{ "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": [ { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.11044921875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1529.0, "completions/mean_length": 472.82919921875, "completions/mean_terminated_length": 340.9787963867187, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.016, "grad_norm": 0.004426147323101759, "learning_rate": 3.1249999999999997e-07, "loss": 0.0444, "num_tokens": 19810059.0, "reward": 0.20038452744483948, "reward_std": 0.36961849331855773, "rewards/accuracy_reward": 0.053515625, "rewards/brier_reward": 0.09974026381969452, "rewards/confidence_one_or_zero": 0.401953125, "rewards/format_reward": 0.24755859375, "rewards/mean_confidence_reward": 0.5259665191173554, "rewards/smce": -0.47275206446647644, "step": 5 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.09716796875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1524.2, "completions/mean_length": 431.33447265625, "completions/mean_terminated_length": 312.50458984375, "completions/min_length": 1.0, "completions/min_terminated_length": 1.0, "epoch": 0.032, "grad_norm": 0.003934022039175034, "learning_rate": 6.249999999999999e-07, "loss": 0.0498, "num_tokens": 39265452.0, "reward": 0.2235296219587326, "reward_std": 0.3799919366836548, "rewards/accuracy_reward": 0.0552734375, "rewards/brier_reward": 0.10921734869480133, "rewards/confidence_one_or_zero": 0.424609375, "rewards/format_reward": 0.2826171875, "rewards/mean_confidence_reward": 0.5619598150253295, "rewards/smce": -0.5068929612636566, "step": 10 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.054296875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1487.0, "completions/mean_length": 312.63193359375, "completions/mean_terminated_length": 242.80802307128906, "completions/min_length": 1.8, "completions/min_terminated_length": 1.8, "epoch": 0.048, "grad_norm": 0.00401897681877017, "learning_rate": 9.374999999999999e-07, "loss": 0.0483, "num_tokens": 57529843.0, "reward": 0.4077006340026855, "reward_std": 0.4407335460186005, "rewards/accuracy_reward": 0.10361328125, "rewards/brier_reward": 0.19856337010860442, "rewards/confidence_one_or_zero": 0.53974609375, "rewards/format_reward": 0.51328125, "rewards/mean_confidence_reward": 0.6942256689071655, "rewards/smce": -0.5910538554191589, "step": 15 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.021484375, "completions/max_length": 1536.0, "completions/max_terminated_length": 1397.2, "completions/mean_length": 214.742578125, "completions/mean_terminated_length": 185.78294677734374, "completions/min_length": 16.2, "completions/min_terminated_length": 16.2, "epoch": 0.064, "grad_norm": 0.002913103671744466, "learning_rate": 1e-06, "loss": 0.037, "num_tokens": 74777895.0, "reward": 0.6778669238090516, "reward_std": 0.39471102356910703, "rewards/accuracy_reward": 0.187109375, "rewards/brier_reward": 0.3357722103595734, "rewards/confidence_one_or_zero": 0.5685546875, "rewards/format_reward": 0.83291015625, "rewards/mean_confidence_reward": 0.7907521605491639, "rewards/smce": -0.6059021592140198, "step": 20 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.01279296875, "completions/max_length": 1536.0, "completions/max_terminated_length": 1216.0, "completions/mean_length": 165.07568359375, "completions/mean_terminated_length": 147.32185974121094, "completions/min_length": 21.6, "completions/min_terminated_length": 21.6, "epoch": 0.08, "grad_norm": 0.002816141350194812, "learning_rate": 1e-06, "loss": 0.0187, "num_tokens": 91500174.0, "reward": 0.7974048018455505, "reward_std": 0.3286789894104004, "rewards/accuracy_reward": 0.21787109375, "rewards/brier_reward": 0.4181051731109619, "rewards/confidence_one_or_zero": 0.476953125, "rewards/format_reward": 0.95888671875, "rewards/mean_confidence_reward": 0.7712665319442749, "rewards/smce": -0.5586693763732911, "step": 25 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00751953125, "completions/max_length": 1536.0, "completions/max_terminated_length": 1278.6, "completions/mean_length": 142.0654296875, "completions/mean_terminated_length": 131.50679779052734, "completions/min_length": 13.6, "completions/min_terminated_length": 13.6, "epoch": 0.096, "grad_norm": 0.0024108965881168842, "learning_rate": 1e-06, "loss": 0.0124, "num_tokens": 107868204.0, "reward": 0.895538580417633, "reward_std": 0.2955020546913147, "rewards/accuracy_reward": 0.25625, "rewards/brier_reward": 0.5535168647766113, "rewards/confidence_one_or_zero": 0.33291015625, "rewards/format_reward": 0.98134765625, "rewards/mean_confidence_reward": 0.6346356272697449, "rewards/smce": -0.39225292205810547, "step": 30 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00380859375, "completions/max_length": 1536.0, "completions/max_terminated_length": 908.2, "completions/mean_length": 126.67451171875, "completions/mean_terminated_length": 121.2887466430664, "completions/min_length": 37.2, "completions/min_terminated_length": 37.2, "epoch": 0.112, "grad_norm": 0.0012922441819682717, "learning_rate": 1e-06, "loss": 0.0069, "num_tokens": 124179495.0, "reward": 0.9732187509536743, "reward_std": 0.2013139009475708, "rewards/accuracy_reward": 0.24404296875, "rewards/brier_reward": 0.7092463850975037, "rewards/confidence_one_or_zero": 0.228125, "rewards/format_reward": 0.9931640625, "rewards/mean_confidence_reward": 0.3768877565860748, "rewards/smce": -0.17112396359443666, "step": 35 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00126953125, "completions/max_length": 1395.8, "completions/max_terminated_length": 780.2, "completions/mean_length": 119.758984375, "completions/mean_terminated_length": 117.95802612304688, "completions/min_length": 47.2, "completions/min_terminated_length": 47.2, "epoch": 0.128, "grad_norm": 0.0008959283004514873, "learning_rate": 1e-06, "loss": 0.0023, "num_tokens": 140359299.0, "reward": 1.0186350703239442, "reward_std": 0.13564346730709076, "rewards/accuracy_reward": 0.2814453125, "rewards/brier_reward": 0.7590552687644958, "rewards/confidence_one_or_zero": 0.203125, "rewards/format_reward": 0.99677734375, "rewards/mean_confidence_reward": 0.246020370721817, "rewards/smce": -0.08598535656929016, "step": 