0.16049768924713134, "signal/frontier_coverage_1/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_1/group_std_mean": 0.20674357414245606, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0020062211668118836, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0020062211668118836, "signal/frontier_coverage_10/centered_abs_mean": 0.15986967086791992, "signal/frontier_coverage_10/group_bin_occupancy": 0.87734375, "signal/frontier_coverage_10/group_std_mean": 0.20592574775218964, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001998370815999806, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001998370815999806, "signal/frontier_coverage_15/centered_abs_mean": 0.1510760545730591, "signal/frontier_coverage_15/group_bin_occupancy": 0.87265625, "signal/frontier_coverage_15/group_std_mean": 0.19448437094688414, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0018884507240727543, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0018884507240727543, "signal/frontier_coverage_20/centered_abs_mean": 0.10296814739704133, "signal/frontier_coverage_20/group_bin_occupancy": 0.877734375, "signal/frontier_coverage_20/group_std_mean": 0.1326945424079895, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0012871018843725324, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0012871018843725324, "signal/frontier_coverage_25/centered_abs_mean": 0.060876214504241945, "signal/frontier_coverage_25/group_bin_occupancy": 0.9328125, "signal/frontier_coverage_25/group_std_mean": 0.07796305119991302, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007609526976011694, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007609526976011694, "signal/frontier_coverage_5/centered_abs_mean": 0.1604850858449936, "signal/frontier_coverage_5/group_bin_occupancy": 0.87890625, "signal/frontier_coverage_5/group_std_mean": 0.20672679245471953, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0020060636568814518, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0020060636568814518, "signal/frontier_ece_reward/centered_abs_mean": 0.0040759890340268615, "signal/frontier_ece_reward/group_bin_occupancy": 0.58046875, "signal/frontier_ece_reward/group_std_mean": 0.005164286121726036, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00040759891853667793, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00040759891853667793, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2600819855928421, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.75, "signal/frontier_entropy_batch_reward/group_std_mean": 0.33443655371665953, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02600819952785969, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02600819952785969, "step": 240 }, { "calibration/aurc": 0.3198998639691154, "calibration/batch_distribution_entropy": 0.9850544929840517, "calibration/batch_entropy_100bins": 0.9723581437579284, "calibration/batch_entropy_10bins": 0.9850544929840517, "calibration/batch_entropy_50bins": 0.9814071704429965, "calibration/batch_uniqueness": 0.954315185546875, "calibration/buffer_distribution_entropy": 0.9989797666145449, "calibration/buffer_entropy_100bins": 0.9989006942795084, "calibration/buffer_entropy_10bins": 0.9989797666145449, "calibration/buffer_entropy_50bins": 0.9989976597501004, "calibration/confidence_entropy": 0.48170084795223217, "calibration/coverage@0%": 0.003515625, "calibration/coverage@1%": 0.046875, "calibration/coverage@10%": 0.16640625, "calibration/coverage@15%": 0.257421875, "calibration/coverage@20%": 0.298046875, "calibration/coverage@25%": 0.34765625, "calibration/coverage@30%": 0.391796875, "calibration/coverage@5%": 0.123828125, "calibration/ece": 0.147533263639873, "calibration/mean_confidence": 0.5226586992664695, "calibration/prompt_uniqueness": 0.84873046875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 836.0, "completions/max_terminated_length": 630.2, "completions/mean_length": 203.60908203125, "completions/mean_terminated_length": 203.47868957519532, "completions/min_length": 104.8, "completions/min_terminated_length": 104.8, "epoch": 0.784, "grad_norm": 0.0009595006122253835, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 824756754.0, "reward": 0.9408976554870605, "reward_std": 0.08104828149080276, "rewards/accuracy_reward": 0.55498046875, "rewards/brier_reward": 0.7876966118812561, "rewards/confidence_uniqueness_reward": 0.9538962721824646, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.0032073096837848423, "rewards/frontier_coverage_0": 0.09118822515010834, "rewards/frontier_coverage_1": 0.09118822515010834, "rewards/frontier_coverage_10": 0.0908893346786499, "rewards/frontier_coverage_15": 0.08770230486989021, "rewards/frontier_coverage_20": 0.06613813489675521, "rewards/frontier_coverage_25": 0.046408718824386595, "rewards/frontier_coverage_5": 0.0911907359957695, "rewards/frontier_ece_reward": 0.001996941142715514, "rewards/frontier_entropy_batch_reward": -0.17872633039951324, "signal/accuracy_reward/centered_abs_mean": 0.089935302734375, "signal/accuracy_reward/group_bin_occupancy": 0.16640625, "signal/accuracy_reward/group_std_mean": 0.11736238300800324, "signal/accuracy_reward/group_zero_std_frac": 0.66875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0449676513671875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0449676513671875, "signal/advantage_abs_mean": 0.06319243982434272, "signal/advantage_pre_scale_abs_mean": 0.06319243982434272, "signal/advantage_pre_scale_std": 0.10230232626199723, "signal/advantage_std": 0.10230232626199723, "signal/brier_reward/centered_abs_mean": 0.11719895899295807, "signal/brier_reward/group_bin_occupancy": 0.866015625, "signal/brier_reward/group_std_mean": 0.1497897356748581, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011719896271824837, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011719896271824837, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01153520792722702, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9546875, "signal/confidence_uniqueness_reward/group_std_mean": 0.015096276812255383, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011535208206623793, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011535208206623793, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0029886286240071057, "signal/frontier_aurc_reward/group_bin_occupancy": 0.723046875, "signal/frontier_aurc_reward/group_std_mean": 0.005069928430020809, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.7357858673203737e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.7357858673203737e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15531314313411712, "signal/frontier_coverage_0/group_bin_occupancy": 0.879296875, "signal/frontier_coverage_0/group_std_mean": 0.19771324694156647, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.001941414293833077, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.001941414293833077, "signal/frontier_coverage_1/centered_abs_mean": 0.15531314313411712, "signal/frontier_coverage_1/group_bin_occupancy": 0.879296875, "signal/frontier_coverage_1/group_std_mean": 0.19771324694156647, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.001941414293833077, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.001941414293833077, "signal/frontier_coverage_10/centered_abs_mean": 0.15472981631755828, "signal/frontier_coverage_10/group_bin_occupancy": 0.879296875, "signal/frontier_coverage_10/group_std_mean": 0.19698039889335633, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001934122759848833, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001934122759848833, "signal/frontier_coverage_15/centered_abs_mean": 0.14061231911182404, "signal/frontier_coverage_15/group_bin_occupancy": 0.876953125, "signal/frontier_coverage_15/group_std_mean": 0.1791936755180359, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017576540820300579, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017576540820300579, "signal/frontier_coverage_20/centered_abs_mean": 0.09040251225233079, "signal/frontier_coverage_20/group_bin_occupancy": 0.880078125, "signal/frontier_coverage_20/group_std_mean": 0.11595978140830994, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001130031398497522, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001130031398497522, "signal/frontier_coverage_25/centered_abs_mean": 0.05589779242873192, "signal/frontier_coverage_25/group_bin_occupancy": 0.937109375, "signal/frontier_coverage_25/group_std_mean": 0.07136962413787842, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006987224216572941, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006987224216572941, "signal/frontier_coverage_5/centered_abs_mean": 0.1553105741739273, "signal/frontier_coverage_5/group_bin_occupancy": 0.879296875, "signal/frontier_coverage_5/group_std_mean": 0.19770981669425963, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.001941382186487317, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.001941382186487317, "signal/frontier_ece_reward/centered_abs_mean": 0.004132912214845419, "signal/frontier_ece_reward/group_bin_occupancy": 0.56796875, "signal/frontier_ece_reward/group_std_mean": 0.005261396989226341, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00041329124360345304, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00041329124360345304, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.25361861288547516, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.74296875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.32895644903182986, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.025361861661076547, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.025361861661076547, "step": 245 }, { "calibration/aurc": 0.21923325914606712, "calibration/batch_distribution_entropy": 0.9794471113491581, "calibration/batch_entropy_100bins": 0.9665456910232415, "calibration/batch_entropy_10bins": 0.9794471113491581, "calibration/batch_entropy_50bins": 0.9744196122590445, "calibration/batch_uniqueness": 0.9528411865234375, "calibration/buffer_distribution_entropy": 0.9989641344717745, "calibration/buffer_entropy_100bins": 0.9989104510581124, "calibration/buffer_entropy_10bins": 0.9989641344717745, "calibration/buffer_entropy_50bins": 0.9989853920103254, "calibration/confidence_entropy": 0.49008329690323293, "calibration/coverage@0%": 0.026953125, "calibration/coverage@1%": 0.026953125, "calibration/coverage@10%": 0.2703125, "calibration/coverage@15%": 0.41015625, "calibration/coverage@20%": 0.53984375, "calibration/coverage@25%": 0.63046875, "calibration/coverage@30%": 0.71015625, "calibration/coverage@5%": 0.069140625, "calibration/ece": 0.1130107877673741, "calibration/mean_confidence": 0.5023502126536943, "calibration/prompt_uniqueness": 0.84384765625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 603.0, "completions/max_terminated_length": 603.0, "completions/mean_length": 203.0861328125, "completions/mean_terminated_length": 203.0861328125, "completions/min_length": 106.8, "completions/min_terminated_length": 106.8, "epoch": 0.8, "grad_norm": 0.000995707232505083, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 841846916.0, "reward": 0.9545474767684936, "reward_std": 0.08173311501741409, "rewards/accuracy_reward": 0.58740234375, "rewards/brier_reward": 0.8068677544593811, "rewards/confidence_uniqueness_reward": 0.9524085998535157, "rewards/format_reward": 1.0, "rewards/frontier_aurc_reward": -0.0028800642117857932, "rewards/frontier_coverage_0": 0.08669406622648239, "rewards/frontier_coverage_1": 0.08669406622648239, "rewards/frontier_coverage_10": 0.08654228895902634, "rewards/frontier_coverage_15": 0.08043843284249305, "rewards/frontier_coverage_20": 0.05822019353508949, "rewards/frontier_coverage_25": 0.046669593453407286, "rewards/frontier_coverage_5": 0.08668917268514634, "rewards/frontier_ece_reward": 0.0021840753732249142, "rewards/frontier_entropy_batch_reward": -0.21913082003593445, "signal/accuracy_reward/centered_abs_mean": 0.086871337890625, "signal/accuracy_reward/group_bin_occupancy": 0.16640625, "signal/accuracy_reward/group_std_mean": 0.1152262806892395, "signal/accuracy_reward/group_zero_std_frac": 0.66875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0434356689453125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0434356689453125, "signal/advantage_abs_mean": 0.06384743526577949, "signal/advantage_pre_scale_abs_mean": 0.06384743526577949, "signal/advantage_pre_scale_std": 0.10191875696182251, "signal/advantage_std": 0.10191875696182251, "signal/brier_reward/centered_abs_mean": 0.11080500036478043, "signal/brier_reward/group_bin_occupancy": 0.853125, "signal/brier_reward/group_std_mean": 0.14265285432338715, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011080499924719333, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011080499924719333, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012248849868774414, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.962890625, "signal/confidence_uniqueness_reward/group_std_mean": 0.015212241560220718, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012248849961906672, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012248849961906672, "signal/format_reward/centered_abs_mean": 0.0, "signal/format_reward/group_bin_occupancy": 0.125, "signal/format_reward/group_std_mean": 0.0, "signal/format_reward/group_zero_std_frac": 1.0, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0, "signal/frontier_aurc_reward/centered_abs_mean": 0.002733212662860751, "signal/frontier_aurc_reward/group_bin_occupancy": 0.7109375, "signal/frontier_aurc_reward/group_std_mean": 0.004572971630841494, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.416515937715303e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.416515937715303e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.14644312858581543, "signal/frontier_coverage_0/group_bin_occupancy": 0.868359375, "signal/frontier_coverage_0/group_std_mean": 0.18806765973567963, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0018305391073226928, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0018305391073226928, "signal/frontier_coverage_1/centered_abs_mean": 0.14644312858581543, "signal/frontier_coverage_1/group_bin_occupancy": 0.868359375, "signal/frontier_coverage_1/group_std_mean": 0.18806765973567963, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0018305391073226928, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0018305391073226928, "signal/frontier_coverage_10/centered_abs_mean": 0.14575624018907546, "signal/frontier_coverage_10/group_bin_occupancy": 0.86875, "signal/frontier_coverage_10/group_std_mean": 0.1871989846229553, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0018219530349597335, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0018219530349597335, "signal/frontier_coverage_15/centered_abs_mean": 0.13003362566232682, "signal/frontier_coverage_15/group_bin_occupancy": 0.862890625, "signal/frontier_coverage_15/group_std_mean": 0.16730130910873414, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0016254203859716655, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0016254203859716655, "signal/frontier_coverage_20/centered_abs_mean": 0.08342465460300445, "signal/frontier_coverage_20/group_bin_occupancy": 0.88125, "signal/frontier_coverage_20/group_std_mean": 0.10799731761217117, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0010428082081489266, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0010428082081489266, "signal/frontier_coverage_25/centered_abs_mean": 0.05322186946868897, "signal/frontier_coverage_25/group_bin_occupancy": 0.92578125, "signal/frontier_coverage_25/group_std_mean": 0.06855758726596832, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006652734009549022, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006652734009549022, "signal/frontier_coverage_5/centered_abs_mean": 0.1464099705219269, "signal/frontier_coverage_5/group_bin_occupancy": 0.868359375, "signal/frontier_coverage_5/group_std_mean": 0.18802581131458282, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0018301247386261822, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0018301247386261822, "signal/frontier_ece_reward/centered_abs_mean": 0.00408116802573204, "signal/frontier_ece_reward/group_bin_occupancy": 0.580859375, "signal/frontier_ece_reward/group_std_mean": 