40 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 0.0, "clip_ratio/low_mean": 0.0, "clip_ratio/low_min": 0.0, "clip_ratio/region_mean": 0.0, "completions/clipped_ratio": 0.00078125, "completions/max_length": 1536.0, "completions/max_terminated_length": 716.4, "completions/mean_length": 119.24130859375, "completions/mean_terminated_length": 118.13349609375, "completions/min_length": 48.8, "completions/min_terminated_length": 48.8, "epoch": 0.144, "grad_norm": 0.0008299688925035298, "learning_rate": 1e-06, "loss": 0.0017, "num_tokens": 156308266.0, "reward": 1.033864641189575, "reward_std": 0.10059063583612442, "rewards/accuracy_reward": 0.30849609375, "rewards/brier_reward": 0.7607093691825867, "rewards/confidence_one_or_zero": 0.15751953125, "rewards/format_reward": 0.99853515625, "rewards/mean_confidence_reward": 0.19957530796527861, "rewards/smce": -0.12014870196580887, "step": 45 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1055.6, "completions/max_terminated_length": 644.2, "completions/mean_length": 119.0482421875, "completions/mean_terminated_length": 118.2185302734375, "completions/min_length": 51.2, "completions/min_terminated_length": 51.2, "epoch": 0.16, "grad_norm": 0.000901242543477565, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 172699192.0, "reward": 1.0454564571380616, "reward_std": 0.09141466915607452, "rewards/accuracy_reward": 0.3259765625, "rewards/brier_reward": 0.7662179708480835, "rewards/confidence_one_or_zero": 0.0939453125, "rewards/format_reward": 0.99873046875, "rewards/mean_confidence_reward": 0.2009649395942688, "rewards/smce": -0.12679803520441055, "step": 50 }, { "epoch": 0.16, "eval_completions/clipped_ratio": 0.001953125, "eval_completions/max_length": 997.5, "eval_completions/max_terminated_length": 347.5, "eval_completions/mean_length": 127.43029403686523, "eval_completions/mean_terminated_length": 124.67033386230469, "eval_completions/min_length": 55.0, "eval_completions/min_terminated_length": 55.0, "eval_loss": 0.0, "eval_num_tokens": 172699192.0, "eval_reward": 1.0254148244857788, "eval_reward_std": 0.1284085400402546, "eval_rewards/accuracy_reward": 0.236328125, "eval_rewards/brier_reward": 0.8164560794830322, "eval_rewards/confidence_one_or_zero": 0.05859375, "eval_rewards/format_reward": 0.998046875, "eval_rewards/mean_confidence_reward": 0.22532617300748825, "eval_rewards/smce": -0.020053687505424023, "eval_runtime": 22.7777, "eval_samples_per_second": 21.951, "eval_steps_per_second": 0.088, "step": 50 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 899.4, "completions/max_terminated_length": 439.8, "completions/mean_length": 119.5984375, "completions/mean_terminated_length": 119.32182464599609, "completions/min_length": 56.4, "completions/min_terminated_length": 56.4, "epoch": 0.176, "grad_norm": 0.0008352047298103571, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 188940424.0, "reward": 1.0325161933898925, "reward_std": 0.08973144143819808, "rewards/accuracy_reward": 0.2689453125, "rewards/brier_reward": 0.7969693899154663, "rewards/confidence_one_or_zero": 0.035546875, "rewards/format_reward": 0.99912109375, "rewards/mean_confidence_reward": 0.24108357727527618, "rewards/smce": -0.039200416207313536, "step": 55 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1097.4, "completions/max_terminated_length": 780.8, "completions/mean_length": 121.203125, "completions/mean_terminated_length": 120.3740463256836, "completions/min_length": 56.4, "completions/min_terminated_length": 56.4, "epoch": 0.192, "grad_norm": 0.0008346716640517116, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 205123144.0, "reward": 1.0596784591674804, "reward_std": 0.09918512254953385, "rewards/accuracy_reward": 0.35859375, "rewards/brier_reward": 0.761845862865448, "rewards/confidence_one_or_zero": 0.0201171875, "rewards/format_reward": 0.99892578125, "rewards/mean_confidence_reward": 0.26954297423362733, "rewards/smce": -0.09118027314543724, "step": 60 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 628.4, "completions/max_terminated_length": 398.6, "completions/mean_length": 121.64189453125, "completions/mean_terminated_length": 121.50375213623047, "completions/min_length": 58.0, "completions/min_terminated_length": 58.0, "epoch": 0.208, "grad_norm": 0.000969071697909385, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 221270965.0, "reward": 1.069624376296997, "reward_std": 0.10735679119825363, "rewards/accuracy_reward": 0.37919921875, "rewards/brier_reward": 0.7605438590049743, "rewards/confidence_one_or_zero": 0.0107421875, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.3172558665275574, "rewards/smce": -0.06722123511135578, "step": 65 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 857.4, "completions/max_terminated_length": 385.4, "completions/mean_length": 124.721484375, "completions/mean_terminated_length": 124.44608459472656, "completions/min_length": 57.6, "completions/min_terminated_length": 57.6, "epoch": 0.224, "grad_norm": 0.0011017704382538795, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 237510193.0, "reward": 1.055229949951172, "reward_std": 0.11658172458410263, "rewards/accuracy_reward": 0.33193359375, "rewards/brier_reward": 0.779213547706604, "rewards/confidence_one_or_zero": 0.00517578125, "rewards/format_reward": 0.99931640625, "rewards/mean_confidence_reward": 0.356457781791687, "rewards/smce": -0.0435761496424675, "step": 70 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 366.0, "completions/max_terminated_length": 366.0, "completions/mean_length": 128.07626953125, "completions/mean_terminated_length": 128.07626953125, "completions/min_length": 58.8, "completions/min_terminated_length": 58.8, "epoch": 0.24, "grad_norm": 0.0010217519011348486, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 254027358.0, "reward": 1.0620179176330566, "reward_std": 0.11173272877931595, "rewards/accuracy_reward": 0.35087890625, "rewards/brier_reward": 0.7738432168960572, "rewards/confidence_one_or_zero": 0.0037109375, "rewards/format_reward": 0.99931640625, "rewards/mean_confidence_reward": 0.3753812491893768, "rewards/smce": -0.03457403443753719, "step": 