0.005205451790243387, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004081168211996555, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004081168211996555, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2899858713150024, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.726171875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.36555504202842715, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.028998588025569916, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.028998588025569916, "step": 250 }, { "epoch": 0.8, "eval_calibration/aurc": 0.41504787957846956, "eval_calibration/batch_distribution_entropy": 0.9368707795829341, "eval_calibration/batch_entropy_100bins": 0.7131708976589626, "eval_calibration/batch_entropy_10bins": 0.9368707795829341, "eval_calibration/batch_entropy_50bins": 0.8007742810610292, "eval_calibration/batch_uniqueness": 0.90234375, "eval_calibration/buffer_distribution_entropy": 0.9990983981684345, "eval_calibration/buffer_entropy_100bins": 0.9989816933775955, "eval_calibration/buffer_entropy_10bins": 0.9990983981684345, "eval_calibration/buffer_entropy_50bins": 0.9990690091378087, "eval_calibration/confidence_entropy": 0.5071529421994212, "eval_calibration/coverage@0%": 0.0546875, "eval_calibration/coverage@1%": 0.0546875, "eval_calibration/coverage@10%": 0.0546875, "eval_calibration/coverage@15%": 0.1015625, "eval_calibration/coverage@20%": 0.2109375, "eval_calibration/coverage@25%": 0.2734375, "eval_calibration/coverage@30%": 0.3515625, "eval_calibration/coverage@5%": 0.0546875, "eval_calibration/ece": 0.1620525753154211, "eval_calibration/mean_confidence": 0.45546636880401825, "eval_calibration/prompt_uniqueness": 0.90234375, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 401.0, "eval_completions/max_terminated_length": 401.0, "eval_completions/mean_length": 208.39224243164062, "eval_completions/mean_terminated_length": 208.39224243164062, "eval_completions/min_length": 114.5, "eval_completions/min_terminated_length": 114.5, "eval_loss": 0.0, "eval_num_tokens": 841846916.0, "eval_reward": 0.7972409278154373, "eval_reward_std": 0.22940341010689735, "eval_rewards/accuracy_reward": 0.4296875, "eval_rewards/brier_reward": 0.7964034378528595, "eval_rewards/confidence_uniqueness_reward": 0.900146484375, "eval_rewards/format_reward": 1.0, "eval_rewards/frontier_aurc_reward": -0.003692921600304544, "eval_rewards/frontier_coverage_0": 0.17483051493763924, "eval_rewards/frontier_coverage_1": 0.17483051493763924, "eval_rewards/frontier_coverage_10": 0.17418187856674194, "eval_rewards/frontier_coverage_15": 0.15721721574664116, "eval_rewards/frontier_coverage_20": 0.09759005531668663, "eval_rewards/frontier_coverage_25": 0.0484907990321517, "eval_rewards/frontier_coverage_5": 0.17479892075061798, "eval_rewards/frontier_ece_reward": 0.0026409561978653073, "eval_rewards/frontier_entropy_batch_reward": -1.0, "eval_runtime": 20.6066, "eval_samples_per_second": 24.264, "eval_signal/accuracy_reward/centered_abs_mean": 0.472412109375, "eval_signal/accuracy_reward/group_bin_occupancy": 0.25, "eval_signal/accuracy_reward/group_std_mean": 0.4934760481119156, "eval_signal/accuracy_reward/group_zero_std_frac": 0.0, "eval_signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.2362060546875, "eval_signal/accuracy_reward/weight": 0.5, "eval_signal/accuracy_reward/weighted_centered_abs_mean": 0.2362060546875, "eval_signal/advantage_abs_mean": 0.21595289930701256, "eval_signal/advantage_pre_scale_abs_mean": 0.21595289930701256, "eval_signal/advantage_pre_scale_std": 0.2268650010228157, "eval_signal/advantage_std": 0.2268650010228157, "eval_signal/brier_reward/centered_abs_mean": 0.17743175104260445, "eval_signal/brier_reward/group_bin_occupancy": 0.90625, "eval_signal/brier_reward/group_std_mean": 0.2296186052262783, "eval_signal/brier_reward/group_zero_std_frac": 0.0, "eval_signal/brier_reward/scaled_weighted_centered_abs_mean": 0.017743175849318504, "eval_signal/brier_reward/weight": 0.10000000149011612, "eval_signal/brier_reward/weighted_centered_abs_mean": 0.017743175849318504, "eval_signal/confidence_uniqueness_reward/centered_abs_mean": 0.040374755859375, "eval_signal/confidence_uniqueness_reward/group_bin_occupancy": 0.40625, "eval_signal/confidence_uniqueness_reward/group_std_mean": 0.04731040354818106, "eval_signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "eval_signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0040374756208620965, "eval_signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "eval_signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0040374756208620965, "eval_signal/format_reward/centered_abs_mean": 0.0, "eval_signal/format_reward/group_bin_occupancy": 0.125, "eval_signal/format_reward/group_std_mean": 0.0, "eval_signal/format_reward/group_zero_std_frac": 1.0, "eval_signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/format_reward/weight": 0.5, "eval_signal/format_reward/weighted_centered_abs_mean": 0.0, "eval_signal/frontier_aurc_reward/centered_abs_mean": 0.004596673185005784, "eval_signal/frontier_aurc_reward/group_bin_occupancy": 0.625, "eval_signal/frontier_aurc_reward/group_std_mean": 0.009125473909080029, "eval_signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 5.745841826865217e-05, "eval_signal/frontier_aurc_reward/weight": 0.012500000186264515, "eval_signal/frontier_aurc_reward/weighted_centered_abs_mean": 5.745841826865217e-05, "eval_signal/frontier_coverage_0/centered_abs_mean": 0.33479782938957214, "eval_signal/frontier_coverage_0/group_bin_occupancy": 0.9765625, "eval_signal/frontier_coverage_0/group_std_mean": 0.41115155816078186, "eval_signal/frontier_coverage_0/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.004184973076917231, "eval_signal/frontier_coverage_0/weight": 0.012500000186264515, "eval_signal/frontier_coverage_0/weighted_centered_abs_mean": 0.004184973076917231, "eval_signal/frontier_coverage_1/centered_abs_mean": 0.33479782938957214, "eval_signal/frontier_coverage_1/group_bin_occupancy": 0.9765625, "eval_signal/frontier_coverage_1/group_std_mean": 0.41115155816078186, "eval_signal/frontier_coverage_1/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.004184973076917231, "eval_signal/frontier_coverage_1/weight": 0.012500000186264515, "eval_signal/frontier_coverage_1/weighted_centered_abs_mean": 0.004184973076917231, "eval_signal/frontier_coverage_10/centered_abs_mean": 0.3334021493792534, "eval_signal/frontier_coverage_10/group_bin_occupancy": 0.9765625, "eval_signal/frontier_coverage_10/group_std_mean": 0.4095291346311569, "eval_signal/frontier_coverage_10/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.004167527076788247, "eval_signal/frontier_coverage_10/weight": 0.012500000186264515, "eval_signal/frontier_coverage_10/weighted_centered_abs_mean": 0.004167527076788247, "eval_signal/frontier_coverage_15/centered_abs_mean": 0.29604823887348175, "eval_signal/frontier_coverage_15/group_bin_occupancy": 0.9765625, "eval_signal/frontier_coverage_15/group_std_mean": 0.3662222847342491, "eval_signal/frontier_coverage_15/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0037006032071076334, "eval_signal/frontier_coverage_15/weight": 0.012500000186264515, "eval_signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0037006032071076334, "eval_signal/frontier_coverage_20/centered_abs_mean": 0.17670203745365143, "eval_signal/frontier_coverage_20/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_20/group_std_mean": 0.22579603269696236, "eval_signal/frontier_coverage_20/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002208775549661368, "eval_signal/frontier_coverage_20/weight": 0.012500000186264515, "eval_signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002208775549661368, "eval_signal/frontier_coverage_25/centered_abs_mean": 0.07929062657058239, "eval_signal/frontier_coverage_25/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_25/group_std_mean": 0.10453899763524532, "eval_signal/frontier_coverage_25/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.000991132837953046, "eval_signal/frontier_coverage_25/weight": 0.012500000186264515, "eval_signal/frontier_coverage_25/weighted_centered_abs_mean": 0.000991132837953046, "eval_signal/frontier_coverage_5/centered_abs_mean": 0.33468862622976303, "eval_signal/frontier_coverage_5/group_bin_occupancy": 0.9765625, "eval_signal/frontier_coverage_5/group_std_mean": 0.4110269472002983, "eval_signal/frontier_coverage_5/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.004183607816230506, "eval_signal/frontier_coverage_5/weight": 0.012500000186264515, "eval_signal/frontier_coverage_5/weighted_centered_abs_mean": 0.004183607816230506, "eval_signal/frontier_ece_reward/centered_abs_mean": 0.005170486867427826, "eval_signal/frontier_ece_reward/group_bin_occupancy": 0.8046875, "eval_signal/frontier_ece_reward/group_std_mean": 0.0065434640273451805, "eval_signal/frontier_ece_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0005170486838323995, "eval_signal/frontier_ece_reward/weight": 0.10000000149011612, "eval_signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0005170486838323995, "eval_signal/frontier_entropy_batch_reward/centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.125, "eval_signal/frontier_entropy_batch_reward/group_std_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_zero_std_frac": 1.0, "eval_signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "eval_signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.0, "eval_steps_per_second": 0.194, "step": 250 }, { "calibration/aurc": 0.2215861278422698, "calibration/batch_distribution_entropy": 0.9739118819418782, "calibration/batch_entropy_100bins": 0.9688909377494104, "calibration/batch_entropy_10bins": 0.9739118819418782, "calibration/batch_entropy_50bins": 0.9753378676276757, "calibration/batch_uniqueness": 0.9523956298828125, "calibration/buffer_distribution_entropy": 0.9990514455271183, "calibration/buffer_entropy_100bins": 0.99894911521825, "calibration/buffer_entropy_10bins": 0.9990514455271183, "calibration/buffer_entropy_50bins": 0.999034838843504, "calibration/confidence_entropy": 0.48981458560224744, "calibration/coverage@0%": 0.014453125, "calibration/coverage@1%": 0.014453125, "calibration/coverage@10%": 0.1328125, "calibration/coverage@15%": 0.220703125, "calibration/coverage@20%": 0.566015625, "calibration/coverage@25%": 0.741796875, "calibration/coverage@30%": 0.8125, "calibration/coverage@5%": 0.062109375, "calibration/ece": 0.14598298681638586, "calibration/mean_confidence": 0.5199221469467401, "calibration/prompt_uniqueness": 0.848095703125, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 749.8, "completions/max_terminated_length": 579.6, "completions/mean_length": 203.56259765625, "completions/mean_terminated_length": 203.433056640625, "completions/min_length": 103.6, "completions/min_terminated_length": 103.6, "epoch": 0.816, "grad_norm": 0.0026411653961986303, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 859030565.0, "reward": 0.9572410106658935, "reward_std": 0.0827606663107872, "rewards/accuracy_reward": 0.5921875, "rewards/brier_reward": 0.7934004306793213, "rewards/confidence_uniqueness_reward": 0.9536818981170654, "rewards/format_reward": 0.99990234375, "rewards/frontier_aurc_reward": -0.00264401335734874, "rewards/frontier_coverage_0": 0.06747592762112617, "rewards/frontier_coverage_1": 0.06747592762112617, "rewards/frontier_coverage_10": 0.06739840060472488, "rewards/frontier_coverage_15": 0.06639842167496682, "rewards/frontier_coverage_20": 0.05490866750478744, "rewards/frontier_coverage_25": 0.04472573772072792, "rewards/frontier_coverage_5": 0.06747474670410156, "rewards/frontier_ece_reward": 0.0018054724670946598, "rewards/frontier_entropy_batch_reward": -0.19107870161533355, "signal/accuracy_reward/centered_abs_mean": 0.09217529296875, "signal/accuracy_reward/group_bin_occupancy": 0.168359375, "signal/accuracy_reward/group_std_mean": 0.12177184522151947, "signal/accuracy_reward/group_zero_std_frac": 0.653125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.046087646484375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.046087646484375, "signal/advantage_abs_mean": 0.06431738883256913, "signal/advantage_pre_scale_abs_mean": 0.06431738883256913, "signal/advantage_pre_scale_std": 0.10477018654346466, "signal/advantage_std": 0.10477018654346466, "signal/brier_reward/centered_abs_mean": 0.11786572337150573, "signal/brier_reward/group_bin_occupancy": 0.863671875, "signal/brier_reward/group_std_mean": 0.15070806741714476, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011786572635173798, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011786572635173798, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011417786404490472, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.95078125, "signal/confidence_uniqueness_reward/group_std_mean": 0.01449128780514002, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011417786590754986, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011417786590754986, "signal/format_reward/centered_abs_mean": 0.000189208984375, "signal/format_reward/group_bin_occupancy": 0.125390625, "signal/format_reward/group_std_mean": 0.0005524271633476019, "signal/format_reward/group_zero_std_frac": 0.996875, "signal/format_reward/scaled_weighted_centered_abs_mean": 9.46044921875e-05, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 9.46044921875e-05, "signal/frontier_aurc_reward/centered_abs_mean": 0.0026362759992480276, "signal/frontier_aurc_reward/group_bin_occupancy": 0.717578125, "signal/frontier_aurc_reward/group_std_mean": 0.00446475874632597, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.295345304650254e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.295345304650254e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15494983196258544, "signal/frontier_coverage_0/group_bin_occupancy": 0.87109375, "signal/frontier_coverage_0/group_std_mean": 0.19886209964752197, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019368728855624795, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019368728855624795, "signal/frontier_coverage_1/centered_abs_mean": 0.15494983196258544, "signal/frontier_coverage_1/group_bin_occupancy": 0.87109375, "signal/frontier_coverage_1/group_std_mean": 0.19886209964752197, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019368728855624795, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019368728855624795, "signal/frontier_coverage_10/centered_abs_mean": 0.15453089773654938, "signal/frontier_coverage_10/group_bin_occupancy": 0.870703125, "signal/frontier_coverage_10/group_std_mean": 0.19833187758922577, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0019316362915560604, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0019316362915560604, "signal/frontier_coverage_15/centered_abs_mean": 0.13622683584690093, "signal/frontier_coverage_15/group_bin_occupancy": 0.862109375, "signal/frontier_coverage_15/group_std_mean": 0.17513113617897033, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017028354573994874, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017028354573994874, "signal/frontier_coverage_20/centered_abs_mean": 0.08619713485240936, "signal/frontier_coverage_20/group_bin_occupancy": 0.88203125, "signal/frontier_coverage_20/group_std_mean": 0.11141703575849533, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0010774641530588269, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0010774641530588269, "signal/frontier_coverage_25/centered_abs_mean": 0.05528386756777763, "signal/frontier_coverage_25/group_bin_occupancy": 0.921875, "signal/frontier_coverage_25/group_std_mean": 0.07089594006538391, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006910483818501234, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006910483818501234, "signal/frontier_coverage_5/centered_abs_mean": 0.15490005612373353, "signal/frontier_coverage_5/group_bin_occupancy": 0.87109375, "signal/frontier_coverage_5/group_std_mean": 0.19879929423332215, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.00193625062238425, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.00193625062238425, "signal/frontier_ece_reward/centered_abs_mean": 0.0038798191118985413, "signal/frontier_ece_reward/group_bin_occupancy": 0.606640625, "signal/frontier_ece_reward/group_std_mean": 