75 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 956.8, "completions/max_terminated_length": 460.6, "completions/mean_length": 131.16923828125, "completions/mean_terminated_length": 130.89476623535157, "completions/min_length": 57.8, "completions/min_terminated_length": 57.8, "epoch": 0.256, "grad_norm": 0.0009919044096022844, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 270449027.0, "reward": 1.0897830486297608, "reward_std": 0.11419396549463272, "rewards/accuracy_reward": 0.41181640625, "rewards/brier_reward": 0.7680457592010498, "rewards/confidence_one_or_zero": 0.00185546875, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.38369869589805605, "rewards/smce": -0.03861977197229862, "step": 80 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 685.8, "completions/max_terminated_length": 455.6, "completions/mean_length": 131.84853515625, "completions/mean_terminated_length": 131.7114471435547, "completions/min_length": 56.0, "completions/min_terminated_length": 56.0, "epoch": 0.272, "grad_norm": 0.000989066786132753, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 286827284.0, "reward": 1.0802953958511352, "reward_std": 0.1197711020708084, "rewards/accuracy_reward": 0.394140625, "rewards/brier_reward": 0.7666484951972962, "rewards/confidence_one_or_zero": 0.00126953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.3936794936656952, "rewards/smce": -0.03813821934163571, "step": 85 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 392.6, "completions/max_terminated_length": 392.6, "completions/mean_length": 133.7263671875, "completions/mean_terminated_length": 133.7263671875, "completions/min_length": 59.8, "completions/min_terminated_length": 59.8, "epoch": 0.288, "grad_norm": 0.0009975044522434473, "learning_rate": 1e-06, "loss": -0.0, "num_tokens": 303172418.0, "reward": 1.0799544334411622, "reward_std": 0.11399292945861816, "rewards/accuracy_reward": 0.38916015625, "rewards/brier_reward": 0.77084881067276, "rewards/confidence_one_or_zero": 0.0013671875, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.39172295331954954, "rewards/smce": -0.0327066283673048, "step": 90 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 597.6, "completions/max_terminated_length": 364.4, "completions/mean_length": 139.4765625, "completions/mean_terminated_length": 139.34026489257812, "completions/min_length": 64.4, "completions/min_terminated_length": 64.4, "epoch": 0.304, "grad_norm": 0.000925487547647208, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 319776530.0, "reward": 1.0868202686309814, "reward_std": 0.11602218449115753, "rewards/accuracy_reward": 0.41484375, "rewards/brier_reward": 0.759091567993164, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.39790862798690796, "rewards/smce": -0.026037893444299697, "step": 95 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1067.8, "completions/max_terminated_length": 479.2, "completions/mean_length": 141.0818359375, "completions/mean_terminated_length": 140.6728057861328, "completions/min_length": 62.6, "completions/min_terminated_length": 62.6, "epoch": 0.32, "grad_norm": 0.0008635123958811164, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 336236632.0, "reward": 1.0830219507217407, "reward_std": 0.11141782402992248, "rewards/accuracy_reward": 0.3953125, "rewards/brier_reward": 0.7712236523628235, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 0.99951171875, "rewards/mean_confidence_reward": 0.4011562764644623, "rewards/smce": -0.047462426498532294, "step": 100 }, { "epoch": 0.32, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 387.5, "eval_completions/max_terminated_length": 387.5, "eval_completions/mean_length": 140.8268051147461, "eval_completions/mean_terminated_length": 140.8268051147461, "eval_completions/min_length": 72.5, "eval_completions/min_terminated_length": 72.5, "eval_loss": 0.0, "eval_num_tokens": 336236632.0, "eval_reward": 1.0537142753601074, "eval_reward_std": 0.20659654587507248, "eval_rewards/accuracy_reward": 0.32421875, "eval_rewards/brier_reward": 0.7832174003124237, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4059121161699295, "eval_rewards/smce": -0.08268068730831146, "eval_runtime": 12.3699, "eval_samples_per_second": 40.421, "eval_steps_per_second": 0.162, "step": 100 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1099.6, "completions/max_terminated_length": 424.2, "completions/mean_length": 141.2322265625, "completions/mean_terminated_length": 140.82349853515626, "completions/min_length": 64.4, "completions/min_terminated_length": 64.4, "epoch": 0.336, "grad_norm": 0.0008358008344657719, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 352714338.0, "reward": 1.0877865314483643, "reward_std": 0.10996164083480835, "rewards/accuracy_reward": 0.41025390625, "rewards/brier_reward": 0.7656147480010986, "rewards/confidence_one_or_zero": 0.00087890625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.40201926827430723, "rewards/smce": -0.03372764438390732, "step": 105 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 597.6, "completions/max_terminated_length": 354.6, "completions/mean_length": 139.06103515625, "completions/mean_terminated_length": 138.92465209960938, "completions/min_length": 64.2, "completions/min_terminated_length": 64.2, "epoch": 0.352, "grad_norm": 0.0008557201945222914, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 369174259.0, "reward": 1.0942952394485475, "reward_std": 0.11433759331703186, "rewards/accuracy_reward": 0.41708984375, "rewards/brier_reward": 0.7716007232666016, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.39787405729293823, "rewards/smce": -0.032202685810625555, "step": 110 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 326.8, "completions/max_terminated_length": 326.8, "completions/mean_length": 138.03349609375, "completions/mean_terminated_length": 138.03349609375, "completions/min_length": 65.0, "completions/min_terminated_length": 65.0, "epoch": 0.368, "grad_norm": 0.0008347496623173356, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 385609770.0, "reward": 1.1024059534072876, "reward_std": 0.10553737431764602, "rewards/accuracy_reward": 0.42998046875, "rewards/brier_reward": 0.7749350905418396, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.40569075345993044, "rewards/smce": -0.06787904426455497, "step": 115 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 373.2, "completions/max_terminated_length": 373.2, "completions/mean_length": 142.7939453125, "completions/mean_terminated_length": 142.7939453125, "completions/min_length": 65.4, "completions/min_terminated_length": 65.4, "epoch": 0.384, "grad_norm": 0.0008544361335225403, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 402015820.0, "reward": 1.083213472366333, "reward_std": 0.10497440248727799, "rewards/accuracy_reward": 0.38974609375, "rewards/brier_reward": 0.7766830205917359, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4167145550251007, "rewards/smce": -0.030062619410455228, "step": 120 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 588.6, "completions/max_terminated_length": 357.6, "completions/mean_length": 144.31806640625, "completions/mean_terminated_length": 144.18206787109375, "completions/min_length": 65.4, "completions/min_terminated_length": 65.4, "epoch": 0.4, "grad_norm": 0.001108017168007791, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 418605509.0, "reward": 1.0979632616043091, "reward_std": 0.10889875888824463, "rewards/accuracy_reward": 0.42548828125, "rewards/brier_reward": 0.7705417275428772, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4027201533317566, "rewards/smce": -0.06595741771161556, "step": 125 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 836.2, "completions/max_terminated_length": 363.0, "completions/mean_length": 147.25498046875, "completions/mean_terminated_length": 146.84811401367188, "completions/min_length": 57.4, "completions/min_terminated_length": 57.4, "epoch": 0.416, "grad_norm": 0.0009119771420955658, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 435183704.0, "reward": 1.1149553060531616, "reward_std": 0.1131348431110382, "rewards/accuracy_reward": 0.4630859375, "rewards/brier_reward": 0.767220401763916, "rewards/confidence_one_or_zero": 0.00068359375, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4106293976306915, "rewards/smce": -0.05815782323479653, "step": 130 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 396.6, "completions/max_terminated_length": 396.6, "completions/mean_length": 148.89404296875, "completions/mean_terminated_length": 148.89404296875, "completions/min_length": 71.6, "completions/min_terminated_length": 71.6, "epoch": 0.432, "grad_norm": 0.0009073733817785978, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 451666715.0, "reward": 1.109913206100464, "reward_std": 0.1156215637922287, "rewards/accuracy_reward": 0.45029296875, "rewards/brier_reward": 0.7695358991622925, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.44832324981689453, "rewards/smce": -0.03289064317941666, "step": 135 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 589.2, "completions/max_terminated_length": 375.2, "completions/mean_length": 151.27353515625, "completions/mean_terminated_length": 151.00367126464843, "completions/min_length": 71.8, "completions/min_terminated_length": 71.8, "epoch": 0.448, "grad_norm": 0.0008107796311378479, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 468305292.0, "reward": 1.0973914384841919, "reward_std": 0.11089750230312348, "rewards/accuracy_reward": 0.42548828125, "rewards/brier_reward": 0.7694926500320435, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.45723145008087157, "rewards/smce": -0.03543643653392792, "step": 140 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 786.6, "completions/max_terminated_length": 562.6, "completions/mean_length": 154.874609375, "completions/mean_terminated_length": 154.73968811035155, "completions/min_length": 71.8, "completions/min_terminated_length": 71.8, "epoch": 0.464, "grad_norm": 0.0008580603753216565, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 485007272.0, "reward": 1.1149845123291016, "reward_std": 0.1036541149020195, "rewards/accuracy_reward": 0.44921875, "rewards/brier_reward": 0.7808510184288024, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.45266503691673277, "rewards/smce": -0.03915540650486946, "step": 145 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 589.0, "completions/max_terminated_length": 355.6, "completions/mean_length": 154.5427734375, "completions/mean_terminated_length": 154.27273559570312, "completions/min_length": 75.6, "completions/min_terminated_length": 75.6, "epoch": 0.48, "grad_norm": 0.0008421405800618231, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 501708158.0, "reward": 1.1118209838867188, "reward_std": 0.11529819369316101, "rewards/accuracy_reward": 0.44970703125, "rewards/brier_reward": 0.7741313695907592, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.44507910013198854, "rewards/smce": -0.019151285663247107, "step": 150 }, { "epoch": 0.48, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 330.0, "eval_completions/max_terminated_length": 330.0, "eval_completions/mean_length": 156.71817016601562, "eval_completions/mean_terminated_length": 156.71817016601562, "eval_completions/min_length": 87.5, "eval_completions/min_terminated_length": 87.5, "eval_loss": 0.0, "eval_num_tokens": 501708158.0, "eval_reward": 1.0748705863952637, "eval_reward_std": 0.22966566681861877, "eval_rewards/accuracy_reward": 0.36328125, "eval_rewards/brier_reward": 0.7864670157432556, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4421289265155792, "eval_rewards/smce": -0.07997188344597816, "eval_runtime": 11.4907, "eval_samples_per_second": 43.513, "eval_steps_per_second": 0.174, "step": 150 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 873.4, "completions/max_terminated_length": 420.2, "completions/mean_length": 159.5671875, "completions/mean_terminated_length": 159.2984588623047, "completions/min_length": 73.0, "completions/min_terminated_length": 73.0, "epoch": 0.496, "grad_norm": 0.000909183407202363, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 518364942.0, "reward": 1.084216022491455, "reward_std": 0.10733449161052704, "rewards/accuracy_reward": 0.3853515625, "rewards/brier_reward": 