0.004969058092683554, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00038798191817477343, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00038798191817477343, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2588703989982605, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.731640625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3298186004161835, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02588704079389572, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02588704079389572, "step": 255 }, { "calibration/aurc": 0.2499557024640589, "calibration/batch_distribution_entropy": 0.971678143936521, "calibration/batch_entropy_100bins": 0.9632596999940832, "calibration/batch_entropy_10bins": 0.971678143936521, "calibration/batch_entropy_50bins": 0.971756536412621, "calibration/batch_uniqueness": 0.9514495849609375, "calibration/buffer_distribution_entropy": 0.9990152665921981, "calibration/buffer_entropy_100bins": 0.9989068990765425, "calibration/buffer_entropy_10bins": 0.9990152665921981, "calibration/buffer_entropy_50bins": 0.9989950321699512, "calibration/confidence_entropy": 0.48584117893588774, "calibration/coverage@0%": 0.066015625, "calibration/coverage@1%": 0.066015625, "calibration/coverage@10%": 0.26171875, "calibration/coverage@15%": 0.350390625, "calibration/coverage@20%": 0.409375, "calibration/coverage@25%": 0.528515625, "calibration/coverage@30%": 0.613671875, "calibration/coverage@5%": 0.2078125, "calibration/ece": 0.11179631257186207, "calibration/mean_confidence": 0.4725055765190076, "calibration/prompt_uniqueness": 0.845556640625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 867.4, "completions/max_terminated_length": 441.0, "completions/mean_length": 207.57265625, "completions/mean_terminated_length": 207.18388671875, "completions/min_length": 101.2, "completions/min_terminated_length": 101.2, "epoch": 0.832, "grad_norm": 0.0009539388702251017, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 876164461.0, "reward": 0.9439595580101013, "reward_std": 0.08230150043964386, "rewards/accuracy_reward": 0.55810546875, "rewards/brier_reward": 0.8166584610939026, "rewards/confidence_uniqueness_reward": 0.9521284937858582, "rewards/format_reward": 0.99970703125, "rewards/frontier_aurc_reward": -0.002334403991699219, "rewards/frontier_coverage_0": 0.11936572343111038, "rewards/frontier_coverage_1": 0.11936572343111038, "rewards/frontier_coverage_10": 0.11920837610960007, "rewards/frontier_coverage_15": 0.10542146414518357, "rewards/frontier_coverage_20": 0.07333532646298409, "rewards/frontier_coverage_25": 0.05698108747601509, "rewards/frontier_coverage_5": 0.1193393424153328, "rewards/frontier_ece_reward": 0.002439719857648015, "rewards/frontier_entropy_batch_reward": -0.20952895283699036, "signal/accuracy_reward/centered_abs_mean": 0.091522216796875, "signal/accuracy_reward/group_bin_occupancy": 0.16953125, "signal/accuracy_reward/group_std_mean": 0.1219845712184906, "signal/accuracy_reward/group_zero_std_frac": 0.64375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0457611083984375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0457611083984375, "signal/advantage_abs_mean": 0.06385182067751885, "signal/advantage_pre_scale_abs_mean": 0.06385182067751885, "signal/advantage_pre_scale_std": 0.10348795205354691, "signal/advantage_std": 0.10348795205354691, "signal/brier_reward/centered_abs_mean": 0.10376252382993698, "signal/brier_reward/group_bin_occupancy": 0.84140625, "signal/brier_reward/group_std_mean": 0.13565710186958313, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01037625279277563, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01037625279277563, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012370881997048855, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.930859375, "signal/confidence_uniqueness_reward/group_std_mean": 0.016567838191986085, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012370882323011756, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012370882323011756, "signal/format_reward/centered_abs_mean": 0.000567626953125, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0016572814900428056, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0002838134765625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0002838134765625, "signal/frontier_aurc_reward/centered_abs_mean": 0.0022006432292982934, "signal/frontier_aurc_reward/group_bin_occupancy": 0.72734375, "signal/frontier_aurc_reward/group_std_mean": 0.003800288261845708, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.7508040693646764e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.7508040693646764e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.1528583884239197, "signal/frontier_coverage_0/group_bin_occupancy": 0.866796875, "signal/frontier_coverage_0/group_std_mean": 0.19648586213588715, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019107298692688345, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019107298692688345, "signal/frontier_coverage_1/centered_abs_mean": 0.1528583884239197, "signal/frontier_coverage_1/group_bin_occupancy": 0.866796875, "signal/frontier_coverage_1/group_std_mean": 0.19648586213588715, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019107298692688345, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019107298692688345, "signal/frontier_coverage_10/centered_abs_mean": 0.15256154537200928, "signal/frontier_coverage_10/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_10/group_std_mean": 0.19609710276126863, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0019070192705839873, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0019070192705839873, "signal/frontier_coverage_15/centered_abs_mean": 0.13187448233366011, "signal/frontier_coverage_15/group_bin_occupancy": 0.861328125, "signal/frontier_coverage_15/group_std_mean": 0.16968309879302979, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001648431015200913, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001648431015200913, "signal/frontier_coverage_20/centered_abs_mean": 0.08502175211906433, "signal/frontier_coverage_20/group_bin_occupancy": 0.893359375, "signal/frontier_coverage_20/group_std_mean": 0.10919729173183441, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0010627718991599977, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0010627718991599977, "signal/frontier_coverage_25/centered_abs_mean": 0.05570260733366013, "signal/frontier_coverage_25/group_bin_occupancy": 0.926171875, "signal/frontier_coverage_25/group_std_mean": 0.07091236412525177, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006962825893424452, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006962825893424452, "signal/frontier_coverage_5/centered_abs_mean": 0.1528185248374939, "signal/frontier_coverage_5/group_bin_occupancy": 0.866796875, "signal/frontier_coverage_5/group_std_mean": 0.1964324086904526, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0019102316349744798, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0019102316349744798, "signal/frontier_ece_reward/centered_abs_mean": 0.003782039554789662, "signal/frontier_ece_reward/group_bin_occupancy": 0.59296875, "signal/frontier_ece_reward/group_std_mean": 0.004850745946168899, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0003782039741054177, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0003782039741054177, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26975165903568266, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.721875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3453064739704132, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02697516605257988, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02697516605257988, "step": 260 }, { "calibration/aurc": 0.31272421109135196, "calibration/batch_distribution_entropy": 0.9737576719497987, "calibration/batch_entropy_100bins": 0.964452611203836, "calibration/batch_entropy_10bins": 0.9737576719497987, "calibration/batch_entropy_50bins": 0.9724989363309756, "calibration/batch_uniqueness": 0.9519146672136859, "calibration/buffer_distribution_entropy": 0.998984872595709, "calibration/buffer_entropy_100bins": 0.9989041474925575, "calibration/buffer_entropy_10bins": 0.998984872595709, "calibration/buffer_entropy_50bins": 0.9989857682450882, "calibration/confidence_entropy": 0.4856939874887142, "calibration/coverage@0%": 0.0203125, "calibration/coverage@1%": 0.0203125, "calibration/coverage@10%": 0.1782014432485323, "calibration/coverage@15%": 0.28532442514677103, "calibration/coverage@20%": 0.43666218199608614, "calibration/coverage@25%": 0.4952933891878669, "calibration/coverage@30%": 0.530078125, "calibration/coverage@5%": 0.05234375, "calibration/ece": 0.1551858467661013, "calibration/mean_confidence": 0.5356993945874529, "calibration/prompt_uniqueness": 0.8437128072970864, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1340.0, "completions/max_terminated_length": 504.0, "completions/mean_length": 207.9849609375, "completions/mean_terminated_length": 207.46531982421874, "completions/min_length": 103.6, "completions/min_terminated_length": 103.6, "epoch": 0.848, "grad_norm": 0.0008241467294283211, "learning_rate": 1e-06, "loss": 0.0015, "num_tokens": 893308595.0, "reward": 0.9326582551002502, "reward_std": 0.07819676548242568, "rewards/accuracy_reward": 0.5345703125, "rewards/brier_reward": 0.8064984679222107, "rewards/confidence_uniqueness_reward": 0.9530541181564331, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.0029445850756019353, "rewards/frontier_coverage_0": 0.12361858934164047, "rewards/frontier_coverage_1": 0.12361858934164047, "rewards/frontier_coverage_10": 0.12343606650829315, "rewards/frontier_coverage_15": 0.10977802127599716, "rewards/frontier_coverage_20": 0.06932718604803086, "rewards/frontier_coverage_25": 0.051479887962341306, "rewards/frontier_coverage_5": 0.12359696924686432, "rewards/frontier_ece_reward": 0.0025736293755471706, "rewards/frontier_entropy_batch_reward": -0.1966809570789337, "signal/accuracy_reward/centered_abs_mean": 0.0755126953125, "signal/accuracy_reward/group_bin_occupancy": 0.165234375, "signal/accuracy_reward/group_std_mean": 0.10611800849437714, "signal/accuracy_reward/group_zero_std_frac": 0.678125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.03775634765625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.03775634765625, "signal/advantage_abs_mean": 0.0583618625998497, "signal/advantage_pre_scale_abs_mean": 0.0583618625998497, "signal/advantage_pre_scale_std": 0.09731595367193221, "signal/advantage_std": 0.09731595367193221, "signal/brier_reward/centered_abs_mean": 0.1124256581068039, "signal/brier_reward/group_bin_occupancy": 0.84296875, "signal/brier_reward/group_std_mean": 0.1462184637784958, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011242566257715225, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011242566257715225, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012062089145183563, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9484375, "signal/confidence_uniqueness_reward/group_std_mean": 0.016294970735907556, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012062089517712593, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012062089517712593, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086533904076, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.0028087024576961992, "signal/frontier_aurc_reward/group_bin_occupancy": 0.715234375, "signal/frontier_aurc_reward/group_std_mean": 0.004810840589925647, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.5108780139125884e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.5108780139125884e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.14909825921058656, "signal/frontier_coverage_0/group_bin_occupancy": 0.873828125, "signal/frontier_coverage_0/group_std_mean": 0.1936970144510269, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0018637283472344279, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0018637283472344279, "signal/frontier_coverage_1/centered_abs_mean": 0.14909825921058656, "signal/frontier_coverage_1/group_bin_occupancy": 0.873828125, "signal/frontier_coverage_1/group_std_mean": 0.1936970144510269, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0018637283472344279, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0018637283472344279, "signal/frontier_coverage_10/centered_abs_mean": 0.14878545105457305, "signal/frontier_coverage_10/group_bin_occupancy": 0.873828125, "signal/frontier_coverage_10/group_std_mean": 0.19328778386116027, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001859818142838776, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001859818142838776, "signal/frontier_coverage_15/centered_abs_mean": 0.1317744418978691, "signal/frontier_coverage_15/group_bin_occupancy": 0.865234375, "signal/frontier_coverage_15/group_std_mean": 0.17112208008766175, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0016471805749461055, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0016471805749461055, "signal/frontier_coverage_20/centered_abs_mean": 0.08432840257883072, "signal/frontier_coverage_20/group_bin_occupancy": 0.895703125, "signal/frontier_coverage_20/group_std_mean": 0.10917071253061295, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001054105069488287, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001054105069488287, "signal/frontier_coverage_25/centered_abs_mean": 0.0555839404463768, "signal/frontier_coverage_25/group_bin_occupancy": 0.929296875, "signal/frontier_coverage_25/group_std_mean": 0.07125073224306107, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006947992835193873, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006947992835193873, "signal/frontier_coverage_5/centered_abs_mean": 0.1490643799304962, "signal/frontier_coverage_5/group_bin_occupancy": 0.873828125, "signal/frontier_coverage_5/group_std_mean": 0.1936515361070633, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0018633048050105572, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0018633048050105572, "signal/frontier_ece_reward/centered_abs_mean": 0.003902303799986839, "signal/frontier_ece_reward/group_bin_occupancy": 0.6140625, "signal/frontier_ece_reward/group_std_mean": 0.004983165673911572, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00039023038698360324, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00039023038698360324, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2635254800319672, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.730859375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.338682496547699, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026352548226714134, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026352548226714134, "step": 265 }, { "calibration/aurc": 0.28139240890818895, "calibration/batch_distribution_entropy": 0.9737581582718506, "calibration/batch_entropy_100bins": 0.9654245703280031, "calibration/batch_entropy_10bins": 0.9737581582718506, "calibration/batch_entropy_50bins": 0.97431982374345, "calibration/batch_uniqueness": 0.9523712158203125, "calibration/buffer_distribution_entropy": 0.9990591931522488, "calibration/buffer_entropy_100bins": 0.9989666108580358, "calibration/buffer_entropy_10bins": 0.9990591931522488, "calibration/buffer_entropy_50bins": 0.9990565252512266, "calibration/confidence_entropy": 0.4984541668490218, "calibration/coverage@0%": 0.0125, "calibration/coverage@1%": 0.0125, "calibration/coverage@10%": 0.088671875, "calibration/coverage@15%": 0.12421875, "calibration/coverage@20%": 0.26640625, "calibration/coverage@25%": 0.45390625, "calibration/coverage@30%": 0.704296875, "calibration/coverage@5%": 0.021484375, "calibration/ece": 0.1461835375649582, "calibration/mean_confidence": 0.577561410029143, "calibration/prompt_uniqueness": 0.859228515625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1036.8, "completions/max_terminated_length": 660.8, "completions/mean_length": 210.7833984375, "completions/mean_terminated_length": 210.2658905029297, "completions/min_length": 111.0, "completions/min_terminated_length": 111.0, "epoch": 0.864, "grad_norm": 0.0010723300511017442, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 910453833.0, "reward": 0.9546422123908996, "reward_std": 0.08574275374412536, "rewards/accuracy_reward": 0.59296875, "rewards/brier_reward": 0.7895945072174072, "rewards/confidence_uniqueness_reward": 0.951865017414093, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.0027264663949608803, "rewards/frontier_coverage_0": 