0.7832805037498474, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.43932910561561583, "rewards/smce": -0.05489234179258347, "step": 155 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 430.6, "completions/mean_length": 159.97548828125, "completions/mean_terminated_length": 159.43768005371095, "completions/min_length": 71.6, "completions/min_terminated_length": 71.6, "epoch": 0.512, "grad_norm": 0.0007752343663014472, "learning_rate": 1e-06, "loss": 0.0013, "num_tokens": 534732563.0, "reward": 1.107197880744934, "reward_std": 0.1062451258301735, "rewards/accuracy_reward": 0.43896484375, "rewards/brier_reward": 0.7758240461349487, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.999609375, "rewards/mean_confidence_reward": 0.4436230480670929, "rewards/smce": -0.03356703966856003, "step": 160 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 649.6, "completions/max_terminated_length": 410.4, "completions/mean_length": 161.52412109375, "completions/mean_terminated_length": 161.38986206054688, "completions/min_length": 79.6, "completions/min_terminated_length": 79.6, "epoch": 0.528, "grad_norm": 0.000873538083396852, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 551442378.0, "reward": 1.1122208118438721, "reward_std": 0.10901667475700379, "rewards/accuracy_reward": 0.45302734375, "rewards/brier_reward": 0.771516752243042, "rewards/confidence_one_or_zero": 0.0005859375, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4236679673194885, "rewards/smce": -0.055811258032917976, "step": 165 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 663.8, "completions/max_terminated_length": 451.6, "completions/mean_length": 162.53212890625, "completions/mean_terminated_length": 162.3982421875, "completions/min_length": 79.8, "completions/min_terminated_length": 79.8, "epoch": 0.544, "grad_norm": 0.0008957363897934556, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 568087443.0, "reward": 1.1219069719314576, "reward_std": 0.10789882093667984, "rewards/accuracy_reward": 0.46884765625, "rewards/brier_reward": 0.7751662611961365, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4372187554836273, "rewards/smce": -0.05467262491583824, "step": 170 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 606.2, "completions/max_terminated_length": 382.6, "completions/mean_length": 159.515625, "completions/mean_terminated_length": 159.3811279296875, "completions/min_length": 77.6, "completions/min_terminated_length": 77.6, "epoch": 0.56, "grad_norm": 0.000858804676681757, "learning_rate": 1e-06, "loss": 0.0011, "num_tokens": 584687539.0, "reward": 1.1200573682785033, "reward_std": 0.10227487087249756, "rewards/accuracy_reward": 0.45849609375, "rewards/brier_reward": 0.7817179918289184, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4568593859672546, "rewards/smce": -0.026216423511505126, "step": 175 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 446.8, "completions/max_terminated_length": 446.8, "completions/mean_length": 158.66416015625, "completions/mean_terminated_length": 158.66416015625, "completions/min_length": 75.0, "completions/min_terminated_length": 75.0, "epoch": 0.576, "grad_norm": 0.0010204884456470609, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 601433508.0, "reward": 1.1217976331710815, "reward_std": 0.1035456970334053, "rewards/accuracy_reward": 0.44609375, "rewards/brier_reward": 0.7975054860115052, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.43276728987693786, "rewards/smce": -0.0471338476985693, "step": 180 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 431.2, "completions/max_terminated_length": 431.2, "completions/mean_length": 159.45615234375, "completions/mean_terminated_length": 159.45615234375, "completions/min_length": 81.8, "completions/min_terminated_length": 81.8, "epoch": 0.592, "grad_norm": 0.0008188558858819306, "learning_rate": 1e-06, "loss": 0.0002, "num_tokens": 617936419.0, "reward": 1.1226669073104858, "reward_std": 0.10481721758842469, "rewards/accuracy_reward": 0.45576171875, "rewards/brier_reward": 0.7895737171173096, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.44277440905570986, "rewards/smce": -0.024372975155711175, "step": 185 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 863.6, "completions/max_terminated_length": 428.8, "completions/mean_length": 164.56435546875, "completions/mean_terminated_length": 164.2962646484375, "completions/min_length": 80.0, "completions/min_terminated_length": 80.0, "epoch": 0.608, "grad_norm": 0.0008857807843014598, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 634850998.0, "reward": 1.1132190942764282, "reward_std": 0.11314407587051392, "rewards/accuracy_reward": 0.437109375, "rewards/brier_reward": 0.7895278453826904, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4769873142242432, "rewards/smce": -0.04614109750837088, "step": 190 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 655.4, "completions/max_terminated_length": 440.0, "completions/mean_length": 161.4076171875, "completions/mean_terminated_length": 161.27352600097657, "completions/min_length": 82.8, "completions/min_terminated_length": 82.8, "epoch": 0.624, "grad_norm": 0.0009903701720759273, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 651472612.0, "reward": 1.1419665813446045, "reward_std": 0.10531192421913146, "rewards/accuracy_reward": 0.49169921875, "rewards/brier_reward": 0.7923331260681152, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.48211621642112734, "rewards/smce": -0.024501551128923892, "step": 195 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 675.2, "completions/max_terminated_length": 435.4, "completions/mean_length": 158.34560546875, "completions/mean_terminated_length": 158.0752716064453, "completions/min_length": 80.6, "completions/min_terminated_length": 80.6, "epoch": 0.64, "grad_norm": 0.0009547388763166964, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 668172215.0, "reward": 1.1136085271835328, "reward_std": 0.10880862027406693, "rewards/accuracy_reward": 0.43544921875, "rewards/brier_reward": 0.7919665455818177, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4641181766986847, "rewards/smce": -0.042938829958438875, "step": 200 }, { "epoch": 0.64, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 319.0, "eval_completions/max_terminated_length": 319.0, "eval_completions/mean_length": 158.68778228759766, "eval_completions/mean_terminated_length": 158.68778228759766, "eval_completions/min_length": 87.5, "eval_completions/min_terminated_length": 87.5, "eval_loss": 0.0, "eval_num_tokens": 668172215.0, "eval_reward": 1.0859458446502686, "eval_reward_std": 0.25144463777542114, "eval_rewards/accuracy_reward": 0.384765625, "eval_rewards/brier_reward": 0.7871337831020355, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4686523377895355, "eval_rewards/smce": -0.08460449427366257, "eval_runtime": 11.2462, "eval_samples_per_second": 44.46, "eval_steps_per_second": 0.178, "step": 200 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 380.4, "completions/max_terminated_length": 380.4, "completions/mean_length": 158.65654296875, "completions/mean_terminated_length": 158.65654296875, "completions/min_length": 79.8, "completions/min_terminated_length": 79.8, "epoch": 0.656, "grad_norm": 0.0008885656134225428, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 684988570.0, "reward": 1.124780035018921, "reward_std": 0.11241506040096283, "rewards/accuracy_reward": 0.46171875, "rewards/brier_reward": 0.787843644618988, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4735654413700104, "rewards/smce": -0.032851331681013105, "step": 205 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 679.4, "completions/max_terminated_length": 461.4, "completions/mean_length": 160.13818359375, "completions/mean_terminated_length": 160.00401000976564, "completions/min_length": 81.2, "completions/min_terminated_length": 81.2, "epoch": 0.672, "grad_norm": 0.0009205000242218375, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 701576577.0, "reward": 1.1205364227294923, "reward_std": 0.10743532329797745, "rewards/accuracy_reward": 0.4419921875, "rewards/brier_reward": 0.7992789030075074, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.47032617330551146, "rewards/smce": -0.03808985371142626, "step": 210 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 880.6, "completions/max_terminated_length": 419.4, "completions/mean_length": 159.65302734375, "completions/mean_terminated_length": 159.384033203125, "completions/min_length": 84.2, "completions/min_terminated_length": 84.2, "epoch": 0.688, "grad_norm": 0.0011366148246452212, "learning_rate": 1e-06, "loss": 0.0006, "num_tokens": 718239936.0, "reward": 1.1395163536071777, "reward_std": 0.10088872164487839, "rewards/accuracy_reward": 0.48984375, "rewards/brier_reward": 0.7893894791603089, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.43761621713638305, "rewards/smce": -0.06050069257616997, "step": 215 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 639.4, "completions/max_terminated_length": 403.6, "completions/mean_length": 163.80947265625, "completions/mean_terminated_length": 163.67565002441407, "completions/min_length": 82.0, "completions/min_terminated_length": 82.0, "epoch": 0.704, "grad_norm": 0.0008274089777842164, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 735125121.0, "reward": 1.1324799537658692, "reward_std": 0.10014449208974838, "rewards/accuracy_reward": 0.48134765625, "rewards/brier_reward": 0.7837143540382385, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.42956447005271914, "rewards/smce": -0.052855022624135016, "step": 220 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 614.2, "completions/max_terminated_length": 386.2, "completions/mean_length": 165.41416015625, "completions/mean_terminated_length": 165.2800323486328, "completions/min_length": 86.6, "completions/min_terminated_length": 86.6, "epoch": 0.72, "grad_norm": 0.001209041103720665, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 751689842.0, "reward": 1.1413109302520752, "reward_std": 0.10785426497459412, "rewards/accuracy_reward": 0.48271484375, "rewards/brier_reward": 0.800009298324585, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4845683515071869, "rewards/smce": -0.05454383902251721, "step": 225 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 887.4, "completions/max_terminated_length": 443.6, "completions/mean_length": 167.75576171875, "completions/mean_terminated_length": 167.4881164550781, "completions/min_length": 84.8, "completions/min_terminated_length": 84.8, "epoch": 0.736, "grad_norm": 0.0009000345598906279, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 768583053.0, "reward": 1.1350015878677369, "reward_std": 0.11322512924671173, "rewards/accuracy_reward": 0.47939453125, "rewards/brier_reward": 0.7908076047897339, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5125478506088257, "rewards/smce": -0.046359357237815854, "step": 230 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 706.8, "completions/max_terminated_length": 509.0, "completions/mean_length": 171.1728515625, "completions/mean_terminated_length": 171.04000854492188, "completions/min_length": 92.4, "completions/min_terminated_length": 92.4, "epoch": 0.752, "grad_norm": 0.000924249179661274, "learning_rate": 1e-06, "loss": 0.0005, "num_tokens": 785295447.0, "reward": 1.1308918714523315, "reward_std": 0.10929270833730698, "rewards/accuracy_reward": 0.47255859375, "rewards/brier_reward": 0.7893296003341674, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4867382884025574, "rewards/smce": -0.07814770340919494, "step": 235 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 629.2, "completions/max_terminated_length": 629.2, "completions/mean_length": 174.9046875, "completions/mean_terminated_length": 174.9046875, "completions/min_length": 91.6, "completions/min_terminated_length": 91.6, "epoch": 0.768, "grad_norm": 0.0008467729785479605, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 802093895.0, "reward": 1.1407481908798218, "reward_std": 0.09571124464273453, "rewards/accuracy_reward": 0.473828125, "rewards/brier_reward": 0.8076701879501342, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.4546728551387787, "rewards/smce": -0.028032467886805534, "step": 240 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 702.6, "completions/max_terminated_length": 