0.0698717974126339, "rewards/frontier_coverage_1": 0.0698717974126339, "rewards/frontier_coverage_10": 0.06982356086373329, "rewards/frontier_coverage_15": 0.06578152254223824, "rewards/frontier_coverage_20": 0.050645779073238376, "rewards/frontier_coverage_25": 0.04575241878628731, "rewards/frontier_coverage_5": 0.0698717974126339, "rewards/frontier_ece_reward": 0.0018079307395964862, "rewards/frontier_entropy_batch_reward": -0.21459771394729615, "signal/accuracy_reward/centered_abs_mean": 0.09521484375, "signal/accuracy_reward/group_bin_occupancy": 0.170703125, "signal/accuracy_reward/group_std_mean": 0.12704967111349105, "signal/accuracy_reward/group_zero_std_frac": 0.634375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.047607421875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.047607421875, "signal/advantage_abs_mean": 0.06586791872978211, "signal/advantage_pre_scale_abs_mean": 0.06586791872978211, "signal/advantage_pre_scale_std": 0.10589962154626846, "signal/advantage_std": 0.10589962154626846, "signal/brier_reward/centered_abs_mean": 0.11946611106395721, "signal/brier_reward/group_bin_occupancy": 0.858203125, "signal/brier_reward/group_std_mean": 0.15361137092113494, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011946610920131207, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011946610920131207, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.013185699470341206, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.944921875, "signal/confidence_uniqueness_reward/group_std_mean": 0.017704489827156066, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001318569970317185, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001318569970317185, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086068242787, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.002779424004256725, "signal/frontier_aurc_reward/group_bin_occupancy": 0.71796875, "signal/frontier_aurc_reward/group_std_mean": 0.004799876641482115, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.4742800198728216e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.4742800198728216e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.1607534945011139, "signal/frontier_coverage_0/group_bin_occupancy": 0.880078125, "signal/frontier_coverage_0/group_std_mean": 0.20486979186534882, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002009418699890375, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002009418699890375, "signal/frontier_coverage_1/centered_abs_mean": 0.1607534945011139, "signal/frontier_coverage_1/group_bin_occupancy": 0.880078125, "signal/frontier_coverage_1/group_std_mean": 0.20486979186534882, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002009418699890375, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002009418699890375, "signal/frontier_coverage_10/centered_abs_mean": 0.16060115694999694, "signal/frontier_coverage_10/group_bin_occupancy": 0.880078125, "signal/frontier_coverage_10/group_std_mean": 0.20467932522296906, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002007514424622059, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002007514424622059, "signal/frontier_coverage_15/centered_abs_mean": 0.14361293911933898, "signal/frontier_coverage_15/group_bin_occupancy": 0.874609375, "signal/frontier_coverage_15/group_std_mean": 0.18320305049419403, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017951617250218988, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017951617250218988, "signal/frontier_coverage_20/centered_abs_mean": 0.08579885065555573, "signal/frontier_coverage_20/group_bin_occupancy": 0.887109375, "signal/frontier_coverage_20/group_std_mean": 0.10999699681997299, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0010724856285378337, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0010724856285378337, "signal/frontier_coverage_25/centered_abs_mean": 0.05766439735889435, "signal/frontier_coverage_25/group_bin_occupancy": 0.919921875, "signal/frontier_coverage_25/group_std_mean": 0.07399031966924667, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007208049762994051, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007208049762994051, "signal/frontier_coverage_5/centered_abs_mean": 0.1607534945011139, "signal/frontier_coverage_5/group_bin_occupancy": 0.880078125, "signal/frontier_coverage_5/group_std_mean": 0.20486979186534882, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002009418699890375, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002009418699890375, "signal/frontier_ece_reward/centered_abs_mean": 0.004102388396859169, "signal/frontier_ece_reward/group_bin_occupancy": 0.62578125, "signal/frontier_ece_reward/group_std_mean": 0.005219202861189843, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004102388396859169, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004102388396859169, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2777667999267578, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.743359375, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3513550639152527, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027776680514216424, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027776680514216424, "step": 270 }, { "calibration/aurc": 0.3722134121437314, "calibration/batch_distribution_entropy": 0.9834287269660139, "calibration/batch_entropy_100bins": 0.9715054222040521, "calibration/batch_entropy_10bins": 0.9834287269660139, "calibration/batch_entropy_50bins": 0.9798710431487379, "calibration/batch_uniqueness": 0.9538380384140865, "calibration/buffer_distribution_entropy": 0.9990343949499687, "calibration/buffer_entropy_100bins": 0.998964414208389, "calibration/buffer_entropy_10bins": 0.9990343949499687, "calibration/buffer_entropy_50bins": 0.9990551451266129, "calibration/confidence_entropy": 0.4800572554592074, "calibration/coverage@0%": 0.017997855392156863, "calibration/coverage@1%": 0.017997855392156863, "calibration/coverage@10%": 0.030505514705882354, "calibration/coverage@15%": 0.05316789215686275, "calibration/coverage@20%": 0.15893075980392157, "calibration/coverage@25%": 0.23793658088235295, "calibration/coverage@30%": 0.3083333333333333, "calibration/coverage@5%": 0.019174325980392155, "calibration/ece": 0.14421647375109334, "calibration/mean_confidence": 0.4686593872662977, "calibration/prompt_uniqueness": 0.8465638020833334, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 1071.2, "completions/max_terminated_length": 780.2, "completions/mean_length": 211.19892578125, "completions/mean_terminated_length": 210.6816436767578, "completions/min_length": 100.4, "completions/min_terminated_length": 100.4, "epoch": 0.88, "grad_norm": 0.0009368134778924286, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 927763582.0, "reward": 0.9198407173156739, "reward_std": 0.0830217108130455, "rewards/accuracy_reward": 0.5150390625, "rewards/brier_reward": 0.7931955099105835, "rewards/confidence_uniqueness_reward": 0.9518650054931641, "rewards/format_reward": 0.99951171875, "rewards/frontier_aurc_reward": -0.0028205557260662316, "rewards/frontier_coverage_0": 0.12802205979824066, "rewards/frontier_coverage_1": 0.12802205979824066, "rewards/frontier_coverage_10": 0.12797902077436446, "rewards/frontier_coverage_15": 0.11462479531764984, "rewards/frontier_coverage_20": 0.06951974853873252, "rewards/frontier_coverage_25": 0.049090569466352464, "rewards/frontier_coverage_5": 0.12802205979824066, "rewards/frontier_ece_reward": 0.0023814529180526734, "rewards/frontier_entropy_batch_reward": -0.2145962357521057, "signal/accuracy_reward/centered_abs_mean": 0.0946044921875, "signal/accuracy_reward/group_bin_occupancy": 0.166796875, "signal/accuracy_reward/group_std_mean": 0.12103023529052734, "signal/accuracy_reward/group_zero_std_frac": 0.665625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.04730224609375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.04730224609375, "signal/advantage_abs_mean": 0.06567731648683547, "signal/advantage_pre_scale_abs_mean": 0.06567731648683547, "signal/advantage_pre_scale_std": 0.10409150719642639, "signal/advantage_std": 0.10409150719642639, "signal/brier_reward/centered_abs_mean": 0.12107095271348953, "signal/brier_reward/group_bin_occupancy": 0.84375, "signal/brier_reward/group_std_mean": 0.1551810085773468, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012107095308601856, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012107095308601856, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.013291874527931213, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.948828125, "signal/confidence_uniqueness_reward/group_std_mean": 0.017739444226026534, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0013291874201968312, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0013291874201968312, "signal/format_reward/centered_abs_mean": 0.000933837890625, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0024258273653686045, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0004669189453125, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0004669189453125, "signal/frontier_aurc_reward/centered_abs_mean": 0.0025796875357627868, "signal/frontier_aurc_reward/group_bin_occupancy": 0.706640625, "signal/frontier_aurc_reward/group_std_mean": 0.004231273150071502, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.224609499739017e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.224609499739017e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17471444010734558, "signal/frontier_coverage_0/group_bin_occupancy": 0.878125, "signal/frontier_coverage_0/group_std_mean": 0.21919819116592407, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002183930575847626, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002183930575847626, "signal/frontier_coverage_1/centered_abs_mean": 0.17471444010734558, "signal/frontier_coverage_1/group_bin_occupancy": 0.878125, "signal/frontier_coverage_1/group_std_mean": 0.21919819116592407, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002183930575847626, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002183930575847626, "signal/frontier_coverage_10/centered_abs_mean": 0.17458401322364808, "signal/frontier_coverage_10/group_bin_occupancy": 0.878125, "signal/frontier_coverage_10/group_std_mean": 0.21903611123561859, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0021823002491146324, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0021823002491146324, "signal/frontier_coverage_15/centered_abs_mean": 0.15611083209514617, "signal/frontier_coverage_15/group_bin_occupancy": 0.88125, "signal/frontier_coverage_15/group_std_mean": 0.196174556016922, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0019513854524120689, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0019513854524120689, "signal/frontier_coverage_20/centered_abs_mean": 0.09129331707954406, "signal/frontier_coverage_20/group_bin_occupancy": 0.901171875, "signal/frontier_coverage_20/group_std_mean": 0.11579755395650863, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0011411665007472038, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0011411665007472038, "signal/frontier_coverage_25/centered_abs_mean": 0.05821162089705467, "signal/frontier_coverage_25/group_bin_occupancy": 0.9296875, "signal/frontier_coverage_25/group_std_mean": 0.07444465011358262, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007276452844962478, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007276452844962478, "signal/frontier_coverage_5/centered_abs_mean": 0.17471444010734558, "signal/frontier_coverage_5/group_bin_occupancy": 0.878125, "signal/frontier_coverage_5/group_std_mean": 0.21919819116592407, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002183930575847626, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002183930575847626, "signal/frontier_ece_reward/centered_abs_mean": 0.003964019613340497, "signal/frontier_ece_reward/group_bin_occupancy": 0.5953125, "signal/frontier_ece_reward/group_std_mean": 0.004973599407821894, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00039640198228880764, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00039640198228880764, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27335782051086427, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.72578125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3443383812904358, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02733578346669674, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02733578346669674, "step": 275 }, { "calibration/aurc": 0.33636947352745883, "calibration/batch_distribution_entropy": 0.9875923943210336, "calibration/batch_entropy_100bins": 0.9730978510331093, "calibration/batch_entropy_10bins": 0.9875923943210336, "calibration/batch_entropy_50bins": 0.9820946235548906, "calibration/batch_uniqueness": 0.9546789530188509, "calibration/buffer_distribution_entropy": 0.9989805664293593, "calibration/buffer_entropy_100bins": 0.9989438784256162, "calibration/buffer_entropy_10bins": 0.9989805664293593, "calibration/buffer_entropy_50bins": 0.99902365883727, "calibration/confidence_entropy": 0.4976681405216764, "calibration/coverage@0%": 0.012890625, "calibration/coverage@1%": 0.012890625, "calibration/coverage@10%": 0.09453125, "calibration/coverage@15%": 0.1390625, "calibration/coverage@20%": 0.237890625, "calibration/coverage@25%": 0.3691635579745597, "calibration/coverage@30%": 0.49496774094911944, "calibration/coverage@5%": 0.033984375, "calibration/ece": 0.13917799622211363, "calibration/mean_confidence": 0.5027692536761388, "calibration/prompt_uniqueness": 0.849334443694719, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 924.8, "completions/max_terminated_length": 565.8, "completions/mean_length": 209.470703125, "completions/mean_terminated_length": 209.0826873779297, "completions/min_length": 95.2, "completions/min_terminated_length": 95.2, "epoch": 0.896, "grad_norm": 0.0021105960477143526, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 945019410.0, "reward": 0.9387855052947998, "reward_std": 0.07869835793972016, "rewards/accuracy_reward": 0.55244140625, "rewards/brier_reward": 0.7936831831932067, "rewards/confidence_uniqueness_reward": 0.9531597137451172, "rewards/format_reward": 0.99970703125, "rewards/frontier_aurc_reward": -0.0027678243117406965, "rewards/frontier_coverage_0": 0.09828835129737853, "rewards/frontier_coverage_1": 0.09828835129737853, "rewards/frontier_coverage_10": 0.09824797809123993, "rewards/frontier_coverage_15": 0.09175494611263275, "rewards/frontier_coverage_20": 0.0582944892346859, "rewards/frontier_coverage_25": 0.04607119113206863, "rewards/frontier_coverage_5": 0.09828835129737853, "rewards/frontier_ece_reward": 0.0016120225423946976, "rewards/frontier_entropy_batch_reward": -0.19465023875236512, "signal/accuracy_reward/centered_abs_mean": 0.079266357421875, "signal/accuracy_reward/group_bin_occupancy": 0.169140625, "signal/accuracy_reward/group_std_mean": 0.11236552149057388, "signal/accuracy_reward/group_zero_std_frac": 0.646875, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0396331787109375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0396331787109375, "signal/advantage_abs_mean": 0.05905200317502022, "signal/advantage_pre_scale_abs_mean": 0.05905200317502022, "signal/advantage_pre_scale_std": 0.09611660242080688, "signal/advantage_std": 0.09611660242080688, "signal/brier_reward/centered_abs_mean": 0.11739996075630188, "signal/brier_reward/group_bin_occupancy": 0.857421875, "signal/brier_reward/group_std_mean": 0.1509625047445297, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011739996634423733, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011739996634423733, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012467708811163903, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.951171875, "signal/confidence_uniqueness_reward/group_std_mean": 0.01622188650071621, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001246770890429616, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001246770890429616, "signal/format_reward/centered_abs_mean": 0.000555419921875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0013209730386734009, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0002777099609375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0002777099609375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0024021160788834095, "signal/frontier_aurc_reward/group_bin_occupancy": 0.707421875, "signal/frontier_aurc_reward/group_std_mean": 0.003959457110613585, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.0026452805032022e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.0026452805032022e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16086428463459015, "signal/frontier_coverage_0/group_bin_occupancy": 0.87578125, "signal/frontier_coverage_0/group_std_mean": 0.2058018058538437, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002010803669691086, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002010803669691086, "signal/frontier_coverage_1/centered_abs_mean": 0.16086428463459015, "signal/frontier_coverage_1/group_bin_occupancy": 0.87578125, "signal/frontier_coverage_1/group_std_mean": 0.2058018058538437, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002010803669691086, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002010803669691086, "signal/frontier_coverage_10/centered_abs_mean": 0.16072534620761872, "signal/frontier_coverage_10/group_bin_occupancy": 0.875390625, "signal/frontier_coverage_10/group_std_mean": 0.205626180768013, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002009066939353943, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002009066939353943, "signal/frontier_coverage_15/centered_abs_mean": 0.14419912099838256, "signal/frontier_coverage_15/group_bin_occupancy": 0.876171875, "signal/frontier_coverage_15/group_std_mean": 0.18417956233024596, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0018024890450760722, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0018024890450760722, "signal/frontier_coverage_20/centered_abs_mean": 0.08435022383928299, "signal/frontier_coverage_20/group_bin_occupancy": 0.896875, "signal/frontier_coverage_20/group_std_mean": 0.10806878358125686, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0010543777840211987, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0010543777840211987, "signal/frontier_coverage_25/centered_abs_mean": 0.05531190261244774, "signal/frontier_coverage_25/group_bin_occupancy": 0.92265625, "signal/frontier_coverage_25/group_std_mean": 0.07079362869262695, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006913988036103547, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006913988036103547, "signal/frontier_coverage_5/centered_abs_mean": 0.16086428463459015, "signal/frontier_coverage_5/group_bin_occupancy": 0.87578125, "signal/frontier_coverage_5/group_std_mean": 0.2058018058538437, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002010803669691086, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002010803669691086, "signal/frontier_ece_reward/centered_abs_mean": 0.003953807940706611, "signal/frontier_ece_reward/group_bin_occupancy": 0.566796875, "signal/frontier_ece_reward/group_std_mean": 0.005080673564225435, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00039538079872727396, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00039538079872727396, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27714616656303404, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.728125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3542065680027008, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02771461717784405, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02771461717784405, "step": 280 }, { "calibration/aurc": 0.38512010083247006, "calibration/batch_distribution_entropy": 0.9765892940303825, "calibration/batch_entropy_100bins": 0.9675763217334217, "calibration/batch_entropy_10bins": 0.9765892940303825, "calibration/batch_entropy_50bins": 0.9757679980721173, "calibration/batch_uniqueness": 0.9527607752803394, "calibration/buffer_distribution_entropy": 0.9989110843370824, "calibration/buffer_entropy_100bins": 0.998887861274374, "calibration/buffer_entropy_10bins": 0.9989110843370824, "calibration/buffer_entropy_50bins": 0.9989638003723323, "calibration/confidence_entropy": 0.4970989885684645, "calibration/coverage@0%": 0.01328125, "calibration/coverage@1%": 0.01328125, "calibration/coverage@10%": 0.026171875, "calibration/coverage@15%": 0.06171875, "calibration/coverage@20%": 0.18214973703522505, "calibration/coverage@25%": 0.28730201198630134, "calibration/coverage@30%": 0.42964927837573386, "calibration/coverage@5%": 0.015625, "calibration/ece": 0.14986459397143637, "calibration/mean_confidence": 0.5094156681015225, "calibration/prompt_uniqueness": 0.8546255121618106, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 9.765625e-05, "completions/max_length": 830.2, "completions/max_terminated_length": 635.2, "completions/mean_length": 210.845703125, "completions/mean_terminated_length": 210.71581420898437, "completions/min_length": 104.2, "completions/min_terminated_length": 104.2, "epoch": 0.912, "grad_norm": 0.0009443348390050232, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 962229766.0, "reward": 0.935080063343048, "reward_std": 0.08064102232456208, "rewards/accuracy_reward": 0.54501953125, "rewards/brier_reward": 0.7987047672271729, "rewards/confidence_uniqueness_reward": 0.9525263547897339, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.0027509741485118866, "rewards/frontier_coverage_0": 0.10241328664124012, "rewards/frontier_coverage_1": 0.10241328664124012, "rewards/frontier_coverage_10": 0.10236898455768824, "rewards/frontier_coverage_15": 0.0977680731099099, "rewards/frontier_coverage_20": 0.06384201645851136, "rewards/frontier_coverage_25": 0.05112010538578034, "rewards/frontier_coverage_5": 0.10241328664124012, "rewards/frontier_ece_reward": 0.002162482589483261, "rewards/frontier_entropy_batch_reward": -0.20416248440742493, "signal/accuracy_reward/centered_abs_mean": 0.083380126953125, "signal/accuracy_reward/group_bin_occupancy": 0.165625, "signal/accuracy_reward/group_std_mean": 0.11096605956554413, "signal/accuracy_reward/group_zero_std_frac": 0.675, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0416900634765625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0416900634765625, "signal/advantage_abs_mean": 0.06246766969561577, "signal/advantage_pre_scale_abs_mean": 0.06246766969561577, "signal/advantage_pre_scale_std": 0.10016400516033172, "signal/advantage_std": 0.10016400516033172, "signal/brier_reward/centered_abs_mean": 0.12096020728349685, "signal/brier_reward/group_bin_occupancy": 0.856640625, "signal/brier_reward/group_std_mean": 0.15659006536006928, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012096020951867103, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012096020951867103, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012567318417131901, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.95625, "signal/confidence_uniqueness_reward/group_std_mean": 0.016150840930640696, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012567318743094801, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012567318743094801, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0026764194015413524, "signal/frontier_aurc_reward/group_bin_occupancy": 0.71484375, "signal/frontier_aurc_reward/group_std_mean": 0.004556700214743614, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.345524310134351e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.345524310134351e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.16329318284988403, "signal/frontier_coverage_0/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_0/group_std_mean": 0.20762727856636048, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.002041164692491293, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.002041164692491293, "signal/frontier_coverage_1/centered_abs_mean": 0.16329318284988403, "signal/frontier_coverage_1/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_1/group_std_mean": 0.20762727856636048, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.002041164692491293, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.002041164692491293, "signal/frontier_coverage_10/centered_abs_mean": 0.16314156353473663, "signal/frontier_coverage_10/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_10/group_std_mean": 0.2074391007423401, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0020392694976180794, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0020392694976180794, "signal/frontier_coverage_15/centered_abs_mean": 0.14723739624023438, "signal/frontier_coverage_15/group_bin_occupancy": 0.87265625, "signal/frontier_coverage_15/group_std_mean": 0.1874768167734146, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001840467471629381, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001840467471629381, "signal/frontier_coverage_20/centered_abs_mean": 0.08508198261260987, "signal/frontier_coverage_20/group_bin_occupancy": 0.896875, "signal/frontier_coverage_20/group_std_mean": 0.10920049995183945, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001063524791970849, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001063524791970849, "signal/frontier_coverage_25/centered_abs_mean": 0.058402543514966966, "signal/frontier_coverage_25/group_bin_occupancy": 0.926953125, "signal/frontier_coverage_25/group_std_mean": 0.07517991065979004, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007300318102352321, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007300318102352321, "signal/frontier_coverage_5/centered_abs_mean": 0.16329318284988403, "signal/frontier_coverage_5/group_bin_occupancy": 0.883203125, "signal/frontier_coverage_5/group_std_mean": 0.20762727856636048, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.002041164692491293, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.002041164692491293, "signal/frontier_ece_reward/centered_abs_mean": 0.004069521045312286, "signal/frontier_ece_reward/group_bin_occupancy": 0.5984375, "signal/frontier_ece_reward/group_std_mean": 0.005161669943481684, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004069521091878414, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004069521091878414, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26892525255680083, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.739453125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34129275679588317, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.02689252570271492, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.02689252570271492, "step": 285 }, { "calibration/aurc": 0.42011991638031904, "calibration/batch_distribution_entropy": 0.984882287813385, "calibration/batch_entropy_100bins": 0.969936746171707, "calibration/batch_entropy_10bins": 0.984882287813385, "calibration/batch_entropy_50bins": 0.9792478123873012, "calibration/batch_uniqueness": 0.954412841796875, "calibration/buffer_distribution_entropy": 0.9990093844910615, "calibration/buffer_entropy_100bins": 0.9989308244058346, "calibration/buffer_entropy_10bins": 0.9990093844910615, "calibration/buffer_entropy_50bins": 0.9990156109170151, "calibration/confidence_entropy": 0.502349615111512, "calibration/coverage@0%": 0.00546875, "calibration/coverage@1%": 0.00546875, "calibration/coverage@10%": 0.00546875, "calibration/coverage@15%": 0.013671875, "calibration/coverage@20%": 0.01953125, "calibration/coverage@25%": 0.11015625, "calibration/coverage@30%": 0.1828125, "calibration/coverage@5%": 0.00546875, "calibration/ece": 0.12506422146177482, "calibration/mean_confidence": 0.5018705025340655, "calibration/prompt_uniqueness": 0.85224609375, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.0001953125, "completions/max_length": 958.4, "completions/max_terminated_length": 652.8, "completions/mean_length": 208.17763671875, "completions/mean_terminated_length": 207.91852722167968, "completions/min_length": 105.2, "completions/min_terminated_length": 105.2, "epoch": 0.928, "grad_norm": 0.0006618878687731922, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 979388321.0, "reward": 0.9288781166076661, "reward_std": 0.07679328620433808, "rewards/accuracy_reward": 0.53291015625, "rewards/brier_reward": 0.7889860391616821, "rewards/confidence_uniqueness_reward": 0.9533542990684509, "rewards/format_reward": 0.9998046875, "rewards/frontier_aurc_reward": -0.0032134697306901217, "rewards/frontier_coverage_0": 0.11046345382928849, "rewards/frontier_coverage_1": 0.11046345382928849, "rewards/frontier_coverage_10": 0.11043328046798706, "rewards/frontier_coverage_15": 0.10390260368585587, "rewards/frontier_coverage_20": 0.0691338375210762, "rewards/frontier_coverage_25": 0.05221138149499893, "rewards/frontier_coverage_5": 0.11046345382928849, "rewards/frontier_ece_reward": 0.001974509679712355, "rewards/frontier_entropy_batch_reward": -0.20209005177021028, "signal/accuracy_reward/centered_abs_mean": 0.076214599609375, "signal/accuracy_reward/group_bin_occupancy": 0.163671875, "signal/accuracy_reward/group_std_mean": 0.1041227251291275, "signal/accuracy_reward/group_zero_std_frac": 0.690625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0381072998046875, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0381072998046875, "signal/advantage_abs_mean": 0.058749759942293166, "signal/advantage_pre_scale_abs_mean": 0.058749759942293166, "signal/advantage_pre_scale_std": 0.09608153104782105, "signal/advantage_std": 0.09608153104782105, "signal/brier_reward/centered_abs_mean": 0.12073529064655304, "signal/brier_reward/group_bin_occupancy": 0.851953125, "signal/brier_reward/group_std_mean": 0.1547304719686508, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.012073528580367565, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.012073528580367565, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.011964846029877662, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.953125, "signal/confidence_uniqueness_reward/group_std_mean": 0.01546640656888485, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0011964846402406693, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0011964846402406693, "signal/format_reward/centered_abs_mean": 0.00037841796875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0011048543266952038, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.000189208984375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.000189208984375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0030647440813481807, "signal/frontier_aurc_reward/group_bin_occupancy": 0.714453125, "signal/frontier_aurc_reward/group_std_mean": 0.005141223315149546, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.830930218100548e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.830930218100548e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15837495028972626, "signal/frontier_coverage_0/group_bin_occupancy": 0.872265625, "signal/frontier_coverage_0/group_std_mean": 0.20259154736995696, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0019796869019046426, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0019796869019046426, "signal/frontier_coverage_1/centered_abs_mean": 0.15837495028972626, "signal/frontier_coverage_1/group_bin_occupancy": 0.872265625, "signal/frontier_coverage_1/group_std_mean": 0.20259154736995696, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0019796869019046426, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0019796869019046426, "signal/frontier_coverage_10/centered_abs_mean": 0.15825226604938508, "signal/frontier_coverage_10/group_bin_occupancy": 0.872265625, "signal/frontier_coverage_10/group_std_mean": 0.20243314504623414, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.001978153456002474, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.001978153456002474, "signal/frontier_coverage_15/centered_abs_mean": 0.14402855336666107, "signal/frontier_coverage_15/group_bin_occupancy": 0.86328125, "signal/frontier_coverage_15/group_std_mean": 0.184735968708992, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0018003570148721337, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0018003570148721337, "signal/frontier_coverage_20/centered_abs_mean": 0.08553653210401535, "signal/frontier_coverage_20/group_bin_occupancy": 0.898046875, "signal/frontier_coverage_20/group_std_mean": 0.1099936231970787, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.001069206721149385, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.001069206721149385, "signal/frontier_coverage_25/centered_abs_mean": 0.058747348934412004, "signal/frontier_coverage_25/group_bin_occupancy": 0.9359375, "signal/frontier_coverage_25/group_std_mean": 