482.4, "completions/mean_length": 173.15537109375, "completions/mean_terminated_length": 173.02240295410155, "completions/min_length": 89.6, "completions/min_terminated_length": 89.6, "epoch": 0.784, "grad_norm": 0.00092527293600142, "learning_rate": 1e-06, "loss": 0.0008, "num_tokens": 818869630.0, "reward": 1.1412487745285034, "reward_std": 0.10290929973125458, "rewards/accuracy_reward": 0.47958984375, "rewards/brier_reward": 0.8030083179473877, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4737207114696503, "rewards/smce": -0.03872436471283436, "step": 245 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 884.0, "completions/max_terminated_length": 459.4, "completions/mean_length": 177.40078125, "completions/mean_terminated_length": 177.0030487060547, "completions/min_length": 94.8, "completions/min_terminated_length": 94.8, "epoch": 0.8, "grad_norm": 0.001135337632149458, "learning_rate": 1e-06, "loss": 0.001, "num_tokens": 835906438.0, "reward": 1.142169713973999, "reward_std": 0.11441078335046768, "rewards/accuracy_reward": 0.48525390625, "rewards/brier_reward": 0.7993808865547181, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4921103477478027, "rewards/smce": -0.03339071050286293, "step": 250 }, { "epoch": 0.8, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 346.0, "eval_completions/max_terminated_length": 346.0, "eval_completions/mean_length": 185.98580932617188, "eval_completions/mean_terminated_length": 185.98580932617188, "eval_completions/min_length": 103.5, "eval_completions/min_terminated_length": 103.5, "eval_loss": 0.0, "eval_num_tokens": 835906438.0, "eval_reward": 1.0858368873596191, "eval_reward_std": 0.2645293176174164, "eval_rewards/accuracy_reward": 0.375, "eval_rewards/brier_reward": 0.7966845631599426, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.4926757961511612, "eval_rewards/smce": -0.1178906224668026, "eval_runtime": 11.9011, "eval_samples_per_second": 42.013, "eval_steps_per_second": 0.168, "step": 250 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 698.0, "completions/max_terminated_length": 469.0, "completions/mean_length": 186.46650390625, "completions/mean_terminated_length": 186.33453979492188, "completions/min_length": 99.8, "completions/min_terminated_length": 99.8, "epoch": 0.816, "grad_norm": 0.0008341626380570233, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 852905775.0, "reward": 1.1321615695953369, "reward_std": 0.09045738875865936, "rewards/accuracy_reward": 0.46025390625, "rewards/brier_reward": 0.804168701171875, "rewards/confidence_one_or_zero": 0.00029296875, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47615723609924315, "rewards/smce": -0.026564698852598666, "step": 255 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 665.6, "completions/max_terminated_length": 482.6, "completions/mean_length": 186.53017578125, "completions/mean_terminated_length": 186.39850463867188, "completions/min_length": 97.2, "completions/min_terminated_length": 97.2, "epoch": 0.832, "grad_norm": 0.0010087019763886929, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 869908228.0, "reward": 1.1526384592056274, "reward_std": 0.10760405957698822, "rewards/accuracy_reward": 0.50556640625, "rewards/brier_reward": 0.7998112916946412, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.4938330054283142, "rewards/smce": -0.040554934181272984, "step": 260 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 671.8, "completions/max_terminated_length": 461.6, "completions/mean_length": 186.29765625, "completions/mean_terminated_length": 186.16553649902343, "completions/min_length": 96.4, "completions/min_terminated_length": 96.4, "epoch": 0.848, "grad_norm": 0.0010384187335148454, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 886850124.0, "reward": 1.1427557468414307, "reward_std": 0.10271692872047425, "rewards/accuracy_reward": 0.4865234375, "rewards/brier_reward": 0.7990883827209473, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.47125096917152404, "rewards/smce": -0.03518822770565748, "step": 265 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 659.0, "completions/max_terminated_length": 433.6, "completions/mean_length": 180.399609375, "completions/mean_terminated_length": 180.26744384765624, "completions/min_length": 87.6, "completions/min_terminated_length": 87.6, "epoch": 0.864, "grad_norm": 0.0007400948088616133, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 903635656.0, "reward": 1.136153483390808, "reward_std": 0.08060802072286606, "rewards/accuracy_reward": 0.458203125, "rewards/brier_reward": 0.8143020153045655, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.4435312509536743, "rewards/smce": -0.03686932548880577, "step": 270 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 652.2, "completions/max_terminated_length": 440.2, "completions/mean_length": 182.7515625, "completions/mean_terminated_length": 182.48705139160157, "completions/min_length": 97.6, "completions/min_terminated_length": 97.6, "epoch": 0.88, "grad_norm": 0.0008531102794222534, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 920581784.0, "reward": 1.1429947853088378, "reward_std": 0.09263245761394501, "rewards/accuracy_reward": 0.4791015625, "rewards/brier_reward": 0.8071859121322632, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.44282519817352295, "rewards/smce": -0.0567308247089386, "step": 275 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 634.0, "completions/max_terminated_length": 419.2, "completions/mean_length": 182.8693359375, "completions/mean_terminated_length": 182.73717346191407, "completions/min_length": 94.2, "completions/min_terminated_length": 94.2, "epoch": 0.896, "grad_norm": 0.0008840742520987988, "learning_rate": 1e-06, "loss": 0.0004, "num_tokens": 937587198.0, "reward": 1.1388145685195923, "reward_std": 0.09899315983057022, "rewards/accuracy_reward": 0.48037109375, "rewards/brier_reward": 0.7974554300308228, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.49169726967811583, "rewards/smce": -0.02959427684545517, "step": 280 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_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": 