0.07478034943342209, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007343418546952307, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007343418546952307, "signal/frontier_coverage_5/centered_abs_mean": 0.15837495028972626, "signal/frontier_coverage_5/group_bin_occupancy": 0.872265625, "signal/frontier_coverage_5/group_std_mean": 0.20259154736995696, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0019796869019046426, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0019796869019046426, "signal/frontier_ece_reward/centered_abs_mean": 0.004087347071617842, "signal/frontier_ece_reward/group_bin_occupancy": 0.5921875, "signal/frontier_ece_reward/group_std_mean": 0.0052358247339725494, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004087347129825503, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004087347129825503, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2739292740821838, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.732421875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3453978717327118, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027392927557229996, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027392927557229996, "step": 290 }, { "calibration/aurc": 0.25755046512623486, "calibration/batch_distribution_entropy": 0.9852682215805295, "calibration/batch_entropy_100bins": 0.9718042997773395, "calibration/batch_entropy_10bins": 0.9852682215805295, "calibration/batch_entropy_50bins": 0.9808351137336755, "calibration/batch_uniqueness": 0.9543346095276247, "calibration/buffer_distribution_entropy": 0.9990938346300986, "calibration/buffer_entropy_100bins": 0.9990056863047949, "calibration/buffer_entropy_10bins": 0.9990938346300986, "calibration/buffer_entropy_50bins": 0.9990975463352042, "calibration/confidence_entropy": 0.5098920029101832, "calibration/coverage@0%": 0.030861668297455968, "calibration/coverage@1%": 0.030861668297455968, "calibration/coverage@10%": 0.19297104329745599, "calibration/coverage@15%": 0.30157396648727985, "calibration/coverage@20%": 0.41333552470645796, "calibration/coverage@25%": 0.49696061643835615, "calibration/coverage@30%": 0.616543847847358, "calibration/coverage@5%": 0.08867416829745597, "calibration/ece": 0.08647428231997596, "calibration/mean_confidence": 0.46917813691430954, "calibration/prompt_uniqueness": 0.8525306788989333, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 996.4, "completions/max_terminated_length": 636.6, "completions/mean_length": 208.5939453125, "completions/mean_terminated_length": 208.20537719726562, "completions/min_length": 102.0, "completions/min_terminated_length": 102.0, "epoch": 0.944, "grad_norm": 0.001066370983608067, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 996499747.0, "reward": 0.9277889251708984, "reward_std": 0.08967502266168595, "rewards/accuracy_reward": 0.53193359375, "rewards/brier_reward": 0.7913672447204589, "rewards/confidence_uniqueness_reward": 0.9521725535392761, "rewards/format_reward": 0.99970703125, "rewards/frontier_aurc_reward": -0.0028002541279420257, "rewards/frontier_coverage_0": 0.111383555829525, "rewards/frontier_coverage_1": 0.111383555829525, "rewards/frontier_coverage_10": 0.11134094297885895, "rewards/frontier_coverage_15": 0.1028158575296402, "rewards/frontier_coverage_20": 0.06852701753377914, "rewards/frontier_coverage_25": 0.047059794515371324, "rewards/frontier_coverage_5": 0.111383555829525, "rewards/frontier_ece_reward": 0.002043371554464102, "rewards/frontier_entropy_batch_reward": -0.2085336834192276, "signal/accuracy_reward/centered_abs_mean": 0.109490966796875, "signal/accuracy_reward/group_bin_occupancy": 0.17421875, "signal/accuracy_reward/group_std_mean": 0.14095723628997803, "signal/accuracy_reward/group_zero_std_frac": 0.60625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0547454833984375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0547454833984375, "signal/advantage_abs_mean": 0.0702172726392746, "signal/advantage_pre_scale_abs_mean": 0.0702172726392746, "signal/advantage_pre_scale_std": 0.11088502258062363, "signal/advantage_std": 0.11088502258062363, "signal/brier_reward/centered_abs_mean": 0.11734042316675186, "signal/brier_reward/group_bin_occupancy": 0.8671875, "signal/brier_reward/group_std_mean": 0.1499061793088913, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011734042689204216, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011734042689204216, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012994713708758354, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.94609375, "signal/confidence_uniqueness_reward/group_std_mean": 0.016800616309046744, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001299471384845674, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001299471384845674, "signal/format_reward/centered_abs_mean": 0.000555419921875, "signal/format_reward/group_bin_occupancy": 0.12578125, "signal/format_reward/group_std_mean": 0.0013209730386734009, "signal/format_reward/group_zero_std_frac": 0.99375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0002777099609375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0002777099609375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0022763639688491823, "signal/frontier_aurc_reward/group_bin_occupancy": 0.7234375, "signal/frontier_aurc_reward/group_std_mean": 0.0037089868914335967, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.8454551284085028e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.8454551284085028e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.17642856538295745, "signal/frontier_coverage_0/group_bin_occupancy": 0.878125, "signal/frontier_coverage_0/group_std_mean": 0.22151066958904267, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0022053570952266454, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0022053570952266454, "signal/frontier_coverage_1/centered_abs_mean": 0.17642856538295745, "signal/frontier_coverage_1/group_bin_occupancy": 0.878125, "signal/frontier_coverage_1/group_std_mean": 0.22151066958904267, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0022053570952266454, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0022053570952266454, "signal/frontier_coverage_10/centered_abs_mean": 0.17627204656600953, "signal/frontier_coverage_10/group_bin_occupancy": 0.877734375, "signal/frontier_coverage_10/group_std_mean": 0.22131376564502717, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.002203400665894151, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.002203400665894151, "signal/frontier_coverage_15/centered_abs_mean": 0.16484588086605073, "signal/frontier_coverage_15/group_bin_occupancy": 0.875, "signal/frontier_coverage_15/group_std_mean": 0.20735318064689637, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0020605736179277303, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0020605736179277303, "signal/frontier_coverage_20/centered_abs_mean": 0.09321919828653336, "signal/frontier_coverage_20/group_bin_occupancy": 0.896484375, "signal/frontier_coverage_20/group_std_mean": 0.11821637749671936, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0011652399785816669, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0011652399785816669, "signal/frontier_coverage_25/centered_abs_mean": 0.05760355591773987, "signal/frontier_coverage_25/group_bin_occupancy": 0.925, "signal/frontier_coverage_25/group_std_mean": 0.07351754009723663, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007200444582849741, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007200444582849741, "signal/frontier_coverage_5/centered_abs_mean": 0.17642856538295745, "signal/frontier_coverage_5/group_bin_occupancy": 0.878125, "signal/frontier_coverage_5/group_std_mean": 0.22151066958904267, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0022053570952266454, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0022053570952266454, "signal/frontier_ece_reward/centered_abs_mean": 0.004139133170247078, "signal/frontier_ece_reward/group_bin_occupancy": 0.583984375, "signal/frontier_ece_reward/group_std_mean": 0.005281085800379515, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00041391332633793353, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00041391332633793353, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2782736957073212, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.751171875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.3528176248073578, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027827369049191475, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027827369049191475, "step": 295 }, { "calibration/aurc": 0.3468888569896286, "calibration/batch_distribution_entropy": 0.9825377517538343, "calibration/batch_entropy_100bins": 0.9696874566081727, "calibration/batch_entropy_10bins": 0.9825377517538343, "calibration/batch_entropy_50bins": 0.9769011276978432, "calibration/batch_uniqueness": 0.95390678655662, "calibration/buffer_distribution_entropy": 0.9992166490828949, "calibration/buffer_entropy_100bins": 0.9990808728443558, "calibration/buffer_entropy_10bins": 0.9992166490828949, "calibration/buffer_entropy_50bins": 0.99918918010415, "calibration/confidence_entropy": 0.49663887683221325, "calibration/coverage@0%": 0.00898666829745597, "calibration/coverage@1%": 0.00898666829745597, "calibration/coverage@10%": 0.01601791829745597, "calibration/coverage@15%": 0.205080418297456, "calibration/coverage@20%": 0.26328583659491195, "calibration/coverage@25%": 0.31799397627201564, "calibration/coverage@30%": 0.36019217832681016, "calibration/coverage@5%": 0.013283543297455968, "calibration/ece": 0.15100879697277722, "calibration/mean_confidence": 0.5372243124816851, "calibration/prompt_uniqueness": 0.8450735927744537, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.000390625, "completions/max_length": 963.6, "completions/max_terminated_length": 730.2, "completions/mean_length": 209.79052734375, "completions/mean_terminated_length": 209.27050170898437, "completions/min_length": 94.6, "completions/min_terminated_length": 94.6, "epoch": 0.96, "grad_norm": 0.0008779728086665273, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 1013588322.0, "reward": 0.9338025331497193, "reward_std": 0.07533612549304962, "rewards/accuracy_reward": 0.5376953125, "rewards/brier_reward": 0.8043188929557801, "rewards/confidence_uniqueness_reward": 0.9524555563926697, "rewards/format_reward": 0.999609375, "rewards/frontier_aurc_reward": -0.003168509528040886, "rewards/frontier_coverage_0": 0.11892200410366058, "rewards/frontier_coverage_1": 0.11892200410366058, "rewards/frontier_coverage_10": 0.11884974390268326, "rewards/frontier_coverage_15": 0.11138282418251037, "rewards/frontier_coverage_20": 0.07219305783510208, "rewards/frontier_coverage_25": 0.05307292938232422, "rewards/frontier_coverage_5": 0.11892200410366058, "rewards/frontier_ece_reward": 0.002490069204941392, "rewards/frontier_entropy_batch_reward": -0.1963996857404709, "signal/accuracy_reward/centered_abs_mean": 0.07672119140625, "signal/accuracy_reward/group_bin_occupancy": 0.165234375, "signal/accuracy_reward/group_std_mean": 0.10641241073608398, "signal/accuracy_reward/group_zero_std_frac": 0.678125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.038360595703125, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.038360595703125, "signal/advantage_abs_mean": 0.0564655676484108, "signal/advantage_pre_scale_abs_mean": 0.0564655676484108, "signal/advantage_pre_scale_std": 0.09460916519165039, "signal/advantage_std": 0.09460916519165039, "signal/brier_reward/centered_abs_mean": 0.11119880378246308, "signal/brier_reward/group_bin_occupancy": 0.847265625, "signal/brier_reward/group_std_mean": 0.14490303993225098, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011119880713522434, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011119880713522434, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01229535136371851, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.947265625, "signal/confidence_uniqueness_reward/group_std_mean": 0.01662969999015331, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.001229535206221044, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.001229535206221044, "signal/format_reward/centered_abs_mean": 0.0007568359375, "signal/format_reward/group_bin_occupancy": 0.1265625, "signal/format_reward/group_std_mean": 0.0022097086068242787, "signal/format_reward/group_zero_std_frac": 0.9875, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.00037841796875, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.00037841796875, "signal/frontier_aurc_reward/centered_abs_mean": 0.002775336196646094, "signal/frontier_aurc_reward/group_bin_occupancy": 0.729296875, "signal/frontier_aurc_reward/group_std_mean": 0.004475030396133661, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.469170187599957e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.469170187599957e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.14989722967147828, "signal/frontier_coverage_0/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_0/group_std_mean": 0.19396860003471375, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0018737154081463815, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0018737154081463815, "signal/frontier_coverage_1/centered_abs_mean": 0.14989722967147828, "signal/frontier_coverage_1/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_1/group_std_mean": 0.19396860003471375, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0018737154081463815, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0018737154081463815, "signal/frontier_coverage_10/centered_abs_mean": 0.14974773824214935, "signal/frontier_coverage_10/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_10/group_std_mean": 0.1937757521867752, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0018718467792496086, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0018718467792496086, "signal/frontier_coverage_15/centered_abs_mean": 0.14086008965969085, "signal/frontier_coverage_15/group_bin_occupancy": 0.8609375, "signal/frontier_coverage_15/group_std_mean": 0.18242167532444, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017607512883841992, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017607512883841992, "signal/frontier_coverage_20/centered_abs_mean": 0.0819431260228157, "signal/frontier_coverage_20/group_bin_occupancy": 0.891015625, "signal/frontier_coverage_20/group_std_mean": 0.10616710931062698, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0010242891265079379, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0010242891265079379, "signal/frontier_coverage_25/centered_abs_mean": 0.05755271688103676, "signal/frontier_coverage_25/group_bin_occupancy": 0.925390625, "signal/frontier_coverage_25/group_std_mean": 0.07354418486356736, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0007194089819677174, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0007194089819677174, "signal/frontier_coverage_5/centered_abs_mean": 0.14989722967147828, "signal/frontier_coverage_5/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_5/group_std_mean": 0.19396860003471375, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0018737154081463815, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0018737154081463815, "signal/frontier_ece_reward/centered_abs_mean": 0.004430179670453072, "signal/frontier_ece_reward/group_bin_occupancy": 0.598828125, "signal/frontier_ece_reward/group_std_mean": 0.005666921101510525, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00044301796588115395, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00044301796588115395, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2533929169178009, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.74296875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.32594398260116575, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.025339291244745255, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.025339291244745255, "step": 300 }, { "epoch": 0.96, "eval_calibration/aurc": 0.43957633866781687, "eval_calibration/batch_distribution_entropy": 0.9190732479256418, "eval_calibration/batch_entropy_100bins": 0.7052278361140918, "eval_calibration/batch_entropy_10bins": 0.9190732479256418, "eval_calibration/batch_entropy_50bins": 0.7865654364102309, "eval_calibration/batch_uniqueness": 0.8994140625, "eval_calibration/buffer_distribution_entropy": 0.9991524655131279, "eval_calibration/buffer_entropy_100bins": 0.9990698119454451, "eval_calibration/buffer_entropy_10bins": 0.9991524655131279, "eval_calibration/buffer_entropy_50bins": 0.9991613986528001, "eval_calibration/confidence_entropy": 0.47934923368158966, "eval_calibration/coverage@0%": 0.0390625, "eval_calibration/coverage@1%": 0.0390625, "eval_calibration/coverage@10%": 0.0390625, "eval_calibration/coverage@15%": 0.078125, "eval_calibration/coverage@20%": 0.09375, "eval_calibration/coverage@25%": 0.2578125, "eval_calibration/coverage@30%": 0.3125, "eval_calibration/coverage@5%": 0.0390625, "eval_calibration/ece": 0.17185171155097956, "eval_calibration/mean_confidence": 0.49174384565028634, "eval_calibration/prompt_uniqueness": 0.8994140625, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 565.0, "eval_completions/max_terminated_length": 565.0, "eval_completions/mean_length": 214.73013305664062, "eval_completions/mean_terminated_length": 214.73013305664062, "eval_completions/min_length": 116.75, "eval_completions/min_terminated_length": 116.75, "eval_loss": 0.0, "eval_num_tokens": 1013588322.0, "eval_reward": 0.8020483553409576, "eval_reward_std": 0.23431198298931122, "eval_rewards/accuracy_reward": 0.435546875, "eval_rewards/brier_reward": 0.8045699149370193, "eval_rewards/confidence_uniqueness_reward": 0.898193359375, "eval_rewards/format_reward": 1.0, "eval_rewards/frontier_aurc_reward": -0.004081238410435617, "eval_rewards/frontier_coverage_0": 0.1887432001531124, "eval_rewards/frontier_coverage_1": 0.1887432001531124, "eval_rewards/frontier_coverage_10": 0.18861982226371765, "eval_rewards/frontier_coverage_15": 0.17859074473381042, "eval_rewards/frontier_coverage_20": 0.10508226417005062, "eval_rewards/frontier_coverage_25": 0.06004101224243641, "eval_rewards/frontier_coverage_5": 0.1887432001531124, "eval_rewards/frontier_ece_reward": 0.0031757288379594684, "eval_rewards/frontier_entropy_batch_reward": -1.0, "eval_runtime": 25.6631, "eval_samples_per_second": 19.483, "eval_signal/accuracy_reward/centered_abs_mean": 0.4764404296875, "eval_signal/accuracy_reward/group_bin_occupancy": 0.25, "eval_signal/accuracy_reward/group_std_mean": 0.4955722391605377, "eval_signal/accuracy_reward/group_zero_std_frac": 0.0, "eval_signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.23822021484375, "eval_signal/accuracy_reward/weight": 0.5, "eval_signal/accuracy_reward/weighted_centered_abs_mean": 0.23822021484375, "eval_signal/advantage_abs_mean": 0.2215503603219986, "eval_signal/advantage_pre_scale_abs_mean": 0.2215503603219986, "eval_signal/advantage_pre_scale_std": 0.2316523641347885, "eval_signal/advantage_std": 0.2316523641347885, "eval_signal/brier_reward/centered_abs_mean": 0.18021617829799652, "eval_signal/brier_reward/group_bin_occupancy": 0.8671875, "eval_signal/brier_reward/group_std_mean": 0.23240404576063156, "eval_signal/brier_reward/group_zero_std_frac": 0.0, "eval_signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01802161755040288, "eval_signal/brier_reward/weight": 0.10000000149011612, "eval_signal/brier_reward/weighted_centered_abs_mean": 0.01802161755040288, "eval_signal/confidence_uniqueness_reward/centered_abs_mean": 0.0399627685546875, "eval_signal/confidence_uniqueness_reward/group_bin_occupancy": 0.3828125, "eval_signal/confidence_uniqueness_reward/group_std_mean": 0.045800674706697464, "eval_signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "eval_signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.003996277111582458, "eval_signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "eval_signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.003996277111582458, "eval_signal/format_reward/centered_abs_mean": 0.0, "eval_signal/format_reward/group_bin_occupancy": 0.125, "eval_signal/format_reward/group_std_mean": 0.0, "eval_signal/format_reward/group_zero_std_frac": 1.0, "eval_signal/format_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/format_reward/weight": 0.5, "eval_signal/format_reward/weighted_centered_abs_mean": 0.0, "eval_signal/frontier_aurc_reward/centered_abs_mean": 0.0053485078969970345, "eval_signal/frontier_aurc_reward/group_bin_occupancy": 0.59375, "eval_signal/frontier_aurc_reward/group_std_mean": 0.010583797935396433, "eval_signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 6.685634980385657e-05, "eval_signal/frontier_aurc_reward/weight": 0.012500000186264515, "eval_signal/frontier_aurc_reward/weighted_centered_abs_mean": 6.685634980385657e-05, "eval_signal/frontier_coverage_0/centered_abs_mean": 0.32940029352903366, "eval_signal/frontier_coverage_0/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_0/group_std_mean": 0.4028378948569298, "eval_signal/frontier_coverage_0/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0041175035876221955, "eval_signal/frontier_coverage_0/weight": 0.012500000186264515, "eval_signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0041175035876221955, "eval_signal/frontier_coverage_1/centered_abs_mean": 0.32940029352903366, "eval_signal/frontier_coverage_1/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_1/group_std_mean": 0.4028378948569298, "eval_signal/frontier_coverage_1/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0041175035876221955, "eval_signal/frontier_coverage_1/weight": 0.012500000186264515, "eval_signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0041175035876221955, "eval_signal/frontier_coverage_10/centered_abs_mean": 0.32909849286079407, "eval_signal/frontier_coverage_10/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_10/group_std_mean": 0.4024873673915863, "eval_signal/frontier_coverage_10/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.004113731323741376, "eval_signal/frontier_coverage_10/weight": 0.012500000186264515, "eval_signal/frontier_coverage_10/weighted_centered_abs_mean": 0.004113731323741376, "eval_signal/frontier_coverage_15/centered_abs_mean": 0.3099505752325058, "eval_signal/frontier_coverage_15/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_15/group_std_mean": 0.3806742876768112, "eval_signal/frontier_coverage_15/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0038743820623494685, "eval_signal/frontier_coverage_15/weight": 0.012500000186264515, "eval_signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0038743820623494685, "eval_signal/frontier_coverage_20/centered_abs_mean": 0.16226129233837128, "eval_signal/frontier_coverage_20/group_bin_occupancy": 0.9375, "eval_signal/frontier_coverage_20/group_std_mean": 0.20643996819853783, "eval_signal/frontier_coverage_20/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.002028266084380448, "eval_signal/frontier_coverage_20/weight": 0.012500000186264515, "eval_signal/frontier_coverage_20/weighted_centered_abs_mean": 0.002028266084380448, "eval_signal/frontier_coverage_25/centered_abs_mean": 0.08894805796444416, "eval_signal/frontier_coverage_25/group_bin_occupancy": 0.953125, "eval_signal/frontier_coverage_25/group_std_mean": 0.11206920817494392, "eval_signal/frontier_coverage_25/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.001111850724555552, "eval_signal/frontier_coverage_25/weight": 0.012500000186264515, "eval_signal/frontier_coverage_25/weighted_centered_abs_mean": 0.001111850724555552, "eval_signal/frontier_coverage_5/centered_abs_mean": 0.32940029352903366, "eval_signal/frontier_coverage_5/group_bin_occupancy": 0.9609375, "eval_signal/frontier_coverage_5/group_std_mean": 0.4028378948569298, "eval_signal/frontier_coverage_5/group_zero_std_frac": 0.0, "eval_signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0041175035876221955, "eval_signal/frontier_coverage_5/weight": 0.012500000186264515, "eval_signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0041175035876221955, "eval_signal/frontier_ece_reward/centered_abs_mean": 0.006212400156073272, "eval_signal/frontier_ece_reward/group_bin_occupancy": 0.8203125, "eval_signal/frontier_ece_reward/group_std_mean": 0.0077107243705540895, "eval_signal/frontier_ece_reward/group_zero_std_frac": 0.0, "eval_signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0006212400330696255, "eval_signal/frontier_ece_reward/weight": 0.10000000149011612, "eval_signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0006212400330696255, "eval_signal/frontier_entropy_batch_reward/centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.125, "eval_signal/frontier_entropy_batch_reward/group_std_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/group_zero_std_frac": 1.0, "eval_signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.0, "eval_signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "eval_signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.0, "eval_steps_per_second": 0.156, "step": 300 }, { "calibration/aurc": 0.26379490171606773, "calibration/batch_distribution_entropy": 0.9777916588549415, "calibration/batch_entropy_100bins": 0.9696650736038007, "calibration/batch_entropy_10bins": 0.9777916588549415, "calibration/batch_entropy_50bins": 0.9777487917828946, "calibration/batch_uniqueness": 0.9531494140625, "calibration/buffer_distribution_entropy": 0.9991595012645323, "calibration/buffer_entropy_100bins": 0.9990516497603645, "calibration/buffer_entropy_10bins": 0.9991595012645323, "calibration/buffer_entropy_50bins": 0.9991591624155308, "calibration/confidence_entropy": 0.5168605188221923, "calibration/coverage@0%": 0.01328125, "calibration/coverage@1%": 0.01328125, "calibration/coverage@10%": 0.302734375, "calibration/coverage@15%": 0.383984375, "calibration/coverage@20%": 0.47890625, "calibration/coverage@25%": 0.535546875, "calibration/coverage@30%": 0.58828125, "calibration/coverage@5%": 0.102734375, "calibration/ece": 0.1243077210540298, "calibration/mean_confidence": 0.5273743331901343, "calibration/prompt_uniqueness": 0.85966796875, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00048828125, "completions/max_length": 934.6, "completions/max_terminated_length": 696.0, "completions/mean_length": 209.55400390625, "completions/mean_terminated_length": 208.90343017578124, "completions/min_length": 96.2, "completions/min_terminated_length": 96.2, "epoch": 0.976, "grad_norm": 0.0010843015043064952, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 1030595275.0, "reward": 0.9408185362815857, "reward_std": 0.08025571703910828, "rewards/accuracy_reward": 0.5572265625, "rewards/brier_reward": 0.797808575630188, "rewards/confidence_uniqueness_reward": 0.952480149269104, "rewards/format_reward": 0.99951171875, "rewards/frontier_aurc_reward": -0.003078281506896019, "rewards/frontier_coverage_0": 0.09664749102666974, "rewards/frontier_coverage_1": 0.09664749102666974, "rewards/frontier_coverage_10": 0.09654867630451917, "rewards/frontier_coverage_15": 0.09212939422577619, "rewards/frontier_coverage_20": 0.06269557178020477, "rewards/frontier_coverage_25": 0.04787970148026943, "rewards/frontier_coverage_5": 0.09664749102666974, "rewards/frontier_ece_reward": 0.0020735796890221534, "rewards/frontier_entropy_batch_reward": -0.20113272964954376, "signal/accuracy_reward/centered_abs_mean": 0.083251953125, "signal/accuracy_reward/group_bin_occupancy": 0.168359375, "signal/accuracy_reward/group_std_mean": 0.11468427479267121, "signal/accuracy_reward/group_zero_std_frac": 0.653125, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0416259765625, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0416259765625, "signal/advantage_abs_mean": 0.06097738593816757, "signal/advantage_pre_scale_abs_mean": 0.06097738593816757, "signal/advantage_pre_scale_std": 0.10007609874010086, "signal/advantage_std": 0.10007609874010086, "signal/brier_reward/centered_abs_mean": 0.10903310328722, "signal/brier_reward/group_bin_occupancy": 0.852734375, "signal/brier_reward/group_std_mean": 0.14228532910346986, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01090331096202135, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01090331096202135, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01266135461628437, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.948046875, "signal/confidence_uniqueness_reward/group_std_mean": 0.016756997257471085, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012661355081945657, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012661355081945657, "signal/format_reward/centered_abs_mean": 0.000909423828125, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.00203030607663095, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0004547119140625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0004547119140625, "signal/frontier_aurc_reward/centered_abs_mean": 0.0027001567650586367, "signal/frontier_aurc_reward/group_bin_occupancy": 0.72109375, "signal/frontier_aurc_reward/group_std_mean": 0.004579899832606316, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.375196174602024e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.375196174602024e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15166537463665009, "signal/frontier_coverage_0/group_bin_occupancy": 0.86796875, "signal/frontier_coverage_0/group_std_mean": 0.19646393954753877, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.001895817252807319, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.001895817252807319, "signal/frontier_coverage_1/centered_abs_mean": 0.15166537463665009, "signal/frontier_coverage_1/group_bin_occupancy": 0.86796875, "signal/frontier_coverage_1/group_std_mean": 0.19646393954753877, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.001895817252807319, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.001895817252807319, "signal/frontier_coverage_10/centered_abs_mean": 0.15149901807308197, "signal/frontier_coverage_10/group_bin_occupancy": 0.86875, "signal/frontier_coverage_10/group_std_mean": 0.1962491124868393, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0018937377724796533, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0018937377724796533, "signal/frontier_coverage_15/centered_abs_mean": 0.1429735541343689, "signal/frontier_coverage_15/group_bin_occupancy": 0.86640625, "signal/frontier_coverage_15/group_std_mean": 0.185446035861969, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017871694406494498, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017871694406494498, "signal/frontier_coverage_20/centered_abs_mean": 0.07984205931425095, "signal/frontier_coverage_20/group_bin_occupancy": 0.888671875, "signal/frontier_coverage_20/group_std_mean": 0.10434311479330063, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0009980257251299917, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0009980257251299917, "signal/frontier_coverage_25/centered_abs_mean": 0.055129478126764296, "signal/frontier_coverage_25/group_bin_occupancy": 0.9265625, "signal/frontier_coverage_25/group_std_mean": 0.07109689712524414, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006891185301356017, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006891185301356017, "signal/frontier_coverage_5/centered_abs_mean": 0.15166537463665009, "signal/frontier_coverage_5/group_bin_occupancy": 0.86796875, "signal/frontier_coverage_5/group_std_mean": 0.19646393954753877, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.001895817252807319, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.001895817252807319, "signal/frontier_ece_reward/centered_abs_mean": 0.004354849411174655, "signal/frontier_ece_reward/group_bin_occupancy": 0.6, "signal/frontier_ece_reward/group_std_mean": 0.00556610943749547, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.0004354849341325462, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.0004354849341325462, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.26732973754405975, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.737890625, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34105202555656433, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.026732975244522096, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.026732975244522096, "step": 305 }, { "calibration/aurc": 0.3769423254121689, "calibration/batch_distribution_entropy": 0.9814860613819866, "calibration/batch_entropy_100bins": 0.9715123709186304, "calibration/batch_entropy_10bins": 0.9814860613819866, "calibration/batch_entropy_50bins": 0.979819228118423, "calibration/batch_uniqueness": 0.9537734806363005, "calibration/buffer_distribution_entropy": 0.9992690569046164, "calibration/buffer_entropy_100bins": 