690.8, "completions/max_terminated_length": 477.0, "completions/mean_length": 182.4890625, "completions/mean_terminated_length": 182.35704650878907, "completions/min_length": 92.6, "completions/min_terminated_length": 92.6, "epoch": 0.912, "grad_norm": 0.0008525560842826962, "learning_rate": 1e-06, "loss": 0.0003, "num_tokens": 954336110.0, "reward": 1.138223648071289, "reward_std": 0.10281311720609665, "rewards/accuracy_reward": 0.4755859375, "rewards/brier_reward": 0.8010599613189697, "rewards/confidence_one_or_zero": 0.0001953125, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.49716113805770873, "rewards/smce": -0.04195708781480789, "step": 285 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 869.8, "completions/max_terminated_length": 438.0, "completions/mean_length": 181.30810546875, "completions/mean_terminated_length": 181.0434326171875, "completions/min_length": 97.0, "completions/min_terminated_length": 97.0, "epoch": 0.928, "grad_norm": 0.0010722881415858865, "learning_rate": 1e-06, "loss": 0.0007, "num_tokens": 971141569.0, "reward": 1.1697998523712159, "reward_std": 0.09429487735033035, "rewards/accuracy_reward": 0.5265625, "rewards/brier_reward": 0.8132358431816101, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.9998046875, "rewards/mean_confidence_reward": 0.5027734339237213, "rewards/smce": -0.04298241809010506, "step": 290 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 1309.6, "completions/max_terminated_length": 562.4, "completions/mean_length": 187.14462890625, "completions/mean_terminated_length": 186.48440856933593, "completions/min_length": 99.2, "completions/min_terminated_length": 99.2, "epoch": 0.944, "grad_norm": 0.000979841803200543, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 988275658.0, "reward": 1.1545918464660645, "reward_std": 0.10092489123344421, "rewards/accuracy_reward": 0.49013671875, "rewards/brier_reward": 0.8196364879608155, "rewards/confidence_one_or_zero": 9.765625e-05, "rewards/format_reward": 0.9994140625, "rewards/mean_confidence_reward": 0.49853025674819945, "rewards/smce": -0.04494751356542111, "step": 295 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 599.4, "completions/max_terminated_length": 599.4, "completions/mean_length": 186.77529296875, "completions/mean_terminated_length": 186.77529296875, "completions/min_length": 98.2, "completions/min_terminated_length": 98.2, "epoch": 0.96, "grad_norm": 0.001181174535304308, "learning_rate": 1e-06, "loss": 0.0001, "num_tokens": 1005166701.0, "reward": 1.1341146945953369, "reward_std": 0.11333226412534714, "rewards/accuracy_reward": 0.46708984375, "rewards/brier_reward": 0.8011467099189759, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 1.0, "rewards/mean_confidence_reward": 0.5431005716323852, "rewards/smce": -0.08158175013959408, "step": 300 }, { "epoch": 0.96, "eval_completions/clipped_ratio": 0.0, "eval_completions/max_length": 577.5, "eval_completions/max_terminated_length": 577.5, "eval_completions/mean_length": 191.35697174072266, "eval_completions/mean_terminated_length": 191.35697174072266, "eval_completions/min_length": 109.5, "eval_completions/min_terminated_length": 109.5, "eval_loss": 0.0, "eval_num_tokens": 1005166701.0, "eval_reward": 1.0900059342384338, "eval_reward_std": 0.2852499932050705, "eval_rewards/accuracy_reward": 0.380859375, "eval_rewards/brier_reward": 0.7991650402545929, "eval_rewards/confidence_one_or_zero": 0.0, "eval_rewards/format_reward": 1.0, "eval_rewards/mean_confidence_reward": 0.5155273675918579, "eval_rewards/smce": -0.13530762493610382, "eval_runtime": 15.8845, "eval_samples_per_second": 31.477, "eval_steps_per_second": 0.126, "step": 300 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 586.8, "completions/max_terminated_length": 586.8, "completions/mean_length": 189.49833984375, "completions/mean_terminated_length": 189.49833984375, "completions/min_length": 99.0, "completions/min_terminated_length": 99.0, "epoch": 0.976, "grad_norm": 0.0009122914634644985, "learning_rate": 1e-06, "loss": 0.0, "num_tokens": 1022172476.0, "reward": 1.151877188682556, "reward_std": 0.09684720933437348, "rewards/accuracy_reward": 0.49482421875, "rewards/brier_reward": 0.8090285301208496, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.99990234375, "rewards/mean_confidence_reward": 0.49103320837020875, "rewards/smce": -0.016386817023158074, "step": 305 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 898.2, "completions/max_terminated_length": 472.8, "completions/mean_length": 196.90341796875, "completions/mean_terminated_length": 196.51105041503905, "completions/min_length": 99.0, "completions/min_terminated_length": 99.0, "epoch": 0.992, "grad_norm": 0.0009238107013516128, "learning_rate": 1e-06, "loss": 0.0012, "num_tokens": 1039372479.0, "reward": 1.168659234046936, "reward_std": 0.08935579359531402, "rewards/accuracy_reward": 0.534765625, "rewards/brier_reward": 0.8028544902801513, "rewards/confidence_one_or_zero": 0.000390625, "rewards/format_reward": 0.99970703125, "rewards/mean_confidence_reward": 0.4400693416595459, "rewards/smce": -0.09543538391590119, "step": 310 }, { "clip_ratio/high_max": 0.0, "clip_ratio/high_mean": 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": 628.0, "completions/max_terminated_length": 628.0, "completions/mean_length": 202.2728042602539, "completions/mean_terminated_length": 202.2728042602539, "completions/min_length": 105.5, "completions/min_terminated_length": 105.5, "epoch": 0.9984, "num_tokens": 1046240895.0, "reward": 1.1533426642417908, "reward_std": 0.09558231383562088, "rewards/accuracy_reward": 0.49169921875, "rewards/brier_reward": 0.8152325451374054, "rewards/confidence_one_or_zero": 0.0, "rewards/format_reward": 0.999755859375, "rewards/mean_confidence_reward": 0.46422119438648224, "rewards/smce": -0.03218200244009495, "step": 312, "total_flos": 0.0, "train_loss": 0.004010988894338087, "train_runtime": 42587.7875, "train_samples_per_second": 0.47, "train_steps_per_second": 0.007 } ], "logging_steps": 5, "max_steps": 312, "num_input_tokens_seen": 1046240895, "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": 4, "trial_name": null, "trial_params": null }