0.9991110384659982, "calibration/buffer_entropy_10bins": 0.9992690569046164, "calibration/buffer_entropy_50bins": 0.9992171115280462, "calibration/confidence_entropy": 0.5204817012954857, "calibration/coverage@0%": 0.0125, "calibration/coverage@1%": 0.0125, "calibration/coverage@10%": 0.04296875, "calibration/coverage@15%": 0.091796875, "calibration/coverage@20%": 0.125, "calibration/coverage@25%": 0.159375, "calibration/coverage@30%": 0.3377079256360078, "calibration/coverage@5%": 0.012890625, "calibration/ece": 0.14285743981539423, "calibration/mean_confidence": 0.47222098782020205, "calibration/prompt_uniqueness": 0.8476286603310352, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00029296875, "completions/max_length": 1175.8, "completions/max_terminated_length": 572.2, "completions/mean_length": 196.46298828125, "completions/mean_terminated_length": 196.07098083496095, "completions/min_length": 89.2, "completions/min_terminated_length": 89.2, "epoch": 0.992, "grad_norm": 0.0010080559877678752, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 1047735536.0, "reward": 0.9261445045471192, "reward_std": 0.07883597165346146, "rewards/accuracy_reward": 0.52568359375, "rewards/brier_reward": 0.7976857542991638, "rewards/confidence_uniqueness_reward": 0.9524794340133667, "rewards/format_reward": 0.99970703125, "rewards/frontier_aurc_reward": -0.0028197201900184155, "rewards/frontier_coverage_0": 0.11574460715055465, "rewards/frontier_coverage_1": 0.11574460715055465, "rewards/frontier_coverage_10": 0.11566894203424453, "rewards/frontier_coverage_15": 0.10948524177074433, "rewards/frontier_coverage_20": 0.069465072453022, "rewards/frontier_coverage_25": 0.05050253868103027, "rewards/frontier_coverage_5": 0.11574025899171829, "rewards/frontier_ece_reward": 0.0019482931587845088, "rewards/frontier_entropy_batch_reward": -0.20381315052509308, "signal/accuracy_reward/centered_abs_mean": 0.080242919921875, "signal/accuracy_reward/group_bin_occupancy": 0.164453125, "signal/accuracy_reward/group_std_mean": 0.10792075693607331, "signal/accuracy_reward/group_zero_std_frac": 0.684375, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.0401214599609375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.0401214599609375, "signal/advantage_abs_mean": 0.06033377572894096, "signal/advantage_pre_scale_abs_mean": 0.06033377572894096, "signal/advantage_pre_scale_std": 0.09838964939117431, "signal/advantage_std": 0.09838964939117431, "signal/brier_reward/centered_abs_mean": 0.11128220856189727, "signal/brier_reward/group_bin_occupancy": 0.86328125, "signal/brier_reward/group_std_mean": 0.14221763908863067, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.01112822126597166, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.01112822126597166, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.01214519441127777, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.947265625, "signal/confidence_uniqueness_reward/group_std_mean": 0.016030562669038774, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012145194690674544, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012145194690674544, "signal/format_reward/centered_abs_mean": 0.000567626953125, "signal/format_reward/group_bin_occupancy": 0.126171875, "signal/format_reward/group_std_mean": 0.0016572814900428056, "signal/format_reward/group_zero_std_frac": 0.990625, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0002838134765625, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0002838134765625, "signal/frontier_aurc_reward/centered_abs_mean": 0.002432570233941078, "signal/frontier_aurc_reward/group_bin_occupancy": 0.731640625, "signal/frontier_aurc_reward/group_std_mean": 0.003922372870147228, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 3.0407129088416696e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 3.0407129088416696e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.15133395195007324, "signal/frontier_coverage_0/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_0/group_std_mean": 0.19293701350688935, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.0018916744738817215, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.0018916744738817215, "signal/frontier_coverage_1/centered_abs_mean": 0.15133395195007324, "signal/frontier_coverage_1/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_1/group_std_mean": 0.19293701350688935, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.0018916744738817215, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.0018916744738817215, "signal/frontier_coverage_10/centered_abs_mean": 0.15115775763988495, "signal/frontier_coverage_10/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_10/group_std_mean": 0.1927188217639923, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0018894718959927559, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0018894718959927559, "signal/frontier_coverage_15/centered_abs_mean": 0.1417694479227066, "signal/frontier_coverage_15/group_bin_occupancy": 0.883984375, "signal/frontier_coverage_15/group_std_mean": 0.1810533493757248, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.001772118080407381, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.001772118080407381, "signal/frontier_coverage_20/centered_abs_mean": 0.07549672797322274, "signal/frontier_coverage_20/group_bin_occupancy": 0.90859375, "signal/frontier_coverage_20/group_std_mean": 0.09715144485235214, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0009437091066502035, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0009437091066502035, "signal/frontier_coverage_25/centered_abs_mean": 0.05469834133982658, "signal/frontier_coverage_25/group_bin_occupancy": 0.93046875, "signal/frontier_coverage_25/group_std_mean": 0.06969960629940034, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006837293040007353, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006837293040007353, "signal/frontier_coverage_5/centered_abs_mean": 0.151319283246994, "signal/frontier_coverage_5/group_bin_occupancy": 0.88515625, "signal/frontier_coverage_5/group_std_mean": 0.19291889667510986, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0018914910964667798, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0018914910964667798, "signal/frontier_ece_reward/centered_abs_mean": 0.004252498596906662, "signal/frontier_ece_reward/group_bin_occupancy": 0.606640625, "signal/frontier_ece_reward/group_std_mean": 0.005385065358132124, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00042524987366050483, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00042524987366050483, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.2715674936771393, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.739453125, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34656343460083006, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027156750857830047, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027156750857830047, "step": 310 }, { "calibration/aurc": 0.2466931830488064, "calibration/batch_distribution_entropy": 0.9841488203743247, "calibration/batch_entropy_100bins": 0.9705053255848448, "calibration/batch_entropy_10bins": 0.9841488203743247, "calibration/batch_entropy_50bins": 0.9789524810008161, "calibration/batch_uniqueness": 0.9543380737304688, "calibration/buffer_distribution_entropy": 0.9994008330875915, "calibration/buffer_entropy_100bins": 0.9991734740669567, "calibration/buffer_entropy_10bins": 0.9994008330875915, "calibration/buffer_entropy_50bins": 0.999291138943704, "calibration/confidence_entropy": 0.49746082053760843, "calibration/coverage@0%": 0.03515625, "calibration/coverage@1%": 0.03515625, "calibration/coverage@10%": 0.1064453125, "calibration/coverage@15%": 0.1630859375, "calibration/coverage@20%": 0.4091796875, "calibration/coverage@25%": 0.572265625, "calibration/coverage@30%": 0.7080078125, "calibration/coverage@5%": 0.08203125, "calibration/ece": 0.11523633209952004, "calibration/mean_confidence": 0.5638644996665658, "calibration/prompt_uniqueness": 0.8515625, "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00048828125, "completions/max_length": 1277.0, "completions/max_terminated_length": 721.5, "completions/mean_length": 193.1613540649414, "completions/mean_terminated_length": 192.50463104248047, "completions/min_length": 91.0, "completions/min_terminated_length": 91.0, "epoch": 0.9984, "num_tokens": 1054527938.0, "reward": 0.9437865018844604, "reward_std": 0.08146928623318672, "rewards/accuracy_reward": 0.56689453125, "rewards/brier_reward": 0.7848476469516754, "rewards/confidence_uniqueness_reward": 0.9534134566783905, "rewards/format_reward": 0.99951171875, "rewards/frontier_aurc_reward": -0.0023456220515072346, "rewards/frontier_coverage_0": 0.06734757125377655, "rewards/frontier_coverage_1": 0.06734757125377655, "rewards/frontier_coverage_10": 0.06731921434402466, "rewards/frontier_coverage_15": 0.06181756407022476, "rewards/frontier_coverage_20": 0.04400276392698288, "rewards/frontier_coverage_25": 0.040297653526067734, "rewards/frontier_coverage_5": 0.06735776551067829, "rewards/frontier_ece_reward": 0.0013854657299816608, "rewards/frontier_entropy_batch_reward": -0.18545609712600708, "signal/accuracy_reward/centered_abs_mean": 0.0810546875, "signal/accuracy_reward/group_bin_occupancy": 0.1669921875, "signal/accuracy_reward/group_std_mean": 0.11190324276685715, "signal/accuracy_reward/group_zero_std_frac": 0.6640625, "signal/accuracy_reward/scaled_weighted_centered_abs_mean": 0.04052734375, "signal/accuracy_reward/weight": 0.5, "signal/accuracy_reward/weighted_centered_abs_mean": 0.04052734375, "signal/advantage_abs_mean": 0.0617546159774065, "signal/advantage_pre_scale_abs_mean": 0.0617546159774065, "signal/advantage_pre_scale_std": 0.10202585905790329, "signal/advantage_std": 0.10202585905790329, "signal/brier_reward/centered_abs_mean": 0.1156514547765255, "signal/brier_reward/group_bin_occupancy": 0.8828125, "signal/brier_reward/group_std_mean": 0.14817585051059723, "signal/brier_reward/group_zero_std_frac": 0.0, "signal/brier_reward/scaled_weighted_centered_abs_mean": 0.011565146036446095, "signal/brier_reward/weight": 0.10000000149011612, "signal/brier_reward/weighted_centered_abs_mean": 0.011565146036446095, "signal/confidence_uniqueness_reward/centered_abs_mean": 0.012430991977453232, "signal/confidence_uniqueness_reward/group_bin_occupancy": 0.9521484375, "signal/confidence_uniqueness_reward/group_std_mean": 0.016900820657610893, "signal/confidence_uniqueness_reward/group_zero_std_frac": 0.0, "signal/confidence_uniqueness_reward/scaled_weighted_centered_abs_mean": 0.0012430992210283875, "signal/confidence_uniqueness_reward/weight": 0.10000000149011612, "signal/confidence_uniqueness_reward/weighted_centered_abs_mean": 0.0012430992210283875, "signal/format_reward/centered_abs_mean": 0.000946044921875, "signal/format_reward/group_bin_occupancy": 0.126953125, "signal/format_reward/group_std_mean": 0.0027621358167380095, "signal/format_reward/group_zero_std_frac": 0.984375, "signal/format_reward/scaled_weighted_centered_abs_mean": 0.0004730224609375, "signal/format_reward/weight": 0.5, "signal/format_reward/weighted_centered_abs_mean": 0.0004730224609375, "signal/frontier_aurc_reward/centered_abs_mean": 0.0020961585105396807, "signal/frontier_aurc_reward/group_bin_occupancy": 0.740234375, "signal/frontier_aurc_reward/group_std_mean": 0.0034951349953189492, "signal/frontier_aurc_reward/group_zero_std_frac": 0.0, "signal/frontier_aurc_reward/scaled_weighted_centered_abs_mean": 2.620198210934177e-05, "signal/frontier_aurc_reward/weight": 0.012500000186264515, "signal/frontier_aurc_reward/weighted_centered_abs_mean": 2.620198210934177e-05, "signal/frontier_coverage_0/centered_abs_mean": 0.14957503974437714, "signal/frontier_coverage_0/group_bin_occupancy": 0.8935546875, "signal/frontier_coverage_0/group_std_mean": 0.19460299611091614, "signal/frontier_coverage_0/group_zero_std_frac": 0.0, "signal/frontier_coverage_0/scaled_weighted_centered_abs_mean": 0.001869688043370843, "signal/frontier_coverage_0/weight": 0.012500000186264515, "signal/frontier_coverage_0/weighted_centered_abs_mean": 0.001869688043370843, "signal/frontier_coverage_1/centered_abs_mean": 0.14957503974437714, "signal/frontier_coverage_1/group_bin_occupancy": 0.8935546875, "signal/frontier_coverage_1/group_std_mean": 0.19460299611091614, "signal/frontier_coverage_1/group_zero_std_frac": 0.0, "signal/frontier_coverage_1/scaled_weighted_centered_abs_mean": 0.001869688043370843, "signal/frontier_coverage_1/weight": 0.012500000186264515, "signal/frontier_coverage_1/weighted_centered_abs_mean": 0.001869688043370843, "signal/frontier_coverage_10/centered_abs_mean": 0.1493651196360588, "signal/frontier_coverage_10/group_bin_occupancy": 0.8935546875, "signal/frontier_coverage_10/group_std_mean": 0.19433680176734924, "signal/frontier_coverage_10/group_zero_std_frac": 0.0, "signal/frontier_coverage_10/scaled_weighted_centered_abs_mean": 0.0018670640420168638, "signal/frontier_coverage_10/weight": 0.012500000186264515, "signal/frontier_coverage_10/weighted_centered_abs_mean": 0.0018670640420168638, "signal/frontier_coverage_15/centered_abs_mean": 0.13645901530981064, "signal/frontier_coverage_15/group_bin_occupancy": 0.8876953125, "signal/frontier_coverage_15/group_std_mean": 0.17800746113061905, "signal/frontier_coverage_15/group_zero_std_frac": 0.0, "signal/frontier_coverage_15/scaled_weighted_centered_abs_mean": 0.0017057376680895686, "signal/frontier_coverage_15/weight": 0.012500000186264515, "signal/frontier_coverage_15/weighted_centered_abs_mean": 0.0017057376680895686, "signal/frontier_coverage_20/centered_abs_mean": 0.07139924541115761, "signal/frontier_coverage_20/group_bin_occupancy": 0.8916015625, "signal/frontier_coverage_20/group_std_mean": 0.09412418305873871, "signal/frontier_coverage_20/group_zero_std_frac": 0.0, "signal/frontier_coverage_20/scaled_weighted_centered_abs_mean": 0.0008924905268941075, "signal/frontier_coverage_20/weight": 0.012500000186264515, "signal/frontier_coverage_20/weighted_centered_abs_mean": 0.0008924905268941075, "signal/frontier_coverage_25/centered_abs_mean": 0.05047915317118168, "signal/frontier_coverage_25/group_bin_occupancy": 0.9228515625, "signal/frontier_coverage_25/group_std_mean": 0.06584775447845459, "signal/frontier_coverage_25/group_zero_std_frac": 0.0, "signal/frontier_coverage_25/scaled_weighted_centered_abs_mean": 0.0006309894379228354, "signal/frontier_coverage_25/weight": 0.012500000186264515, "signal/frontier_coverage_25/weighted_centered_abs_mean": 0.0006309894379228354, "signal/frontier_coverage_5/centered_abs_mean": 0.14956200122833252, "signal/frontier_coverage_5/group_bin_occupancy": 0.8935546875, "signal/frontier_coverage_5/group_std_mean": 0.1945861652493477, "signal/frontier_coverage_5/group_zero_std_frac": 0.0, "signal/frontier_coverage_5/scaled_weighted_centered_abs_mean": 0.0018695250037126243, "signal/frontier_coverage_5/weight": 0.012500000186264515, "signal/frontier_coverage_5/weighted_centered_abs_mean": 0.0018695250037126243, "signal/frontier_ece_reward/centered_abs_mean": 0.0042242975905537605, "signal/frontier_ece_reward/group_bin_occupancy": 0.6025390625, "signal/frontier_ece_reward/group_std_mean": 0.005414322717115283, "signal/frontier_ece_reward/group_zero_std_frac": 0.0, "signal/frontier_ece_reward/scaled_weighted_centered_abs_mean": 0.00042242975905537605, "signal/frontier_ece_reward/weight": 0.10000000149011612, "signal/frontier_ece_reward/weighted_centered_abs_mean": 0.00042242975905537605, "signal/frontier_entropy_batch_reward/centered_abs_mean": 0.27094003558158875, "signal/frontier_entropy_batch_reward/group_bin_occupancy": 0.7294921875, "signal/frontier_entropy_batch_reward/group_std_mean": 0.34401288628578186, "signal/frontier_entropy_batch_reward/group_zero_std_frac": 0.0, "signal/frontier_entropy_batch_reward/scaled_weighted_centered_abs_mean": 0.027094004675745964, "signal/frontier_entropy_batch_reward/weight": 0.10000000149011612, "signal/frontier_entropy_batch_reward/weighted_centered_abs_mean": 0.027094004675745964, "step": 312, "total_flos": 0.0, "train_loss": 0.004430588132611144, "train_runtime": 59873.197, "train_samples_per_second": 0.334, "train_steps_per_second": 0.005 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1054527938, "num_train_epochs": 1, "save_steps": 60, "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 }